Analysis of the 2025 Bundestag elections. Part 4 of 4: Changes in the German political spectrum

This is the last out of four papers on the 2025 German federal elections continuing our analysis of the 2009, 2013, 2017 and 2021 Bundestag elections. First, we apply the model from [Tangian 2022b] to construct the 2025 German political spectrum understood as a contiguous party ordering, i.e., such that the neighboring parties have close policy profiles. For this purpose, we consider the parties that took part in the 2025 federal elections, define their policy profiles as 38-dimensional vectors of their Yes/No answers to 38 policy questions from the German voting advice application Wahl-O-Mat ('Support for Ukraine'?-Yes/No, 'General speed limit on motorways?'-Yes/No, etc.), and contiguously order them by means of Principal Component Analysis. The circular party ordering obtained is cut, resulting in a horseshoe-shaped left-right ideological axis with the far-left and far-right ends approaching each other. Among other things, the one-dimensionality of the political spectrum looks as a precondition for the voters' single-peaked preferences that guarantee the election consistency. Second, using similar data from the 2009, 2013, 2017 and 2021 German federal elections, we construct political spectra for these years as well and trace the changes. Since the set of contesting parties varies from one election to another, and the Wahl-O-Mat questions vary as well, we only dispose five party orderings with a relatively small core of 13 parties that participated in all five elections. To locate the five spectra on a common scale, we consider 60 parties that have ever participated in elections and order them basing on five spectra on subsets of 24, 29, 31, 37 and 28 parties, respectively. This is done in terms of collective choice: find a group preference on 60 alternatives given five individual preferences on five incomplete alternative subsets; so we adapt the Condorcet and Borda approaches. Then the five political spectra are stretched onto this unified party ordering by constrained least squares, adjusting the distances between the parties in each spectrum. All of these enable to adequately visualize party reshuffles in the political space. In particular, we see that, among the major German parties, the SPD fluctuates by far the most between left and right. This political inconsistency can deter voters, especially floating voters without a firm self-identification with a particular party, and may explain the SPD's failure in the 2025 elections, when the party received the historical minimum of 16.4% of the votes, having lost 9.3 percent points compared with the 2021 elections.