The party program database used in Chapter 3 consists of programmatic documents issued by the parties, and program declarations at party congresses. Several existing accounts of party stances have relied on the views of party supporters or the evaluations of experts to determine the parties' stances. However, supporter views may differ considerably from the parties' stances. In fact, they cannot be assumed to correspond to either the party's official stances or its addressed constituency. Using supporter stances to measure party views also makes it impossible to determine whether the parties are responsive. Therefore, I emphasize content analysis of actual party programs and statements, as this method allows us to examine what the parties' actual messages were.

All programmatic documents available for each party after 1989 were coded in their entirety, translated by the author from the original language. Paragraphs in the party programs were the unit of analysis. This approach loses some reliability and accuracy, in comparison with coding sentences. It also biases the results toward the null hypothesis (that no change occurs in the party programs), since it may gloss over more minute changes within the program. Nevertheless, coding the party programs by paragraph perhaps best reflects the intentions of their framers, since most of the party programs examined were divided into paragraphs, each with a specific policy statement or message. The paragraphs were coded both for content (within one of twenty categories) and for emphasis (on a scale of 1 to 4), where appropriate. If more than one firm statement was expressed within a paragraph, these ideas were also coded in an auxiliary column.