The featured study collected data on traffic stops in a midwestern community during a 1-year period. The police agency voluntarily initiated a data-collection project for the purpose of gaining a better understanding of officer decisionmaking during traffic enforcement encounters. A new departmental policy required all officers to complete a short form for each vehicle stop. The data pertained to the date, time, and location of the stop, the reason for the stop, driver demographics, actions taken during a stop, and the traffic stop disposition. Data were collected on 30,514 stops from February 12, 2001, to February 11, 2002. Assessed in conjunction with other data collection efforts in similar studies, as well as the research experiences of the authors of this paper and other scholars, the current data were used to illustrate the need for methodological revisions in future traffic-stop studies. Most traffic stop studies have relied on officer-reported information, either in the form of specific-purpose data or existing data sources. There are issues of reliability and validity associated with this method of data collection. Based on such data, most of the research that has used officer reported data have failed to establish conclusively that racial profiling was a real and significant social issue. The current study found, however, some race-based disparities in officer traffic stops. This finding was mediated by the equally important roles of gender and age in shaping police decisionmaking. This study, as well as other research, has lacked sufficiently rich data to make clear determinations that observed disparities were the product of race-based impropriety in the use of police discretion. This paper advises that given the complexity associated with police decisionmaking in traffic enforcement encounters, data analysis procedures should simultaneously account for multiple variables. Multivariate techniques provide a more complete understanding of the many complex factors that influence officer conduct. Multivariate modeling also provides the opportunity for the researcher to understand intervening and spurious relationships between variables in an effort to derive the best possible explanatory model. In addition to using multivariate analytical techniques, other data sources might be used to enhance explanatory models, such as census data and community-level data on demographics, poverty, unemployment, disorder, population density, etc. Attitudes of police officers concerning crime and demographics may also contribute to officer decisionmaking. 10 notes and 51 references