Ryan Hill and Dr. Josh Gubler, Political Science Department
Introduction
This experiment tested how social norms affect individual incentives to conserve common-pool resources. We have tested this question with a simple field experiment on BYU campus. We observed individuals using public water sources after being treated with different inducements to conserve water. We are interested to see if individuals change their behavior in response to reminders of social norms immediately before making a decision to conserve scarce water. If there is an effect, does it vary based on the wording and nature of the reminder?
We hypothesized that individuals would reduce their water use the most when they are reminded that their peers are trying to conserve water. This hypothesis is consistent with the common literature in psychology and behavioral economics. However, the results of our experiment fail to identify any treatment that effectively reduces water at a significant level. In the following sections we explain our experimental method and present the extended results. We offer some explanation of why we were unable to identify any effective intervention that could reduce water use.
Method
Our hypothesis was best tested in an environment that would require an immediate decision to conserve a common pool resource (CPR). The most common resource that we consume every day is water, and it is one of the most critical CPRs to conserve. Our research will specifically investigate everyday decisions to conserve water, but there are wider applications of the knowledge in higher demand water markets such as agriculture and industry. To test our hypothesis, we chose to conduct a field experiment on Brigham Young University (BYU) campus that would test the everyday consumption of water from public drinking fountains. We ran the experiment on BYU campus during the fall semester of 2012. This was a unique opportunity to study water conservation because the state of Utah and most of the western United States had been experiencing a very serious drought for the previous 36 months. The summer of 2012 was particularly dry, experiencing the lowest rainfall in over 50 years. We took advantage of these conditions by encouraging students on campus to reduce the amount of water that they were drinking from the water fountains. During a period of three weeks during the semester, we posted flyers directly above drinking fountains that had different reminders to conserve water.
Each flyer had a different encouragement to conserve water. The four flyers contained three treatments and a neutral placebo that encouraged the individual using the fountain to conserve water. The treatments encouraged the subjects to conserve for the following reasons: 1) to help the local economy, 2) to preserve the local ecosystem 3) to conserve because their classmates were doing their best to conserve. Each of the treatments contained the same wording as the placebo, but with the treatment wording inserted into the message. Based on existing literature, we expected that the treatment referring to classmate behavior would be the most effective at reducing individual water use.
We collected our data by randomly assigning the flyers to two different drinking fountains in a prominent building on BYU campus. Each day before one of the busy passing periods between classes, an observer posted a flyer above the fountain that was selected by computer randomization. They then discreetly observed the drinking fountain with a stopwatch and timed the duration of water use. Instead of directly recruiting subjects to participate, we simply observed the normal flow of students using that particular drinking fountain. The advantage of this data collection technique was that it carefully mimicked everyday activity. Field experiments are characterized by high external validity. Subjects were unaware that they were participating in research, so their behavior was as close to normal activity as possible. The only manipulation was that each subject was confronted with a blatant request to conserve the water that they had intended to drink as normal. This brief but important inducement allowed us to infer that any change in behavior was in fact caused by the manipulation.
Results
The initial analysis of the data was performed with an ordinary least squares regression. This allowed us to test for any significant differences in the mean drinking time between treatment groups. We found that none of the treatments had a statistically significant impact on drinking time. We interpret the results with two possible explanations: 1) we lacked statistical power to differentiate between our competing hypotheses and/or 2) the results reflect the true state of the world that the treatments have no effect in this environment.
Unfortunately we were unable to gather enough data to find any significant differences in mean drinking time between treatments. We were limited in time and resources, and it may be that with a larger sample we could see significant differences in treatments even if the magnitudes are small. It may also be that the treatments were ineffective, and a large sample would strengthen this assertion. Many of the subjects did not take the time to read the flyers, and even if they did, the drinking times were short enough that we may see very little variance in drinking time.
Discussion
Our results point to one of the most difficult aspects of the tragedy of the commons. In many cases, resources are depleted not by significant free-riding by a few agents, but by minor freeriding by many agents. It is likely that most of our subjects responded to the inducement with flippancy, justifying that their decision had a miniscule effect on the overall water situation. Therefore they were unwilling to change their behavior to save only a cup or less of water. Unfortunately every student that takes a drink every day makes the same decision, and thousands of cups of water add up over time. Our results then cast a gloomy picture over the potential of peer pressure in fixing the tragedy of the commons in certain situations. Free-riding still poses a challenge to the management of CPRs when many agents are participating in the market. Future research will hopefully expand on the findings of this study. Although our study had ambiguous results, it is important to note that it was conducted in a very limited environment. Other studies may be able to better capture causal effects with strong treatments, manipulation checks, randomization, and larger sample sizes. The extension of this important body of experimental literature will better inform us of the optimal methods of managing common-pool resources.