Making Nostalgia Great Again: How Party Platforms Sell the Past

Abstract

This study analyzes nostalgic rhetoric in U.S. party platforms from 1972-2024 using a fine-tuned DistilBERT transformer model applied to more than 68,000 sentences. Two hypotheses were tested: that Republicans use more nostalgic rhetoric than Democrats, and that challengers use more nostalgic rhetoric than incumbents. Results provide modest support for a partisan difference. Both a pattern-based and statistical test support Republicans using more nostalgic rhetoric in party platforms, however, the strength of the results are limited. Little evidence is found in support of the challenger hypothesis. Methodologically, this research adds to the growing literature demonstrating the value of transformer-based models for large scale text analysis in Political Science, achieving accuracy comparable to human coding. By combining modern computational methods with theories of political communication, this study contributes to the understanding of how nostalgia manifests in politics.