Mean rule has been popularly used to aggregate consumer ratings of online products. This study applied social choice theory to evaluate the Condorcet efficiency of the mean rule, and to investigate the effect of sample size (number of voters) on the agreement or disagreement between the mean and majority rules. The American National Election Survey data (1968) were used, where three candidates competed for the presidency, and the numerical thermometer scores were provided for each candidate. Random sampling data with varied sample sizes were drew from the survey, and then were aggregated according to the majority rule, the mean rule, and other social choice rules. The results show that the sample winner of the mean rule agrees with the sample majority winner very well; as sample size increases, the sample mean rule even converges faster to the correct population majority winner and ordering than does the sample majority rule. The implications for using aggregation rules for online product rating were also discussed.
展开▼
