Paloyn, A. R., Rogan, S., & Siminski, P. (2016). The causal effects of the Peer Assisted Study Sessions (PASS) on educational outcomes. Retrieved from https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0ahUKEwjwoqG7uazLAhVGuIMKHXJuD4UQFggiMAA&url=http%3A%2F%2Fwww.uow.edu.au%2F~siminski%2FFinal%2520Report%2520-%2520Public%2520Copy.pdf&usg=AFQjCNH96hWntRMHTwkOanTB-n6YmCtEdg&sig2=MbSrUzNY-TR6Ba3PJsbWLA

This report summarises the results of a HEPPP-funded research project on the effects of the Peer Assisted Study Sessions (PASS) on educational outcomes. The study used a randomised encouragement design (RED), which avoids the potential problem of selection bias that pervades non-experimental evaluations.  Globally, this is the first large-scale experiment on the effectiveness of PASS or related Supplemental Instruction programmes.The study population consists of 6954 student subject observations from 14 first-year courses at the University of Wollongong in Australia in 2014 and 2015. Following the RED approach, a randomly selected sub-group was offered a large, near-cash incentive to participate in PASS. Whilst PASS participation is voluntary and unrestricted, participation was 0.47 sessions (19 percent) greater for the incentivised group compared to the non-incentivised group. This inducement effect is larger for students from low-SES areas (0.89 sessions). But the overall inducement effect is smaller than anticipated, which limits the statistical power of the main analysis, especially for subgroups. We also varied the size of the incentive greatly between semesters, but this did not meaningfully change the size of the inducement effect. The design of effective incentives for student populations warrants further research. The experiment suggests that one hour of PASS improved grades by 0.065 standard deviations (1.26 marks on a raw 100-point scale), which is consistent with the non-experimental literature. However, this estimate is not statistically significant, reflecting limited statistical power. The estimated effect is largest and statistically significant for students in their first semester at university (0.153 standard deviations or almost 3 marks per hour of PASS). This particular sub-group analysis was not in our preanalysis plan, and so it should be treated as a suggestive–rather than a confirmatory –result. Nevertheless, it remains plausible given issues around transitioning into a university environment, including the more independent, self-directed study skills and time management required in tertiary study, as well as the need for structure and social support. We had intended to study heterogeneity of effects for a number of other subgroups (by socioeconomic status, rural and indigenous backgrounds, age, sex, domestic/international status, and high school grades), but this was not feasible because of limited statistical power.

To download the complete annotated bibliography of 1,100+ citations of postsecondary peer cooperative learning programs, click on the following link, http://z.umn.edu/peerbib