Under ideal circumstances, a sound longitudinal analysis of gang-problem localities would use units of 1 year, or possibly even 1 month, to record and analyze changes in prevalence trends. Unfortunately for the purposes of precise analysis, the use of smaller time units was not possible in the present study because the nature of the base data makes it difficult or impossible to ascertain the exact day, month, or year when youth gang problems emerged or reemerged in a given locality.

As a consequence, this study uses the decade as its major time unit and calculates incidence and trend data on a 10-year basis instead of using smaller time units. Three decades are used—the 1970's (1970 through 1979), the 1980's (1980 through 1989), and the 1990's (1990 through 1995). As noted in the second chapter, the first known citation of gang problems in a given locality (a "new" gang locality) is assigned to the decade during which it is reported, whatever the day, month, and year of that reporting. Thus, a locality that first reported gang problems in June 1980 and one that first reported gang problems in June 1989 are both tabulated as new gang localities in the 1980's.

One problem in using the decade as a time unit—a problem shared with many other studies—is that the major Federal census is conducted only once every 10 years. This means, for example, that a study that uses population data and is conducted near the end of the decade will face a high likelihood of encountering outdated figures. In the United States, a city with a given population in census year 1990 will almost certainly show a smaller or larger population in census year 2000. The Bureau of the Census conducts smaller scale intradecade surveys for some data, and the use of these and/or extrapolations based on earlier data is feasible in some instances. Such use was not feasible in the present case, and this study assigns to each decade the population figures published near the beginning of that decade.