In the most similar systems research design (a comparative study), research sites are selected to hold constant all important explanatory variables with the exception of the single variable whose effect is being analyzed. In multistrata similar systems research designs, several sets of comparable neighborhoods were selected to ensure variation in neighborhood wealth and racial composition. A close match to the distribution of neighborhood types served by police across the country is thus obtained. This gives the studies greater face validity but requires more complex analysis schemes to compensate for the potential confounding effects of multiple neighborhood types. Examples of these and other studies are cited. Some researchers have attacked citizen surveys in performance assessment, stating, for example, that citizens' reported satisfactions with police services and with parks and recreation services in Los Angeles were only weakly related to objective measures of those services. Moreover, citizens have been accused of paying little attention to services and failing to perceive differences in service quality. Other comments on citizens' perceptions of police services are noted. Endnotes, tables, and about 35 references are included.