The current study also has limitations. First, this study deals with issues of social control in communities, and one of the challenges of this type of research is to distinguish individual-level effects from community-level effects. The reason for this is that what we study may be related to the structural characteristics of ecological units such as face blocks, census tracts, neighborhoods, and so on. Because of many factors such as social inequality, there may be a significant variation from neighborhood to neighborhood in 174 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY terms of the social, structural, and ecological characteristics. As a result, we may not be able to inform whether the findings are due simply to individual level traits (e.g., heightened sensitivity to disorder, support for authority) or to actual differences in neighborhood disorder and crime, when multilevel analysis is not employed. However, because our data set was not intended for multilevel analysis, there are no level II variables available. This limitation may have an impact on the interpretation and implication of the results. Second, community policing is the central concept in this study; however, it may not be adequately measured, due to the availability of variables in the data set. According to the literature, community policing seems to have at least the following characteristics: partnership with community, personalized policing, decentralized place, proactive policing, crime prevention, order restoration and maintenance, problem solving, interagency cooperation, unisolated patrol, permanent beats, and so on. The current study only covered partnership with community and crime prevention (may also implicitly include proactive policing and problem solving), a portion of the content domain of community policing. This may also present a problem for how to accurately interpret the results. For instance, it is hard to know whether an insignificant result reflects the truth or is simply due to the incomplete measurement. This would also work in the other way (i.e., a significant result may not fully confirm a hypothesized relationship). Third, the external validity of the study needs to be further tested because of the data limitations. (1) Because the data were collected in one metropolitan area in the United States, the results based on them may not be generalizable to the areas, whose demographic, socioeconomic, and ecological structures and characteristics are significantly different from those of Colorado Springs. (2) As typical of social science data, many variables used in the model are Likert-type variables. Ordinal level variables as manifest variables in a structural-equation model tend to bias correlations/covariances toward 0 and generate unreliable standard errors (Rigdon 2003). Therefore, the results should be interpreted with caution, even if our bootstrapping showed certain robustness of the results. (3) The sample size of the data (n = 710), though not too small, is at the lower end of what is acceptable. The small sample size may cause larger sample to sample variation and standard errors. As a result, the estimates may not be as stable as what would be with a large sample size. Finally, the current analysis mainly focuses on the roles of the police and citizens and a limited number of factors, when discussing the mechanisms of social control in the community. However, a more systematic discussion should include a larger context and more relevant parties in the social-control process such as the "Big 6" introduced by Trojanowicz.

What potential limitations were noted by the author/s? Did you see any other potential limitations that were not noted? Explain your answer.