Zusammenfassung (übersetzt):
In this study we analyze some methods that can be used to perform grouping of unit that make up a public service delivery system (e.g., schools, health centers and police stations). The methods discussed here are: i) one based on some spatial partition of the units (for example, grouping by administrative region of a city); ii) what we call ordered indicator, in which the system units are grouped according to some partition of the distribution (e.g., deciles) of the weighted average of the variables of interest of the units; and iii) clusters that use a distance measure across these multiple variables to group the units of the system so that there is a high homogeneity of the units within the groups and, at the same time, a high heterogeneity among the groups. An empirical exercise using information from the municipal public school system of Rio de Janeiro is presented in the text for comparison of the methods in focus.