Heterogeneity in Perceptions of National Economic Conditions

Raymond Duch, Harvey D. Palmer, Christopher J. AndersonAmerican 2000 · DOI

TL;DR

This paper challenges the common assumption that individual-level "noise" in public opinion about the economy cancels out when aggregated, leading to accurate collective measures. Instead, it demonstrates that systematic biases stemming from individual differences in political attitudes, personal experiences, and information levels distort aggregate economic perceptions, impacting our understanding of economic voting and government accountability.

What Problem Does This Paper Address?

A long-standing debate in political science concerns how voters perceive national economic conditions and how these perceptions influence their voting behavior. Macro-level studies often show a strong link between economic indicators and election outcomes, suggesting voters are rational and informed. However, micro-level studies using individual survey data present a more mixed picture, with many voters appearing uninformed or influenced by factors other than objective economic performance.

A common explanation for this discrepancy is that individual survey responses contain a lot of "noise" – random errors or non-attitudes – which, when aggregated, cancels out, leaving a clear signal of public opinion. This paper questions this assumption, arguing that the "noise" might not be random but systematic, meaning aggregation doesn't necessarily purify the signal. If these systematic biases exist, then aggregate measures of public opinion, often used in macro-level analyses, are not as objective as assumed and can lead to misleading conclusions about economic voting and government accountability.

Methods

The authors employ a two-stage statistical approach:

  1. Individual-Level Analysis (Ordered Probit Models): They first build ordered probit models to explain individual voters' retrospective and prospective evaluations of the national economy. These models incorporate various factors hypothesized to cause systematic variation (heterogeneity) in perceptions, including:
  2. Information: Political sophistication and media usage.
  3. Political Attitudes: Party identification and its interaction with sophistication.
  4. Personal Financial Experience: Individual financial situation and its interaction with sophistication.
  5. Group Self-Interest: Demographic characteristics like income, occupation, age, race, and gender.

    The goal here is to determine if these factors significantly influence how individuals perceive the economy, thus demonstrating that the "noise" is systematic.

  6. Aggregate-Level Analysis (Purging and Comparison): In the second stage, they use the insights from the individual-level models to "purge" the aggregate measures of public opinion. This involves creating hypothetical aggregate measures where the influence of specific subjective factors (political attitudes, information, personal financial experience) is removed. By comparing these "purged" aggregate measures to the "raw" aggregate measures (what is typically used in research), they can quantify the distortion caused by systematic individual-level biases.

Datasets

The primary dataset used is the American National Election Study (ANES), specifically data from the 1992 presidential election for the initial individual-level analysis. For the aggregate-level analysis, they utilize ANES survey data from 1980 to 1998, including pilot studies from 1991, 1993, and 1995.

Key Insights and Ideas

  • "Noise" Isn't Always Random: The core idea is that individual-level variations in economic perceptions aren't just random errors. They are systematically influenced by a voter's political leanings, personal financial situation, and how much information they have.
  • Aggregation Doesn't Magically Fix It: The paper strongly argues against the idea that simply averaging individual opinions cancels out all the "noise." If the noise is systematic, aggregation can actually preserve and even amplify these biases at the collective level.
  • Subjectivity Matters More Than We Think: Voters don't just look at objective economic data. Their political identity, personal experiences (even if not directly tied to national policy), and how they consume media all shape their view of the economy.
  • Time-Varying Distortions: The degree and direction of these biases aren't constant. They fluctuate from year to year, making it difficult to apply a one-size-fits-all correction to aggregate economic data.
  • Implications for Accountability: If voters' perceptions are systematically biased, then the link between a government's actual economic performance and public approval (and thus, accountability) becomes muddied. It's harder to tell if voters are rewarding/punishing the government for its policies or for other subjective reasons.

Key Results

  • Significant Predictors: Political attitudes (party identification) and personal financial experience were found to be the most influential factors driving systematic variation in economic perceptions. Information levels also played a significant role.
  • Distortion is Real: The comparison between raw and purged aggregate measures showed substantial differences, indicating that aggregation does not eliminate these individual-level distortions.
  • Magnitude Varies: The "total distortion" (combined effect of all subjective factors) varied considerably over time, sometimes making aggregate perceptions more positive and sometimes more negative than they would be if purely objective.
  • Information Depresses Optimism: In many instances, lower levels of political sophistication and media usage led to more pessimistic evaluations of the national economy.
  • Political Attitudes Inflate Optimism: Conversely, partisan attachments and positive personal financial experiences tended to inflate aggregate evaluations of the economy.

Limitations and Future Work

  • Causality: While the study identifies strong correlations, establishing definitive causality between subjective factors and economic perceptions is complex. The authors acknowledge that reciprocal relationships might exist (e.g., economic perceptions influencing party identification), though their auxiliary analyses suggest their core findings hold.
  • Measurement of "Objective" Economy: The paper focuses on subjective perceptions. Defining and measuring the "objective" economic conditions against which these perceptions are compared is a separate, complex task.
  • Generalizability: While the study uses a broad time series (1980-1998), the specific findings might be sensitive to the particular economic and political contexts of those years.
  • Future Work: The authors suggest that future research needs to account for this time-varying measurement bias in aggregate-level models of economic voting. Ignoring it can lead to misleading statistical inferences and a potentially inaccurate understanding of government accountability. They also highlight the need to understand how these biases interact with other factors, like media exposure, to shape public opinion.

Summary updated

Abstract

Raymond M. Duch, Harvey D. Palmer, Christopher J. Anderson, Heterogeneity in Perceptions of National Economic Conditions, American Journal of Political Science, Vol. 44, No. 4 (Oct., 2000), pp. 635-652
BibTeX
@inproceedings{duch2000heterogeneity,
  author = {Raymond Duch and Harvey D. Palmer and Christopher J. Anderson},
  title = {{Heterogeneity in Perceptions of National Economic Conditions}},
  booktitle = {American},
  year = {2000},
  doi = {10.2307/2669272},
}
Figure