The Charitable-Giving Value of Man's Best Friend: A Field Experiment

In the treated group, we present a dog along with a donation solicitor at a well-know street donation site. In the control group, only a donation solicitor will be present.

2017-11-27

2017-11-29

The portion of pedestrians who make a donation (extensive margin)

The average amounts of donations collected within each 2-hour time increment at each location (intensive margin)

We will partner with the Salvation Army to run a natural field experiment in Greenville, NC from 11/27/17 to 11/29/17 (10:00am to 8:00pm each day) at three different collection sites (three entrances of a local Walmart). We will bring and remove dogs randomly in 1-hour increments at each of the 3 sites. These dogs will be those of the experimenters and the experimenter's faculty colleagues that are well-known to the experimenters and can be trusted in the public around people. Students will be hired as temporary research assistants to support data collection by "ringing bells" at each of the 3 collection sites as well as recording the number of donations made. The experimenters will also observe and oversee student participants.

The experimenters will record the number of pedestrians and the number of donors at each of the 6 sites. Dividing the number of donors by the number of pedestrians gives out first key outcome variable. The experimenters will switch out the locked "kettles" (donation container) every 1 hours with empty kettles and will deliver them to the Salvation Army. The Salvation Army will sum the money for each 1-hour site-session.

The experimenters will record the entire solicitation processes with car dash cameras at each of the 3 sites. The cameras will be located in cars parked at a distance to avoid being seen by pedestrians or capturing pedestrians facial looks. The video will be analyzed to calculate the extensive margin. Also, the video allows us to extract data about the characteristics of individual pedestrians, such as gender, visible age and ethnical group, interaction with the dogs, etc.

The experimenters will also record license plate associated with randomly-selected donors and non-donors (and based on parking lot exit of choice). License plate information will be recorded on a university-purchased Microsoft Surface tablet and uploaded each evening directly to the secure ECU pirate port server. The data will then be deleted from the portable device. License plate data will be used to record (i) vehicle value, (ii) vehicle MPG and (iii) recorded traffic citations, which are all publicly available data. These 3 indicators will provide useful proxy data for (i) income, (ii) eco-preferences and (iii) risk preferences, respectively. This entirely will be purely observational.

Randomization done in office by a computer

1-hour increments at each solicitation location

Yes

90 = 10 periods per day per location * 3 locations * 3 days

Total number of pedestrians

45 clusters in the treated group and 45 clusters in the control group

Along the intensive margin, assuming the mean donation amount per 1-hour period is $17.67 in the control group with a standard deviation of $9.51, our sample size of 45 clusters per group allows us to detect an effect size of $5.62 at the 0.05 significance level with a power of 0.8. The mean and standard deviation are calculated using historical salvation army solicitation data at the same locations on the same dates.