• Keine Ergebnisse gefunden

Evidence from a Survey Experiment 1

Hypothesis 1: Prominent mention of a quality seal has a positive effect on respondents’ willingness to donate compared to the control group

3.2. Experimental Treatments

Individuals were randomly assigned to one of the three conditions: control, impact, or seal.

Each individual was presented with a fundraising scenario followed by a request to indicate their personal donation. The aim was to provide enough information without overloading the participants. The control group represented the basic scenario used for comparison, which consisted exclusively of emotional fundraising elements. The impact and seal group scenarios were identical to the control treatment but extended with additional informative elements.

With regard to the marketing techniques implemented within the three treatments, certain very basic instruments were employed. A photo was included, as this medium fosters emotions very effectively; the emotional impact was further enhanced by displaying a (sad-looking) child. The image was selected to trigger negatively framed emotions such as guilt and pity, which are known to have a positive effect on donation likelihood. The singularity effect was also considered. The scenario described the situation of children in Sub-Saharan Africa.

Respondents were then given the opportunity to resolve their “emotional imbalance” through a donation. However, the marketing aspect was not the major focus of the study; it was merely adapted to simulate a realistic fundraising request.

22 Control Group

The fundraising request for the control group began with an introduction explaining the general context and describing the widespread problem of school absenteeism in the region of Sub-Saharan Africa, where many parents regard education as unimportant.

The Sub-Saharan African region has seen a remarkable rise in school enrolment over the past decade. However, more than 21.6 million children of lower secondary school age in the region still do not attend school, and many are expected never to have access to formal education. Children from very poor families, orphans, and girls are particularly underprivileged, despite the fact that education is one possible way to escape from poverty.

The valuation section continued with a description of the aid program seeking to reduce school absenteeism organized by the fictitious charity “Initiative Help4Children”. We chose a fictitious organization for logistical reasons and so that we could implement the necessary ceteris paribus comparisons without deceiving subjects (e.g., describing an organization that has or does not have a quality seal, depending on the treatment). As is usual for CVM studies, the project, the implementing institution, and the resulting benefits were explained.

Respondents were told that the program is dependent on donations and that study participants can actively contribute to making a change.

It may be that some parents regard education as useless. Imagine the fictitious organization “Initiative Help4Children”, which informs parents about the income options of successful school graduates. This information conveys important incentives for parents to send their children to school and to support their education. This campaign, however, can only take place with the help of your contribution. Your donation will ensure that as many households as possible can benefit from the aid campaign.

After the context was explained, each respondent was asked about his or her willingness to donate: “Do you want to donate to the ‘Initiative Help4Children’ campaign now?” If respondents indicated a positive WTD, a further question appeared, requesting them to fill in the respective donation amount.

23 Impact Treatment

The impact group featured the same basic set-up as the control group. However, the fundraising appeal was extended by the additional information that aid programs can often be ineffective despite well-intended implementation efforts. Evaluation institutions and their services, as well as the benefits of impact evaluations, were explained to participants.

There is often a lack of information with regard to whether a certain aid project is suitable to help people in need. By implementing scientific randomized studies, one can determine which program has the greatest benefit for aid recipients for a given amount of money. These studies can be compared to experiments in the field of medicine in which a treatment is tested on a randomly chosen group of people. These results are then compared to those from the control group that was not exposed to the treatment.

Because program evaluations are relatively new and probably unknown to most survey participants, it was important to explain their function plausibly to ensure understanding. The text emphasized that these randomized, scientifically conducted studies test the cost-effectiveness of aid programs. The text went on to state that J-PAL has evaluated several aid programs that seek to decrease school absenteeism in Africa. According to their findings, the most cost-effective treatment is informing the parents about the benefits of education.

The most promising measure to improve school participation in poor regions is to inform parents about the benefits of education for their children. This information campaign results in a significant increase in school participation compared to other measures for a given amount of money. Thus, 100 USD spent to inform parents about the benefits of education increases school participation by 21 children per year. If same amount were instead invested in supplying free school uniforms, for example, school participation would increase by only 1 child per year.

The subsequent fundraising request was the same as that of the control group.

24 Quality Seal Treatment

This treatment also featured the same basic set-up as the control group, followed by an additional transparency disclosure. The introductory section, picture, description of one aspect of the current educational situation in Sub-Saharan Africa, presentation of “Initiative Help4Children” and its “Info-Campaign” aid program, and the plea for financial help were as described above. This was followed by a brief passage on organizational effectiveness. The text noted that “Initiative Help4Children” is certified by the Austrian Charity Quality Seal (for the survey conducted in Innsbruck) or the German “Donation Quality Seal of the DZI” (for the surveys conducted in Germany). Both seals assure ethical behavior in fundraising and donation-funded activities in accordance with objective and verifiable standards.

It is often impossible to know whether an organization is suitable to help people in need. Some organizations might have a greater effect on poverty alleviation because their administration is more accountable and efficient. Consequently, institutes that evaluate organizations in terms of quality and accountability have been established.

Imagine that the fictitious organization “Initiative Help4Children” is certified by the Austrian Charity Quality Seal/Donation Quality Seal of the DZI. Certified organizations are proven to comply with objective and verifiable standards with regard to fundraising activities and the administration of donations. The donation quality seal ensures that your donation is actually channelled to those in need and that the organization manages donations carefully and responsibly.

The additional information signals trustworthiness, efficiency, and the reliable use of funds.

The subsequent fundraising request was the same as that of the control group.

25 3.3. Data collection

The sample included students from all departments of the universities of Innsbruck, Mannheim, and Marburg who subscribed to the “Email Newsletter on Social-Scientific Surveys”. The newsletter included the link to the survey, information on the topic, the duration of the survey (10 minutes), and the fact that answers would be strictly anonymous.

Students were not compensated for answering the questions.f

Although a student sample is unrepresentative of the population as a whole, students were chosen as the target population because they represent a homogenous group. This facilitates the detection of patterns regarding donation behavior in comparison to a highly fragmented target sample. Furthermore, students are an interesting group for analysis: Despite lower average donations by current students, it is important for charities to establish a relationship at an early stage and create donor loyalty, as contributors do not readily switch their charitable allegiances later in life.

Our study began in Innsbruck in July 2014. We decided to oversample the impact and control treatment, as we were initially most interested in these differences. The surveys in Mannheim and Marburg were conducted in the first quarter of 2015. Combining the surveys implemented in the three locations, we had a total of 578 complete and 279 incomplete responses. The number of respondents assigned to each survey and treatment is displayed in Table 1.

26

Table 1: Observations by treatment and completeness

Observations Innsbruck Mannheim Marburg Total

Completed surveys

Control 126 43 18 187

Impact 151 58 16 225

Seal 76 72 18 166

Total (3 Treatments) 353 173 52 578

Incomplete surveys

Control 76 3 7 86

Impact 104 8 12 124

Seal 56 10 3 69

Total (3 Treatments) 236 21 22 279

Notes: A total of 857 individuals participated in the surveys conducted in Innsbruck, Mannheim, and Marburg; 578 people completed the questionnaire fully, whereas 279 observations were incomplete (i.e., respondents stopped the survey during/after answering the valuation section of the respective treatment). The majority of observations stemmed from the survey conducted in Innsbruck, amounting to 589 respondents in total (353 complete and 236 incomplete answers). Student participants from Mannheim accounted for 194 observations in total (173 complete and 21 incomplete answers). The remaining 74 observations (52 complete and 22 incomplete answers) were generated by respondents from Marburg. With regard to sample size by treatment, most observations were assigned to the impact condition (349), followed by the control group (273) and the seal group (235).