Using household surveys to implement field experiments: the willingness to donate to food banks

ABSTRACT

This paper demonstrates that it is possible to jointly produce household surveys and field experiments by incorporating field experiments into the structure of the financial incentives used to enhance response rates. We use the opportunity to donate the financial incentive to a food bank to illustrate how the strategy would work.

KEYWORDS:

• Value of charity
• field experiment

JEL CLASSIFICATION:

• H44

Acknowledgement

Thanks are due to Shauna Mortensen for assistance in preparing this manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Respondents are given the option to take the survey online, by phone, or in person. The majority (64%) who completed the survey during this experiment took it online.

² The survey could be taken online, by telephone, or with a scheduled personal interview. A multinomial-logit model considers the demographic factors influencing the choice of online, telephone, or in person response rates indicated that whites were significantly more likely to use online or telephone, and education was a significant positive factor in the selection of the online mode.

³ See Smith et al. (2015) for more detail on the experiment and results for response rates.

⁴ While early research considering binary response, dependent variable models favoured probit or logit, Angrist and Pischke (2009) and Wooldridge (2010, Chapter 15) conclude that the linear probability (OLS) model generally provides reliable estimates of the directions and magnitudes of the average effects of independent variables.

⁵ To develop this estimate, consider a simple model for each survey participant’s decision to donate to the food bank. Equation (1) specifies a linear, indirect utility function for the utility realized if a respondent donates his financial incentive to the food bank and (2) if he does not .

This formulation is similar to the logic that Hanemann (1984) originally outlined to derive the welfare properties of discrete choice models. We assume that the marginal utility of income , designated by , remains constant across the two decisions. represents the monetary incentive. Since the model describes the choice for an individual who has agreed to participate, it recognizes that taking the survey implies the time commitment, represented here by . With this treatment, a decision to donate the incentive implies that the respondent’s ‘out of pocket’ cost is the time for the survey. By contrast, a decision to keep the incentive offsets some of the time cost of doing the survey. The and terms capture the difference in well-being a person experiences with the decision to donate versus not donate to the food bank.

Additional information

Funding

This work was supported by the National Science Foundation [grant number BCS-1026865]; and Central Arizona-Phoenix Long-Term Ecological Research (CAP LTER).