Robert Francis and Daniel Nielson, Political Science

Introduction

A recent trend in both development research and practice has dealt with increasing the transparency of government and international development institutions. While donors and stakeholders are able to put significant legal and social pressure on large development organizations, this same pressure has not materialized in regards to NGOs. They are not legally required to make publicly available information on activities and finances, which means that such transparency must be exogenously motivated. The research question behind this project is therefore to investigate what factors could motivate NGOs to act transparently.

Methodology

This project takes advantage of a database of roughly fifty thousand NGOs in India, obtained from India’s Ministry of Planning. This study consisted of sending each of the NGOs an email containing a paragraph describing the effort to create a database of NGO operations as well as a description of why such an effort is important. Each email also contained an invitation to participate, brief instructions for completing the survey, and a link to the survey itself. Additionally, each email contained a control (no additional text) or a paragraph-long textual treatment providing an additional motivation to contribute to the project. Four motivation treatments were used. The first, “Coordination,” is a description of how contributing the NGO’s information will help improve the coordination of activities between organizations, boosting effectiveness and efficiency. The second, a “Charge for Transparency,” is a simple description of transparency as a normative issue and why it is an important principle for NGOs to exemplify. The third, “Donors’ Interest,” describes how providing operation information will both raise the profile of the organization and make it more visible to donors, as well as make the organization more attractive to donors, as they favor transparent organizations. The last is a “Social Proof,” which asserts that most NGOs are transparent, and that the vast majority of Indian NGOs, when asked whether they will participate in the study, say that they will. The length of the motivation treatments was between 62 and 74 words.

Additionally, a Commitment treatment was used, which consisted of a very short email briefly describing the project and asserting that the respondent will receive an email containing the survey in a few days. The commitment email also contained a link to a one-question Qualtrics survey that asked whether they will be willing to complete the full survey when it arrives. Half of all NGOs received this treatment as part of a factorial design.

Results

In this study I used outcome measures of survey completeness, namely binary variables for starting and completing the survey. The first test I conducted was to determine the effect of receiving the Commitment email on completing the NGO information survey. Of the 20,000 NGOs that received the treatment email, 434 respondents completed the commitment survey included in the email, of whom 98% indicated that they would participate in the full survey. However, a t test between the treatment and control groups indicated that NGOs receiving the commitment email were no more likely than those who did not to start or complete the full NGO information survey.

The second test I conducted was to determine the impact of the motivation treatments. Roughly 8,000 emails were successfully sent to NGOs in the control group and each of the treatment groups. Additionally, two rounds of reminder emails were sent out to each of the NGOs that had not completed the survey within a week of the initial email being sent. Difference of means tests were conducted using the start and finish rates between the control group and each of the treatment groups. The results are presented in Table 1.

Table 1: Response Rates of Motivation Treatments and Control

Treatment	N	Start rate	p value	Finish rate	p value
Control	8064	0.0425		0.0155
Coordination	7801	0.031	0.0001	0.0109	0.01
Charge	7858	0.029	0.0001	0.0092	0.0003
Donor	7946	0.035	0.021	0.0128	0.154
Proof	7935	0.031	0.0001	0.0103	0.0038

The control group had a higher start rate than any of the groups in which a motivation treatment was included in the body of the email. The differences between the treatment groups and the control are all statistically significant. The finish rates for the motivation treatments are likewise all lower than the Control finish rate. However, the difference between the Donor finish rate and the Control finish rate is not statistically significant, having a p value of 0.154. The differences between the Control finish rate and the finish rates of the other motivation treatments are all significant at the 1% significance levels. The final statistical test I ran was to determine whether the start and finish rates of the Donor treatment are statistically different from the other motivation treatments. While the start and finish rates are higher for the Donor treatment than the other treatments in an absolute sense, the difference is only statistically significant in comparison to the Charge for Transparency treatment.

Discussion

An observation of the results of this study reveals one primary limitation in measuring the relative treatment effects of each of the motivation treatments. Each motivation treatment yielded a lower start and finish rate than the control. As the only difference between treatment and control groups was the addition of a 60-70 word paragraph to the email sent to each NGO, the time required to read an additional paragraph likely dissuaded many of the organizations from reading the end of the email and clicking on the link to the survey. A longer email likely results in attrition, and as all of the treatments were one paragraph longer than the control, the additional length caused a decrease in response. The control group is therefore not a “pure” control that is comparable against the motivation treatments. A more useful test is the comparison between the various treatments. The treatment-control comparison suggested that the Donor treatment is more motivating to NGOs than peer effects, cognitive dissonance, or coordination. The direct comparison between treatments, however, showed that the start and finish rates are only statistically different between the Donor and Charge treatments.

Conclusion

This study was able to provide tentative evidence that NGOs are more motivated by financial considerations than they are by factors such as cognitive dissonance, concern over coordination with other organizations, or comparisons with other organizations when it comes to providing information regarding their organization. In situations where coercive or legal force is not an option – where transparency is optional and voluntary – an exogenous motivation is necessary in order to promote a response. This study provides support to the theory that either deterrence or reassurance is necessary to promote a response – endogenous motivations are not sufficient, or at least are not most effective, when inducing transparency.