Study
The study by Falk, Kosse, and Pinger (2020) is a long-term follow-up of the sample in Study 1 (Kosse et al. 2016). The same methodology described in Study 1 (above) was used to conduct the follow-up study. The sample consisted of children (n = 479) who were able to be followed through grades 9 and 10. Participant attrition was not related to treatment arm or grade-point average (GPA).
Children averaged 7.8 years old and were in grade 2 or 3 when they were interviewed along with their parents at baseline, and they were in grades 9 or 10 when they were interviewed again 6 years later at the end of 2018. Mothers completed questionnaires at both time points about their socioeconomic background, school choice for their child, and their child’s GPA. Stratified random treatment assignment was used to ensure groups had an equal number of youths who were low income, had parents who didn’t attend “high track” educational programs, and who had a single parent. There was no statistically significant difference between the treatment and control group in pre-intervention GPA.
In Germany, students starting at around 10 years old are formally sorted and separated into one of three educational tracks (high track, middle track, and low track) that determine their eventual degree and career options. The outcome of interest in this study was high-track attendance, defined as any school program that would lead to an upper secondary school degree qualifying for university studies (German Abitur). Students who were sorted into the middle or lower track were defined as “low track.” The program effects were estimated using intent-to-treat and controlling for GPA in elementary school, sex, and age. The study conducted subgroup analyses to examine separate treatment effects for students who stayed in their attendance track versus those who reversed tracks, and for students who experienced grade retention versus those who did not.
Study
Kosse and colleagues (2016) evaluated the effects of the Baloo and You program on prosociality in a randomized study of low-socioeconomic status (SES) children (aged 7 to 9) in Bonn and Cologne, Germany. All families with children born between September 2003 and August 2004 and one third of families with children born between September 2002 and August 2003 were contacted via postal mail. Interested parents completed a non-binding letter of intent and a questionnaire about household socioeconomic characteristics. Responses (n = 1,626) were used to categorize families into either low- or high-SES households reflecting level of material, educational, and time resources available. A household was classified as low SES if 1) it was low income (i.e., the income level was lower than the 30 percent quantile of the German income distribution, which was equivalent to 1,065 euros at the time of the study), or 2) low education (neither parent had education beyond the highest secondary school level), or 3) had single parent status (parent not living with a partner). Households that did not meet any of the criteria were classified as high SES.
Of the 590 low-SES families who participated in the first wave of data collection (fall 2011) and who consented to have their addresses sent to the mentoring program, 212 were randomly selected to receive the intervention after stratification into 14 groups based on a combination of location (Bonn or Cologne) and SES criteria (low income and/or low education and/or single parent). Additionally, because Bonn had a larger supply of mentors, more families were assigned to the intervention group in Bonn, compared with Cologne. The remaining 378 families were assigned to the control group. A third group consisted of 122 randomly selected high-SES families who provided wave 1 data. All families were invited to participate in wave 2 data collection after the 1-year intervention period (607 of 712 families provided data; 85 percent) and in wave 3 data collection 2 years after the end of the intervention (509 of 607 provided data; 83.9 percent). No additional information on the demographic characteristics of the sample was provided.
Seventy-four percent of children in the intervention group were successfully matched with mentors, 8 percent had matches that were initiated but closed early, and 18 percent were not matched due to mentor shortages. Children in the intervention group were provided with a mentor for up to 12 months. The average relationship duration for the study sample was 9.5 months: more than 50 percent of matches met for 12 months, and more than 80 percent met for at least 6 months. On average, each child met with his or her mentor 22.8 times, typically for an afternoon.
Data on prosociality was collected through interviews with the child and the accompanying parent (who was the mother in 95 percent of cases). A measure of prosociality was generated by combining responses for three facets of prosociality: altruism, trust, and other regarding behavior. Altruism was assessed using three incentivized versions of dictator games. In the first version, the children were provided with two paper bags, one for themselves and another for an unknown child from their same city. The children were also provided with two paper stars, the experimental currency, and informed that the stars could later be exchanged for toys in four categories of visibly increasing objective value so that more stars would allow them to select toys from higher categories. The children were then asked to choose how to distribute the stars between the available bags (i.e., to place both stars in one bag or one star in each bag). In the second version, the children were provided with two paper bags, one for themselves and another for an unknown child from a different German town, and six stars. They were asked to divide the stars between the two bags. In the third version, the second bag was for an unknown child in Africa, and again the children were asked to divide the six stars between the two bags. The average share of stars given to the other child was used to operationalize altruism. Trust was assessed with a 3-item children-report measure adapted from the German Social–Economic Panel Study. Each statement (e.g., “One can trust other people.”) was read out loud by the interviewer, and the child was asked to indicate agreement on a five-point Likert response scale ranging from “totally disagree” to “totally agree.” Other regarding behavior was assessed using the five-item, parent-report, Prosocial Scale of the Strengths and Difficulties Questionnaire. Sample items included “My child is considerate of other people’s feelings” and “My child is helpful if someone is hurt, upset, or feeling ill.”
Interviews were conducted in central locations in Bonn and Cologne (waves 1 and 2) and at the participants’ homes (wave 3) by trained interviewers hired by a professional surveying company. Interviewers were blinded to the purpose of the study and group assignments. During the experiment, the accompanying parent completed a survey to provide information on the socioeconomic background of the family, the personality and attitudes of the child, the personality of the parent, and details on how the parent spent time with the child, including joint activities.
Intent-to-treat analysis was used to determine the differences in prosociality between children in the intervention group and those in the control group (i.e., low SES). Additional analyses compared prosociality between children in the intervention group and those in the high-SES group and between children in the low- and high-SES groups. There were no baseline differences in prosociality between children in the intervention and control groups. Furthermore, loss to follow up (30 percent of children in the intervention group, 30 percent in the control group, and 20 percent in the high-SES group were lost to follow up at wave 3) was unrelated to baseline levels of prosociality or group assignment.
