Bar chart showing logistic regression models achieved up to 73% explanatory power (R²) in predicting Trump support in 2020 election

Economic Perceptions and Immigration Attitudes: Electoral Outcomes 2016-2020

Quantitative analysis using logistic regression on American National Election Studies (ANES) data covering 6,173 voter responses across 2016 and 2020 presidential elections. Models achieved McFadden's R² of 0.73, revealing an 11.62-point coefficient reversal in how economic perceptions affected Trump support, while immigration attitudes intensified 44% as a stable predictor (all p<.001).

This study examines how economic perceptions and immigration attitudes influenced Donald Trump's electoral support across the 2016 and 2020 presidential elections. Using logistic regression analysis on American National Election Studies (ANES) data covering 6,173 voters, the research reveals how incumbency fundamentally transformed the political meaning of economic dissatisfaction.

Key discovery: Economic concerns that helped Trump win as a challenger in 2016 (+5.03 coefficient) became his biggest liability as incumbent in 2020 (-6.59 coefficient)—an unprecedented 11.62-point reversal. Meanwhile, immigration attitudes intensified as a stable predictor of support, increasing 44% in effect size between elections.

Research Question

How did economic perceptions and immigration attitudes influence Trump's electoral support differently between 2016 and 2020?

Key Finding

Economic perceptions completely reversed between 2016 and 2020

Economic dissatisfaction completely reversed its effect between elections—from helping Trump as a challenger (+5.03 coefficient) to hurting him as incumbent (-6.59 coefficient). This 11.62-point swing provides empirical evidence for how incumbency transforms electoral dynamics.

Meanwhile, immigration concerns intensified as a stable predictor (+1.66 to +2.39), operating independently of Trump's performance in office.

Research Methodology

Dataset: American National Election Studies (ANES) Time Series

Sample: 6,173 respondents (2020), 2,750 respondents (2016)

Analysis: Logistic regression with Basic and Full model specifications

Software: R for statistical analysis, Python for visualization

Variables:

• Dependent: Binary vote choice (Trump vs. not Trump)

• Independent: Economic perceptions (4-point scale), Immigration attitudes (4-point scale)

• Controls: Democracy satisfaction, political ideology, political information

Key Findings

1. Economic Perceptions: The Incumbency Reversal

• 2016: Strong disapproval → +5.03 (p<.001) [Helped Trump]

• 2020: Strong disapproval → -6.59 (p<.001) [Hurt Trump]

• Swing: 11.62 points

2. Immigration Attitudes: Stable and Intensifying

• 2016: Extremely likely immigrants take jobs → +1.66 (p<.001)

• 2020: Extremely likely immigrants take jobs → +2.39 (p<.001)

• Change: +44% increase in effect size

3. Strong Model Performance

• 2016 Full Model: McFadden's R² = 0.58

• 2020 Full Model: McFadden's R² = 0.73

• All main effects: p < .001

Discussion & Implications

Immigration concerns intensified as a stable predictor across both elections

This research makes three important contributions to understanding electoral behavior:

First, the 11.62-point coefficient reversal provides rare empirical evidence of incumbency effects using within-candidate comparison. Most studies compare different candidates or use cross-sectional data, making causal inference difficult. By analyzing the same candidate in different positions, this design isolates how accountability transforms voter behavior—a methodological advantage that strengthens causal claims.

Second, the distinction between performance-dependent factors (economic perceptions) and identity-dependent factors (immigration attitudes) has implications for understanding political coalition stability. Performance-based support is conditional and vulnerable to governing challenges, while identity-based support proves more durable. This suggests candidates building coalitions on grievances face structural risks once they assume office and become accountable for outcomes.

Third, the increasing model fit from 2016 (R² = 0.58) to 2020 (R² = 0.73) reveals something about contemporary American politics. A smaller set of factors now explains more variance in voting decisions, suggesting increased polarization has made elections more predictable but potentially less deliberative. When economic perceptions and cultural attitudes overwhelm other considerations, crosscutting coalitions become harder to build.

The 2020 election occurred during the COVID-19 pandemic, which likely mediated economic perceptions. Future research should explicitly model crisis contexts to understand how extraordinary events interact with incumbency accountability. Additionally, demographic interactions—how these effects vary by education, race, or geography—would provide insight into coalition dynamics within Trump's electoral base.

This study demonstrates that rigorous quantitative methods can illuminate fundamental questions about democratic accountability, coalition formation, and how political meaning shifts with institutional position.

Technical Details

Sample Size: 6,173 respondents (2020), 2,750 respondents (2016)

Time Period: 2016 and 2020 U.S. presidential elections

Data Source: American National Election Studies (ANES) Time Series

Statistical Method: Logistic regression with maximum likelihood estimation

Software: R for statistical analysis, Python (Matplotlib) for visualization

Key Statistics:

• Economic reversal: +5.03 (2016) to -6.59 (2020), 11.62-point swing

• Immigration intensification: +1.66 (2016) to +2.39 (2020), 44% increase

• Model performance: R² = 0.58 (2016) to 0.73 (2020)

• All main effects: p < .001 (highly significant)

About This Research

Completed for POSC 201: Political Research Design at Chapman University, April 2024. Demonstrates advanced quantitative analysis capabilities and rigorous statistical methodology applied to electoral behavior research.