No abstract

The effect of seller reputation on seller success in peer-to-peer online markets has been investigated in dozens of studies by means of the analysis of digital trace data. A recent meta-analysis synthesizing evidence from over a hundred studies corroborates that sellers with a better reputation sell more products at higher prices. However, the meta-analysis also shows that these reputation effects exhibit excess variation that cannot be attributed to sampling error. Moreover, there is still little consensus on how the size of a reputation effect should be interpreted and what might cause its variation. Here we use a meta-analytic model selection approach and multi-model inference on two subsets of 406 coefficient estimates to identify potential moderators of reputation effects. We identify contextual, product-related, and method-related moderators. Our results show that, among others, geographical region, product condition, sample size, and type of regression model have a bearing on the size of the reputation effect. The moderating effect of the geographical region suggests that reputation effects are substantially larger in the Chinese context than in the European or US contexts. The moderating effect of product condition—estimates based on new products are larger than estimates based on used products—is unexpected and worthwhile investigating further. The moderating effects of sample size and model type could be related to study quality. We do not find evidence for publication bias as a potential explanation for the effects of method-related moderators.