Field Experiments

Definition

Studying phenomena in social sciences is complicated. A researcher typically wants to provide a rigorous assessment of the direct link between two variables that in turn are linked with many (in principle infinite) other variables in the real world. This complication motivates the use of an experiment, which applies the scientific method to potentially assess the direct link between the variables while keeping all other variables under control.

In general, in an experiment, the researcher is interested in knowing if (and how much) variable x causes variable y. For this reason, participants to the experiments are usually randomly split in two groups not interacting with each other, called treatment and control groups, in which variable x is, respectively, present (or emphasized) and absent (or minimized). The random assignment to either [Page 877]group ensures that each participant has the same probability of being placed in the treatment or control group. The two groups, whose participants then share similar characteristics, are subsequently compared in terms of variable y to judge the impact of x on y. Randomization helps to ensure that the variation observed on variable y is not influenced by other variables than x.

The peculiarity of the field experiment is that it takes place in a real-life environment under discretion of the researcher. For instance, a field experiment run at school may check if incentives to eat fruits and vegetables help children improve their dietary habits, therefore making them healthier. Children are split in two groups, where one group (the treatment group) is subject to the incentives and the other (the control group) is not. The comparison of the two groups can provide insights on the effectiveness of the incentive and whether it varies with age and gender. In addition, it would be interesting to understand whether the effects persist in the long run, that is, when the children become adults.

Field, Laboratory, and Natural Experiments

Field experiments differ from laboratory and natural experiments in two key dimensions: the environment (artificial vs. real) and the control of the researcher (present vs. absent). Field experiments, working on a real environment under researcher’s control, can be seen as a compromise between laboratory experiments and natural experiments. In fact, they share the control of the environment with the former and a real-life outcome with the latter.

...