Field experiments occupy an important middle ground between laboratory experiments and naturally occurring field data. The underlying idea behind most field experiments is to make use of randomization in an environment that captures important characteristics of the real world. Distinct from traditional empirical economics, field experiments provide an advantage by permitting the researcher to create exogenous variation in the variables of interest, allowing us to establish causality rather than mere correlation.
In relation to a laboratory experiment, a field experiment potentially gives up some of the control that a laboratory experimenter may have over her environment in exchange for increased realism Field experiments can be a useful tool for each of these purposes. For example, Anderson and Semester (2003)2 collect facts useful for constructing a theory about consumer reactions to nine-dollar endings on prices.
They explore the effects of different price endings by conducting a natural field experiment with a retail catalogue merchant. Randomly selected customers receive one of three catalogue versions that show different prices for the same product. Systematically changing a product’s price varies the presence or absence of a nine-dollar price ending. For example, a cotton dress may be offered to all consumers, but at prices of 34, 39, and 44 dollars, respectively, in each catalogue version.
They find a positive effect of a nine-dollar price on quantity demanded, large enough that a price of 39 dollars actually produced higher quantities than a price of 34 dollars. Their results reject the theory that consumers turn a price of 34 dollars into 30 dollars by either truncation or rounding. This finding provides empirical evidence on an interesting topic and monstrance the need for a better theory of how consumers process price endings.
Another example, Karl and List (2007) is an example of a natural field experiment designed to measure key parameters of a theory. In their study, they explore the effects of ‘price’ changes on charitable giving by soliciting contributions from more than 50,000 supporters of a liberal organization. They randomize subjects into several different groups to explore whether solicited respond to upfront monies used as matching funds. They find that simply announcing that a match is available inconsiderably increases the revenue per solicitation – by 19 percent.
In addition, the match offer significantly increases the probability that an individual donates – by 22 percent. Yet, while the match treatments relative to a control group increase the probability of donating, larger match ratios -3:1 dollars (that is, 3 dollars match for every 1 dollar donated) and 2:1 dollar – relative to smaller match ratios (1:1 dollar) have no additional impact. 3 NEED AND TYPES OF FIELD EXPERIMENTS: Within economics, much experimental research has taken the form of laboratory economics economics By Himalayan
FIELD EXPERIMENTS IN ECONOMICS By: Ravish Soda Work using laboratory experiments has offered a variety of insights. For example, Smith’s research illustrated the robustness of market mechanisms in reaching an equilibrium price, showed the effect of institutions on allocations, and explored the formation and dissolution of asset bubbles in markets, among other lessons. However, results in laboratory economics are inevitably subject to questions over the extent to which they generalize to non-laboratory settings.
One concern is that such experiments are often done with college students as subjects. Thus field experiments play a very important role here because of the mere fact that they do not involve students or a sample but participants drawn from the market of interest, I. E. , the market where the experiment has to be conducted. Of course, a plausible concern about laboratory experiments in economics, whether the participants are students or others, is the extent to which the results are influenced by the laboratory setting.
A “framed field experiment” resolves such issues by conducting a structured experiment in the natural environment of the subject rather than in the laboratory. Framed fife led experiments have also been used in smaller-scale settings. Peter Boom (1972)4 was an early experimenter to depart from traditional laboratory economics methods with an experiment on the willingness to pay for a public good? in this case, a highly anticipated new television show that was being broadcast on Swede’s one television station in 1969.
More recent examples of framed field experiments within economics include my study with Jason Shorter exploring the efficacy of the contingent valuation method to estimate economic values of non arrested goods and services (List and Shorter, 1998); Lucking-Reilly (1999) study testing theoretical predictions concerning various auction formats; and Fryer’s (2010) study of students’ responses to an experiment offering some of them financial incentives for good academic performance.
More recently, the wave of field experiments or “randomized control trials” executed in developing countries are often framed field experiments, which typically are geared toward advancing policy (for example, Kramer, Miguel, and Thornton, 2009; Duffel, Dupes, Kramer, and Sine, 2006). Of course, a plausible concern is that when subjects know they are participating in an experiment, whether as part of an experimental group or as part of a control group, they may react to that knowledge in a way that leads to bias in the results.
A “natural field experiment” occurs in the environment where the subjects are naturally undertaking certain tasks, and where the subjects do not know that they are participants in an experiment. Such an experiment combines the most attractive elements of the experimental method and naturally occurring data: randomization ND realism. By combining randomization and realism in this manner, natural field experiments provide a different parameter estimate than do laboratory, rate factual, and framed field experiments.
One possible source of bias in these other experimental approaches is that generally the subjects who choose to participate in the experiment are those who expect to gain the most (perhaps because they believe they are likely to get good results from the treatment). As a result, the estimated causal effect from these other experimental types, while valid, might not generalize to he target population of interest?which of course includes the subpopulation (often the majority) that did not volunteer for the experiment when offered the opportunity. Hospice about whether they will participate, the treatment effect obtained from natural field experiments is, in the best-case scenario, an estimate that is both causal and broadly generalize (in AY-Badly and List, 2011)5. Put simply, since participants in the natural field experiment are a representative, randomly chosen, non-self-selected subset of the treatment population of interest, the causal effect obtained from this type of experiment is the average causal effect for the full population?not for a nonrandom subset that choose to participate.