Big Data Segmentation and Slacktivism

Create a one page review for each article. Each review should be one page, double spaced. The review should be a ½ page summary of the article, and a ½ page application of the material to the real world.

Consumer Behaviorconsumer BEHAVIOR

Journal of Intelligent & Fuzzy Systems 38 (2020) 6159–6173 6159
DOI:10.3233/JIFS-179698
IOS Press

The impact of big data market segmentation
using data mining and clustering techniques

Fahed Yosepha,b,∗ , Nurul Hashimah Ahamed Hassain Malimb, Markku Heikkiläc,
Adrian Brezulianud, Oana Gemane and Nur Aqilah Paskhal Rostamb

aFaculty of Social Sciences, Business and Economics, Åbo Akademi University, Turku, Finland
bDepartment of School of Computer Sciences, Universiti Sains Malaysia, Penang, Malaysia
cFaculty of Social Sciences, Business and Economics, Åbo Akademi University, Turku, Finland
dFaculty of Electronics, Telecommunications and Information Technology,
Gheorghe Asachi Technical University, Iaşi, Romania
eDepartment of Health and Human Development, Stefan cel Mare University, Suceava, Romania

Abstract. Targeted marketing strategy is a prominent topic that has received substantial attention from both industries and
academia. Market segmentation is a widely used approach in investigating the heterogeneity of customer buying behavior
and profitability. It is important to note that conventional market segmentation models in the retail industry are predominantly
descriptive methods, lack sufficient market insights, and often fail to identify sufficiently small segments. This study also
takes advantage of the dynamics involved in the Hadoop distributed file system for its ability to process vast dataset. Three
different market segmentation experiments using modified best fit regression, i.e., Expectation-Maximization (EM) and K-
Means++ clustering algorithms were conducted and subsequently assessed using cluster quality assessment. The results of
this research are twofold: i) The insight on customer purchase behavior revealed for each Customer Lifetime Value (CLTV)
segment; ii) performance of the clustering algorithm for producing accurate market segments. The analysis indicated that the
average lifetime of the customer was only two years, and the churn rate was 52%. Consequently, a marketing strategy was
devised based on these results and implemented on the departmental store sales. It was revealed in the marketing record that
the sales growth rate up increased from 5% to 9%.

Keywords: Market segmentation, data mining, customer lifetime value (CLTV), RFM model (recency frequency monetary)

1. Introduction is the key success to brand loyalty, repeat store visits,
and ultimately, sales conversions. This relationship

The retail industry collects enormous volumes of has been affected by recent economic and social. The
POS data. However, this RAW POS data has min- retail industry is prompted to be more strategic in
imal use if it’s not properly processed to generate their planning and to develop a deep understanding
retail insights, optimize marketing efforts and drive of its consumers as well as their competitors. Under-
decisions. The retailer’s relationship with customers standing customers’ behavior as well as establishing a

loyal relationship with customers has become the cen-
∗Corresponding author. Fahed Yoseph, Faculty of Social Sci- tral concern and strategic goal for most retailers [1]

°ences, Business and Economics, Abo Akademi University, Turku, interested in tracking and managing their customerFinland, and Deparment f School of Computer Sciences, Uni-
lifetime value on a systematic basis [44]. Market seg-versiti Sains Malaysia, 11800, Penang, Malaysia. E-mail:

fyoseph@abo.fi. mentation is the process to divide the market base

mailto:fyoseph@abo.fi

https://1064-1246/20/$35.00

6160 F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques

of potential customers into similar or homogeneous
groups or segments that possess mutual characteris-
tics helps marketers to gather individuals with similar
choices and interests [2]. This enables retailers to
avoid selling unprofitable and irrelevant products
with regards to their marketing purpose, which will
result in better management of the available resources
through the selection of suitable market segment and
the primary focus of specific promising segments
[3–15].

Furthermore, as far as the research scope is
concerned, there has been number of studies that
examine customer purchase behavior and lifetime
value among different products based on a variety
of market segmentation with demographic variables
and characteristics. Instead of addressing individual
consumers based on their purchasing behavior, most
market segmentation studies merely considered the
overview of consumers’ historical data to produce
assumptions of what makes consumers similar to one
another. It is significant to highlight that this method
hides critical facts about individual consumers.

Among those customer lifetime value models, a
highly regarded model cited by many experts is the
Pareto/NBD Counting Your Customers proposed by
Schmittlein, Morrison, and Colombo (1987). The
model investigates customer purchase behavior in
settings where customer purchase dropout is unob-
served. However, the model is powerful for analyzing
customer purchase behavior, but it has been proven
to be empirically complex to implement due to the
computational challenges, and only a handful of
researches claim to have implemented it [44].

Based on previous studies of market segmenta-
tion on the retail domain, Recency, Frequency, and
Monetary (RFM) has been extensively employed as
this model can divide customers into groups which,
therefore, enables retailers to decide on ways to fully
utilize their limited resources in providing effective
customer service through the categorization of cus-
tomers. Nonetheless, RFM also has its own limitation
[4] where it only focuses on customers’ best scores
in addition to providing less meaningful scoring on
recency, frequency and monetary for most consumers
(Wei, Lin, and Wu, 2010). Moreover, RFM analy-
sis is not able to prospect for new customers, as it
mainly concerns the organization’s current customers
[6] and that it is not considered as a precise quan-
titative analysis model as the importance of each
RFM measure is different among other industries
[16–20]. The current research foresees an enhanced
user-friendly market segmentation modeling method,

which is more advanced and effective than conven-
tional RFM method. The integration of Customer
Lifetime Value and newly proposed RFM variants
(PQ) (T) into a closed-loop model represents dif-
ferent variation in customer purchase behavior. The
enhanced model has the capability to simultane-
ously analyze millions of raw POS data, identify
groups of customers by criteria the retailer may
never have considered. This goldmine knowledge is
expected to help marketers avoid the assumptions
when doing customer deep-dive and trend analysis,
which subsequently tapped marketers to device tar-
geted marketing campaign resulting in sales growth
and higher ROI. The RFMPQ and RFMT dataset con-
centrate on the idea of identifying the purchasing
power history of an individual customer or segment.
P variable represents the average purchasing power
per customer per all transactions, Q variable repre-
sents the average purchasing power per product, and
T represents the change of consumer buying behav-
ior or trend using change rate. The enhanced RFM
model also incorporates CLTV for predicting future
cash flow attributed to the customer’s shopping period
with the retailer [8], followed by applying a mod-
ified best-fit regression technique, and K-means++
and Expectation-Maximization (EM) clustering algo-
rithms to analyze the customer buying behavior as
well as to assess the clustering technique’s perfor-
mance using cluster quality assessment. The analysis
can also identify marketers’ area of focus and ensure
the highest quality of customer service.

2. Installing and using the microsoft word
template

Market segmentation is the process of categorizing
large homogenous market into similar or homoge-
neous smaller groups who share characteristics such
as income, shopping habits, lifestyle, age, and per-
sonality traits [9]. These segments are relevant to
marketing and sales and can be used to optimize
products, customer service and advertising to differ-
ent consumers [6]; It is seen that many companies
across the retail industry have identified customer
service as a market key differentiator and tend to
segment their customers for positive customer expe-
rience and service delivery [10]. There are three
types of market segmentation bases, namely demo-
graphic, geographic, and behavioral. Demographic
segmentation is the most commonly used variable
when segmenting a market. It has the ability to

F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques 6161

provide the retailer with a clear vision with the
future advertising plans, a precise customer shop-
ping profile and focuses on the measurable factors
of consumers and their households. Furthermore, this
segment is primarily descriptive in terms of gender,
race, age, income, lifestyle, and family status [11].
In contrast, behavioral segmentation divides cus-
tomers based on their attitude toward products. Many
marketers consider behavioral variables such as occa-
sions, benefits, usage rate, customer status, readiness,
loyalty status, and attitude towards a product as the
ideal starting points for creating market segmentation
[11]. According to behavioral segmentation, con-
sumer behavior is the segmentation process based
on their evaluation and buying activities, as well as
the use and disposal of goods to recognize consumer
needs. These criteria can provide a thorough under-
standing of consumer behavior as they reason from
social psychology, anthropology, economics, sociol-
ogy, and psychology that influence consumers on
their purchasing decision of products [21–25]. To
get a sense of the overall customer lifetime value
for the customer-base, [45] proposed a framework to
integrate customers’ distribution with the iso-value
curves, by grouping customers on the basis of RFM
characteristics and to understand the factors that trig-
ger consumer’s defection. [48] proposed analytical
model for consumer engagement, related to the subse-
quent stages of the consumer life-cycle like customer
development, customer acquisition, and customer
retention. The authors concluded that the availabil-
ity of data is vital to the development of advanced
analysis in each consumer’s stage. However, sev-
eral organizational issues of analytics for consumer
engagement remain, which constitute barriers to
implementing analytics for customer engagement.
In order to solve the problems of consumer behav-
ior that evolved with time, this research examines
the behavioral, demographic segmentation model and
identifies customer behavior using model Customer
Life Time Value (CLTV) and Recency, Frequency and
Monetary (RFM) model.

2.1. Customer life time value (CLTV) model

Customer Lifetime Value (CLTV) is an important
metric to measure the total worth or profit to a busi-
ness obtained from a customer over the whole period
of their relationship with the retailer [8]. The liter-
ature defines the customers churn as the extinction
of the contract between the firm and the customer,
where customer retention refers to the collection of

activities organizations take to reduce the number of
customer’s defections.

Churn rate and retention rate critical matrix for
any company and considered primary components
of the future CLTV. Where CLTV is an estimation
of the average profit, a customer is expected to gen-
erate before he or she churn [48]. The concept of
retention and churn is often correlated with industry
life-cycle. When the industry is in the growth phase
of its life-cycle, sales increase exponentially. How-
ever, customer churn is the most challenging task for
the retailer industry. In this perspective, more insight
is needed to know the reason for customer churn in a
dynamic industry.

The three main components of CLTV are customer
acquisition, customer expansion, and customer reten-
tion [46]. Nevertheless, it is crucial to consider COGS
(Cost of Goods Sold) and acquisition cost to square
off the real CLTV. The basic model to calculate CLTV
is presented in Equation (1).

�n ptCLTV = (r)t (1)
t=1 (1 + d)t

The above CLTV formula is more of a proxy for
an average customer who stays for X period of time
and pays Y total amount of money. The t represents
a specific period of time, while (t = 1) represents the
first year, and (t = 2) denotes the second year. The n
represents the total time period the customer will stay
with the retailer before churn occurs. The r represents
the month over retention rate. Pt is the profit that
the customer/customers will contribute or generate
to the Retailer in the Period t, and finally, d refers to
the churn rate. Additionally, the customer’s loyalty
can be calculated using the Retention Rate formula,
as illustrated in Equation (2). Based on the Retention
Rate formula, CE denotes to the number of customers
at the end of each time period, where, CN is the total
number of new customers acquired in the chosen time
period, and CS denotes to the number of customers
at the start of the time period. � �

(CE – CN)
Retention rate = × 100 (2)

Management of consumer retention requires the
tools that allow decision-makers to assess the risk
of each consumer to defect and understanding the
factors that trigger consumers’ defection [47]. Cus-
tomer retention strategy also known as a loyalty
rate is the collection of activities a retailer uses to
maintain on a long-term relationship basis by engag-
ing existing customers to increase profitability by

6162 F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques

Table 1
Criteria of customers in each segment

Segment Criteria of Customers

Best The average purchase amount > the total average
purchase amount

The average purchase frequency of
customer > the total average frequency

Spender The average purchase amount > the total average
purchase amount

The Average frequency of customer < the total average frequency

Frequent The average purchase amount < the total average purchase amount

The average frequency of customer > the total
average frequency

Uncertain The average purchase amount < the total average purchase amount

The average frequency of customer < the total average frequency

increasing the number of repeat customers. CLTV
represents a greater improvement compared to the tra-
ditional RFM analysis as the frequency of customer’s
purchases, and the amount of customers’ average pur-
chase is used for segmenting customers-base. CLTV
matrix classifies customer purchase behavior using
different segments, namely Best, Spender, Frequent,
and Uncertain classified by Marcus (1998). Table 1
illustrates the criteria of each segment that were clas-
sified by Marcus (1998).

2.2. Recency, frequency, and monetary (RFM)
model

RFM is a standard statistical marketing model
for customer behavior segmentation assess consumer
lifetime value. The model is very popular in the
retail industry as it groups customers based on their
shopping power history – how recently, how often,
and how much did the customer buy. RFM model
helps retailers group customers into various segments
or categories to identify customers who are more
likely to respond to marketing promotions and future
customer personalization services [17]. The R sym-
bolizes recency refers to the interval between the time
since last purchase the customer made. The F sym-
bolizes the frequency of consumer behavior in a time
period, and the M symbolizes monetary referring to
the amount of money consumption in a period [18].
Quintiles scoring is the most commonly used scor-
ing in the RFM method in arranging customers in
ascending or descending order or (Best to Worst).
Customers are grouped into five equal groups where
the best group receives the highest score of (5), and

the worst receives the lowest score of (1) [1]. The
RFM score is the weighted average of its individual
components and is calculated as portrayed in equation
3 and 4 to derive a continuous RFM Score. Finally,
these scores can be re-scaled to the 0 –1 range [17].

RFM score = (recency score ×

recency weight) + (frequency score

× frequency weigh + (monetary score

x monetary weight)) (3)

Rescaled RFM score = (RFM score

− minimum RFM Score)/(Maximum RFM

score − minimum RFM score) (4)

2.3. Market segmentation using data mining,
RFM, CLTV models and clustering
techniques

Market segmentation helps to differentiate and cus-
tomize marketing strategies into segments. Market
Segmentation is a significant key in data mining,
where data mining is used to interrogate segmenta-
tion data to create data-driven behavioral information
segments that are applied to detect meaningful pat-
terns and rules underlying consumer behavior [19].
Furthermore, [26] and [27] were among the stud-
ies that performed market segmentation using data
mining, RFM, CLTV, and clustering technique to
form a decision-making system. [28] proposed clus-
tering and profiling of customers using customer
relationship management (CRM) and RFM for rec-
ommendations were proposed. On the other hand,
data mining was conducted on historical data of cus-
tomer’s sales using the RFM model with K-Means
algorithm where results have outlined recommen-
dations to perform customer relationship strategy.
Also, f (2016) proposed a three-dimensional mar-
ket segmentation model based on customer lifetime
value, customer activity, and customer satisfaction.
For more accuracy, the author grouped customers into
several different groups. RFM, Kano, and BG/NBD
models obtained the corresponding variables.

Furthermore, the market segmentation model helps
enterprises to maximize their profits. In [29–35], cus-
tomers were classified into various clusters using
RFM technique and association rules were mined to
identify high-profit customers. RFM statistical Tech-
niques and Clustering methods for Customer Value

F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques 6163

Analysis were combined in [26] for a company’s
online selling. As far as the methodology is con-
cerned, there is no standard convention for measuring
customer purchase behavior as each literature differs
in examining customer purchase behavior. Neverthe-
less, the K-Means algorithm is noted as an extensively
used clustering algorithm in previous research due
to its simplicity and speed in working with a large
amount of data [36]. Despite its strength, it has been
recorded that the K-Means method uses a random dis-
tribution for the seeding positions and does not main-
tain the same result each time that it is run [37]. New,
improved K-means algorithm called K-means++,
which uses sophisticated seeding procedure for the
initial choice of the center positions and often twice
as fast as the standard k-means. Contrastingly, Neha,
Kirti, and Kanika (2012) noted that K-means and
(EM) Expectation-Maximization algorithms are the
two most commonly used based algorithms for the
identification of growth patterns. Even though EM
is similar to the K-Means algorithm, this algorithm
is based on two different steps iterated until there
are no more changes in the current hypothesis [29].
Expectation (E) refers to computing the probability
that each datum is a member of each class. Maximiza-
tion (M) refers to altering the parameters of each class
to maximize those probabilities. Eventually, they con-
verge, although not necessarily correct. Furthermore,
EM algorithm is embedded with a significant feature
where it can be applied to problems with observed
data that provide “partial” information only [30].
Based on several comparative studies of EM and K-
Means methods [31–34], it was observed that EM
outperformed K-Means and results were improved
when they were hybridized. The current study inte-
grates two dynamics models, namely CLTV and RFM
models, with the addition of new RFM variants, i.e.,
P, Q and T to cater the weakness and inaccuracy of
consumer modeling that are caused by the limita-
tions RFM. In addition, this study applies K-means++
and Expectation Maximization (EM) clustering algo-
rithms to offer the retail industry with effective analy-
sis of customer buying behavior through the combina-
tion of customer profitability and product profitability
in creating a strategic marketing campaign as
explained previously in the introduction [38–40].

2.4. Mining big data

Data mining is the process of extracting infor-
mation from large data sets and transform it into
an understandable form for further use. Data min-

ing can be used in such a case where the database
is large, and the classification of such data is dif-
ficult [35]. The term Big data is often used for very
large databases whose size in terabytes to many PETA
bytes and it is beyond the ability of commonly used
Relational Database Management (RDBM) to pro-
cess the data within a tolerable elapsed time. Patel,
Birla, & Nair (2012) have done a lot of experiment
on the big data problem. The result was the finding
Hadoop Distributed File System (HDFS) for storage
and map-reduce method for parallel processing on a
large volume of data. However, the research in Big
Data analysis using data mining especially with clus-
tering methods is still considered to be young, and
therefore attracts many researchers to conduct fur-
ther research in this potential area [37, 38] proposed
a fast-parallel k-means clustering algorithm based on
Map Reduce, which has been widely embraced by
both academia and industry. They used to speed up,
scale-up, and size up to evaluate the performances of
their proposed algorithm. Their finding showed that
the proposed model could process very large dataset
on commodity (Low-cost) hardware effectively.

Hadoop is becoming a commodity for every data-
driven organization, where data is larger and comes
in many formats, mining and extracting intelligence
from data has always been a challenge [39]. The
new dynamic in the database has brought new chal-
lenges to the current analytical models and traditional
databases and emphasize the need for a paradigm shift
in data extraction and data analysis. Such challenges
are the performance of the data retrieval and the vari-
eties of data sources for which the format of the
relational databases may no longer be the best option.
[39] stated that Traditional database systems fall short
in handling scalability to boost the performance effi-
ciency and dealing with Big Data effectively and
thus the adoption of based systems such as Hadoop
is increasing. Hadoop is an open-source framework
for data-intensive distributed system processing of
large-scale data, based on Map Reduce programming
model and a distributed file system called Hadoop
Distributed File system (HDFS).

Map Reduce programming model is a methodol-
ogy that deals with implementation and generating
large datasets, making Hadoop the preferred as a solu-
tion to the problems in the traditional Data Mining
[41].

The main components of Hadoop are Hadoop
distributed file system (HDFS) a high bandwidth
clustered storage allows writing an application that
rapidly processes massive data in parallel, which is

6164 F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques

vital for large files. Map Reduce, is the heart of
Hadoop. HDFS is high bandwidth clustered storage,
while Map-Reduce processing enormous pieces of
data and divide the input dataset into independent
smaller pieces and be distributed amongst multiple
machines referred to as nodes to parallel process them
[42].

3. Research method

The methodology of this work, which involved
Five phases, is outlined in Fig. 1. The first phase
focused on the implementation of our POS database
on a single node Hadoop distributed file system.

The experiments were performed on the following
system: Single node using Java: Hardware: Intel Core
i5, 8GB RAM, CPU 2.4 GHz Software: Java with
JDK 1.8 Hadoop implementation: Software: Ubuntu
16.10, Java with JDK 1.8, Hadoop 2.7.0. The sec-
ond phase focused on (ETL) Extract, Transform, and
Load involving data preprocessing steps, i.e., data
cleansing, features selection, and data transforma-
tion. Since the dataset that was used in the current
research was different than those in existing litera-
ture, a controlled experiment was performed where
the work of [20] was replicated as the baseline of this
research.

The hybrid approach (RFMPQ & CLTV) was
included in the third and fourth phase, where differ-
ent methods were employed stepwise. As the data was
transformed into three different variants, i.e., RFM,
PQ, and T, the first processing step differed in one of
them.

The classification was used to categorize cus-
tomer purchase behavior into CLTV matrix based
on the RFM and PQ dataset, while modified best-fit
regression was performed on the T dataset to find
the customer purchase trend (curve). Even though
[20] only employed K-means technique, the present
experiment was extended to include the utilization of
K-means++ and (EM).

Subsequently, the outputs were fed into the clus-
tering algorithms, i.e., K-means++ and EM at the
fourth phase for further demographical segmentation.
The accuracy of these clustering algorithms was mea-
sured during the final phase using the cluster quality
assessment that was introduced by Draghici & Kuklin
(2003). Additionally, the retention rate was calcu-
lated, and human judgment was also included as a
measure of the effectiveness of this method for a
marketing campaign.

Fig. 1. Methodologies for the hybrid of classification and cluster-
ing of market segmentation.

The Methodologies for this research is illustrated
in Fig. 1.

3.1. Proposed data transformation using RFM
model, RFMPQ, RFMT

In this study we are using Apache Hadoop on
single-node Hadoop cluster using Ubuntu Linux
12.04 64 Bits Server Edition was preferred as
the operating system and KVM (virtual memory)
was selected as virtualization environment. Hadoop
(HDFS) node was accessed via Secure Shell (SSH).
In this study, no parameter or optimization adjust-
ment was made on the operating system to cause
performance improvement. This type of Hadoop
implementation serves the purpose and sufficient to
have a running Hadoop environment in order to con-
duct our experiment. The market segmentation model
uses retail POS data acquired from a medium-sized
retailer from the State of Kuwait. The POS data con-
tains Three years (2012 – 2015) of customers initial
and repeated purchases who made their purchases
at different geographical branches. Each transac-
tion represented a product purchased, with each line
consists of a cashier number, store-code, item-code,
brand- code, product (quantity) sold, product price,
date and time of the transaction, sub-total, grand total
as well as the customer’s demographic information.

F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques 6165

Since the data was in separate compilations based on
the stores’ geographic locations, a common data for-
mat with consistent definitions for more descriptive
keys and fields was developed to merge the informa-
tion. In this phase, the string variables are converted to
numeric variables, and subsequently, missing values
were checked and replaced default or mean values
manually. In this research, the PQ variants will be
used to describe customer’s purchase power in differ-
ent demographic and behavioral eras and customer’s
attractiveness to a specific product and service. For
the T variant customers from the best segment are
used to identify customer purchase curve, and we cal-
ibrate the market segmentation model using repeated
transactions for 3220 customers over two years’
period.

The age attribute was grouped into four (i.e., ages
from 1–17, ages from 18–24, ages from 25–34, ages
from 35–44, ages from 45–54 and ages above 55).
The age group analysis was based on the premise
that a typical customer’s needs would change as
they age. The customer’s age was classified into
six categories, where each category was identified
using a unique number. Category 1 = (1–17), Cat-
egory 2 = (18–24), Category 3 = (25–34), Category
4 = (35–44), Category 5 = (45–54), and senior Cate-
gory 6 = (55 +). The Gender attribute was encoded as
1 for Male customers, 2 for Female customers, and 3
for Companies. Furthermore, the demographic con-
cept hierarchy method such as city and country was
replaced by higher-level concept nationality. Citizens
of Middle Eastern nationalities, Asian nationalities,
USA, and Canada, were assigned unique numeric
(binary) value. Other nationalities were grouped
based on continents, namely Europeans and Africans.
One exception was made for British nationalities
due to the high volume of purchases. To ensure the
maximum accuracy of RFM scores, the values of
five-dimension attributes from the POS Data were
necessary.

The attributes are described in Table 2 as follows:
The next step involved the calculation of RFM

scores as well as the newly proposed variation PQ
and T. The implementation of CLVT, retention rate,
and RFMPQ and T are developed using advanced
PL/SQL programming language. It must be noted
that the RFMPQ score refers to the weighted aver-
age of its individual components in which the scoring
analysis typically involves grouping customers into
equal buckets (quantiles) sizes. As far as this study is
concerned, the grouping procedure was applied inde-
pendently to the five RFMPQ component measures.

Table 2
Attributes of RFMPQ

CUSTOMER ID Customer unique identifier used to
capture customer’s related
information.

TRX DATE: Transaction date used to capture
customer’s Recency (R).

TXH COUNT Number of Transaction used to capture
the Frequency (F) number of each
transaction made by a customer.

TRX TOTAL SALE The total amount of each transaction
used to capture the Monetary (M)
value made by the customer.

TRXUNIT PRICE The average purchase power (Monetary)
used to capture Average Monetary (P)
per customer.

TRX QTY The average purchase power (quantity –
Q) used to capture the Average Items
purchased per customer.

Customers were grouped according to the respec-
tive measure into classes of equal sizes. The derived
R, F, M, P, and Q groups became the components
for the RFMPQ cell assignment. RFMPQ groups are
aggregated, with appropriate client-defined weights,
and the scores were the weighted average of its com-
ponents.

The next step involved where values and scores
of RFMPQ variables were determined and used as
inputs of clustering algorithms. According to this
research and the client’s requests, RFMPQ variables
had equal weights (1:1:1).

The second variant proposed in this research was
T variable, which represented the trend of customer
purchase behavior using change rate. This study pro-
poses a combination of two analytical data mining
steps. For finding T, the change in consumer purchase
behavior trend, it uses an enhanced best-fit regression
algorithm. Then T dataset is then put into the unsu-
pervised clustering algorithms, to split the consumers
into different groups based on patterns dissimilarities.
The variable T should answer a very important ques-
tion if the consumer is at high risk of shifting his or
her service to another retailer. One of the most com-
mon indicators of high-risk consumers is a drop off
in purchase power and a decrease of visits. One of the
most common indicators of high-risk consumers is a
drop off in purchase power and a decrease of visits.
Major limitations of market segmentation and RFM
models is ignoring the behavioral changes of con-
sumers during the time period of analysis. Although
the recency variable acted as one of the indicators
of such consumer behavior, it was affected by cus-
tomers’ transient behavior, and it was only based

6166 F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques

on the last purchase date. Therefore, introducing a
new analysis variable was essential for the retailer to
narrow down high-risk consumers.

Based on the perspective of a retailer, average val-
ues change according to customers’ purchase and
their satisfaction with the retailer in terms of price
and services. If these average purchase amounts of
a customer decrease continuously, it could be con-
cluded that this customer is or was on the verge of
shifting his business to another retailer. A similar
conclusion can also be derived from a customer who
had an increased average purchase value, showing
that the customer was very profitable and should be
ranked accordingly. Therefore, these customers with
a variety must be treated accordingly. A computable
parameter was developed through the introduction of
some advanced programming. Firstly, the purchase
amounts of all customers in each selected time period
of analysis were required.

3.2. Market segmentation using K-means++ and
EM clustering methods

The drawback of the classic k-means algorithm is
that the user needs to define the centroid point and
offers no accuracy guarantees. This has become more
critical when clustering documentation because each
center point is represented by a word and the distance
calculation between words is not a trivial task. To
overcome this problem, a k-means++ was introduced
in order to find a good initial center point. K-means++
is a simple probabilistic means of initializing k-
means clustering that not only has the best known
theoretical guarantees on the expected outcome qual-
ity but works very well in practice. According to
[43–51] k-means++ algorithm is another variation of
the k-means algorithm, a new approach to select ini-
tial cluster centers by random starting centers with
specific probabilities are used. The steps used in
this algorithm are described below. In this regard,
the essential component required is the preserva-
tion of the diversity of seeds while ensuring that
the outliers remain robust. The primary concern of
the k-means problem is to identify cluster centers
that minimize intra-class variance by reducing the
distances from each clustered data point. This can
be achieved through an effective and well-designed
cluster-initialization technique. k-means++ was pro-
posed in 2007 by [42] for choosing initial values
(seeds) for the classic k-means clustering algorithm
to avoid poor results found by the k-means clustering
algorithm. K-means++ algorithm initializes means

more intelligently so that there is a distribution of
cluster means that is roughly even relative to the data.
K-means++ accomplishes this by selecting the first
cluster center at random and then drawing the next
sample from a distribution that puts a heavy proba-
bility weight where there are data and no close-by
cluster center. The execution of K-Means++ and EM
algorithms is carried out using WEKA tools, then
the generated results are exported into excel for easy
comparative analysis. We evaluate the performance
of the market basket based on the mean calculated
across three years forecast customers’ sales transac-
tion. The K-means++ algorithm equation’s method
is explained below. The K-means++ algorithm comes
with a theoretical guarantee to find a solution that is O
(log k) competitive to the optimal k-means solution.
It is also fairly simple to describe and implement.

Expectation-maximization (EM) algorithm is an
iterative estimation algorithm, a method similar to the
K-Means algorithm introduced by Dempster, Laird,
& Rubin (1977). The Expectation-Maximization
algorithm is an important tool of a statistical and pow-
erful method for obtaining the maximum likelihood
estimation of the parameters of an underlying distri-
bution when data contains null and missing values
to generate an accurate hypothesis. There are three
steps involved in EM technique, and the first was
the EM clustering initialization. Every class j, of M
classes (clusters), is formed by a vector parameter
(θ), composed by two parameters the mean (�j) and
the covariance matrix (Pj) which defined the Gaus-
sian probability normal distribution as features used
to describe or classify the observed and unobserved
entities of the data set x. The Expectation Maximiza-
tion (EM) algorithm was aimed to approximate the
parameter vector (θ) of the real distribution. Clus-
ter Convergence was the third step in M-step. After
every iteration was performed, a convergence inspec-
tion was conducted to verify whether the difference of
the attributes vector of an iteration to the previous iter-
ation was smaller than an acceptable error tolerance
given by parameter.

3.3. Quality assessment

One type of cluster quality assessment as suggested
in [41] was performed in this study is to compare
between the size (diameter) of the clusters versus the
distance to the nearest cluster, the inter-cluster dis-
tance versus the size of the cluster is conducted in this
study. This process can also be understood in terms
of the distance between members of each cluster and

F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques 6167

Fig. 2. Cluster quality accessed by ratio of distance to the nearest
cluster and cluster diameter: source (Draghici, 2003).

the center of the cluster, and the diameter (size) of
the smallest sphere containing the cluster. If the inter-
cluster distance was much larger than the size of the
clusters, then the clustering algorithm is considered
to be trustworthy. The formula is explained as; min
(Dij)/max (di) where Dij was the distance between
cluster I and cluster j, where both i and j were from
(1 to 6) and di is the diameter.

Figure 2 shows the quality can be assessed sim-
ply by looking at the cluster diameter. Therefore, the
cluster can be formed by heuristic even when there
is no similarity between clustered patterns. This is
occurring because the algorithm forces K clusters to
be created.

4. Results and discussion on RFM and
RFMPQ dataset

The main objective driven by the customer pyramid
classification is to determine the segment in which
customers spend more with the retailer over a period
of time and also the segment that is less costly to
maintain. The distribution of customers after using
CLTV matrix on the RFMPQ dataset is illustrated
in Fig. 3 with respect to their frequency and average
monetary values. The findings generated by customer
value matrix classification revealed that there were
7024 customers from who were categorized under
the Best segment, 25153 customers fell under the
Spender segment, 39107 customers were grouped in
the Frequent segment, and 39624 customers were
classified in the Uncertain segment. These findings
indicated that most customers were shoppers with a
limited and high budget.

Based on the results in Fig. 4, it can be observed
that the analysis of K-means++ and EM clustering
algorithms illustrated that both algorithms are agree-
able on the gender and age segments. Cluster 1 is

Fig. 3. Effective customer pyramid classification.

the most beneficial segment of customers. The best
spenders in cluster 1 are predominantly females from
the age group (25–34) and (35–44). The cluster qual-
ity assessment of RFM dataset is shown in Table 3,
clearly shows the K-means++ algorithm with the size
of cluster 328.4529657 compare with the EM algo-
rithm 271.6329114 is far more accurate, because it’s
largest value of inter-cluster distance divided by the
size of the cluster.

Since classification, according to the CLTV matrix
was already applied prior to clustering, discussion on
the results of this experiment is divided into subsec-
tion according to the quadrant of the CLTV matrix.
The first section will elaborate on the demographic
clustering for gender and age, where summaries for
both gender and age are provided for each segment.
The following section will discuss the cluster aver-
age shopping (visits) frequency, the cluster average
monetary, and the cluster average spending per visit.

4.1. RFMPQ results using K-means++ and EM
on best, spender, frequent, and uncertain
segments

Results in Fig. 5. shows the summary for all four
segments, namely best, spender, frequent and uncer-
tain where most customers were females and their
average age according to the cluster. In general, K-
means++ is seen as the best clustering method except
in the Best segment; the accuracy is very close due to
the less number of customers. Each cluster will be dis-
cussed in further details with respect to the nationality
in the following sections.

Figure 6 shows the four clusters for Uncertain seg-
ment as generated by K-means++ and EM Clustering.
Total customers in this segment are 39624. The anal-
ysis in both clustering algorithms showed that cluster
1 was the most beneficial cluster (segment) because

6168 F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques

Fig. 4. The average of four clusters generated by the K-means++ algorithm and EM algorithm for gender and age.

it was more superior to the other clusters in terms of
the demographic variables.

Average age results portrayed that customers were
between the age of (45–54) are the most beneficial
customers, and the average gender indicated most
customers were male customers. In terms of the
average nationality, citizens of the Philippines were
considered as the best customers in average monetary
when using K-Means++ and citizens of India were the
best customers based on EM Clustering.

4.2. Summary of the market segmentation on
RFMPQ dataset

With regards to the Market Segmentation on
RFMPQ dataset, it can be summarized that Best seg-
ment was the most profitable segment with its Total
Average Spending Per Segment of (1109.37) com-
pared to other segments. Cluster 2 was the most
profitable cluster in the Best segment with most
female customers and the age groups of (25–34) and
(35–44). Meanwhile, the result for average national-
ity portrayed that citizens of Kuwait and UAE were
the best customers in average monetary and it was dis-
covered that both K-means++ and EM showed similar
accuracy. The Spender segment was the second most
profitable segment with the Total Average Spending
Per Segment of (834.01). Cluster 4 of this segment
was the most profitable cluster with an average mon-
etary of (924.19).

Females customers, those between the ages of
(25–34) and (35–44) as well as the citizens of Kuwait
and Qatar were the best customers, and it was also
shown that K-Means++ algorithm was the most accu-

rate. Next, the Frequent segment was the third most
profitable segment with the Total Average Spending
Per Segment of (349.57), and it was noted that cluster
1 was the most profitable cluster with an average mon-
etary of (397.81). Female customers, the age group of
(35–44) and citizens of the UAE and Bahrain were the
best customers. K-means++ algorithm was regarded
as the most accurate algorithm for this cluster in Fre-
quent segment.

The Uncertain segment was the least profitable seg-
ment with its Total Average Spending Per Segment
(168.59). Cluster 1 was the least profitable clus-
ter with an average monetary of (239.14). Female
customers, those in the age group of (45–44) and cit-
izens of the Philippines and some Arab nationalities
were the best customers, and K-means++ algorithm
was the most accurate algorithm for this cluster
in Uncertain segment. Based on the comprehensive
information extracted from the RFMPQ dataset, it is
concluded that (a) the newly proposed PQ method
can fulfill both robust classification and robust seg-
mentation for market segmentation model, even in
the dataset that has noisy data. Also, it is noted that
(b) The (PQ) is the most important variable, and it has
also been proven as a crucial parameter for clustering.

4.3. Predictive customer lifetime value

F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques 6169

Fig. 5. Summary for all segments in demographic clustering (Gender).

Fig. 6. Four clusters generated by K-Means++ and EM clustering for uncertain segment.

is essential to consider COGS (Cost of Goods Sold) customers’ expected lifetime and potential monetary
and acquisition cost to square off the real CLTV. value from new purchases. Furthermore, based on the
Predictive Customer Lifetime Value (CLTV): Estima- clients’ request, the static discount rate is set 25% in,
tion of the customer’s future value also considers the of which 25% is minced from the total sales. The

6170 F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques

Fig. 7. The result from the strategies implemented using ranking matrix.

acquisition cost, which is also set based on clients’
request, which is 400 becomes the estimated figure
or percentage of how much is spent to attract new
customers.

4.4. Predictive customer lifetime value

Predictive CLTV Matrix projects the new cus-
tomers’ expenses over their entire lifetime with the
retailer. Most retailers measure CLTV in dollars or
based on the retailer’s local currency spent by the cus-
tomer over their entire relationship with the retailer,
i.e., from the first to last transaction. Nevertheless, it
is essential to consider COGS (Cost of Goods Sold)
and acquisition cost to square off the real CLTV.
Predictive Customer Lifetime Value (CLTV): Estima-
tion of the customer’s future value also considers the
customers’ expected lifetime and potential monetary
value from new purchases. Furthermore, based on the
clients’ request, the static discount rate is set 25% in,
of which 25% is minced from the total sales. The
acquisition cost, which is also set based on clients’
request, which is 400 becomes the estimated figure
or percentage of how much is spent to attract new
customers.

4.5. Customer lifetime value ranking matrix

CLTV ranking groups customers into quad-
rants according to their profitability and retention
propensity. CLTV ranking can assist marketers in
performing more effective market segmentation [17].
This is accomplished through the incorporation of
RFMPQ and RFMT values with the CLTV to rank
the total sales quarterly and yearly using the ranking
matrix analytical function, as shown in Fig. 7.

Experimental results on CLTV using Ranking
Matrix, which is presented in Fig. 7 reflects the
success of the implementation of the market segmen-

tation data mining model. Important to note, the test
POS dataset is from 2012 to 2015. The client noticed
a drop in sales and a decline in foot traffic in the sec-
ond half of 2011. However, the development of our
Market Segmentation model started in 2013, and our
market segmentation model was implemented in the
fourth quarter of 2014. The analysis indicated that
the client had lost customers and sales immediately
dropped in quarter 1 of 2015 as a direct result from
the shift of mass marketing strategy to target mar-
keting strategy using market segmentation methods.
This is considered as a healthy result due to the shift-
ing from mass marketing to more customized market
segmentation. Nonetheless, in quarter 2 and onwards,
the client has gained a small increase in sales, and a
new group of customers has been retained. Results
also revealed that the retailer started to turn profitable,
starting with quarter 3 and big growth in quarter 4 of
2015.

A consistent result was observed from each quar-
ter in 2015. Thus, the client strengthened its market
position and made slow the effects of a rapidly weak-
ening overall retail market. It is generally portrayed
that the average lifetime of the customer is only two
years, and the churn rate is 52% (see Table 6), result-
ing in the development of a comprehensive marketing
strategy. Results have shown that the incorporation of
RFMPQ model and CLTV.

Matrix was the best method to categories and
classified different consumer segments and different
potential consumer segments by long term profitabil-
ity. Each of the market segments will be further
investigated and analyzed to provide specific charac-
terizations, better understanding and identification of
opportunities, as well as the profitability and the exis-
tence of risks as each segment no longer encompasses
the entire market. Based on the segments identified,
the retailer should create a tailored campaign for each
segment and offer separate services that distinguish

F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques 6171

these segments to maximize responses according to
each segment’s behavior. Finally, based on the exper-
iment results, it can be positively concluded that the
newly proposed RFM (PQ) and (T) were far more
efficient than the traditional RFM in classifying and
clustering the segmentation of customer value via
RFM attributes. Kindly note that the implementation
of this work was done in 2015; however, for some
circumstances, our documentation and publication
process was delayed for quite some time.

5. Conclusions

This study has introduced a new technique of
CLTV and proposed three new RFM variations P,
Q, and T. Firstly, the new RFM variations P, Q, and
T, which have some advantages over the standard
RFM model. The advantage of these newly pro-
posed methods is that the RFMPQ method takes into
consideration the changes in the customer’s average
shopping power over time as well as the average pur-
chase power of products purchased over time by a
given customer. “The new proposed PQ provided key
attributes describing customer’s purchase behavior in
different demographic eras. The P variant helped to
identify which segment of customers spends more
over a period of time and costs less to maintain. The
Q variant helped to identify customer’s attractiveness
to a specific product and service, and also helped
to identify best-selling products, which resulted in
increase of sales potential in terms of number of units
of best-selling products that can be sold.” The market
segments were constructed through several clustering
algorithms such as K-Means++ and EM. To convert
this idea into a computable dynamic parameter, newly
modified best-fit regression algorithm and RFMT
variable were proposed. The new modified regression
algorithm demonstrated a high degree of accuracy
in strengthening and reinforcing the effect of the
customer’s most recent T purchase while justify-
ing the importance of previous consumer’s purchase.
However, more modified regression algorithm can be
further extended to highlight key customer’s trends
using new demographic variables, like profession,
income level, spending methods, and online spend-
ing. Results of the application of the new RFM PQ T
variations have reflected their effectiveness for mar-
ket segmentation and their ability to offer the retail
industry with intelligent analysis to combine cus-
tomer and product profitability. This type of analysis
can also identify the area in which marketers should

focus their attention to as well as ensuring the highest
level of quality and customer service. Additionally,
these models help retailers identify VIP consumers
who are on the verge of shifting taken their business
to another competitor. Products that are highly prof-
itable and purchased by the most profitable segments
can also be easily identified. Results of the analysis
results as well as the marketing experts have agreed
that the classification of customer purchase behavior
using CLTV matrix against RFMPQ dataset revealed
the most accurate and crucial information about cus-
tomer purchase behavior.

Furthermore, results have also indicated that age
and gender variables provided an accurate result with
an estimated analysis accuracy of 75%. Nevertheless,
the nationality variable presented a low percentage
of accuracy in both algorithms, possibly due to the
missing and noisy data related to these variables. The
results of applying the new RFM PQ and T variations
have portrayed their effectiveness which resulted in
the sales growth rate of up to 6% for market segmen-
tation and their ability to offer the SMR industry with
intelligent analysis to combine customer and product
profitability.

Secondly, the retail is data-driven industry and pro-
cess large POS transactions Increase in the collection
of data is often seen as bottlenecks for Big Data anal-
ysis. Many retailers face the challenge of keeping
data on a platform, which gives them a single con-
sistent view. Although Hadoop is not the main focus
of this study, however, in this study, the capabilities
of Hadoop was investigated. Hadoop implementa-
tion provided a highly scalable data distribution and
lighting data analysis performance.

Finally, it has been proven that sophisticated
statistical modeling methods can provide useful
information for experts, but at the same time they
are costly and complex to implement and are likely
to present a challenge to the implementation of
marketing strategies. This study proposed a sim-
ple yet powerful approach to market segmentation.
The results have shown the model provides easy
to implement and affordable market segmentation
methodologies that deliver substantial value rela-
tive to the amount of effort involved. The analysis
results discovered from the (Best, Spender, Fre-
quent, and Uncertain) provided the client a guided
vision on how to calculate customer lifetime value
and retention.

Further search recommendation is to examine the
implementation of large POS dataset on multiple
clusters and scaling the cluster by adding extra nodes

6172 F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques

across several inexpensive servers. Finally, ignor-
ing consumer behavior purchase behavior trends can
be disastrous. Further research is needed for more
Explanatory data mining model like Market Basket
Analysis to help to identify what the profile of the
future customer might look like from a product per-
spective.

Funding

This research received no external funding.

Acknowledgments

We are grateful for the support given by the Uni-
versity Sains Malaysia where Y.A.F completed his
Masters on this research, under the supervision of
N.H.A.H.M.

References

[1] A.B. Patel, M. Birla and U. Nair, Addressing Big
Data Problem Using Hadoop and Map Reduce, in 2012
Nirma University International Conference on Engineering
(NUiCONE) (2012), pp. 1–5.

[2] A. Chorianopoulos and K. Tsiptsis, Data Mining Techniques
in CRM: Inside Customer Segmentation, no. 21. John Wiley
& Sons., 2010.

[3] A. David and S. Vassilvitskii, k-means++: The advantages
of careful seeding, Proceedings of the eighteenth annual
ACM-SIAM symposium on Discrete algorithms. Society for
Industrial and Applied Mathematics, 2007.

[4] A.P. Dempster, N.M. Laird and D.B. Rubin, Maximum
Likelihood from Incomplete Data via the EM Algorithm,
J of the R Stat Soc Ser B 39(1) (1977), 1–38.

[5] A. Kak, Expectation-Maximization Algorithm for Cluster-
ing Multidimensional Numerical Data, Purdue University,
2017.

[6] A. Neha, A. Kirti and G. Kanika, Comparative Analysis of
k-means and Enhanced K-means clustering algorithm for
data mining, Int J Sci Eng Res 3(3), 2012.

[7] B. Sohrabi and A. Khanlari, Customer Lifetime Value (CLV)
Measurement Based on RFM Model, Iran Account Audit
Rev 14(47) (2007), 7–20.

[8] B. Tammo HA, et al., Analytics for customer engagement,
Journal of Service Research 13(3) (2010), 341–356.

[9] C.C. Aggarwal, Data Mining: The textbook. New York,
USA: Springer, 2015.

[10] C. Bauckhage, Lecture Notes on Data Science : k-Means
Clustering Is Gaussian Mixture Modeling, B-IT, University
of Bonn We, 2015.

[11] C. Marcus, A Practical Yet Meaningful Approach to
Customer Segmentation, J Consum Mark 15(5) (1998),
494–504.

[12] D. Arthur and S. Vassilvitskii, k-means ++ : The Advan-
tages of Careful Seeding, in SODA ’07 Proceedings of the

eighteenth annual ACM-SIAM symposium on Discrete algo-
rithms, (2007), pp. 1027–1035.

[13] D. Birant, Data Mining Using RFM Analysis Knowledge-
Oriented Applications in Data Mining, Prof Kimi InTech,
2011.

[14] D.R. Kishor and N.B. Venkateswarlu, Hybridization of
Expectation-Maximization and K-Means Algorithms for
Better Clustering Performance, Cybern Inf Technol 16(2)
(2016), 16–34.

[15] F. Jiashuang, Y. Suihuai, Y. Mingjiu, et al., Optimal selection
of design scheme in cloud environment: A novel hybrid
approach of multi-criteria decision-making based on F-ANP
and F-QFD, DOI: 10.3233/JIFS-190630, Journal: Journal of
Intelligent & Fuzzy Systems, vol. Pre-press, no. Pre-press,
(2019), pp. 1-18.

[16] F. Peter, S. Bruce, G.S. Hardie and K. Lok Lee, Count-
ing your customers the easy way: An alternative to
the Pareto/NBD model, Marketing Science 24(2) (2005),
275–284.

[17] G. Lefait and K. Tahar, Customer Segmentation Archi-
tecture Based on Clustering Techniques, in 2010 Fourth
International Conference on Digital Society, (2010), pp.
243–248.

[18] G. Sunil, C.F. Mela and J.M. Vidal-Sanz, The Value of A
‘Free’ Customer, 2006.

[19] I. He and C. Li, The Research and Application of Customer
Segmentation on E-commerce Websites, in 2016 6th Inter-
national Conference on Digital Home (ICDH), (2016), pp.
203–208.

[20] I. Kolyshkina, E. Nankani, S.J. Simoff and S.M. Denize,
RDMA Aware Networks Programming User Manual, in
Australian & New Zealand Marketing Academy. Confer-
ence (ANZMAC), (2010), pp. 1–7.

[21] I. Yeh, K. Yang and T. Ting, Knowledge Discovery on RFM
Model using Bernoulli Sequence, Expert Syst Appl 36(3)
(2009), 5866–5871.

[22] J.A. McCarty and M. Hastak, Segmentation Approaches in
Data-Mining: A Comparison of RFM, CHAID, and Logistic
Regression, J Bus Res 60 (2007), pp. 656–662.

[23] J.L. Schafer, Analysis of Incomplete Multivariate Data, First
edit. USA: CRC Press LLC, 1997.

[24] J. Maryani and D. Riana, Clustering and Profiling of
Customers Using RFM For Customer Relationship Manage-
ment Recommendations, in 5th International Conference
on Cyber and IT Service Management (CITSM), (2017), pp.
1–7.

[25] J.R. Miglautsch, Thoughts on RFM scoring, J Database
Mark Cust Strateg Manag 8(1) (2000), 67–72.

[26] J. Wei, S. Lin and H. Wu, A review of the application of RFM
model, African J Bus Manag 4(19) (2010), 4199–4206.

[27] K. Dipanjan, S. Garla and C. Goutam, Comparison of
Probabilistic-D and k-Means Clustering in Segment Pro-
files for B2B Markets SAS Global Forum 2011 Management
Comparison of Probabilistic-D and k-Means Clustering in
Segment Profiles for B2B Markets, in SAS Global Forum,
Las Vegas, NV, USA, 2011, no. April.

[28] K.R. Kashwan and C.M. Velu, Customer Segmentation
Using Clustering and Data Mining Techniques, Int J Comput
Theory Eng 5(6), 2013.

[29] M. Cleveland, N. Papadopoulos and M. Laroche, Identity,
Demographics and Consumer Behaviors International mar-
ket Segmentation Across Product Categories, Int Mark Rev
28(3) (2011), 244–266.

[30] M. Khajvand, Z. Kiyana, A. Sarah and A. Somayeh, Esti-
mating Customer Lifetime Value Based on RFM Analysis of

F. Yoseph et al. / The impact of big data market segmentation using data mining and clustering techniques 6173

Customer Purchase Behavior : Case Study, Procedia Com-
put Sci 3(2011) (2010), 57–63.

[31] M. Rezaeinia and R. Rahmani, Recommender System
Based on Customer Segmentation (RSCS), Kybernetes
45(7) (2016), 1129–1157.

[32] M. Safari, Customer Lifetime Value to Managing Marketing
Strategies in the Financial Services, Int Lett Soc Humanist
Sci 42 (2015), 164–173.

[33] N. Li, L. Zeng, Q. He and Z. Shi, Parallel Implementation
of Apriori Algorithm Based on MapReduce, in 2012 13th
ACIS International Conference on Software Engineering,
Artificial Intelligence, Networking and Parallel/Distributed
Computing, (2012), pp. 236–241.

[34] O. Geman, et al., book chapter: From Fuzzy Expert Sys-
tem to Artificial Neural Network: Application to Assisted
Speech Therapy, INTECH, Artificial Neural Networks.
Models and Applications, Edited by Joao Luis Garcia Rosa,
Published: October 19th 2016, DOI: 10.5772/61493, ISBN:
978-953-51-2705, 2016.

[35] O. Geman, et al., book chapter Mathematical Models Used
in Intelligent Assistive Technologies: Response Surface
Methodology in Software Tools Optimization for Medical
Rehabilitation, Intel Syst Ref Library, Vol. 170, Hari-
ton Costin et al. (Eds): Recent Advances in Intelligent
Assistive Technologies: Paradigms and Applications, 978-
3-030-30816-2, 463829 1 En, (4), 2019

[36] P.K. Srimani and M.S. Koti, Evaluation of Principal Com-
ponents Analysis (PCA) and Data Clustering Technique
(DCT) on Medical Data, Int J Knowl Eng 3(2) (2012),
202–206.

[37] P. Mishra and S. Dash, Developing RFM Model for Cus-
tomer Segmentation in Retail Industry, Int J Mark Hum
Resour Manag 1(1) (2010), 58–69.

[38] R.A.I.T. Daoud, B. Belaid, A. Abdellah and L. Rachid,
Combining RFM Model and Clustering Techniques for
Customer Value Analysis of a Company selling online,
in 2015 IEEE/ACS 12th International Conference of
Computer Systems and Applications (AICCSA), 2015,
pp. 1–6.

[39] R. Soudagar, Customer Segmentation and Strategy Defini-
tion in Segments Case Study: An Internet Service Provider
in Iran, Lulea University of technology Department, 2012.

[40] R. Sethy and M. Panda, Big Data Analysis using Hadoop
: A Survey International Journal of Advanced Research in
Big Data Analysis using Hadoop : A Survey, Int J Adv Res
Comput Sci Softw Eng 5(7), 2015.

[41] S. Chen, Cheetah : A High Performance, Custom Data Ware-
house on Top of MapReduce, in Proceedings of the VLDB
Endowmen, (2010), pp. 1459–1468.

[42] S. Draghici and A. Kuklin, Data Analysis Tools for DNA
Microarrays, 1st editio., no. June 2003. CRC Press, 2003.

[43] S. Mohammad, S. Hosseini, A. Maleki and M.R. Gho-
lamian, Cluster Analysis using Data Mining Approach to
Develop CRM Methodology to Assess the Customer Loy-
alty, Expert Syst Appl 37(2010) (2010), 5259–5264.

[44] S. Zahra, M.A. Ghazanfar, A. Khalid, M.A. Azam, U.
Naeem and A. Prugel-bennett, Novel Centroid Selection
Approaches for KMeans-Clustering Based Recommender
Systems, Inf Sci (Ny) 320 (2015), 156–189.

[45] T.K. Hun and R. Yazdanifard, The Impact of Proper Market-
ing Communication Channels on Consumer’s Behavior and
Segmentation Consumers, Asian J Bus Manag 2(2) (2014),
155–159.

[46] T. White, Hadoop: The Definitive Guide, Third Edit.
Sebastopol, United States: O’Reilly Media, Inc, USA, 2012.

[47] T. Upadhyay, V. Atma and D. Vishal, Customer Profiling and
Segmentation using Data Mining Techniques, Int J Comput
Sci Commun 7(2016) (2016), 65–67.

[48] V. Vandenbulcke, F. Lecron, C. Ducarroz and F. Fouss,
Customer Segmentation Based On a Collaborative Recom-
mendation System : Application to A Mass Retail Company,
in Proceedings of the 42nd Annual Conference of the Euro-
pean Marketing Academy, 2013, pp. 1–7.

[49] V.S. Jose, M. Carl, F. Mela and S. Gupta, The value of a free
customer. No. wb092903. Universidad Carlos III de Madrid.
Departamento de Econom ́ia de la Empresa, 2009.

[50] W. Dong, Y. Cai and T. Kwok Leung, Making EEMD more
effective in extracting bearing fault features for intelligent
bearing fault diagnosis by using blind fault component sep-
aration, DOI: 10.3233/JIFS-169523, Journal: Journal of
Intelligent & Fuzzy Systems 34(6) (2018), 3429–3441.

[51] W. Zhao, H. Ma and Q. He, Parallel K -Means Clustering
Based on MapReduce, in In IEEE International Conference
on Cloud Computing, (2009), pp. 674–679.

Copyright of Journal of Intelligent & Fuzzy Systems is the property of IOS Press and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder’s express written permission. However, users may print, download, or email
articles for individual use.

PLOS ONE

Check for
updates

b OPEN ACCESS

Citation: Chou EY, Hsu DY, Hemon E (2020) From

slacktivism to activism: Improving tt”e

commitment power of e-pledges for prosocial

causes. PLoS ONE 15(4): e0231314. https://doi.

org/10.1371�ournal.pone.0231314

Editor: Valerio Capraro, Middlesex University,

UNITED KINGDOM

Received: October 20, 2019

Accepted: March 20, 2020

Published: ,llpril 29, 2020

access article distributed under tt”e terms of the

Creative Commons Attribution License, which

permits unrestricted use, distribution, and

reproduction in any medium, provided tt”e original

author and source are credited.

Data Availability Statement Al I data files are

available from the Open Science Database

database at https://osf.io/zdfr3/files/.

Funding: Tue authors received no specific funding

for this work

Competing interests: Tue authors have declared

that no competing interests exist

RESEARCH ARTICLE

From slacktivism to activism: Improving the
commitment power of e-pledges for prosocial
causes

Eileen Y. Chou 1*, Dennis Y. Hsu2 , Eileen Hernon3

1 Batten School of Leadership and Public Policy, University of Virginia, Charlottesville, Virginia, United States
of America, 2 Faculty of Business and Economics, The University of Hong Kong, Hong Kong, Republic of
China, 3 University of Virginia, Char1ottesville, Virginia, United States of America

* eileen.chou@virginia.edu

Abstract

Prosocial organizations increasingly rely on e-pledges to promote their causes and secure

commitment. Yet their effectiveness is controversial. Epitomized by UNICEF’s “Likes Don’t

Save Lives” campaign, the threat of slacktivism has led some or ganizations to forsake social

media as a potential platform for garnering commitment. We proposed and investigated a

novel e-pledging method that may enable organizations to capitalize on the benefits of e

pledging without compromising on its mass outreach potential. In two pilot studies, we first

explored whether and why conventional e-pledges may not be as effective as intended.

Building on those insights, we conducted one field and two lab experiments to tes t our pro

posed e-pledge intervention. Importantly, the field study demonstrated the effectiveness of

the intervention for commitment behavior across a 3-month period. The laboratory experi

ments provided a deeper and more refined mechanism understanding of the effect and

ruled out effort, novelty, and social interaction mindset as alternative explanations for why

the intervention may be effective. As technological innovations continue to redefine how

people interact with the world, this research sheds Ii ght on a promising method for trans

forming a simple virtual acknowledgment into deeper commitment-and, ideally, to action.

Introduction

In 2014, Indonesian political analyst Denny Januar Ali amassed more than 2.5 million retweets
that pledged to support Indonesian presidential candidate Joko “Jokowi” Widodo and to
replace discrimination with love [l]. In 2016, Facebook COO Sheryl Sandberg rallied the plat
form’s 1.6 billion users to redirect their 6-billion-times-a-day habit of clicking “Like” by sup
porting an online campaign committed to defeating ISIS recruiters [2]. The power of social
media as an efficient and massive information dispenser testifies to its capacity to mobilize
support on an unprecedented level. Proponents of using social media as a key campaign plat
form also tout its potential trickle-down effect: By raising awareness, people are more likely to
engage in activities that may indirectly help the given cause. As a result, political and social
groups increasingly dedicate resources to these new online efforts [3].

PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 / 21

https://doi.org/10.1371

mailto:eileen.chou@virginia.edu

https://osf.io/zdfr3/files

https://doi

PLOS ONE Commitment power of e-pledges

Despite the rapid growth of social media campaigns, the notion of slacktivism-defined as
“feel-good online activism with little meaningful social or political impact” (4–5]-challenges
the value of these efforts. Slacktivism highlights the ease with which people can “click it and
forget it.” Based on this assumption, some even argue that engaging in online campaie s may gn

lead people to believe that they have already contributed to the cause, without doing anything
meaningful. Indeed, a recent poll revealed that only 3% of active social media users cited online
campaigns as a key motivator in their donation decisions ( 6]. In a field study in collaboration
with Heifer International, Lacetera, Macis, and Mele showed that an online campaign engaged
almost 6.4 million online users (in the form ofe”likes” or “shares”), yet only 30 made an actual
donation [7]. The concern that slacktivism could encroach on tangible support has led, for
instance, to UNICEF Sweden’s “Likes Don’t Save Lives” campaign [8].

Rather than focusing on an impact evaluation of online campaigns, we contend that the
more pressing and realistic issue is how we can improve the existing platform to secure greater
commitment, thereby allowing organizations to capitalize on the platform’s power. With this
goal in mind, we focused on e-pledges-one of the most common methods used by online
campaigns-and shed light on the prevalent phenomenon of slacktivism. We aimed to answer
two questions: (a) why are e-pledgers less motivated to follow through on their pledges, and
(b) what type of intervention might increase e-pledges’ commitment power?

We conducted five studies to tackle these questions, with the results of each study informing
the next. Pilot Study lA demonstrated the presence and prevalence of slacktivism by directly
comparing the effectiveness of conventional e-pledges with their traditional counterparts. Pilot
Study lB provided a layperson’s perspective as to why conventional e-pledges are ineffective.
These insights then served as the foundation for a novel e-pledging method that seeks to
strengthen pledgers’ commitment to prosocial causes. Based on the aggregated results, we rea
soned that e-pledges that can better activate the pledger’s sense of public self-awareness and
personal accountability [9] would be more effective for securing commitment behavior. There
fore, we built on the existing literature and tested our prediction-that instructing participants
to pledge with both their own name and those of someone important to them-would be an
effective e-pledge intervention. Study 1, a field experiment, tested the effect of the intervention
against two common forms of conventional e-pledging methods: clicking on the “Like” icon
(as often used on social media platforms) and typing their name. Results from Study l then
prompted us, in Studies 2 and 3, to explore explanations why the intervention might work and
rule out alternative explanations (10]. We present all measures in this paper and data analyses
occurred after the predetermined data collection period (see online supplemental material for
the actual experimental materials).

This research offers insights with potential practical and theoretical advancements. First,
we empirically investigated a promising solution to a pervasive problem: the increasing, yet
ineffective, reliance onee-pledges as a way to secure prosocial commitment. By revealing lay
people’s perspectives on why e-pledges might contribute to slacktivism, the aggregated trend
in our data played a crucial role in developing an ecologically valid intervention. Second, we
ruled out several closely related alternative explanations to the effectiveness of the intervention.
By investigating both whether the intervention would work and why, our research lays a foun
dation for future research to generate flexible and improved e-pledge methods. In tum, this
work augments the scope of its practical implications. Third, this paper integrates theoretical
perspectives on social influence and objective-self-awareness, and presents a set of empirically
driven research questions for future studies. Lastly, given the continuing surge of social media,
a rapidly changing demographic, and e-pledges’ potential to reach a broad audience, our
research aims to shed light on how to effectively transform virtual acknowledgement into
deeper commitment-and, ideally, action.

PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 2 / 21

https://doi.org/10.1371

PLOS ONE Commitment power of e-pledges

Understanding what e-pledging is and the source of its ineffectiveness

A pledge is a person’s solemn promise to commit to a cause that he or she deems worthy ( 11].

In essence, pledges serve as a means of social control.Yet unlike formal contracts or sanctions,

failure to honor a pledge has minimal punitive consequences. Therefore, the commitment

power of pledges often relies on ( a) the normative expectation that people will be held account

able after making the pledge ( 12] and (b) self-investment and identification with the cause (4,

13, 14]. Together, these critical forces motivate individuals to fulfill their pledge.

People traditionally confirm their pledge by signing their name on a piece of paper or pub

licly announcing their commitment to the cause. An e-pledge, which we define as a virtual

promise to honor a commitment, serves the same objective function as traditional pledges.

The only substantive difference is the method by which people pledge: Instead of signing their

name by hand on paper, would-be-pledgers indicate their commitment electronically, either

on a social media platform (e.g., Facebook, Twitter) or through an online portal (e.g., Change.

goiv, Redcross.org).

However, as a plethora of slacktivism anecdotes suggest, e-pledges may not be as effective

in their abili ty to secure commitment as their traditional counterparts. We posit that while e

pledges and traditional pledges serve the same objective functions, they diverge in the psycho

logical weight they may evoke in the pledger. Indeed, past research posits two potential drivers

of this ineffectiveness: (a) The online pledge in general is perceived to be less trustworthy or

mobilizing (regardless of how it was signed) or (b) the method used to pledge (e.g., “Like”

clicking, name initials typing, etc.) dilutes the commitment effect. We expand on these two

drivers below.

On the one hand, it could be that people consider online pledges to be less persuasive or

trustworthy. In line with this notion, prior research has found that people perceive electroni

cally transmit ted docwnents to be less trustworthy (15]. Therefore, it could be that people are

willing to pledge their support, but question whether the campaign itself warrants further

involvement . As a result, they stop short of actual action.

On the other hand, it could be that the pledging process itself is less effective for motivating

people to take the desirable action. In recent e-signature research, compared with participants

who signed by hand, e-signers were less likely to obey the terms of the contract they signed

(16]. Similarly, conswners who typed their names (versus signing by hand) were less likely to

make a purchase afterward [ 17]. These findings suggest that conventional methods of e-pledg

ing may be the reason for its ineffectiveness, independent of the cause or campaign being

promoted.

To address slacktivism within the e-pledge domain, we first need to demonstrate empiri

cally that it is indeed an issue and then try to understand the underlying source and mecha

nism of the problem. To this end we conducted two pilot studies that served two purposes.

First, Pilo t Study lA and 1B s provided an empirical assessment oflaypeople’s engagement in

and perception of slacktivism-specifically, its presence, prevalence, and severity. Second, we

presented participants in P ilot Study 1B with the two potential sources of e-pledge’s ineffec

tiveness grounded in prior research, and gauged which they deemed more consequential.

Based on the results, we were able to derive the most suitable way to improve the overall effec

tiveness of e-pledging for securing commitment behaviors.

Pilot Study 1A: Are conventional e-pledges effective?

Pilot Study lA set out to demonstrate whether conventional e-pledges are indeed less effective

than the traditional way of pledging in a pro social domain that supports scientific advance

ment. To do so, we investigated whether three different forms of pledging could influence

PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 3/ 21

https://doi.org/10.1371

https://Redcross.org

PLOS ONE Commitment power of e-pledges

subsequent commitment behavior in variant degrees. We included two common forms of e
pledging-a checked box (as often seen on platforms such as Facebook and Twitter) and typed
full name ( as often seen on platforms such as Change.gov and Redcross.org)-and compared
their impact to the traditional way of pledging with a handwritten signature. We then mea
sured their subsequent commitment behavior to the clause.

Methods

Participants and procedure. Ninety-three undergraduate students (mean agee= 20.38,
SD = 3.63; 51 % female) participated in the study in exchange for a snack and the chance to win
a $50 bonus. We obtained IRB approval from the University ofVirginia to conduct this study,
with written consent from the participants.

Participants completed a two-stage study on a laptop preloaded with the study programed
in Qualtrics. In the first stage, participants were informed of a cover story that the study was
interested in decisions made under time pressure. The then played three rounds of “Where’s ey

Waldo?.” Each round presented participants with a large image and asked them to locate the
figurine “Waldo.” Participants had up to 30 seconds per round to find Waldo. We included
this first stage and a cover story to minimize potential demand effect of participants succumb
ing to how they think the experimenter would want them to behave.

Pledee- sie nine manipulation. Ue n completing the Where’s Waldo task, participants g g g po

then learned that they would have the option to sign a pledge to support evidence-based behav
ioral research at their institution. The pledge read as follows:

Please read the following petition regarding behavioral scientific research, and sign if you

agree. Otherwise, leave this blank and move to the next page.

To create a better tomorrow, we must start today and draft evidence-based policies. Investing

time.focus, and money in understanding the social and psychological implications of public

and private policies is crucial in their eventual effectiveness.

Join us at the Behavioral and Science Policy Association (behavioralpolicy.org) in helping to

develop a rigorous, comprehensive, and evidence-based behavioral research. No matter what

you do, let your actions be seen.

Participants were then randomly assigned to sign the pledge in one of three ways. Partici
pants were asked to either “Take the pledge by click on the Like button below” (Like condition),
“Take the pledge by typing your initials below” (initials condition), or “Take the pledge by sign

ing your name with the cursor in the space below” (traditional pledge condition). Everyone read
the exact same pledge. The only difference was how they signed the pledge.

Commitment behavior. After the pledge, participants were told that the experimenters
would like to gain insight into how to improve the participant-recrnitment process at their
institution. Their responses would allow the experimenters to enhance behavioral research.
Participants were then given the ope rtunity to provide as many or as few ways of improving po

how participants were being recruited. In essence, this task provided participants with an
opportunity to support evidence-based behavioral research-which adhered to the pledge that
they had signed. We then measured the nwnber of suggestions each participant provided,
which served as the behavioral measure of commitment.

The instruction made it clear that the participants were under no obligation to either sign
the pledge or provide any suggestions to the experimenters. Regardless of their behaviors and

PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 4 / 21

https://doi.org/10.1371

Home

https://Change.gov

PLOS ONE Commitment power of e-pledges

Table I. Poisson regression analysis on commitment behavior, Pilot Study IA.

Variable

Pledge Condi tion

B l SE 9S% CI

Check Box – .86 “‘
“‘ “‘ .23 [- 1 . 3 1 , – .40)

Type Initials H
– . 64 . 24 [- 1 . 1 2, – . 1 6)

Hand-signed’ – – –

Intercept 3.64,t,,t,, t . 1 2 [ .35, .85)

‘p<.05; "p<.0 1 ; ' "p<.001 'Hand-signed condit ion served a s the reference group https://doi.org/1 0.1 371/joumal.pone.0231 31 4.t001 the pledging condition to which they were randomly assigned, participants then provided their demographic infonnation and were thanked and excused. Results All participants signed the pledge to help advance behavioral research. However, more than half of the participants (53.8%) did not provide suggestions. This prompted us to conduct a Poisson regression analysis to gain a more detailed understanding of how much people actu ally helped. We used a Poisson regression analysis because it allows us to preserve the mean ingfulness of the zeros in our data and because the dependent variable is a count variable. Table l presents the full results of the regression analysis along. As predicted, the pledge-sign ing manipulation had a significant impact on commitment behavior .i(2) = 1 6.59, p < .001. Parameter estimation with the signed by hand condition as the reference group indicated that both the checked box (B = -.86, SE = .23, x2 = 13. 92, p < .001e) and the typed initials condition (B = -.64, SEe= .2 4, x2 = 6.94, p = .008) differed significantly from the handwritten condition. Pairwise comparisons further revealed that those who signed the pledge by hand volunteered more suggestions (M = 1.82, SD = 2.27) than those who pledged via checking a box (M = .77, SD = 1.33; p < .001, Cohen's d = .56) or typing their initials (M = .96, SD = 1.26; p = .008, Cohen's d = .46). The checked box condition did not significantly differ from the typed name condition (p = .45). Discussion Results from Pilot Study lA reveal that although everyone received the same text in the pledge, how they signed it significantly affected whether and how much they hele d to further the pe cause. In short, common forms of e-pledging are indeed less effective at securing commitment than the traditional form of signing pledges by hand. This discrepancy further highlights the importance of bolstering and solidifying e-pledges' effectiveness as a commitment tool. Pilot Study 1B set out to further understand why this effect occurs. Pilot Study 1 B: Why conventional e-pledges are ineffective We conducted an online survey study using the Amazon Mechanical Turk platform (MTurk). The objective of this study is to understand laypeople's perception of the shortcomings of con ventional e-pledges. As these online participants came from the population frequently targeted PLOS ONE I https://doi .org/1 0. 1 371 /journal .pone.0231 31 4 April 29, 2020 5 / 21 https://doi.org/10.1371 https://doi.org/1 PLOS ONE Commitment power of e-pledges for online campaigns and e-pledges, we reason that their responses would be a valuable and valid source of information. Methods Particieants and procedure. Three hundred and one participants recruited from the p MTurk online platform completed the survey online ( 4 4% female; mean agee= 35.10, SDe= 9.9 1 ). We obtained IRB approval from the University of Virginia to conduct this study, with written consent from the participants. After entering their MTurk ID, participants read the definition of slacktivism, which was defined as "a phenomenon in which people pledge support for a cause on social media without following up with actual behaviors that contribute to the cause." They then responded to two blocks of survey questions in sequence to measure their perceptions of (1) the prevalence and severity of slacktivism and (2) why e-pledges may fail to work. We describe each of the blocks in more detail below. Participants also had the opportunity to provide open-ended comments on slacktivism at the end of the survey. They were not obligated to provide any responses to this question; in our final data, ouly 96 partici pants provided any comments, half of which were not related to slacktivism. Therefore, we did not submit the data to any systematic qualitative analysis. Measures of the prevalence and severity of slacktivism. After reading the definition of slacktivism, participants were told that the experimenters would like to learn more about their perception of the prevalence and severity of slacktivism. Participants then responded to three questions: whether they had personally engaged in slacktivism in the previous 6 months (1 = definitely not to 5 = definitely yes); how prevalent a problem they think slacktivism is (1 = not at all to 5 = very prevalent); and how serious a toll it takes on society at large ( 1 = not at all to 5 = a great deal) . Measures of whe e-pledees mae fail. After completing the block on slacktivism prevay g y lence and severity, participants were then asked to reflect on reasons why e-pledges may be ineffective. We grounded these reasons in past research, which highlight two competing forces that may contribute to slacktivism. Participants first reviewed five reasons that may have con tributed to e-pledging's failure to secure commitment and indicated how much the thoughtey each contributed to slacktivism (using a 5-point scale: l = not at all to 5 = very much so). Two of the reasons focused on the pledger's reaction to the pledge ( "People feel less accountable to the pledges and petitions they sign online '; "People feel less guilty for breaking online pledges and petitions"), and three that concerned the pledge itself ("There are too many online pledges and petitions around"; "People often question the veracity of the on line pledges and petitions"; and "Online pledges and petitions are less emotionally appealing"). After rating each of the five rea sons, participants then ranked order them from 1 = most important to 5 = least important. Results Severity of the issue. We submitted participants' ratings of their personal involvement in slacktivism, the prevalence of slacktivism, and the severity of slacktivism to a series of one sample t-tests. Results revealed that all three were significantly more than the mide int of the po response scale (t(300) > 5.86, p < .001). Most notably, 58.8% of participants indicated that they most likely or had definitely committed slacktivism in the previous 6 months, and 80.1 % indicated that it is a prevalent or very prevalent issue. In total, 46.9% of participants agreed that slacktivism is a serious to very serious issue for society. Whe e-pledees do not work. Because participants provided individual ratings for each of y g the five reasons, we employed a repeated measures AN OVA on the rating data to test equality of means of the five reasons. As all of the participants rated the same five statements, a repeated PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 6 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges 4.S 3.S 3 2 . S 1 . 5 Less accoun1ability Fed ing l ess guilty Too many e-pledges Too man)' fake Xot emotion.illy pledges appealing Fig I . Ratings of factors that contributed to slacktivism, Pilot Study 1 B. https://doi.org/1 0.1 371/joumal.pone.0231 31 4.g001 measures ANO VA would allow us to investigate whether there were overall systematic differ ences across how people responded to those statements. Results revealed a significant differ ence in lay perception of what contributes to slacktivism (F( 4, 1200) = 60.7 4, p < .001, ,J2 = .1 6; Fig 1 ). The overall significant effect further granted us the ability to assess pairwise differences between the statements. Subsequent pairwise comparisons showed that "less-accountability" (M = 4.16, SD = .96) and "less guilt" (M = 4.10, SD = 1.00) were significantly different from the remaining three reasons (p5 < .001 ). However, these two reasons were not perceived to be con tributing differently (p = .34). We then examined forced ranking data. A Friedman nonparametric test demonstrated that there were overall differences in the rankings of why people thought e-pledgers shirk (x2 ( 4) = 327.87, p < .001 ). Participants' rankings identified "lacking accountability" (M = 1 .95) and "feeling less guilty" (M = 2.35) as the two highest rated reasons for why e-pledgers shirk. The remaining reasons, in order of rank, were "too many pledges available" (M = 3.1 6), "online pledges are often fake" (M = 3.72), and "not as emotionally appealing" (M = 3.82). Friedman follow-up tests revealed that "less accountability" was ranked to be significantly more impor tant than "feeling less guilty (p < .001), and each were ranked as significantly more important than the third-ranked item, "too many pledges available" (p < .001 ). Discussion This exploratory study yielded several insights. First, it provided empirical support for the prevalence of slacktivism. Second, not only do people recognize the severity of the issue, they also acknowledge that they themselves have been culprits. This discrepancy between willing ness to pledge and subsequent commitment behavior further accentuates the importance of improving e-pledges' effectiveness as a commitment tool. More importantly, the pilot study provided directions for how we could improve conven tional e-pledges. Our results identified ( 1 ) weak accountability and (2) lack of emotional investment as major contributors to slacktivism; both reasons rest more on the method of e- PLOS ONE I https://doi .org/1 0. 1 371 /journal .pone.0231 31 4 April 29, 2020 7 / 21 https://doi.org/10.1371 https://doi.org/10.1371/joumal.pone.0231 PLOS ONE Commitment power of e-pledges pledging itself than on the pledges in general. These findings suggest that to strengthen e pledges and curb s]acktivism, we would need to think about ways to boost accountability and heighten the sense of emotional consequences of the pledge. These findings enabled us to design potential interventions that may be more effective. Study 1 : E-pledging to volunteer in the field setting Results from our two pilot studies served as a springboard for potential ways to improve e pledges' effectiveness. When developing interventions, it is essential to keep any constraints and boundaries in mind. Furthermore, the intervention should preserve the current system's strengths while improving it. We aimed to preserve two parameters. First, two strengths of e pledges are their efficiency and ease of administering (1 8]. Any modification, therefore, should be neither cumbersome for p]edgers nor difficult for campaigns to disseminate. Similarly, the proposed method should be as generalizable as possible across different causes; otherwise, too much tailoring to the cause or specific campaign may limit broad usage. With these caveats in mind, we focused our attention on how to boost psychological accountability and a sense of nonnative pressure when people sign an e-p]edge. We reasoned that one way to improve an e-p]edge's commitment power is to raise the sense of public self awareness as people e-p]edge. Public self-awareness is the state in which people focus on the impressions the make on others, based on their behavior and appearance [9]. In such a state, ey people observe their own behaviors from the vantage point of real others and seek social approval [ 1 9-2 1 ]. In turn, public self-awareness induces greater accountability and leads peo ple to act in line with perceived social norms and personal standards. Notably for our purpose, public self-awareness can be activated using different accountabil ity cues (21-24]. For instance, a classic example of an accountability cue is the mirror manipu lation [23] : placing an individual in front of a mirror so that their image was visible throughout the experiment served to heighten participants' self-awareness. Likewise, the pres ence of a camera can induce public self-awareness (25]. For instance, being in front of a web cam with other people who, potentially, are watching the webcam feed can also heighten accountability (20]. We propose that asking people to pledge with their own name plus the names of someone important to them would be a two-pronged approach to creating such an accountability cue. First, this self-other e-p]edge could heighten psychological accountability by introducing a vir tual audience as people e-p]edge (20, 26-28] . The familiar nature of the other person's name would also lead e-p]edgers to perceive the "virtual audience" as more legitimate: People respond better when they envisage having to explain their actions to a friend (a legitimate audience) than to a random stranger [an illegitimate stranger; 29-31] . Furthermore, the sense of social surveillance, achieved through real or imagined presence of others during acts, has been shown to fuel public self-awareness (20, 32-33] . In turn, we predict that a heightened sense of public self-awareness should lead to more socially desirable behavior (20, 3 4]. Second, this self-other e-p]edge could emphasize core personal standards and normative expectations, because the introspective process by which people contemplate who is important to them could draw attention to the self [32, 35] . When people are more focused on the self, they become more attuned to important personal standards and fee] pressured to act in line with their core identity (36]. Beca use people have an inherent desire to see themselves as con sistent across behaviors, this activated sense of self compels them to comply with the behaviors targeted by the pledge (12]; several empirical studies have documented the mobilizing power of self-focus in securing prosocial behavior (21 ], which ]end indirect support to our prediction. PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 8i/ 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges Taken together, we set out to test a novel e-pledging method aimed at increasing public self-awareness: self-other pledging. This method requires that e-pledgers pledge not only with their own name but also with the name of someone important to them and whom the respect.ey It is also important to note that although pledgers are using two names to pledge their support, they are making the pledge on their own behalf and not the other person's. In effect, e-pledgers are dedicating their effort to the other person they included in the pledge. To test the effectiveness of our proposed intervention in a context with high ecological validity, we conducted a field experiment in collaboration with a university-affiliated student volunteer center that sponsors various programs in the community. One of the major chal lenges most nonprofit organizations face nowadays is high volunteer attrition (37] . Because these organizations greatly depend on volunteers to function, securing volunteer commibnent is critical. Our goal was to investigate whether our proposed e-pledge intervention would func tion better at securing volunteering commitment than conventional methods. At the beginning of the fall semester, students who had registered for one of the center's vol unteering programs were asked to take an e-pledge of commibnent. They were randomly assigned to one of three methods of e-pledging: two conventional methods (signing with their name or by clicking on "Like") and our proposed method (combined self-other names). We predicted that at the end of the semester (three months later), those who pledged with com bined names would volunteer significantly more hours than those who pledged with either of the conventional methods. Methods Participants recruitment process. At the time of data collection, the center sponsored 27 volunteering programs around the community, such as youth mentoring, tax services, housing improvement, and medical services. Each of these programs was headed by a program direc tor, who oversaw the management of their volunteers throughout the duration of the semester. We worked directly with program directors at the volunteer center because of their proximity to the organization process. Eighteen program directors agreed to work with us. During the time of our data collection, the 1 8 program directors were tasked to manage 1 40 volunteers in total. Per our arrangement with the program directors, we were permitted to embed an e pledge at the end of a standardized and mandatory online survey that all volunteers receive at the beginning of the semester. The online survey was administered by the program directors (38]. We were also given access to the timecards at the end of the semester, along with basic demographic information of the volunteers (i.e., age, year at school, tenure with the program, and gender). We were not pennitted to contact the volunteers during the course of the semes ter or after the semester had ended. We obtained IRB approval from the University of Virginia to conduct this study, with written consent from the participants. At the end of the semester, eight volunteers terminated their involvements with the volun teering programs all together (four from the Like condition, one from the initials condition, and three from the combined self-other initials condition; see O nline Supplemental Material S l Table for more detail). Therefore, our final data set comprised of 132 volunteers across 18 volunteering programs (77% female; mean agee= 19.51, SD = 1.1e4; tenure with programe= 1 .76 semesters, SD = .87). We did not have access to the reasons why those eight volunteers tenni nated their involvements, nor did we have access to the nwnber of hours they worked prior to terminating their involvement with the programs. We conducted a Chi-square analysis to assess whether attrition was related to experimental condition. Results revealed that there was no significant relationship between condition and attrition (x2(2) = 1 .79, p = .40). We also con ducted a correlation analysis to verify that the random assignment to experimental conditions PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 9 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges and volunteering programs were not correlated. The Pearson Correlation Coefficient between the volunteering program and experimental condition is 0.01 4 (p-value = .865). This result suggests that both variables are not correlated with each other. Pledee manipulation. The program directors informed us that it is customary for all volg unteers to complete an online survey at the beginning of the semester that asked volunteers to provide basic demographic information, such as age and gender, and their proposed schedule. This provided us with the opportunity to institute an e-pledge. At the end of the survey that they would normally complete, volunteers were asked to pledge that they would honor their commitment throughout the 3-month period. All of the participants read the same content of the pledge: "On my honor, I pledge to uphold the values of a good volunteer. I will consistently attend my volunteer shifts on time, I will do my best to work enthusiastically and maintain a positive attitude, and I am committed to seeing my work through until the end of the semester." Although the pledge's content was identical, we randomly assigned volunteers to one of three e-pledge formats: two conventional methods ( clicking on "Like" or typing their initials) and the combined self-other initials e-pledge. We randomized participants to different e pledging conditions regardless of the program. Following prior research, we asked participants to use initials instead of their full names in order to protect their identity and maintain ano nymity [see 1 6]. Those in the "Like" e-pledee condition read the above statement and were asked to "Pleaseg sign the pledge by clicking on the "Like" symbol". Those in the "Self-initials" e-pledee condition read the pledge above and were asked to g "Please sign the pledge by entering your initials." Those in the proposed intervention-the "Combined self-other initials" e-pledee condig tion read the pledge above and then were asked to "Think of a person who is extremely impor tant to you and helped you to become the person you are today, then sign the pledge with both your own initials plus the initials of that person." All of the volunteers opted to take the pledge. The extent of our involvement with the pro grams or the volunteers ended after that initial online survey. Volunteers went about their rou tine and did not receive any reminders or follow-up surveys from us. At the end of the semester, we obtained the logs of volunteer hours from the program directors; the logs tracked volunteers' actual work hours across the semester, which served as our dependent variable. Control variables. We consulted with program directors to understand the demographic variables that have systematically affected volunteering hours in the past. Based on that infor mation, we controlled for each volunteer's age, sex, and any previous volunteer experience with the nonprofit organization. These information were part of the mandatory survey that volunteers filled out at the beginning of the semester. Analytical approach. We anale ed the data in two ways. We first used path modeling to yz analyze whether the e-pledge manipulation had a significant impact on volunteering hours. Because (a) different programs may require different levels of involvement, with some being more time-intensive than others (e.g., daily after-school tutoring versus cleaning up a river bank on weekends), and (b) theee-pledging conditions were randomly assigned to volunteers regardless of the program, volunteering hours in our experiment were not independent across programs. Therefore, we modeled the programs' nonindependence using the cluster command in Mplus 7 (39] , which adjusts standard errors for nested data structures. We also conducted Poisson regression analyses to further understand whether those who pledged with combined iuitials would volunteer significantly more hours than those who PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 0 / 21 https://doi.org/10.1371 hmteer Tenure �i- - - - - - - - - - - -o. PLOS ONE Commitment power of e-pledges 0. 1 2 (.04) Pledging Condition \ ohmteering Hours 0. 1 6 (. 1 2) 0 . 18 (.0 )* 0.02 (.08) 0.49 I'- - - - - - - - - - - - -olunteer Sex (.10)* • -0. 1 3 (.06) olunteer Age p<0.00 1 ; *p<0.0 1 ; p<0.05 R.,v[SEA = 0 .04; CFI: 0.96; Chi-square(6)= .44; p = 0.28 Fig 2. Path model analysis, Study 1 . https://doi.org/1 0.1 371/joumal.pone.0231 31 4.g002 pledged using conventional methods (i.e., "L ike" clicking and name initials typing), above and beyond the control variables. We adopted a Poisson regression analysis because the dependent variable is a count variable. Results We conducted a single-indicator nested modeling analysis clustered on programs, controlling for volunteer's age, sex, and their previous volunteer experience with the nonprofit organiza tion. Fig 2 presents the path coefficient. We first examine the overall model to determine whether it confirmed that the pledging condition had a significant effect on hours volunteered. This analysis revealed that model fit indices satisfied the goodness of fit standard [CFI = 1 .00, RMSEA = 0.00, x2(6) = 5.53, p = .47; Fig 2; 40] . We then focused on the impact of the interven tion. As predicted, those who e-pledged with combined initials volunteered significantly more hours throughout the semester than the those who pledged with conventional methods (B = .1i1 , SE = .04, p = .006; CFI = 1.00, RMSEA = 0.00, x2(6) = 5.05, p = .53). Poisson regression analysis shed more light on the self-o ther initials effect (see Online Sup plemental Material S2 Table) . Model l included only the experimental conditions. Model 2 showed that the effect strengthen after including program as a control variable. Model 3 included all remaining control variables-volunteers' gender, age, and tenure. Results showed that, after controlling for the control variables, volunteers in both "Like" and "self-init ials" conditions volunteered significantly fewer hours than those who pledged via "self-other ini tials." Volunteers who took the "like" e-pledge (M = 7.29 hrs, SD = 4.93) worked significantly fewer hours than those who took the "self-o ther" e-pledge (M = 9.09 hrs, SD = 5.45; B = -.56, SE = .25, p = .02). Those who pledged with their self-initials (M = 8.12 hrs, SD = 4.80) also PLOS ONE I https://doi .org/1 0. 1 371 /journal .pone.0231 31 4 April 29, 2020 1 1 / 21 https://doi.org/1 https://doi.org/10.1 PLOS ONE Commitment power of e-pledges worked fewer hours than those e-pledged with "self-other" initials (B = -5 4, SE = .27, p = .05). This indicates the robustness of the self-other initials effect. Discussion Study 1 provides the first validation of our self-other initials intervention. By asking volunteers to pledge with both their own initials and another important other's, they volunteered 24.69% longer over a span of 3 months than those who pledged with the "Like" button, and 11. 94% longer than those who pledged with self-initials only. This field study's longitudinal design also boosts the external validity of the effect. These findings, while supportive of our predictions with strong external validity, are less controlled due to the nature of field studies. For instance, the number of volunteers who participated in this study was outside of our control. Additionally, as a compromise we had to make to gain entry to the organization, we agreed to make the experimental design as nonintrusive as possi ble. Hence, the only volunteering information we were able to retrieve was the number of hours. To address these concerns and enhance the causal inference of our findings, we con ducted Studies 2 and 3 with better experimental control to complement our initial findings. More importantly, results from Study 1 reveal that although everyone received the same pledge, how they signed affected whether and how much they helped in a tangible manner to further the cause. Yet, results from Study 1 did not speak to why pledging with the name of someone important to them made a significant difference in behavior. On the one hand, par ticipants might have instinctively dedicated their effort to the person the named in theirey pledge. On the other hand, simply recalling a person might have been sufficient to improve commitment. Study 2 set out to refine the intervention by explicitly asking participants to ded icate their efforts to that person who is very important to them. Study 2 : E-p ledging to act Study 2 strived to achieve four goals: First, we aimed to gain greater control and reduce the noise that is inherent in field experiments. To do so, we created an experimental design in which participants had the opportunity toee-pledge their commitment to the same cause, via different e-pledging methods. We viewed this as a more conservative test of theee-pledge inter vention, because people were all pledging to the same cause and they were randomly assigned to the various e-pledging methods. Second, we set out to better understand why pledging with people's own initials plus that of someone important and impactful to them led to greater commitment in Study l. To do so, we created two combined-initials condition. The first asked participants to think of the name of someone in their life who comes to mind, then take the pledge with their own initials plus the initials of that person. We termed this the self& top-of-mind condition. The second asked par ticipants to think of someone in their life that they would like to dedicate their efforts to, and then take the pledge with their own initials plus the initials of that person. We term this the self-dedication condition. By comparing the commitment effects of these two different com bined-initials conditions, and contrasting them to the commitment of those who pledged with just their own initials, we can gain a deeper understanding of the underlying psychological mechanism that led participants to be more committed in a given cause. Third, we aimed to rule out pledging effort as an important potential alternative explana tion; that is, the intervention required that e-pledgers exert more effort than participants using conventional methods for e-pledges. It is conceivable that the amount of effort required to e pledge may have driven the positive effect on commitment. To control for this, we measured how long it took participants to take the pledge. We then examined if there were any PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 2 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges systematic differences across conditions, and, whether that would explain the predicted differ ence in commitment. Fourth, while it was described as optional, everyone in Study 1 signed the pledge. In Study 2, we made it even more explicit that signing the pledge was optional, in an effort to minimize any potential demand effect. We also tracked whether people opted out of signing the pledge across different e-pledge conditions. Methods Participants and design. We aimed to recruit 330 participants from MTurk with an IP address based in the US in exchange for $0.81. At the end of the predetennined data collection period, we yielded 329 valid responses (mean agee= 35.35, SD = 1 1.21; 41 .77% female). We obtained IRB approval from the University of Virginia to conduct this study, with written con sent from the participants. Participants were randomly assigned to one of three conditions: (a) pledging by their own initials (self-initials), (b) pledging with their own initials plus the initials of a person in their life who comes to mind (self & top-of-mind), and (c) pledging with their initials plus the initials of someone to whom they would like to dedicate their efforts to (self dedication initials). E-pledee manipulation. Participants first read a two-paragraph statement about childg hunger in the U.S. For instance, the learned that one in five children in the U.S. lack proper ey nutrition and access to food at some point during the year ( 41] . Participants then read the fol lowing pledge: Any action you take will work toward the same goal-to confront child hunger and give our future generations the nourishment they need to thrive. Let us join efforts and commit to help raise awareness for the cause in the next few weeks. Participants were then given the option of pledging their support in one of three ways: self initials, self & top-of-mind, or self-dedication initials. Those in the "Self-initials" e-pledee condition read the pledge above and were asked to g "Please sign the pledge by entering your initials." Those in the proposed intervention-the "Self & Top-of-Mind" e-pledee condition read g the pledge above and then were asked to "Think of someone who comes to mind, then sign the pledge with both your own initials plus the initials of that person." Those in the proposed intervention-the "Self-Dedication initials" e-pledee conditiong read the pledge above and then were asked to "Think of a person who is extremely important to you and helped you to become the person you are today, then dedicate the pledge that person with both your own initials plus the initials of that person." Commitment behavior. After signing the pledge, participants were given the opportunity to list the concrete steps they would take to help end child hunger. The instruction also stressed that they could list as many or as few steps as they wished. It was also made clear that the par ticipants' compensation for the experiment was not tied to the number of commitment actions they would take. All of the participants provided their demographic information at the end of the survey and were paid the next day. Results Opt-out rate. We began by examining how many participants chose not to pledge as a function of the three e-pledging conditions. Twenty-seven participants opted out of signing PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 3/ 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges the pledge. Results revealed that whereas 1 1 .57% of those in the self-initials (n = 1 4) and 10.67% in the self & top-of-mind conditions chose not to pledge (n = 1 1 ), only 1.90% of the participants in the self-dedication condition declined to pledge (n = 2; ,r(2) = 8.18, p = .01). The significantly lower attrition rate demonstrated that people were not deterred by the self dedication condition. The results remained significant if we were to exclude participants who did not sign the pledge of the data. Commitment behavior. Results from one-way ANOVA revealed that the pledging manipulation affected participants ' commitment to the cause, F(2, 326) = 5.01 , p = .007, partial rf = .03. Post hoc analysis revealed that those in the self-dedication initials condition (M = 3.33, SD = 2.63) generated significantly more concrete actions that they would take than those in the self-initials (M = 2.73, SD = 1 .71 ; p = .03; Cohen's d = .27) or the self & top-of mind (M = 2.43, SD = 1.83; p = .002; Cohen's d = .39) conditions. We also examined whether it would take longer to sign the pledge across the three condi tions, and whether this would partiaJly explain the commitment effect. One-way ANOV A revealed that the pledging manipulation did indeed affect how long it took to complete the pledge, F(2, 326) = 8.13, p < .001 , partial 1/ = .04. Post hoc analysis revealed that those in the self-dedication (M = 21.48 s, SD = 34.00) condition took significantly longer to pledge than those in the self-initials (M = 9.55 s, SD = ll.36;ep < .001) and those in the self & top-of-mind conditions (M = 15.31 s, SD = 13. 78; p = .04). Those in the self & top-of-mind condition also took longer to pledge than the self-initials condition (p = .05). To assess whether the effort it took to take the pledge would explain the commitment effect, we conducted a MANOVA analysis with pledging effort as a covariate. The results revealed that the pledging commitment remained significant and strong (F(2, 325) = 5.17, p = .006, par tial ,J2 = .03), despite including pledging effort as a covariate. Pledging effort did not have a sig nificant effect on commitment, F( l , 325) = .3 4, p = .55. Together, these results suggest that pledging effort alone may not have been sufficient to exp lain why the self-dedication initials would deepen people's commitment to the cause they pledged to support. Discussion Results from Study 2 extend our understanding in three ways. First, the findings provided a deeper understanding of the effectiveness of the proposed intervention. It is worth noting that the effect remained robust even when all of the participants read about the same nonprofit organization and received the same pledge: the only difference was how they signed the pledge. Second, we showed that the self-dedication initials' effect on commitment was independent of the effort exerted to sign the pledge. This is supported both by how long it took to sign the pledge and by including a condition in which participants pledged with their own initials plus the initials of someone who came to mind. Third, results on the differential opting-out rate further demonstrate that the self-dedication pledging method did not systematically deter peo ple from engaging in the pledge. To the extent that the intervention could be practically imple mented in the field, it is reassuring to see that this method may have encouraged, rather than discouraged, potential pledgers to sign the pledge. Study 3 : E-p ledging to commit Study 3 had two main goals: First, we set out to replicate the results from Study 2 in a more controlled setting. Second, we aimed to rule out pledging effort as an important potential alter native explanation in a different way than in Study 2. To do so, we included a condition in which participants were asked to include two computer-generated letters in their pledge. These computer-generated random letters would increase the effort it took for participants to PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 14 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges e-pledge without increasing self-focus, unlike the proposed intervention. Same as in Study 2, we tracked whether participants opted out of signing the pledge across different e-pledge conditions. Methods Participants and design. We aimed to recruit 360 participants from MTurk with IP addresses based in the US in exchange for $0.76. At the end of the predetermined data collec tion period, we received 348 complete responses (mean agee= 35. 1 8, SD = 10.34; 38.79% female). We obtained IRB approval from the University ofVirginia to conduct this study, with written consent from the participants. Participants were randomly assigned to one of three conditions: (1 ) pledging with their own iuitials, (2) pledging with their own initials plus two random letters, and (3) pledging with their initials plus the initials of someone important to them. E-pledee manipulation. Participants first read a two-paragraph description of a U.S. nong profit organization, the Boys and Girls Club of America (BGCA). They read about BGCA's functions and the various ways people could get involved in improving a child's life. Partici pants then read the following pledge: Any action you take will work toward the same goal-to strengthen and empower children in need. Join the Boys and Girls Clubs of America campaign by committing yourself to help raise awareness for the foundation in the next few weeks. Participants were then given the option of pledging their support in one of the three ways: self-initials, self-initials plus initials of someone very important to them (self-dedication ini tials), or self-initials plus two randomly generated letters (self & random letters). The phrasing of the self-initials and the self-dedication initials pledges were the same as in Study 2. Those in the self & random letters condition read "Please sign the pledge by entering your initials plus the two letters shown below." These two letters were randomly generated by the computer program for each participant. For instance, someone with the initials EC would pledge with her initials plus two letters that were randomly generated by the program (i.e., ECNM). Commitment. Participants responded to three questions that captured their commitment to the cause: "I will tell other people of the Boys and Girls Club"; "I feel very committed to mis sions of the Boys and Girls Club"; and "I feel very connected to the children being served by the Boys and Girls Club." We averaged responses to form the commitment scale (ae= .83). All of the participants provided their demographic information at the end of the survey and were paid the next day. Results Oe t-out rate. We began by examining how many participants chose not to pledge as a p function of the three e-pledging conditions. Results revealed that whereas 10.74% of those in the self-initials (n = 13) and 1 3.15% of the self & random letters conditions (n = 15) chose not to pledge, only 3.53% of the participants in the self-dedication initials condition (n = 4) declined to pledge (x2(2) = 6.82, p = .03). The significantly lower attrition rate demonstrated that people were not deterred by the self-dedication initials condition. Results remained signif icant if we were to include excluded participants who did not sign the pledge. Commitment. Results from one-way ANOVA revealed that the pledging manipulation affected participants' commitment to the cause, F(2, 3 45) = 6.48, p = .002, partial rf = .03. Post hoc analysis revealed that those in the self-dedication initials condition (M = 3.47, SD = 1 .06) PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 5 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges were significantly more committed to the cause than those in the self-initials (M = 2.99, SD = 1.03; p = .00 1 ; Cohen's d = .45) or self & random letters (M = 3.09, SD = 1.09; p = .008; Cohen's d = .35) conditions. We also examined whether it would take longer to sign the pledge across the three condi tions, and whether this would partially explain the commitment effect. One-way ANOVA revealed that the pledging manipulation did indeed affect how long it took to complete the pledge, F(2, 3 45) = 6.95, p = .001 , partial 172 = .03. Post hoc analysis revealed that those in the self-dedication initials (M = 1 9.1 8 s, SD = 22.48) and self & random letters (M = 15.76 s, SD = 30.02) conditions took significantly longer to pledge than those in the self-initials condi tion (M = 8.57 s, SD = 10.30; Ps < .01 for both conditions). It did not take significantly longer for people to pledge using self-other or self & random letters (p = .24). To assess whether the effort it took to take the pledge would explain the commitment effect, we conducted a MANO VA analysis with effort as a covariate. The results revealed that the 2pledging commitment remained significant and strong (F(2, 344) = 5.31 , p = .005, partial ,, = .03), even with effort as a covariate. Importantly, effort did not have a significant effect on commitment, F( l, 3 44) = 3.17, p = .07. Together, these results suggest that effort alone may not have been sufficient to explain why the self-dedication initials would deepen people's commit ment to the cause they pledged to support. General d iscussion Comedian Seth Me ers once said, "If you make a Face book page we will 'like' it-it's the least ey we can do. But it 's also the most we can do." While meant to be satirical, his statement largely reflects a reality of social media: Conventional e-pledges and online campaigns are the new and increasingly ubiquitous reality for nonprofit organizations and advocacy groups [ 42- 43], yet their convenience and broad reach may be offset by their ineffectiveness for securing com mitted behaviors. With the goal to improve this de amic, we conducted two pilots and three yn experiments to empirically examine whether a novel and simple e-pledge intervention-self other initials-is more effective for securing commitment behavior than other types of com mon e-pledges. To that point, we refined the pledge by asking participants to explicitly dedi cate their thoughts to that other person, and ruled out effort, novelty, and social interaction mindset as alternative explanations. These results provide a new and enhanced method that may enable organizations to capitalize on the benefits of e-pledging without compromising on social media's mass outreach effect. Notably, the positive link between self-other e-pledging and volunteering commitment over a span of 3 months further enhances confidence in the external and ecological validity of our findings (Study 1). The self-other e-pledge's ability to secure commitment behavior through explicit dedication also highlights the importance of enforcing accountability cues in the e-pledging process (Studies 2 and 3). Together, our findings aimed to explain whether and how a common and increasingly prevalent method for obtaining commitment-e-pledging can effectively be strengthened to secure a wider range of prosocial behaviors that are other wise difficult to motivate [ 44 ]. It is important to point out that regardless of theee-pledge condition participants were assigned to, virtually all of them self-elected to take the pledge in Study 1, and a significant por tion of the participants did so in Studies 2 and 3. Nevertheless, the actual commitment behav ior varied significantly across e-pledging conditions. The discrepancy between signing the pledge and taking action provides further empirical evidence for slacktivism, consistent with survey responses from our Pilot Study lB. This also suggests that participants' recognition of the importance of these social causes and that at least at the time of pledging, they may have PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 6 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges intended to honor their pledges. Therefore, the effectiveness of our intervention was unlikely to be a result of participants' insincerity or an experimental demand effect. Even so, we consis tently showed that small yet carefully planned interventions-such as how people pledged-can greatly impact their committing behaviors (see 45] . Specifically, compared to those who pledged with a "Like " symbol-a widely used method of e-pledging across social media-those who pledged with combined initials volunteered 24.69% longer across the span of 3 months (Study 1). Theoretical contributions and frameworks for future research Our research makes several theoretical contributions and initiates new research directions. First of all, our findine shed light on laypersons perspective of slacktivism, speak to and integs grate theoretical perspectives on social influence and public self-awareness, and provide sup port for a different and more effective method of e-pledging. These results form the basis for fruitful research endeavors. We describe the linkage to previous literature and outline possible future directions below. Results from the pilot studies showed that laypeople acknowledged theee-pledging process as the main contributor to slacktivism, citing lack of accountability and consequence as major factors. Building on these results and drawing from public self-awareness research, we pre dicted that an effective intervention would incorporate an accountability cue in the moment people signed the pledge in the virtual enviromnent. In a sense, our proposed intervention embedded a virtual mirror as people signed the pledge. Results from Studies 1 -3 provide empirical support for our proposed intervention. Namely, the self-other intervention required participants to generate their own standard. Research on public self-awareness theory postu lates that people would then measure their subsequent behavior by the standard they have just created. This heightened sense of self-regard would therefore compel them to follow through with the e-pledge. While we have ruled out several possible alternative explanations, we acknowledge that oth ers may exist as well. One such explanation might be that the intervention evoked a "signifi cant-other transference." Namely, contextual cues not ouly work to activate a sense of self, but different cues can affect which version of the self is activated [ 46, 47]. For instance, cues in a family setting can elicit affective, motivational, and behavioral responses associated with one's "family-oriented self," and this activation of the relational self can spill over to a different con text (46, 48- 49] . Therefore, it is possible that e-pledgers became more committed because it activated the relational self. The positivity and intimacy derived from transference can boost the pledger's likelihood of carrying out the pledged behavior. Future research may shed more light in this direction by investigating this and related alternative explanations. Our results also have point to broader questions for research on conventional e-pledges and slacktivism. Prior research has demonstrated that people often divide a superordinate goal into incremental subgoals. How people interpret these subgoals, in turn, has direct implications for goal pursuit. Specifically, when people interpret their success in fulfilling a subgoal as a possible substitute for the superordinate goal, they are less likely to pursue the superordinate goal [ 50] . Applying that to the current context, one reason that contributes to slacktivism could be due to pledgers feeling that they have already taken steps to support the cause. A more comprehen sive understanding of the interplay between these forces would be a promising direction for future research. Along similar lines, it is possible that people may simply disengage from commitment and responsibility after e-pledging; research on moral licensing suggests this potentially trouble some consequence. Namely, e-pledgers could feel that they have already obtained the feel- PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 7 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges good "moral credits" just by clicking on the pledge, and therefore free themselves to engage in opposite behavior away from the commitment (5 1]. Beyond this possibility, we encourage future research to examine potential negative consequences that may result from these exter nally induced psychological barriers. Practical implications There are several ways our research informs organizations that seek to promote their causes. We have focused on the self-other pledge as one way to awaken and reinforce accountability. Organizations may want to further tailor their pledging prompt by more explicitly asking peo ple to generate names of others who are relevant in that particular domain. For example, St. Jude's campaign could ask e-pledgers to dedicate their efforts to others who had coura geously battled an illness, either personally or in the signer's role as a caregiver. Similarly, the American Red Cross could suggest that the "other" be someone who generously and faithfully provides assistant to people in need in their communities. By tailoring these messages to the essence of their cause, organizations may be even better served by this intervention. Our findings also highlight a more cost-effective method for reaching a broad volunteer base and keeping them engaged. Currently, nonprofit organizations often send reminders of the recipient's prior commitment. However, not only is this practice costly, but reminding people about previous pledges may also backfire. Namely, research has demonstrated that peo ple sometimes react to reminders of past commitment failures by veering away from their orig inal moral standards (52-5 4]. If theee-pledger needs to spend a lot of effort to make up for the failure (i.e., physical distance is far or time commitment is high), then reminding them of the e-pledge they made may inadvertently push them further away and motivate them to disen gage. This puts organizations in a difficult position. Our research suggests that it may be more effective to incorporate an intervention at the moment of e-pledging rather than waiting until the behavior has already occurred [see 22 and 24]. In short, our research provides a more cost efficient method for nonprofit organizations to secure long-term commitment. This research also has broad societal implications. In 201 9, UC Berkeley lost its #2 U.S. News public university ranking because it provided data on the number of alwnni who pledged to donate rather than the actual donation rate (55]. In an ideal world the number of alwnni who pledge to donate would be equal to the nwnber of actual donations. In reality, the discrep ancy had a consequential impact on the university, with unknown future ramifications. By sys tematically investigating the psychological mechanism that can strengthen people's commibnent to the e-pledges they make, and by identifying easily administered e-pledge inter ventions like the one outl ined in this research, we envision this line of work potentially empowers nonprofit organizations to further their causes and secure long-term committed support. Concl usion Our findings have high practical importance for motivating prosocial behaviors-i.e., behav iors that cannot be coerced or stipulated, yet are crucial to a society's well-being. By identifying a simple, low-cost intervention, our findine suggest ways for nonprofit and advocacy groupsgs to benefit from the convenience and efficiency e-pledges offer without compromising actual and long-term commitment. As social media and web-based technologies continue to redefine how people interact with the world, our findings move us one step closer to a broader under standing of how to transform a simple virtual acknowledgement into deeper commitment and, ideally, action. PLOS ONE I https://doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 8 / 21 https://doi.org/10.1371 PLOS ONE Commitment power of e-pledges Supporting information S1 Table. Percentage of volunteers by e-pledge condition per program in Study 1. (DOCX) S2 Table. Poisson regression analysis on hours volunteered. (DOCX) S1 Data. Online supplemental materials: Experimental materials. (DOCX) Acknowledgments The authors thank JKM for his invaluable guidance. Author Contributions Conceptualization: Eileen Y. Chou, Dennis Y. Hsu, Eileen Hernon. Data curation: Eileen Y. Chou. Formal analysis: Eileen Y. Chou, Dennis Y. Hsu. Investigation: Eileen Y. Chou. Methodology: Eileen Y. Chou, Dennis Y. Hsu. Resources: Eileen Y. Chou. Writing - original draft: Eileen Y. Chou. Writing - review & editing: Eileen Y. Chou, Dennis Y. Hsu, Eileen Hemon. References 1 . L uckerson V. These are the 1 0 most popular tweets o f all time. Time Magazine. 201 6 . http://time.com / 4263227/most-popular-tweets/. 2. Yadron, D. Facebook's Sheryl Sandberg: 'Likes' can help stop Isis recruiters. https://www.theguardian. com/technology/201i6/jan/20/facebook-davos-isis-sheryl-sandberg. 201i6. 3. Obar JA, Zube P, L ampe C. Advocacy 2.0: An analysis of how advocacy groups in the United States perceive and use social media as tools for facilitating civic engagement and collective action. Journal of Information Policy. 201 2; 2: 1 -25. 4. Kristofferson K, White K, Peloza J. The nature of slacktivism: How the social observability of an initial a ct of token support affects s ubsequent prosocial action. Journal of Cons umer Research. 2014; 40:1 1 i49-1i1 66. 5. Morozov E. From s lacktivism to activism, Foreig n Policy. 201 9. https://foreignpolicy.com/2009 /09/05/ from-slacktivi sm-to-activi smi/. 6. American Red Cross: Holiday Poll. 201i4. httpJ/www.redcross .org/images/MEDIA_ Custom ProductCatalog/rn40440098_ARC-Holida y-Pol l-201i4 . 7. L acetera N, Macis M, Mele A. Viral altruism? Charitable giving and social contagion in online networks. Sociological Science. 201 5; 3:202-238. 8. UN ICEF Sweden "Like Don't Save Lives" Promotiona l Campaign. 201 3. http://unicef.se/. 9. Prentice-Dunn S, Rogers RW Effects of public and private self-awareness on deindividuation and aggres sion. Journal of Personality and Social Psychology. 1 982; 43(3) :503-5 1i3. 1 0. All three data fi les a re uploaded on the Open Science Database database, and will be made publical ly available upon acceptance of the manuscript for publication. 1 1 . Merriam-Webster's Online Dictionary. 201 9. https://www.merriam-webster.c om/dictionary/pledge. 12. Cialdini RB, Goldstein NJ. Social influence: Compliance and conformity. Ann ual Review of Psych ology. 2004; 55:59 1 -621. https://doi.org/1 0.1 1 46/annurev.psych.55.090902.1 4201 5 PMID: 1 4744228 PLOS ONE I https J/doi.org/10.1371 /journal.pone.0231314 April 29, 2020 1 9 / 21 https://httpsJ/doi.org/10.1371 https://doi.org/10.1146/annurev.psych.55.090902.142015 https://www.merriam-webster.com/dictionary/pledge http://unicef.se https://httpJ/www.redcross.org/images/MEDIA https://foreignpolicy.com/2009/09/05 https://www.theguardian http://time.com P LOS O N E Commitment power of e-pledges 1 3. Grant AM, Mayer DM. Good s oldiers and good actors: Prosocial and impression management motives as interactive predictors of affiliative citizenship behaviors, Journal of Applied Psychology. 2009; 94(4): 900--9 1 2. https://doi.org/1 0.1 037/a001i3770 PMID: 1 9594233 1 4. Schwartz D, Loewenstein G. (201 7). The chill of the moment: Emotions and proenvironmental behavior. Journal of Public Policy & Marketing. 20 1 7; 36: 255---268. 1 5. Chou EV. Paperless and soulless: E-signatures diminish the signer's presence and decrease accep tance. Social Psychological and Personality Science. 201 5; 6(3):343--35 1 . 1 6. Chou EV. What's in a n ame? The toll e-signatures take on individual honesty. Journal of Experimental Socia l Psychology. 201 6; 61 :84-95. 1 7. Kettle KL, Haubl G. The signature effect: Signing influences consumption related behavior by priming self-identity. Journa l of Consumer Research. 201 1i; 38;474-489. 1 8. Ba rnett WS. Effectiveness of early educational intervention. Science. 201 1 ; 333(6045): 9 75---978. https://doi.org/10.1 1 26 /science.1 2 04534 PM ID: 21 852490 19. Solomon MR, Schopler J. Self-consciousness and clothing. Personality and Social Psychology Bulletin. 1i982; 8(3):50�14. 20. van Bommel M, van Prooijen J-W, Elffers H, Van Lange PAM. Be aware to ca re: Public self-awareness leads to a reversal of the bystander effect. Journal of Experimental Social Psychology. 201 2; 48:926- 9 30. 21 . Bryan CJ, Walton GM, Rogers T, Dweck CS. Motivating voter turnout by invoking the self. Proceeding s of the National Academy of Sciences of the United States of America. 201i 1 ; 1 08: 1 2653--1 2656. https://doi.org/1i0.1 073/pnas.1 1 033431 08 PMID: 21 768362 22. Capraro V, Jagfeld G, Klein R, Mui M, van de Pol I. I ncreasing altruistic and cooperative behaviour with simple moral nudges. Scientific Reports 201 9 ; 9, 1 1 880--1i 1 89 1 . https ://doi.org/1 0. 1 038/s41 598-01 9- 48094-4 PMID: 31 i41i71 i06 23. Diener E, Wallbom M. Effects of self-awareness on antinormative behavior. Journal of Research in Per sonality, 1 976; 1 0(1i): 1 07-1i1 1 . 24. Shu ll, Mazar N, Gino F, Ariely D, Bazerman MH. Signing at the beginning makes ethics salient and decreases dishonest self-reports in comparison to signing at the end. PNAS, 201 2; 38: 1 51 9 7-1 5200. 25. Davies M F Mirror and camera self-focusing effects on complexity of private and public aspects of iden- tity. Perceptual and Motor Skills. 2005; 1 00(3): 895---898. 26. Guerin B. Social Facilitation. New York: Cambridge University Press. 1i993. 27. Zajonc RB. Social facilitation. Science. 1 965; 1 49: 1 4�269. 28. Zajonc RB, Sales SM. Social facilitation of dominant and subordinate responses. Journal of Experimen tal Social Psychology. 1 966; 2: 1 60--1 68. 29. Cvetkovich G. Cognitive accommodation, language, and social responsibility. Social Psycholog y. 1 978; 2:149-1 55. 30. Lerner JS, T etlock PE. Accounting for the effects of accountability. Ps ychological Bul letin. 1 999; 1 25 (2) :255---275. https ://doi.org/1 0. 1 037/0033-2909.1 252.255 PMID: 1 0087938 31 . TylerT. The psychology of leg itimacy: A relational perspective on voluntary deference to authorities. Personality and Social Psychological Review. 1 997; 1 :323--245. 32. Duval S, Wicklund RA. Effects of objective self-awareness on attribution of causality. Journal of Experi mental Social Psychology. 1 973; 9( 1 ): 1 7-31 . 33. Gervais WM, Norenzayan A. Like a camera in the s ky? Thin king about God increases public self-aware ness and socially desirable responding, Journal of Experimental Social Psychology. 201 2; 48(1 ): 298- 302 34. Sproul l L, Subramani R, Kiesler S, Walker J, Waters K. When the interface is a face. Human-Computer Interaction. 1 996; 1 1 : 97-1 24 35. Fenig stein A, Scheier MF, Bus s AH. Public and private self-consciousnes s: Assessment and theory. Journal of Consulting and Clinical Psychology. 1 975;43:522-527. 36. Gino F, Desai D. Memory lane and morality: How childhood memories promote prosocial behavior. Journal of Personality and Social Psychology. 201 2; 1 02: 7 43--758. https://doi.org/1 0. 1 037/a0026565 PM ID: 221 81 000 37. Allen JA, Mueller SL The revolving door: A closer look at major factors in volunteers' intention to quit. Journal of Community Psychology. 201 3; 4 1 (2) : 1 3�1i55. 38. Each Qualtrics survey was created prior to the program directors distributing them. Because student volunteers choose their own programs, not every program had the same number of participants. As s uch, while Qualtrics a lgorithm random distributed the conditions within each volunteering program, the PLOS ONE I https://doi.org/10.1371 /journal.pone.0231 314 April 29, 2020 20i/ 21 https://doi.org/10.1371 https://doi.org/1 https://doi.org/10.1037/0033-2909.1252.255 https://doi.org/10.1038/s41598-019 https://doi.org/1 https://doi.org/1 https://doi.org/10.1037/a001 P LOS O N E Commitment power of e-pledges aggregated number of volunteers per condition was not evenly distributed. Because we randomly assigned the pledge conditions within and across each program, we addressed this i ssue by ( 1 ) using nonparametric test that does not assume equal sample size across conditions (2) using nested cluster ing analysis to analyze the data. See online supplemental material S 1 Table for the breakdown of exper imental conditions by program. 39. Muthen LK, Muthen BO. Mplus user's guide. Sixth Edition. Los Angeles, CA: Muthen & Muthen. 2007. 40. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1 999; 6: 1 --55. 41 . Feeding America. Feeding American Impact Report 201 9. Avaialble from https://www.feedingamerica. org/about-us/donor-impact-report. 42. Guengerich G. Online fundraising trends. 201 6. httpsJ/upleaf.com/nonprofit-resources/online fundraising/onl ine-fundraising-trerds. 43. Morrison K. Report: Nonprofits rely heavily on social media to raise awareness. 201 6. httpJ/www. adweek.com/socialtimes/report-nonprofits-rely-heavily-on-social -media-to-raise-awareness/638229. 44. Penner LA, Dovidi o J F, Pil iavin JA, Schroeder DA. Prosocial behavior: Multi level perspectives. Annual Review of Psychology. 2005; 56:365-392. https://doi.org/1 0.1i 1 46/annurev.psych.56.091i 1 03.0701 41 PMID: 1 5709940 45. Baca-Motes K, Brown A, Gneezy A, Keenan EA, Nelson LO. Commitment and behavior change: Evi dence from the field. Journal of Consumer Research. 201 2; 39:1 070-1 084. 46. Anderson SM, Chen S. The relational self: An interpersonal social-cognitive theory. Psychological Review. 2002; 1 09(4):61 9-645. https://doi .org/1 0. 1 037/0033-295x.1 09.4.61 9 PMID: 1 2374322 47. Baldwin MW, Carrell SE, Lopez OF. Priming relationship schemas: My advisor and the Pope are watch ing me from the back of my mind. Journal of Experimental Social Psychology. 1 990; 26:435-454. 48. Anderson SM, Glassman NS. Responding to significant others when they are not there: Effects on inter personal inference, motivation, and affect In Sorrentino R. M. & Higgins E. T. (Eds.), Handbook of moti vation and cognition ( Vol 3, pp. 262-321i) . New York: Guilford Press. 1 996. 49. Berk MS, Andersen SM. The impact of past relationships on interpersonal behavior: Behavioral confir mation in the social-cognitive process of transference. Journal of Personality and Social Psychology. 2000; 79:546--562. PM ID: 1 1 045738 50. F ishbach A, Dhar R, Zhang Y. Subgoals as substitutes or complements: The role of goal accessibility. Journal of Personality and Social Psychology. 2006; 91 :232-242. httpsJ/doi.org/1 0. 1 037/0022-351 4. 9 1 2.232 PM ID: 1 6881 761 51 . Merritt AC, Effron DA, Monin B. Moral self-licensing: When being good frees us to be bad. Social and Personality Psychology Compass. 201 0; 4(5) :344-357. 52. Duval TS. & Duval VH. Level of perceived coping ability and attribution for negative events. Journal of Socia l and Oinical Psychology. 1 987; 5:452-468. 53. Duval TS, Silvia PJ. Self-awareness, probability of improvement, and the self-serving bias. Journal of Personality and Social Psychology. 2002; 82( 1 ):49-61 . https://doi.org/1 0. 1 037//0022-351 4.82.1 .49 PMID: 1 1 81 1 633 54. Si lvia PJ, Duval TS. Objective self-awareness theory: Recent progress and enduring problems. Person ality and Social Psychology Review. 2001i; 5(3):230-241 . 55. U.S. News. Updates to 5 schools' 201 9 Best Colleges Rankings Data. 201 9 https://www.usnews.com/ education/blogs/college-rankings-blog/articles/201 i9-07-25/updates-to-5-schools-20i1 9-best-col leges rankings-data. PLOS ONE I httpsJ/doi .org/1 0. 1 371 /journal .pone.0231 31 4 April 29, 2020 21 / 21 https://httpsJ/doi.org/10.1371 https://www.usnews.com https://doi.org/10.1037//0022-3514.82.1.49 https://httpsJ/doi.org/10.1037/0022-3514 https://doi.org/10.1037/0033-295x.109.4.619 https://doi.org/10.1 https://adweek.com/socialtimes/report-nonprofits-rely-heavily-on-social-media-to-raise-awareness/638229 https://httpsJ/upleaf.com/nonprofit-resources/online https://www.feedingamerica Copyright of PLoS ONE is the property of Public Library of Science and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder' s express w1itten permission . However, users may print, download, or email articles for individual use . Structure Bookmarks PLOS ONE PLOS ONE Check for updates OPEN ACCESS b Citation: Chou EY, Hsu DY, Hemon E (2020) From slacktivism to activism: Improving tt"e commitment power of e-pledges for prosocial causes. PLoS ONE 15(4): e0231314. . org/10.1371Łournal.pone.0231314 https://doi Editor: Valerio Capraro, Middlesex University, UNITED KINGDOM Received: October 20, 2019 Accepted: March 20, 2020 Published: ,llpril 29, 2020 Copyright: © 2020 Chou et al. This is an open access article distributed under tt"e terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided tt"e original author and source are credited. Data Availability Statement Al I data files are available from the Open Science Database database at /. https://osf.io/zdfr3/files Funding: Tue authors received no specific funding for this work Competing interests: Tue authors have declared that no competing interests exist RESEARCH ARTICLE From slacktivism to activism: Improving the commitment power of e-pledges for prosocial causes From slacktivism to activism: Improving the commitment power of e-pledges for prosocial causes Eileen Y. Chou *, Dennis Y. Hsu, Eileen Hernon Eileen Y. Chou *, Dennis Y. Hsu, Eileen Hernon 1 2 3 1 Batten School of Leadership and Public Policy, University of Virginia, Charlottesville, Virginia, United States of America, 2 Faculty of Business and Economics, The University of Hong Kong, Hong Kong, Republic of China, 3 University of Virginia, Char1ottesville, Virginia, United States of America * eileen.chou@virginia.edu Abstract Abstract Prosocial organizations increasingly rely on e-pledges to promote their causes and secure commitment. Yet their effectiveness is controversial. Epitomized by UNICEF's "Likes Don't Save Lives" campaign, the threat of slacktivism has led some or ganizations to forsake social media as a potential platform for garnering commitment. We proposed and investigated a novel e-pledging method that may enable organizations to capitalize on the benefits of epledging without compromising on its mass outreach potential. ruled out effort, novelty, and social interaction mindset as alternative explanations for why the intervention may be effective. As technological innovations continue to redefine how people interact with the world, this research sheds Ii ght on a promising method for transforming a simple virtual acknowledgment into deeper commitment-and, ideally, to action. Introduction Introduction In 2014, Indonesian political analyst Denny Januar Ali amassed more than 2.5 million retweets that pledged to support Indonesian presidential candidate Joko "Jokowi" Widodo and to replace discrimination with love [l]. In 2016, Facebook COO Sheryl Sandberg rallied the platform's 1.6 billion users to redirect their 6-billion-times-a-day habit of clicking "Like" by supporting an online campaign committed to defeating ISIS recruiters [2]. The power of social media as an efficient and massive information dispe PLOS ONE I /journal.pone.0231314 April 29, 2020 1 / 21 https://doi.org/10.1371 Despite the rapid growth of social media campaigns, the notion of slacktivism-defined as "feel-good online activism with little meaningful social or political impact" (4--5]-challenges the value of these efforts. Slacktivism highlights the ease with which people can "click it and forget it." Based on this assumption, some even argue that engaging in online campaies may gn lead people to believe that they have already contributed to the cause, without doing anything meaningful. Indeed, a recent poll revealed that only 3% of active social media users cited online campaigns as a key motivator in their donation decisions ( 6]. In a field study in collaboration with Heifer International, Lacetera, Macis, and Mele showed that an online campaign engaged almost 6.4 million online users (in the form ofe"likes" or "shares"), yet only 30 made an actual donation [7]. The concern that sl Rather than focusing on an impact evaluation of online campaigns, we contend that the more pressing and realistic issue is how we can improve the existing platform to secure greater commitment, thereby allowing organizations to capitalize on the platform's power. With this goal in mind, we focused on e-pledges-one of the most common methods used by online campaigns-and shed light on the prevalent phenomenon of slacktivism. We aimed to answer two questions: (a) why are e-pledgers less motivated to follow thr (b) what type of intervention might increase e-pledges' commitment power? We conducted five studies to tackle these questions, with the results of each study informing the next. Pilot Study lA demonstrated the presence and prevalence of slacktivism by directly comparing the effectiveness of conventional e-pledges with their traditional counterparts. Pilot Study lB provided a layperson's perspective as to why conventional e-pledges are ineffective. These insights then served as the foundation for a novel e-pledging method that seeks to strengthen pledgers' commitment to prosocial This research offers insights with potential practical and theoretical advancements. First, we empirically investigated a promising solution to a pervasive problem: the increasing, yet ineffective, reliance onee-pledges as a way to secure prosocial commitment. By revealing laypeople's perspectives on why e-pledges might contribute to slacktivism, the aggregated trend in our data played a crucial role in developing an ecologically valid intervention. Second, we ruled out several closely related alternative PLOS ONE I /journal.pone.0231314 April 29, 2020 2/ 21 https://doi.org/10.1371 Understanding what e-pledging is and the source of its ineffectiveness Understanding what e-pledging is and the source of its ineffectiveness A pledge is a person's solemn promise to commit to a cause that he or she deems worthy ( 11]. In essence, pledges serve as a means of social control.Yet unlike formal contracts or sanctions, failure to honor a pledge has minimal punitive consequences. Therefore, the commitment power of pledges often relies on ( a) the normative expectation that people will be held accountable after making the pledge ( 12] and (b) self-investment and identification with the cause (4, 13, 14]. Together, these critical forces People traditionally confirm their pledge by signing their name on a piece of paper or publicly announcing their commitment to the cause. An e-pledge, which we define as a virtual promise to honor a commitment, serves the same objective function as traditional pledges. The only substantive difference is the method by which people pledge: Instead of signing their name by hand on paper, would-be-pledgers indicate their commitment electronically, either on a social media platform (e.g., Facebook, Twitter) or Redcross.org However, as a plethora of slacktivism anecdotes suggest, e-pledges may not be as effective in their ability to secure commitment as their traditional counterparts. We posit that while epledges and traditional pledges serve the same objective functions, they diverge in the psychological weight they may evoke in the pledger. Indeed, past research posits two potential drivers of this ineffectiveness: (a) The online pledge in general is perceived to be less trustworthy or mobilizing (regardless of how it was signed) or (b) the method used to pledge (e.g., "Like" clicking, name initials typing, etc.) dilutes the commitment effect. We expand on these two drivers below. On the one hand, it could be that people consider online pledges to be less persuasive or trustworthy. In line with this notion, prior research has found that people perceive electronically transmitted docwnents to be less trustworthy (15]. Therefore, it could be that people are willing to pledge their support, but question whether the campaign itself warrants further involvement. As a result, they stop short of actual action. On the other hand, it could be that the pledging process itself is less effective for motivating people to take the desirable action. In recent e-signature research, compared with participants who signed by hand, e-signers were less likely to obey the terms of the contract they signed (16]. Similarly, conswners who typed their names (versus signing by hand) were less likely to make a purchase afterward [ 17]. These findinsuggest that conventional methods of e-pledging may be the reason for its ineffectiveness, independent of the cause or campaign being promoted. gs To address slacktivism within the e-pledge domain, we first need to demonstrate empirically that it is indeed an issue and then try to understand the underlying source and mechanism of the problem. To this end we conducted two pilot studies that served two purposes. First, Pilot Study lA and 1B s provided an empirical assessment oflaypeople's engagement in and perception of slacktivism-specifically, its presence, prevalence, and severity. Second, we presented participants in Pilot Study 1B with the two po Pilot Study 1A: Are conventional e-pledges effective? Pilot Study 1A: Are conventional e-pledges effective? Pilot Study lA set out to demonstrate whether conventional e-pledges are indeed less effective than the traditional way of pledging in a pro social domain that supports scientific advancement. To do so, we investigated whether three different forms of pledging could influence PLOS ONE I /journal.pone.0231314 April 29, 2020 3/ 21 https://doi.org/10.1371 subsequent commitment behavior in variant degrees. We included two common forms of epledging-a checked box (as often seen on platforms such as Facebook and Twitter) and typed full name ( as often seen on platforms such their impact to the traditional way of pledging with a handwritten signature. We then measured their subsequent commitment behavior to the clause. as Change.gov and Redcross.org)-and compared Methods Methods Participants and procedure. Ninety-three undergraduate students (mean agee= 20.38, SD = 3.63; 51 % female) participated in the study in exchange for a snack and the chance to win a $50 bonus. We obtained IRB approval from the University ofVirginia to conduct this study, with written consent from the participants. Participants completed a two-stage study on a laptop preloaded with the study programed in Qualtrics. In the first stage, participants were informed of a cover story that the study was interested in decisions made under time pressure. Thethen played three rounds of"Where's ey Waldo?." Each round presented participants with a large image and asked them to locate the figurine "Waldo." Participants had up to 30 seconds per round to find Waldo. We included this first stage and a cover story to minimize potential demand effect of participants succumbing to how they think the experimenter would want them to behave. Pledee-sieninemanipulation. Uen completing the Where's Waldo task, participants ggg po then learned that they would have the option to sign a pledge to support evidence-based behavioral research at their institution. The pledge read as follows: Please read the following petition regarding behavioral scientific research, and sign if you agree. Otherwise, leave this blank and move to the next page. To create a better tomorrow, we must start today and draft evidence-based policies. Investing time.focus, and money in understanding the social and psychological implications of public and private policies is crucial in their eventual effectiveness. Join us at the Behavioral and Science Policy Association (behavioralpolicy.org) in helping to Join us at the Behavioral and Science Policy Association (behavioralpolicy.org) in helping to develop a rigorous, comprehensive, and evidence-based behavioral research. No matter what you do, let your actions be seen. Participants were then randomly assigned to sign the pledge in one of three ways. Participants were asked to either "Take the pledge by click on the Like button below" (Like condition), "Take the pledge by typing your initials below" (initials condition), or "Take the pledge by signing your name with the cursor in the space below" (traditional pledge condition). Everyone read the exact same pledge. The only difference was how they signed the pledge. Commitment behavior. After the pledge, participants were told that the experimenters would like to gain insight into how to improve the participant-recrnitment process at their institution. Their responses would allow the experimenters to enhance behavioral research. Participants were then given the opertunity to provide as many or as few ways of improving po how participants were being recruited. In essence, this task provided participants with an opportunity to support evidence-based behavioral research-which adhered to the pledge that they had signed. We then measured the nwnber of suggestions each participant provided, which served as the behavioral measure of commitment. The instruction made it clear that the participants were under no obligation to either sign the pledge or provide any suggestions to the experimenters. Regardless of their behaviors and PLOS ONE I /journal.pone.0231314 April 29, 2020 4/ 21 https://doi.org/10.1371 Table I. Poisson regression analysis on commitment behavior, Pilot Study IA. Table I. Poisson regression analysis on commitment behavior, Pilot Study IA. Table I. Poisson regression analysis on commitment behavior, Pilot Study IA. Variable Pledge Condition Variable Pledge Condition B l SE 9S% CI TR Check Box -.86 "' "'"' .23 [1.31 , -.40) - TR Type Initials H -. 64 .24 [-1.12, -.16) TR Hand-signed' - - - Intercept Intercept 3.64,t,,t,,t .12 [.35, .85) 'p<.05; "p<.01; '"p<.001 'Hand-signed condition served as the reference group .1371/joumal.pone.0231 314.t001 .1371/joumal.pone.0231 314.t001 https://doi.org/1 0 the pledging condition to which they were randomly assigned, participants then provided their demographic infonnation and were thanked and excused. Results Results All participants signed the pledge to help advance behavioral research. However, more than half of the participants (53.8%) did not provide suggestions. This prompted us to conduct a Poisson regression analysis to gain a more detailed understanding of how much people actually helped. We used a Poisson regression analysis because it allows us to preserve the meaningfulness of the zeros in our data and because the dependent variable is a count variable. Table l presents the full results of the regression analysis along. As predicted, the pledge-sign ing manipulation had a significant impact on commitment behavior .i(2) = 16.59, p < .001. Parameter estimation with the signed by hand condition as the reference group indicated that both the checked box (B = -.86, SE = .23, x= 13. 92, p < .001e) and the typed initials condition (B = -.64, SEe= .24, x= 6.94, p = .008) differed significantly from the handwritten condition. Pairwise comparisons further revealed that those who signed the pledge by hand volunteered more suggestions (M = 1.82, SD=2.27) than thos 2 2 Discussion Discussion Results from Pilot Study lA reveal that although everyone received the same text in the pledge, how they signed it significantly affected whether and how much they heled to further the pe cause. In short, common forms of e-pledging are indeed less effective at securing commitment than the traditional form of signing pledges by hand. This discrepancy further highlights the importance of bolstering and solidifying e-pledges' effectiveness as a commitment tool. Pilot Study 1B set out to further understand why this effect occurs. Pilot Study 1 B: Why conventional e-pledges are ineffective Pilot Study 1 B: Why conventional e-pledges are ineffective We conducted an online survey study using the Amazon Mechanical Turk platform (MTurk). The objective of this study is to understand laypeople's perception of the shortcomings of conventional e-pledges. As these online participants came from the population frequently targeted PLOS ONE I /journal.pone.0231314 April 29, 2020 5/ 21 https://doi.org/10.1371 for online campaigns and e-pledges, we reason that their responses would be a valuable and valid source of information. Methods Methods Particieants and procedure. Three hundred and one participants recruited from the p MTurk online platform completed the survey online (44% female; mean agee= 35.10, SDe= 9.91). We obtained IRB approval from the University of Virginia to conduct this study, with written consent from the participants. After entering their MTurk ID, participants read the definition of slacktivism, which was defined as "a phenomenon in which people pledge support for a cause on social media without following up with actual behaviors that contribute to the cause." They then responded to two blocks of survey questions in sequence to measure their perceptions of (1) the prevalence and severit ey were not obligated to provide any responses to this question; in our final data, ouly 96 participants provided any comments, half of which were not related to slacktivism. Therefore, we did not submit the data to any systematic qualitative analysis. Measures of the prevalence and severity of slacktivism. After reading the definition of slacktivism, participants were told that the experimenters would like to learn more about their perception of the prevalence and severity of slacktivism. Participants then responded to three questions: whether they had personally engaged in slacktivism in the previous 6 months (1 = definitely not to 5 = definitely yes); how prevalent a problem they think slacktivism is (1 = not at all to 5 = very prevalent); and how seri Measures of whee-pledees maefail. After completing the block on slacktivism preva y gy lence and severity, participants were then asked to reflect on reasons why e-pledges may be ineffective. We grounded these reasons in past research, which highlight two competing forces that may contribute to slacktivism. Participants first reviewed five reasons that may have contributed to e-pledging's failure to secure commitment and indicated how much thethought ey each contributed to slacktivism (using a 5-point scale: l = not at all to 5 = very much so). Two of the reasons focused on the pledger's reaction to the pledge ("People feel less accountable to the pledges and petitions they sign online'; "People feel less guilty for breaking online pledges and petitions"), and three that concerned the pledge itself ("There are too many online pledges and petitions around"; "People often question the veracity of the on line pledges and petitions"; and "Online pledges and pe Results Results Severity of the issue. We submitted participants' ratings of their personal involvement in slacktivism, the prevalence of slacktivism, and the severity of slacktivism to a series of onesample t-tests. Results revealed that all three were significantly more than the mideint of the po response scale (t(300) > 5.86, p < .001). Most notably, 58.8% of participants indicated that they most likely or had definitely committed slacktivism in the previous 6 months, and 80.1 % indicated that it is a prevalent or very prevalent issue. In total, 46.9% of participants agreed that slacktivism is a serious to very serious issue for society. Whee-pledees do not work. Because participants provided individual ratings for each of y g the five reasons, we employed a repeated measures AN OVA on the rating data to test equality of means of the five reasons. As all of the participants rated the same five statements, a repeated PLOS ONE I /journal.pone.0231314 April 29, 2020 6/ 21 https://doi.org/10.1371 4.S 3.S 3 2.S 1.5 Less accoun1ability Less accoun1ability Less accoun1ability Fed ing less guilty Too many e-pledges Too man)' fake Xot emotion.illy TR pledges appealing Fig I. Ratings of factors that contributed to slacktivism, Pilot Study 1 B. Fig I. Ratings of factors that contributed to slacktivism, Pilot Study 1 B. .g001 .g001 https://doi.org/10.1371/joumal.pone.0231 314 measures ANO VA would allow us to investigate whether there were overall systematic differences across how people responded to those statements. Results revealed a significant difference in lay perception of what contributes to slacktivism (F( 4, 1200) = 60.7 4, p < .001, ,J= .16; Fig 1). The overall significant effect further granted us the ability to assess pairwise differences between the statements. Subsequent pairwise comparisons showed that "less-accountability" (M = 4.16, SD = .96) and "less guilt" 2 We then examined forced ranking data. A Friedman nonparametric test demonstrated that there were overall differences in the rankings of why people thought e-pledgers shirk (x2 ( 4) = 327.87, p < .001). Participants' rankings identified "lacking accountability" (M = 1.95) and "feeling less guilty" (M = 2.35) as the two highest rated reasons for why e-pledgers shirk. The remaining reasons, in order of rank, were "too many pledges available" (M = 3.16), "online pledges are often fake" (M = 3.72), and "not as e Discussion Discussion This exploratory study yielded several insights. First, it provided empirical support for the prevalence of slacktivism. Second, not only do people recognize the severity of the issue, they also acknowledge that they themselves have been culprits. This discrepancy between willingness to pledge and subsequent commitment behavior further accentuates the importance of improving e-pledges' effectiveness as a commitment tool. More importantly, the pilot study provided directions for how we could improve conventional e-pledges. Our results identified (1) weak accountability and (2) lack of emotional investment as major contributors to slacktivism; both reasons rest more on the method of e- PLOS ONE I /journal.pone.0231314 April 29, 2020 7 / 21 https://doi.org/10.1371 pledging itself than on the pledges in general. These findings suggest that to strengthen epledges and curb s]acktivism, we would need to think about ways to boost accountability and heighten the sense of emotional consequences of the pledge. These findings enabled us to design potential interventions that may be more effective. Study 1: E-pledging to volunteer in the field setting Study 1: E-pledging to volunteer in the field setting Results from our two pilot studies served as a springboard for potential ways to improve epledges' effectiveness. When developing interventions, it is essential to keep any constraints and boundaries in mind. Furthermore, the intervention should preserve the current system's strengths while improving it. We aimed to preserve two parameters. First, two strengths of epledges are their efficiency and ease of administering (18]. Any modification, therefore, should be neither cumbersome for p]edgers nor diffic With these caveats in mind, we focused our attention on how to boost psychological accountability and a sense of nonnative pressure when people sign an e-p]edge. We reasoned that one way to improve an e-p]edge's commitment power is to raise the sense of public selfawareness as people e-p]edge. Public self-awareness is the state in which people focus on the impressions themake on others, based on their behavior and appearance [9]. In such a state, ey people observe their own behaviors from the vantage point of real others and seek social approval [ 19-21]. In turn, public self-awareness induces greater accountability and leads people to act in line with perceived social norms and personal standards. Notably for our purpose, public self-awareness can be activated using different accountability cues (21-24]. For instance, a classic example of an accountability cue is the mirror manipulation [23]: placing an individual in front of a mirror so that their image was visible throughout the experiment served to heighten participants' self-awareness. Likewise, the presence of a camera can induce public self-awareness (25]. For instance, being in front of a webcam with other people who, potentially, are watc We propose that asking people to pledge with their own name plus the names of someone important to them would be a two-pronged approach to creating such an accountability cue. First, this self-other e-p]edge could heighten psychological accountability by introducing a virtual audience as people e-p]edge (20, 26-28]. The familiar nature of the other person's name would also lead e-p]edgers to perceive the "virtual audience" as more legitimate: People respond better when they envisage having to explain their Second, this self-other e-p]edge could emphasize core personal standards and normative expectations, because the introspective process by which people contemplate who is important to them could draw attention to the self [32, 35]. When people are more focused on the self, they become more attuned to important personal standards and fee] pressured to act in line with their core identity (36]. Because people have an inherent desire to see themselves as consistent across behaviors, this activated sense of sel PLOS ONE I /journal.pone.0231314 April 29, 2020 8i/ 21 https://doi.org/10.1371 Taken together, we set out to test a novel e-pledging method aimed at increasing public self-awareness: self-other pledging. This method requires that e-pledgers pledge not only with their own name but also with the name of someone important to them and whom therespect. ey It is also important to note that although pledgers are using two names to pledge their support, they are making the pledge on their own behalf and not the other person's. In effect, e-pledgers are dedicating their effort to the other person they included in the pledge. To test the effectiveness of our proposed intervention in a context with high ecological validity, we conducted a field experiment in collaboration with a university-affiliated student volunteer center that sponsors various programs in the community. One of the major challenges most nonprofit organizations face nowadays is high volunteer attrition (37]. Because these organizations greatly depend on volunteers to function, securing volunteer commibnent is critical. Our goal was to investigate whether our pr At the beginning of the fall semester, students who had registered for one of the center's volunteering programs were asked to take an e-pledge of commibnent. They were randomly assigned to one of three methods of e-pledging: two conventional methods (signing with their name or by clicking on "Like") and our proposed method (combined self-other names). We predicted that at the end of the semester (three months later), those who pledged with combined names would volunteer significantly more hours than thos Methods Methods Participants recruitment process. At the time of data collection, the center sponsored 27 volunteering programs around the community, such as youth mentoring, tax services, housing improvement, and medical services. Each of these programs was headed by a program director, who oversaw the management of their volunteers throughout the duration of the semester. We worked directly with program directors at the volunteer center because of their proximity to the organization process. Eighteen program directors a (38]. We were also given access to the timecards at the end of the semester, along with basic demographic information of the volunteers (i.e., age, year at school, tenure with the program, and gender). We were not pennitted to contact the volunteers during the course of the semester or after the semester had ended. We obtained IRB approval from the University of Virginia to conduct this study, with written consent from the participants. At the end of the semester, eight volunteers terminated their involvements with the volunteering programs all together (four from the Like condition, one from the initials condition, and three from the combined self-other initials condition; see Online Supplemental Material Sl Table for more detail). Therefore, our final data set comprised of 132 volunteers across 18 volunteering programs (77% female; mean agee= 19.51, SD = 1.1e4; tenure with programe= 1.76 semesters, SD = .87). We did not have access to t PLOS ONE I /journal.pone.0231314 April 29, 2020 9/ 21 https://doi.org/10.1371 and volunteering programs were not correlated. The Pearson Correlation Coefficient between the volunteering program and experimental condition is 0.014 (p-value = .865). This result suggests that both variables are not correlated with each other. Pledee manipulation. The program directors informed us that it is customary for all vol g unteers to complete an online survey at the beginning of the semester that asked volunteers to provide basic demographic information, such as age and gender, and their proposed schedule. This provided us with the opportunity to institute an e-pledge. At the end of the survey that they would normally complete, volunteers were asked to pledge that they would honor their commitment throughout the 3-month period. All of the participants read the same content of the pledge: "On my honor, I pledge to uphold the values of a good volunteer. I will consistently attend my volunteer shifts on time, I will do my best to work enthusiastically and maintain a positive attitude, and I am committed to seeing my work through until the end of the semester." attitude, and I am committed to seeing my work through until the end of the semester." Although the pledge's content was identical, we randomly assigned volunteers to one of three e-pledge formats: two conventional methods ( clicking on "Like" or typing their initials) and the combined self-other initials e-pledge. We randomized participants to different epledging conditions regardless of the program. Following prior research, we asked participants to use initials instead of their full names in order to protect their identity and maintain anonymity [see 16]. Those in the "Like" e-pledee condition read the above statement and were asked to "Please g sign the pledge by clicking on the "Like" symbol". sign the pledge by clicking on the "Like" symbol". Those in the "Self-initials" e-pledee condition read the pledge above and were asked to g "Please sign the pledge by entering your initials." "Please sign the pledge by entering your initials." Those in the proposed intervention-the "Combined self-other initials" e-pledee condi g tion read the pledge above and then were asked to "Think of a person who is extremely important to you and helped you to become the person you are today, then sign the pledge with both your own initials plus the initials of that person." tion read the pledge above and then were asked to "Think of a person who is extremely important to you and helped you to become the person you are today, then sign the pledge with both your own initials plus the initials of that person." All of the volunteers opted to take the pledge. The extent of our involvement with the programs or the volunteers ended after that initial online survey. Volunteers went about their routine and did not receive any reminders or follow-up surveys from us. At the end of the semester, we obtained the logs of volunteer hours from the program directors; the logs tracked volunteers' actual work hours across the semester, which served as our dependent variable. Control variables. We consulted with program directors to understand the demographic variables that have systematically affected volunteering hours in the past. Based on that information, we controlled for each volunteer's age, sex, and any previous volunteer experience with the nonprofit organization. These information were part of the mandatory survey that volunteers filled out at the binning of the semester. eg Analytical approach. We analeed the data in two ways. We first used path modeling to yz analyze whether thee-pledge manipulation had a significant impact on volunteering hours. Because (a) different programs may require different levels of involvement, with some being more time-intensive than others (e.g., daily after-school tutoring versus cleaning up a river bank on weekends), and (b) theee-pledging conditions were randomly assigned to volunteers regardless of the program, volunteering hours in our experiment were not independent across programs. Therefore, we modeled the programs' nonindepe We also conducted Poisson regression analyses to further understand whether those who pledged with combined iuitials would volunteer significantly more hours than those who PLOS ONE I /journal.pone.0231314 April 29, 2020 10/ 21 https://doi.org/10.1371 0.12 (.04) Pledging Condition \ohmteering Hours Pledging Condition \ohmteering Hours 0.16 (.12) 0.18 (.0 )* 0.02 (.08) 0.49 I'------------ I'------------ - olunteer Sex (.10)* • -0.13 (.06) olunteer Age p<0.001; *p<0.01; p<0.05 R.,v[SEA = 0.04; CFI: 0.96; Chi-square(6)= .44; p = 0.28 Fig 2. Path model analysis, Study 1. /joumal.pone.0231 314.g002 /joumal.pone.0231 314.g002 https://doi.org/10.1 371 pledged using conventional methods (i.e., "Like" clicking and name initials typing), above and beyond the control variables. We adopted a Poisson regression analysis because the dependent variable is a count variable. Results Results We conducted a single-indicator nested modeling analysis clustered on programs, controlling for volunteer's age, sex, and their previous volunteer experience with the nonprofit organization. Fig 2 presents the path coefficient. We first examine the overall model to determine whether it confirmed that the pledging condition had a significant effect on hours volunteered. This analysis revealed that model fit indices satisfied the goodness of fit standard [CFI = 1.00, RMSEA = 0.00, x(6) = 5.53, p = .47; Fig 2 2 2 Poisson regression analysis shed more light on the self-other initials effect (see Online Supplemental Material S2 Table). Model l included only the experimental conditions. Model 2 showed that the effect strengthen after including program as a control variable. Model 3 included all remaining control variables-volunteers' gender, age, and tenure. Results showed that, after controlling for the control variables, volunteers in both "Like" and "self-initials" conditions volunteered significantly fewer hours t PLOS ONE I .1371 /journal.pone.0231314 April 29, 2020 11 / 21 https://doi.org/1 0 worked fewer hours than those e-pledged with "self-other" initials (B = -54, SE = .27, p = .05). This indicates the robustness of the self-other initials effect. Discussion Discussion Study 1 provides the first validation of our self-other initials intervention. By asking volunteers to pledge with both their own initials and another important other's, they volunteered 24.69% longer over a span of 3 months than those who pledged with the "Like" button, and 11. 94% longer than those who pledged with self-initials only. This field study's longitudinal design also boosts the external validity of the effect. These findings, while supportive of our predictions with strong external validity, are less controlled due to the nature of field studies. For instance, the number of volunteers who participated in this study was outside of our control. Additionally, as a compromise we had to make to gain entry to the organization, we agreed to make the experimental design as nonintrusive as possible. Hence, the only volunteering inform More importantly, results from Study 1 reveal that although everyone received the same pledge, how they signed affected whether and how much they helped in a tangible manner to further the cause. Yet, results from Study 1 did not speak to why pledging with the name of someone important to them made a significant difference in behavior. On the one hand, participants might have instinctively dedicated their effort to the person thenamed in their ey pledge. On the other hand, simply recalling a person might have been sufficient to improve commitment. Study 2 set out to refine the intervention by explicitly asking participants to dedicate their efforts to that person who is very important to them. Study 2: E-pledging to act Study 2: E-pledging to act Study 2 strived to achieve four goals: First, we aimed to gain greater control and reduce the noise that is inherent in field experiments. To do so, we created an experimental design in which participants had the opportunity toee-pledge their commitment to the same cause, via different e-pledging methods. We viewed this as a more conservative test of theee-pledge intervention, because people were all pledging to the same cause and they were randomly assigned to the various e-pledging methods. Second, we set out to better understand why pledging with people's own initials plus that of someone important and impactful to them led to greater commitment in Study l. To do so, we created two combined-initials condition. The first asked participants to think of the name of someone in their life who comes to mind, then take the pledge with their own initials plus the initials of that person. We termed this the self& top-of-mind condition. The second asked participants to think of someone in their life t Third, we aimed to rule out pledging effort as an important potential alternative explanation; that is, the intervention required that e-pledgers exert more effort than participants using conventional methods for e-pledges. It is conceivable that the amount of effort required to epledge may have driven the positive effect on commitment. To control for this, we measured how long it took participants to take the pledge. We then examined if there were any PLOS ONE I /journal.pone.0231314 April 29, 2020 12/ 21 https://doi.org/10.1371 systematic differences across conditions, and, whether that would explain the predicted difference in commitment. Fourth, while it was described as optional, everyone in Study 1 signed the pledge. In Study 2, we made it even more explicit that signing the pledge was optional, in an effort to minimize any potential demand effect. We also tracked whether people opted out of signing the pledge across different e-pledge conditions. Methods Methods Participants and design. We aimed to recruit 330 participants from MTurk with an IP address based in the US in exchange for $0.81. At the end of the predetennined data collection period, we yielded 329 valid responses (mean agee= 35.35, SD = 11.21; 41.77% female). We obtained IRB approval from the University of Virginia to conduct this study, with written consent from the participants. Participants were randomly assigned to one of three conditions: (a) pledging by their own initials (self-initials), (b) pl E-pledee manipulation. Participants first read a two-paragraph statement about child g hunger in the U.S. For instance, thelearned that one in five children in the U.S. lack proper ey nutrition and access to food at some point during the year (41]. Participants then read the following pledge: Any action you take will work toward the same goal-to confront child hunger and give our future generations the nourishment they need to thrive. Let us join efforts and commit to help raise awareness for the cause in the next few weeks. Participants were then given the option of pledging their support in one of three ways: selfinitials, self & top-of-mind, or self-dedication initials. Those in the "Self-initials" e-pledee condition read the pledge above and were asked to g "Please sign the pledge by entering your initials." Those in the proposed intervention-the "Self &Top-of-Mind" e-pledee condition read g the pledge above and then were asked to "Think of someone who comes to mind, then sign the pledge with both your own initials plus the initials ofthat person." Those in the proposed intervention-the "Self-Dedication initials" e-pledee condition g read the pledge above and then were asked to "Think of a person who is extremely important to you and helped you to become the person you are today, then dedicate the pledge that person with both your own initials plus the initials of that person." Commitment behavior. After signing the pledge, participants were given the opportunity to list the concrete steps they would take to help end child hunger. The instruction also stressed that they could list as many or as few steps as they wished. It was also made clear that the participants' compensation for the experiment was not tied to the number of commitment actions they would take. All of the participants provided their demographic information at the end of the survey and were paid the next day. Results Results Opt-out rate. We began by examining how many participants chose not to pledge as a function of the three e-pledging conditions. Twenty-seven participants opted out of signing PLOS ONE I /journal.pone.0231314 April 29, 2020 13/ 21 https://doi.org/10.1371 the pledge. Results revealed that whereas 11.57% of those in the self-initials (n = 14) and 10.67% in the self & top-of-mind conditions chose not to pledge (n = 11), only 1.90% of the participants in the self-dedication condition declined to pledge (n = 2; ,r(2) = 8.18, p = .01). The significantly lower attrition rate demonstrated that people were not deterred by the selfdedication condition. The results remained significant if we were to exclude participants who did not sign the pledge of the data. Commitment behavior. Results from one-way ANOVA revealed that the pledging manipulation affected participants' commitment to the cause, F(2, 326) = 5.01, p = .007, partial rf = .03. Post hoc analis revealed that those in the self-dedication initials condition ys (M = 3.33, SD = 2.63) generated significantly more concrete actions that they would take than those in the self-initials (M = 2.73, SD = 1.71; p = .03; Cohen's d = .27) or the self & top-ofmind (M = 2.43, SD = 1.83; p = .002; Cohen's d = .39) conditions. We also examined whether it would take longer to sign the pledge across the three conditions, and whether this would partiaJly explain the commitment effect. One-way ANOV A revealed that the pledging manipulation did indeed affect how long it took to complete the pledge, F(2, 326) = 8.13, p < .001, partial 1/ = .04. Post hoc analysis revealed that those in the self-dedication (M = 21.48 s, SD = 34.00) condition took significantly longer to pledge than those in the self-initials (M = 9.55 s, SD = ll.36;ep < To assess whether the effort it took to take the pledge would explain the commitment effect, we conducted a MANOVA analysis with pledging effort as a covariate. The results revealed that the pledging commitment remained significant and strong (F(2, 325) = 5.17, p = .006, partial ,J= .03), despite including pledging effort as a covariate. Pledging effort did not have a significant effect on commitment, F(l, 325) = .34, p = .55. Together, these results suggest that pledging effort alone may not have been su 2 Discussion Discussion Results from Study 2 extend our understanding in three ways. First, the findings provided a deeper understanding of the effectiveness of the proposed intervention. It is worth noting that the effect remained robust even when all of the participants read about the same nonprofit organization and received the same pledge: the only difference was how they signed the pledge. Second, we showed that the self-dedication initials' effect on commitment was independent of the effort exerted to sign the pledge. This i Study 3: E-pledging to commit Study 3 had two main goals: First, we set out to replicate the results from Study 2 in a more controlled setting. Second, we aimed to rule out pledging effort as an important potential alternative explanation in a different way than in Study 2. To do so, we included a condition in which participants were asked to include two computer-generated letters in their pledge. These computer-generated random letters would increase the effort it took for participants to PLOS ONE I /journal.pone.0231314 April 29, 2020 14/ 21 https://doi.org/10.1371 e-pledge without increasing self-focus, unlike the proposed intervention. Same as in Study 2, we tracked whether participants opted out of signing the pledge across different e-pledge conditions. Methods Methods Participants and design. We aimed to recruit 360 participants from MTurk with IP addresses based in the US in exchange for $0.76. At the end of the predetermined data collection period, we received 348 complete responses (mean agee= 35.18, SD = 10.34; 38.79% female). We obtained IRB approval from the University ofVirginia to conduct this study, with written consent from the participants. Participants were randomly assigned to one of three conditions: (1) pledging with their own iuitials, (2) pledging with E-pledee manipulation. Participants first read a two-paragraph description of a U.S. non g profit organization, the Boys and Girls Club of America (BGCA). They read about BGCA's functions and the various ways people could get involved in improving a child's life. Participants then read the following pledge: Any action you take will work toward the same goal-to strengthen and empower children in need.Join the Boys and Girls Clubs of America campaign by committing yourself to help raise awareness for the foundation in the next few weeks. Participants were then given the option of pledging their support in one of the three ways: self-initials, self-initials plus initials of someone very important to them (self-dedication initials), or self-initials plus two randomly generated letters (self & random letters). The phrasing of the self-initials and the self-dedication initials pledges were the same as in Study 2. Those in the self & random letters condition read "Please sign the pledge by entering your initials plus the two letters shown below Commitment. Participants responded to three questions that captured their commitment to the cause: "I will tell other people of the Boys and Girls Club"; "I feel very committed to missions of the Boys and Girls Club"; and "I feel very connected to the children being served by the Boys and Girls Club." We averaged responses to form the commitment scale (ae= .83). All of the participants provided their demographic information at the end of the survey and were paid the next day. Results Oet-out rate. We began by examining how many participants chose not to pledge as a p function of the three e-pledging conditions. Results revealed that whereas 10.74% of those in the self-initials (n = 13) and 13.15% of the self & random letters conditions (n = 15) chose not to pledge, only 3.53% of the participants in the self-dedication initials condition (n = 4) declined to pledge (x2(2) = 6.82, p = .03). The significantly lower attrition rate demonstrated that people were not deterred by the self-dedication initials condition. Results remained significant if we were to include excluded Commitment. Results from one-way ANOVA revealed that the pledging manipulation affected participants' commitment to the cause, F(2, 345) = 6.48, p = .002, partial rf = .03. Post hoc analysis revealed that those in the self-dedication initials condition (M = 3.47, SD = 1.06) PLOS ONE I /journal.pone.0231314 April 29, 2020 15/ 21 https://doi.org/10.1371 were significantly more committed to the cause than those in the self-initials (M = 2.99, SD = 1.03; p = .001; Cohen's d = .45) or self & random letters (M = 3.09, SD = 1.09; p = .008; Cohen's d = .35) conditions. We also examined whether it would take longer to sign the pledge across the three conditions, and whether this would partially explain the commitment effect. One-way ANOVA revealed that the pledging manipulation did indeed affect how long it took to complete the pledge, F(2, 345) = 6.95, p = .001, partial 17= .03. Post hoc analysis revealed that those in the self-dedication initials (M = 19.18 s, SD = 22.48) and self & random letters (M = 15.76 s, SD = 30.02) conditions took significantly longer to pledge 2 To assess whether the effort it took to take the pledge would explain the commitment effect, we conducted a MANO VA analysis with effort as a covariate. The results revealed that the 2 pledging commitment remained significant and strong (F(2, 344) = 5.31, p = .005, partial ,, = .03), even with effort as a covariate. Importantly, effort did not have a significant effect on commitment, F(l, 344) = 3.17, p = .07. Together, these results suggest that effort alone may not have been sufficient to explain why the self-dedication initials would deepen people's commitment to the cause they pledged to support. General discussion General discussion Comedian Seth Meers once said, "If you make a Face book page we will 'like' it-it's the least ey we can do. But it's also the most we can do." While meant to be satirical, his statement largely reflects a reality of social media: Conventional e-pledges and online campaigns are the new and increasingly ubiquitous reality for nonprofit organizations and advocacy groups [ 42-43], yet their convenience and broad reach may be offset by their ineffectiveness for securing committed behaviors. With the goal to improve this deamic, we conducted two pilots and three yn experiments to empirically examine whether a novel and simple e-pledge intervention-selfother initials-is more effective for securing commitment behavior than other types of common e-pledges. To that point, we refined the pledge by asking participants to explicitly dedicate their thoughts to that other person, and ruled out effort, novelty, and social interaction mindset as alternative explanations. These results provide a new and enhanced method that may enable organizations to capitalize on the benefit Notably, the positive link between self-other e-pledging and volunteering commitment over a span of 3 months further enhances confidence in the external and ecological validity of our findings (Study 1). The self-other e-pledge's ability to secure commitment behavior through explicit dedication also highlights the importance of enforcing accountability cues in thee-pledging process (Studies 2 and 3). Together, our findings aimed to explain whether and how a common and increasingly prevalent method for obtai It is important to point out that regardless of theee-pledge condition participants were assigned to, virtually all of them self-elected to take the pledge in Study 1, and a significant portion of the participants did so in Studies 2 and 3. Nevertheless, the actual commitment behavior varied significantly across e-pledging conditions. The discrepancy between signing the pledge and taking action provides further empirical evidence for slacktivism, consistent with survey responses from our Pilot Study lB. T PLOS ONE I /journal.pone.0231314 April 29, 2020 16/ 21 https://doi.org/10.1371 intended to honor their pledges. Therefore, the effectiveness of our intervention was unlikely to be a result of participants' insincerity or an experimental demand effect. Even so, we consistently showed that small yet carefully planned interventions-such as how people pledged-can greatly impact their committing behaviors (see 45]. Specifically, compared to those who pledged with a "Like" symbol-a widely used method of e-pledging across social media-those who pledged with combined initials volunteered 24. Theoretical contributions and frameworks for future research Theoretical contributions and frameworks for future research Our research makes several theoretical contributions and initiates new research directions. First of all, our findineshed light on laypersons perspective of slacktivism, speak to and inte gs grate theoretical perspectives on social influence and public self-awareness, and provide support for a different and more effective method of e-pledging. These results form the basis for fruitful research endeavors. We describe the linkage to previous literature and outline possible future directions below. Results from the pilot studies showed that laypeople acknowledged theee-pledging process as the main contributor to slacktivism, citing lack of accountability and consequence as major factors. Building on these results and drawing from public self-awareness research, we predicted that an effective intervention would incorporate an accountability cue in the moment people signed the pledge in the virtual enviromnent. In a sense, our proposed intervention embedded a virtual mirror as people signed the pledge. While we have ruled out several possible alternative explanations, we acknowledge that others may exist as well. One such explanation might be that the intervention evoked a "significant-other transference." Namely, contextual cues not ouly work to activate a sense of self, but different cues can affect which version of the self is activated [ 46, 47]. For instance, cues in a family setting can elicit affective, motivational, and behavioral responses associated with one's "family-oriented self," and this Our results also have point to broader questions for research on conventional e-pledges and slacktivism. Prior research has demonstrated that people often divide a superordinate goal into incremental subgoals. How people interpret these subgoals, in turn, has direct implications for goal pursuit. Specifically, when people interpret their success in fulfilling a subgoal as a possible substitute for the superordinate goal, they are less likely to pursue the superordinate goal [ 50]. Applying that to the curre Along similar lines, it is possible that people may simply disengage from commitment and responsibility after e-pledging; research on moral licensing suggests this potentially troublesome consequence. Namely, e-pledgers could feel that they have already obtained the feel- PLOS ONE I /journal.pone.0231314 April 29, 2020 17 / 21 https://doi.org/10.1371 good "moral credits" just by clicking on the pledge, and therefore free themselves to engage in opposite behavior away from the commitment (51]. Beyond this possibility, we encourage future research to examine potential negative consequences that may result from these externally induced psychological barriers. Practical implications Practical implications There are several ways our research informs organizations that seek to promote their causes. We have focused on the self-other pledge as one way to awaken and reinforce accountability. Organizations may want to further tailor their pledging prompt by more explicitly asking people to generate names of others who are relevant in that particular domain. For example, St. Jude's campaign could ask e-pledgers to dedicate their efforts to others who had courageously battled an illness, either personally or in th Our findings also highlight a more cost-effective method for reaching a broad volunteer base and keeping them engaged. Currently, nonprofit organizations often send reminders of the recipient's prior commitment. However, not only is this practice costly, but reminding people about previous pledges may also backfire. Namely, research has demonstrated that people sometimes react to reminders of past commitment failures by veering away from their original moral standards (52-54]. If theee-pledger needs to sp This research also has broad societal implications. In 2019, UC Berkeley lost its #2 U.S. News public university ranking because it provided data on the number of alwnni who pledged to donate rather than the actual donation rate (55]. In an ideal world the number of alwnni who pledge to donate would be equal to the nwnber of actual donations. In reality, the discrepancy had a consequential impact on the university, with unknown future ramifications. By systematically investigating the psychological mechanism that can strengthen people's commibnent to the e-pledges they make, and by identifying Conclusion Conclusion Our findings have high practical importance for motivating prosocial behaviors-i.e., behaviors that cannot be coerced or stipulated, yet are crucial to a society's well-being. By identifying a simple, low-cost intervention, our findinesuggest ways for nonprofit and advocacy groups gs to benefit from the convenience and efficiency e-pledges offer without compromising actual and long-term commitment. As social media and web-based technologies continue to redefine how people interact with the world, our findings move us one step closer to a broader understanding of how to transform a simple virtual acknowledgement into deeper commitmentand, ideally, action. PLOS ONE I /journal.pone.0231314 April 29, 2020 18/ 21 https://doi.org/10.1371 Supporting information Supporting information S1 Table. Percentage of volunteers bye-pledge condition per program in Study 1. S1 Table. Percentage of volunteers bye-pledge condition per program in Study 1. (DOCX) S2 Table. Poisson regression analysis on hours volunteered. S2 Table. Poisson regression analysis on hours volunteered. (DOCX) S1 Data. Online supplemental materials: Experimental materials. S1 Data. Online supplemental materials: Experimental materials. (DOCX) Acknowledgments Acknowledgments The authors thank JKM for his invaluable guidance. Author Contributions Author Contributions Conceptualization: Eileen Y. Chou, Dennis Y. Hsu, Eileen Hernon. Data curation: Eileen Y. Chou. Conceptualization: Eileen Y. Chou, Dennis Y. Hsu, Eileen Hernon. Data curation: Eileen Y. Chou. Formal analis: Eileen Y. Chou, Dennis Y. Hsu. Investigation: Eileen Y. Chou. Methodology: Eileen Y. Chou, Dennis Y. Hsu. Resources: Eileen Y. Chou. ys Writing -original draft: Eileen Y. Chou. Writing -review & editing: Eileen Y. Chou, Dennis Y. Hsu, Eileen Hemon. References References 1. 1. 1. Luckerson V. These are the 10 most popular tweets of all time. Time Magazine. 2016. / 4263227/most-popular-tweets/. http://time.com 2. 2. Yadron, D. Facebook's Sheryl Sandberg: 'Likes' can help stop Isis recruiters. . com/technology/201i6/jan/20/facebook-davos-isis-sheryl-sandberg. 201i6. https://www.theguardian 3. 3. Obar JA, Zube P, Lampe C. Advocacy 2.0: An analysis of how advocacy groups in the United States perceive and use social media as tools for facilitating civic engagement and collective action. Journal of Information Policy. 2012; 2:1-25. 4. 4. Kristofferson K, White K, Peloza J. The nature of slacktivism: How the social observability of an initial act of token support affects subsequent prosocial action. Journal of Consumer Research. 2014; 40:11i49-1i166. 5. 5. 5. Morozov E. From slacktivism to activism, Foreign Policy. 2019. https://foreignpolicy.com/2009/09/05/ from-slacktivi sm-to-activi smi/. 6. 6. 6. American Red Cross: Holiday Poll. 201i4. httpJ/www.redcross.org/images/MEDIA_ Custom ProductCatalog/rn40440098_ARC-Holida y-Poll-201i4 . 7. 7. 7. Lacetera N, Macis M, Mele A. Viral altruism? Charitable giving and social contagion in online networks. Sociological Science. 2015; 3:202-238. 8. 8. UNICEF Sweden "Like Don't Save Lives" Promotional Campaign. 2013. . http://unicef.se/ 9. 9. Prentice-Dunn S, Rogers RW Effects of public and private self-awareness on deindividuation and aggression. Journal of Personality and Social Psychology. 1982; 43(3):503-51i3. 10. 10. All three data files are uploaded on the Open Science Database database, and will be made publically available upon acceptance of the manuscript for publication. 11. 11. Merriam-Webster's Online Dictionary. 2019. https://www.merriam-webster.com/dictionary/pledge. 12. 12. Cialdini RB, Goldstein NJ. Social influence: Compliance and conformity. Annual Review of Psychology. 2004; 55:591-621. PMID: 14744228 https://doi.org/10.1146/annurev.psych.55.090902.142015 PLOS ONE I /journal.pone.0231314 April 29, 2020 19/ 21 httpsJ/doi.org/10.1371 13. 13. 13. Grant AM, Mayer DM. Good soldiers and good actors: Prosocial and impression management motives as interactive predictors of affiliative citizenship behaviors, Journal of Applied Psychology. 2009; 94(4): 900--912. 3770 PMID: 19594233 https://doi.org/10.1037/a001i 14. 14. 14. Schwartz D, Loewenstein G. (2017). The chill of the moment: Emotions and proenvironmental behavior. Journal of Public Policy & Marketing.2017; 36: 255---268. 15. 15. Chou EV. Paperless and soulless: E-signatures diminish the signer's presence and decrease acceptance. Social Psychological and Personality Science. 2015; 6(3):343--351. 16. 16. Chou EV. What's in a name? The toll e-signatures take on individual honesty. Journal of Experimental Social Psychology. 2016; 61 :84-95. 17. 17. Kettle KL, Haubl G. The signature effect: Signing influences consumption related behavior by priming self-identity. Journal of Consumer Research. 2011i; 38;474-489. 18. 18. 18. Barnett WS. Effectiveness of early educational intervention. Science. 2011; 333(6045): 975---978. 0.1126/science.1204534 PMID: 21852490 https://doi.org/1 19. 19. Solomon MR, Schopler J. Self-consciousness and clothing. Personality and Social Psychology Bulletin. 1i982; 8(3):50Ł14. 20. 20. van Bommel M, van Prooijen J-W, Elffers H, Van Lange PAM. Be aware to care: Public self-awareness leads to a reversal of the bystander effect. Journal of Experimental Social Psychology. 2012; 48:926930. - 21. 21. 21. Bryan CJ, Walton GM, RogersT, Dweck CS. Motivating voter turnout by invoking the self. Proceedings of the National Academy of Sciences of the United States of America. 201i1; 1 08: 12653--12656. 0.1073/pnas.1103343108 PMID: 21768362 https://doi.org/1i 22. 22. Capraro V, Jagfeld G, Klein R, Mui M, van de Pol I. Increasing altruistic and cooperative behaviour with simple moral nudges. Scientific Reports 2019; 9, 11880--1i1891. 48094-4 PMID: 31i41i71i06 https://doi.org/10.1038/s41598-019 - 23. 23. Diener E, Wallbom M. Effects of self-awareness on antinormative behavior. Journal of Research in Personality, 1976; 10(1i):107-1i11. 24. 24. Shull, MazarN, Gino F, Ariely D, Bazerman MH. Signing at the beginning makes ethics salient and decreases dishonest self-reports in comparison to signing at the end. PNAS, 2012; 38:15197-15200. 25. 25. Davies MF Mirror and camera self-focusing effects on complexity of private and public aspects of identity. Perceptual and Motor Skills. 2005; 100(3): 895---898. - 26. 26. Guerin B. Social Facilitation. New York: Cambridge University Press. 1i993. 27. 27. Zajonc RB. Social facilitation. Science. 1965; 149:14Ł269. 28. 28. Zajonc RB, Sales SM. Social facilitation of dominant and subordinate responses. Journal of Experimental Social Psychology. 1966; 2: 160--168. 29. 29. Cvetkovich G.Cognitive accommodation, language, and social responsibility. Social Psychology. 1978; 2:149-155. 30. 30. Lerner JS, T etlock PE. Accounting for the effects of accountability. Psychological Bulletin. 1999; 125 (2):255---275. PMID: 10087938 https://doi.org/10.1037/0033-2909.1252.255 31. 31. 31. TylerT. The psychology of legitimacy: A relational perspective on voluntary deference to authorities. Personality and Social Psychological Review. 1997; 1 :323--245. 32. 32. Duval S, Wicklund RA. Effects of objective self-awareness on attribution of causality. Journal of Experimental Social Psychology. 1973; 9(1):17-31. 33. 33. Gervais WM, Norenzayan A. Like a camera in the sky? Thinking about God increases public self-awareness and socially desirable responding, Journal of Experimental Social Psychology. 2012; 48(1 ): 298302 - 34. 34. Sproull L, Subramani R, Kiesler S, Walker J, Waters K. When the interface is a face. Human-Computer Interaction. 1996; 11: 97-124 35. 35. 35. Fenigstein A, Scheier MF, Buss AH. Public and private self-consciousness: Assessment and theory. Journal of Consulting and Clinical Psychology. 1975;43:522-527. 36. 36. 36. Gino F, Desai D. Memory lane and morality: How childhood memories promote prosocial behavior. Journal of Personality and Social Psychology. 2012; 1 02: 7 43--758. 0.1037/a0026565 PMID: 22181000 https://doi.org/1 37. 37. 37. AllenJA, Mueller SL The revolving door: A closer look at major factors in volunteers' intention to quit. Journal of Community Psychology. 2013; 41 (2):13Ł1i55. 38. 38. Each Qualtrics survey was created prior to the program directors distributing them. Because student volunteers choose their own programs, not every program had the same number of participants. As such, while Qualtrics algorithm random distributed the conditions within each volunteering program, the PLOS ONE I /journal.pone.0231 314 April 29, 2020 20i/ 21 https://doi.org/10.1371 aggregated number of volunteers per condition was not evenly distributed. Because we randomly assigned the pledge conditions within and across each program, we addressed this issue by ( 1) using nonparametric test that does not assume equal sample size across conditions (2) using nested clustering analysis to analyze the data. See online supplemental material S 1 Table for the breakdown of experimental conditions by program. 39. 39. 39. Muthen LK, Muthen BO. Mplus user's guide. Sixth Edition. Los Angeles, CA: Muthen & Muthen. 2007. 40. 40. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1999; 6:1--55. 41. 41. Feeding America. Feeding American Impact Report 201 9. Avaialble from . org/about-us/donor-impact-report. https://www.feedingamerica 42. 42. Guengerich G. Online fundraisingtrends. 2016. httpsJ/upleaf.com/nonprofit-resources/online fundraising/online-fundraising-trerds. 43. 43. 43. Morrison K. Report: Nonprofits rely heavily on social media to raise awareness. 2016. httpJ/www. . adweek.com/socialtimes/report-nonprofits-rely-heavily-on-social-media-to-raise-awareness/638229 44. 44. Penner LA, DovidioJF, PiliavinJA, Schroeder DA. Prosocial behavior: Multilevel perspectives. Annual Review of Psychology. 2005; 56:365-392. 146/annurev.psych.56.091i103.070141 PMID: 15709940 https://doi.org/10.1i 45. 45. Baca-Motes K, Brown A, Gneezy A, Keenan EA, Nelson LO. Commitment and behavior change: Evidence from the field. Journal of Consumer Research. 2012; 39:1070-1084. 46. 46. Anderson SM, Chen S. The relational self: An interpersonal social-cognitive theory. Psychological Review. 2002; 109(4):619-645. PMID: 12374322 https://doi.org/10.1037/0033-295x.109.4.619 47. 47. Baldwin MW, Carrell SE, Lopez OF. Priming relationship schemas: My advisor and the Pope are watching me from the back of my mind. Journal of Experimental Social Psychology. 1990; 26:435-454. 48. 48. Anderson SM, Glassman NS. Responding to significant others when they are not there: Effects on interpersonal inference, motivation, and affect In Sorrentino R. M. & Higgins E.T. (Eds.), Handbook of motivation and cognition ( Vol 3, pp. 262-321i). New York: Guilford Press. 1996. 49. 49. Berk MS, Andersen SM. The impact of past relationships on interpersonal behavior: Behavioral confirmation in the social-cognitive process of transference. Journal of Personality and Social Psychology. 2000; 79:546--562. PMID: 11045738 50. 50. Fishbach A, Dhar R, Zhang Y. Subgoals as substitutes or complements: The role of goal accessibility. Journal of Personality and Social Psychology. 2006; 91 :232-242. . httpsJ/doi.org/10.1037/0022-3514 912.232 PMID: 16881761 51. 51. 51. Merritt AC, Effron DA, Monin B. Moral self-licensing: When being good frees us to be bad. Social and Personality Psychology Compass. 2010; 4(5):344-357. 52. 52. Duval TS. & Duval VH. Level of perceived coping ability and attribution for negative events. Journal of Social and Oinical Psychology. 1987; 5:452-468. 53. 53. Duval TS, Silvia PJ. Self-awareness, probability of improvement, and the self-serving bias. Journal of Personality and Social Psychology. 2002; 82( 1 ):49-61. PMID: 11811633 https://doi.org/10.1037//0022-3514.82.1.49 54. 54. Silvia PJ, Duval TS. Objective self-awareness theory: Recent progress and enduring problems. Personality and Social Psychology Review. 2001i; 5(3):230-241. 55. 55. U.S. News. Updates to 5 schools' 2019 Best Colleges Rankings Data. 2019 education/blogs/college-rankings-blog/articles/201i9-07-25/updates-to-5-schools-20i19-best-collegesrankings-data. https://www.usnews.com/ PLOS ONE I /journal.pone.0231314 April 29, 2020 21 / 21 httpsJ/doi.org/10.1371 Copyright of PLoS ONE is the property of Public Library of Science and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express w1itten permission. However, users may print, download, or email articles for individual use.

Turn in your highest-quality paper
Get a qualified writer to help you with

“ Big Data Segmentation and Slacktivism ”

Get high-quality paper

Guarantee! All work is written by expert writers!

Still stressed from student homework?

Get quality assistance from academic writers!

Order now