Research as Leisure: Experimental Evidence on Voluntary Contributions to Science

Treatment Groups:

Scientific Contribution Treatment: One explanation for voluntary contributions to science we will test in this project is that people may derive utility from the belief that they are contributing to a welfare enhancing public good (e.g., Bergstrom et al, 1986; Stiglitz, 1999). More specifically, they may volunteer their time to science because they derive a "warm glow" from participating actively in scientific discovery (Andreoni et al, 1996). To test this explanation, the scientific contribution of the research project will be emphasized in one of the project pages.

Value of Labor Supply Treatment: A related explanation for voluntary contributions to science is that people believe scientific discovery is important, and that their labor supply is more valuable than the money they could donate to buy the equivalent contribution (Andreoni et al, 1996). In this explanation, volunteers may not receive a warm glow from the act of contributing, but rather, they may believe scientific discovery is important and that their labor supply is the best way they can contribute to it. To test this explanation, the value of volunteer labor supply in the project in terms of the amount of paid labor it is substituting for will be emphasized in one of the project pages.

Control Group:
The control group will be sent a version of the project with a basic description and with none of the treatments described above.

References:

Andreoni, James, et al. "Charitable contributions of time and money." University of Wisconsin–Madison Working Paper (1996).

Bergstrom, Theodore, Lawrence Blume, and Hal Varian. "On the private provision of public goods." Journal of Public Economics 29.1 (1986): 25-49.

Stiglitz, Joseph E. "Knowledge as a global public good." Global public goods 1.9 (1999): 308-326.

2017-06-02

2017-07-18

Number of contributors, number of contributions made per contributor, quality of contributions

Quality of contributions will be measured as the accuracy of the contributions made

To avoid contamination of our study, we will make the design public once the study has been completed.

In order to test why certain types of people have a desire to contribute to science, and whether this desire can be increased through certain non-monetary incentives and research task modifications, we will run a natural field experiment on Zooniverse, the largest voluntary crowd science platform in the world. Specifically, we will post four actual scientific research projects on the Zooniverse platform for platform participants to assist with. The projects will all have the same scientific content but will vary in how they are presented to participants. These variations are described in detail below. The project will be to classify photos of rangelands taken in Northern Kenya as part of a large rangeland condition crowd-sourcing effort.

Participants will not be aware that these projects are being used to investigate questions related to the economics of science in addition to the publicly stated objective of the projects. To eliminate concerns about participant selection into a particular project based on an experimental treatment, and to ensure participants are not aware that there are several versions of the same project (i.e., different experimental treatments), each version will be emailed to a randomly selected subset of Zooniverse participants who will be invited to contribute to the projects. Zooniverse will send these emails on our behalf to their population of users. The total population of users who will receive an email is 31,000. Each participant will receive one version of the project. We will control for participant characteristics by administering a short survey on participant age, gender, employment status, educational level, and career goals before they contribute to the project.
Treatment Groups:

Scientific Contribution Treatment: One explanation for voluntary contributions to science we will test in this project is that people may derive utility from the belief that they are contributing to a welfare enhancing public good (e.g., Bergstrom et al, 1986; Stiglitz, 1999). More specifically, they may volunteer their time to science because they derive a "warm glow" from participating actively in scientific discovery (Andreoni et al, 1996). To test this explanation, the scientific contribution of the research project will be emphasized in one of the project pages.

Value of Labor Supply Treatment: A related explanation for voluntary contributions to science is that people believe scientific discovery is important, and that their labor supply is more valuable than the money they could donate to buy the equivalent contribution (Andreoni et al, 1996). In this explanation, volunteers may not receive a warm glow from the act of contributing, but rather, they may believe scientific discovery is important and that their labor supply is the best way they can contribute to it. To test this explanation, the value of volunteer labor supply in the project in terms of the amount of paid labor it is substituting for will be emphasized in one of the project pages.

Control Group:
The control group will be sent a version of the project with a basic description and with none of the treatments described above.
References:

Andreoni, James, et al. "Charitable contributions of time and money." University of Wisconsin–Madison Working Paper (1996).

Bergstrom, Theodore, Lawrence Blume, and Hal Varian. "On the private provision of public goods." Journal of Public Economics 29.1 (1986): 25-49.

Stiglitz, Joseph E. "Knowledge as a global public good." Global public goods 1.9 (1999): 308-326.

Randomization done by crowd science organization using computer program

Individual contributors

No

31,000 individuals will be sent an email, we expect 3,100 will contribute based on uptake averages on the platform

31,000 individuals will be sent an email, we expect 3,100 will contribute based on uptake averages on the platform

10,300 receiving the value of contributor output treatment, 10,300 receiving the value of contributor input treatment, 10,300 receiving the control page

To achieve power on the quality of output based on the pilot quality outcome measured as average agreement in responses across respondents, we need 699 classifications per project page. On our pilot, we achieved 630 classifications with 5,000 emails send out so we expect that with slightly over 10,000, we will achieve between 1,200 and 1,300 per page. The pilot control page average agreement was 60%, with a standard deviation of 0.114. With 699 classifications, we would need the treatment groups to achieve quality levels that differ from the control by 1.4 percentage points to achieve power of 90%.