Questions for reflection and review 2

Lord, are you there?

Describe the common findings in past research on elite perceptual accuracy. What’s the “good news” and the “bad news” here?

Good news

• Politicians are not completely clueless. Their perceptions of constituents’ ideology are positively correlated with reality: more conservative places are generally seen as more conservative, and more liberal places as more liberal.

• On average, they have the right ordering of municipalities and issues, even if their levels are off.

Bad news

• At the aggregate level they show clear conservative overestimation: they systematically think their publics are more right-wing / more conservative than survey-based “ground truth” indicates (the regression lines sitting above the 45° line in the “General Municipal Ideology” and “Conjoint Profiles” plots).

• They also exhibit strong projection—they assume that citizens’ attitudes are similar to their own policy views and to the views of people who resemble them.

• These biases can distort representation even when politicians are trying to be responsive.

• Describe the study involving individual rather than general estimates of constituents’ preferences. How did this study work? What did it find in terms of the persistence of the major forms of error?

How the study worked

• Instead of asking for a single estimate of “where is your municipality on a 0–10 left–right scale?”, politicians evaluated many individual conjoint profiles.

• Each profile randomly varied characteristics (party, age, gender, immigrant status, housing, education, etc.) and policy or ideological positions.

• For each profile, politicians were asked how much public support that specific profile or its positions would receive.

• These predictions were compared to ground-truth public opinion gathered from citizen surveys using the same conjoint design.

Key findings about persistence of error

• The big, aggregate conservative overestimation seen for general municipal ideology does not carry over to the individual profile level; if anything, the figures suggest a slight move in the opposite direction.

• Tests for privilege bias (systematically downgrading minority / less-privileged profiles) show very weak evidence: most coefficients for race, immigrant status, housing, etc. are small and not statistically significant.

• In contrast, projection effects remain strong, especially for policy attitudes: politicians consistently think the public is more supportive of positions held by people like themselves (same party, age, duration in office, etc.).

• Describe the study involving tokens placement rather than mean estimates of constituents’ preferences. How did this study work? What did it find in terms of the persistence of the major forms of error?

• Politicians saw a 0–10 left–right scale and were asked to allocate 20 tokens, each representing 5% of residents, across the bins.

• This produced a full perceived distribution of citizen ideology, not just a single point.

• Researchers compared:

  • The point estimate (where the politician had earlier placed “average citizens”), and

  • The implied average from the token distribution
    to an MRP-based “ground truth” distribution from citizen surveys.

• They also randomly varied question order: some politicians did the point estimate first, others did the distribution task first.

When politicians placed tokens across a 0–10 scale, they were much more accurate, with much less conservative bias and weaker projection. Doing distributions first improved later point estimates.

• Describe “midpoint hedging” — what do the figures suggest about the prevalence of “midpoint hedging” among politicians?

Politicians pull their estimates toward the middle of the scale when uncertain, causing overestimation of minority support and underestimation of majority support. Figures show this is very common

• What is a “domain-general” mechanism of misperception? And how is it different from a “domain-specific” mechanism?

A universal cognitive bias (like midpoint hedging) that appears across all issues.
Domain-specific biases only appear on particular topics (e.g., immigration, crime).
Slides show misperceptions are mostly domain-general.
A key shaped U-shaped error pattern

• Summarize our current knowledge of elite perceptual accuracy. What findings are most robust? Which findings in the past literature are most questionable?

Robust: overall correlation with reality, conservative bias in point estimates, strong projection, and U-shaped error (underestimate majorities, overestimate minorities).
Questionable: strong privilege bias and claims that politicians are wildly inaccurate.

• Describe the implications of recent studies of elite perceptual accuracy for our more general thinking about the two “pathways to representation”

Misperceptions can distort responsiveness, since politicians often misread public support. But because errors are cognitive and predictable, better information improves accuracy. The selection pathway (electing like-minded reps) becomes more important.

Describe the two “pathways to representation.” What are the differences between the two pathways?

1. The congruence pathway. Politicians share the views of their constituents and act on those views in their policy making

2. The knowledge pathway. Politicians know what their constituents want and act on those preferences in their policy making

What is “congruence”?

the degree to which politicians’ positions match the public’s positions

Examples of congruence

On issues like climate change, active transportation, and living wage, politicians’ support increases as public support rises → high congruence (steep solid line).

• On COVID-19 freedom policies or property taxes, politicians’ views do not track changes in public support → low congruence (flat line)

What is “perceptual accuracy”?

how accurately politicians know what the public wants

Examples of perceptual accuracy

• On issues like federal transit funds or handgun bans, knowledge bars (teal) are high → politicians correctly estimate public support.

• However, even with high knowledge, congruence can be low (e.g., local immigration, business tax breaks), meaning knowledge ≠ representation.

What is the relationship between politicians’ and citizens’ ideology, as reported in Ideology in Canadian Municipal Politics?

• Politicians tend to be slightly to the left of the average constituent in their municipality.

• There is a strong positive slope between citizen ideology and politician ideology → more progressive municipalities elect more progressive politicians.

• Ideology is the strongest predictor of congruence: the closer a constituent’s views are to the politician’s, the more likely they align on policy

When we think about representation, what’s the difference between interpreting the “slope” of the relationship and interpreting the “intercept” of the relationship?

Slope:

• Shows responsiveness.

• A steep slope indicates that politicians’ support increases as public support increases.

• High slope = politicians respond to public opinion.

Intercept:

• Shows baseline position.

• Indicates where politicians stand when public support is at zero or low.

• Helps identify directional bias (e.g., politicians may be systematically more left-leaning or more supportive overall than their electorate).

Describe the data and methods used by Lucas et al. to measure policy preferences in Canadian municipalities

• Lucas et al. use MRP (Multilevel Regression and Poststratification) to estimate public support for a wide range of municipal policies (visible in the congruence figure).

• Individual survey responses (binary 0/1 support) are plotted against MRP-estimated public support for those policies.

• Politicians’ opinions are collected through surveys of municipal councillors.

• They then compare politicians’ responses to MRP-based public preferences to calculate congruence and knowledge.

What is Lucas et al.’s argument about the two pathways to representation?

• Both pathways—congruence and knowledge—play a role in municipal representation.

• Politicians often have high perceptual accuracy (they correctly understand what citizens want), but this does not always produce congruence.

• Representation varies by issue:

  • On some issues, congruence is high because politicians share citizens’ views.

  • On others, they know public preferences but do not act on them.

• Therefore, substantive representation in municipal politics is conditional and uneven, not automatic

Explain how to interpret the results in the congruence_by_issue_facet.pdf figure included with this week’s slides

Interpretation rules:

• Vertical axis: politician’s probability of support.

• Horizontal axis: public support (MRP estimate).

• Solid line: actual politician responsiveness.

• Dotted line: perfect congruence (ideal responsiveness).

Interpretation:

• Steep solid line (close to dotted line):
  Politicians respond strongly to public opinion → high congruence (e.g., climate change, living wage).

• Flat solid line:
  Politicians’ positions do not change with public support → low congruence (e.g., COVID-19 freedom, property taxes).

• Points scattered around the line show individual citizens’ opinions for that issue

When it comes to variation in congruence, what factors seem to predict higher or lower congruence?

Factors predicting higher congruence:

• Ideological proximity (strongest predictor).

• Age (older constituents align more).

• Income (higher-income constituents align more).

• White respondents show slightly higher congruence.

Factors not predicting congruence:

• Housing tenure (homeowner vs renter).

• Gender.

• Education level.

• Time living in the municipality.

Issue-level patterns:

• Issues with clear partisan or ideological structure show higher congruence