The sample states slightly overrepresent both opposition-party registration of candidates and postelectoral conflicts, but in most cases, the sample does not overrepresent the universe of all of Mexico's municipalities by more than 5 percent (Table A.1). Postelectoral conflicts were coded from a wide range of media sources, including those listed in Appendix B.1 with regard to coding of the independent variable “localized social conflicts,” as well as the national and local newspapers cited in the bibliography. Multiple opposition-party mobilizations in one municipality were rare, but when they occurred, in every case I entered only the mobilization by the non-PRI contender (PAN or PRD), as there were not enough PRI or third opposition-party mobilizations to allow for the model to retain statistical significance when they were included as separate dependent variable categories. In the dozen or so cases with mobilizations by both the PAN and the PRD, the higher vote-getter among the runner-up parties was credited with the conflict, as I considered that party to be the main postelectoral contender (and usually there was a large margin between second- and third-place finishers). Just as electoral contention was either PRI-PAN or PRI-PRD but almost never PAN-PRI-PRD (at least not until the late 1990s), postelectoral contention also followed this pattern during the period under study. The selection of the ten sample states and specification of the time periods is addressed in Chapter 5.