Public Opinion and Voting Behaviour
1.Opinions and Belief Systems: Readings
A.The Importance of Knowing “What Goes with What”: Reinterpreting the Evidence on Policy Attitude
Stability, Freeder et al. (2018)
The study focuses on how stable citizens’ policy attitudes really are, and what explains the instability that
surveys show.
Research question
• How many citizens hold meaningful and stable policy attitudes rather than random or superficial
ones?
• Is the instability of survey responses mainly due to measurement error (bad questions, random
noise) or to genuine instability in people’s opinions?
• Does knowing “what goes with what” (which issue positions belong to which party/ideology) explain
who has stable attitudes and who does not?
Main argument
Earlier work split between 2 views:
1. Converse and Zaller: many people lack real attitudes; instability reflects real opinion change or
random sampling from conflicting considerations
2. Achen, Ansolabehere et al.: people do have stable attitudes, but survey measures are noisy; once
you correct for measurement error (e.g. multi-item scales) attitudes look stable
The authors argue this debate missed a key factor: placement knowledge, i.e. knowing which party or
candidate is more liberal or conservative on specific issues (“what goes with what”).
People with this placement knowledge tend to align their own views with their party or ideology (by
following party cues or choosing parties that match their issue views), which produces stable attitudes over
time. Therefore, much of the instability in survey data is not just measurement error; it comes from people
who lack this knowledge and thus have weakly anchored, genuinely unstable opinions.
,Data and method
• Panel surveys from the US and UK: ANES 1972–76 and 1992–96, British Election Studies 1992–97
and 1997–2001, Patterson 1976 panel, plus a 2015–16 SSI panel
• Placement knowledge: respondent correctly ranks parties/candidates on the liberal–conservative
side of an issue in both panel waves
• Opinion stability: over-time correlations of multi-item issue scales; also “crystallized attitudes”
(staying on same side of the scale) and absolute change in position
• General political knowledge: standard factual knowledge scales (Zaller-style).
Core empirical findings
1. Multi-item scales and general knowledge help only moderately
• Adding more items into an issue scale (e.g. 1→4 items) modestly increases stability (correlations rise
by about 0.18 on four-item economic scales), but responses remain far from perfectly stable
• Higher general political knowledge also modestly increases stability (about 0.3 difference in
correlations between lowest and highest quintiles), but even the most knowledgeable show only
moderate stability.
2. Placement knowledge strongly predicts stability
• In the 1972–76 ANES economic scale, respondents who correctly placed candidates on all
four economic issues had very high stability (correlation ≈ 0.88 between waves); those who got all
four wrong had low stability (≈ 0.36), even using multi-item scales
• Across multiple US and UK panels, moving from zero correct placements to full placement
knowledge on four-item scales raises stability by about 0.48 on average (e.g. from ~0.34 to ~0.82),
far larger than the effects of more items or more general political knowledge
• This pattern also appears for individual issues (abortion, busing, marijuana, etc.): people who can
correctly place parties/candidates show higher test-retest correlations than those who cannot,
although “easy issues” show some stability even among the ignorant.
3. Instability is mainly in the opinions, not just the measurement
• Even after using multi-item scales (which should cut measurement error), respondents with low
placement knowledge still show low stability
• This breaks the “observational equivalence” between measurement error and real instability: the
instability is concentrated among people who lack “what goes with what” knowledge, indicating that
many citizens’ attitudes themselves are weak and shifting.
4. Placement knowledge matters more than general political knowledge
• In pooled regression models, an extra correct item of general knowledge modestly increases the
probability of crystallized attitudes, but an extra correct placement item is about 2.5–3 times more
influential on stability for multi-item scales and 7–9 times more influential for single items
• Within-respondent models (fixed effects) show that for the same person, issues on which they have
correct placement knowledge are more stable than issues where they lack it, ruling out many
alternative explanations (like general attentiveness).
5. Stability is strongest when people agree with their own party
• The stabilizing effect of placement knowledge is much larger when people both know their
party/candidate’s position and agree with it, and keep the same party/candidate over time
, • If respondents know where their party stands but disagree with it, their attitudes are much less
stable on most issues (with some exceptions like abortion, busing, marijuana where strong “easy
issue” attitudes can be stable even in disagreement).
6. How many citizens have stable attitudes?
• Using ANES 1972–2012 surveys, somewhat less than half of the US public, on average, can correctly
place both parties and candidates on policy questions; in the 1992–96 panel, only 19% correctly
place parties and candidates on all five items
• Combining this with the stability data, the authors estimate that in the US roughly 25–50% of citizens
have moderately stable attitudes (correlation ≥ 0.70) on key policy scales, depending on the panel
and issue set
• In Britain, the share with stable attitudes is higher, about 60% in the mid-1990s but falling to about
40% by the late 1990s as party differences shrink.
Takeaways
• Policy attitude instability in survey data is not mainly a measurement artifact; it is largely genuine
instability among citizens who do not know “what goes with what” – which issue positions go with
which party or ideological camp
• Stable attitudes are concentrated among people who know party/candidate positions and whose
own views line up with their preferred party, and this group is only about 20–40% of Americans on
many issues (somewhat more in Britain)
• For democratic accountability, this means that a sizable minority (but not a majority) of citizens have
well-anchored, stable policy views, while many others have weak and fluctuating opinions that can
shift over time.
B.Are Cultural and Economic Conservatism Positively Correlated? A Large-Scale Cross-National Test, Malka
et al. (2019)
The article asks whether cultural and economic conservatism usually “go together” in mass publics
worldwide and argues that they often do not, finding instead a common pattern where cultural right views
combine with economic left views (and vice versa).
Research question
• Are cultural and economic conservatism typically positively correlated within national mass publics
(i.e., do right-wing cultural attitudes go with right-wing economic attitudes)?
• How does this relationship vary across countries (e.g. post-Communist vs other, low vs high
development, traditional vs progressive) and across individuals (especially by political engagement)?
• Do psychological and social factors like needs for security and certainty and social class help explain
why cultural and economic attitudes sometimes align and sometimes oppose each other?
Main argument
• Standard political science sees little natural psychological reason for people to line up consistently
left or right across issues; when they do, it is usually because elites and media package issues into
right–left bundles that politically engaged citizens adopt.
• Political psychology’s “rigidity of the right” model instead claims a common psychological basis—
needs for security and certainty—should normally push people toward both cultural and economic
conservatism, so the two should usually be positively correlated.
• The authors argue that, once you look globally, neither story fully holds: needs for security and
certainty and social class often push people toward cultural right but economic left (protection in
, both domains), while elite right–left discourse pushes toward cultural right plus economic right, and
the balance between these forces varies strongly by context and engagement.
• As a result, attitude structure is often better described as a protection–freedom dimension (cultural
and economic protection vs cultural and economic freedom) than a single right–left dimension, and
the familiar right–left alignment is a special, context-dependent case.
Data, concepts, and methods
• Data: 229 national samples from the World Values Survey, covering 99 countries on all inhabited
continents, surveyed between 1989 and 2014
• Cultural attitudes:
- Sexual morality (index of abortion and homosexuality items)
- Immigration attitudes
- Women’s role in the workforce.
• Economic attitudes:
- Social welfare (index of views on income inequality and government responsibility for providing
for people)
- Preference for private vs government ownership of business/industry
• Political engagement: composite of political interest and the importance of politics in one’s life
• Needs for security and certainty: value-based scale (tradition, security, conformity vs self-direction,
stimulation) from the Schwartz value survey; analysed for 2 later waves
• Country-level variables: post-Communist status, Human Development Index (development), and
national traditionalism (average sexual morality traditionalism)
• Methods:
- Within-country bivariate correlations between cultural and economic attitude measures
- Multilevel (three-level) random-coefficient regressions: individuals nested in survey-years nested
in countries, modelling economic attitudes from cultural attitudes, demographics, engagement,
and country-level moderators; plus cross-level interaction tests
- Additional multilevel regressions by wave to see how needs for security and certainty and social
class relate separately to cultural vs economic attitudes.
Core empirical findings
1. Globally, cultural and economic conservatism are not usually positively correlated
• Across 6 cultural–economic pairings (3 cultural × 2 economic measures), the average within-country
correlation is small and negative in each case.
• For each pairing, significantly negative correlations across countries are more frequent than
significantly positive ones; for example, immigration conservatism vs business-ownership
conservatism is significantly negative in about 58% of countries and significantly positive in only
about 7%.
• Multilevel regressions confirm: pooled slopes relating cultural to economic conservatism are
generally near zero or small and negative; there is no evidence of a typical positive relationship in
the world as a whole.
2. Protection–freedom attitude organization is widespread
• It is more common for cultural right attitudes (e.g. traditional sexual morality, anti-immigration,
traditional gender roles) to go with economic left attitudes (support for welfare, more government
ownership) and for cultural left attitudes to go with economic right attitudes.
1.Opinions and Belief Systems: Readings
A.The Importance of Knowing “What Goes with What”: Reinterpreting the Evidence on Policy Attitude
Stability, Freeder et al. (2018)
The study focuses on how stable citizens’ policy attitudes really are, and what explains the instability that
surveys show.
Research question
• How many citizens hold meaningful and stable policy attitudes rather than random or superficial
ones?
• Is the instability of survey responses mainly due to measurement error (bad questions, random
noise) or to genuine instability in people’s opinions?
• Does knowing “what goes with what” (which issue positions belong to which party/ideology) explain
who has stable attitudes and who does not?
Main argument
Earlier work split between 2 views:
1. Converse and Zaller: many people lack real attitudes; instability reflects real opinion change or
random sampling from conflicting considerations
2. Achen, Ansolabehere et al.: people do have stable attitudes, but survey measures are noisy; once
you correct for measurement error (e.g. multi-item scales) attitudes look stable
The authors argue this debate missed a key factor: placement knowledge, i.e. knowing which party or
candidate is more liberal or conservative on specific issues (“what goes with what”).
People with this placement knowledge tend to align their own views with their party or ideology (by
following party cues or choosing parties that match their issue views), which produces stable attitudes over
time. Therefore, much of the instability in survey data is not just measurement error; it comes from people
who lack this knowledge and thus have weakly anchored, genuinely unstable opinions.
,Data and method
• Panel surveys from the US and UK: ANES 1972–76 and 1992–96, British Election Studies 1992–97
and 1997–2001, Patterson 1976 panel, plus a 2015–16 SSI panel
• Placement knowledge: respondent correctly ranks parties/candidates on the liberal–conservative
side of an issue in both panel waves
• Opinion stability: over-time correlations of multi-item issue scales; also “crystallized attitudes”
(staying on same side of the scale) and absolute change in position
• General political knowledge: standard factual knowledge scales (Zaller-style).
Core empirical findings
1. Multi-item scales and general knowledge help only moderately
• Adding more items into an issue scale (e.g. 1→4 items) modestly increases stability (correlations rise
by about 0.18 on four-item economic scales), but responses remain far from perfectly stable
• Higher general political knowledge also modestly increases stability (about 0.3 difference in
correlations between lowest and highest quintiles), but even the most knowledgeable show only
moderate stability.
2. Placement knowledge strongly predicts stability
• In the 1972–76 ANES economic scale, respondents who correctly placed candidates on all
four economic issues had very high stability (correlation ≈ 0.88 between waves); those who got all
four wrong had low stability (≈ 0.36), even using multi-item scales
• Across multiple US and UK panels, moving from zero correct placements to full placement
knowledge on four-item scales raises stability by about 0.48 on average (e.g. from ~0.34 to ~0.82),
far larger than the effects of more items or more general political knowledge
• This pattern also appears for individual issues (abortion, busing, marijuana, etc.): people who can
correctly place parties/candidates show higher test-retest correlations than those who cannot,
although “easy issues” show some stability even among the ignorant.
3. Instability is mainly in the opinions, not just the measurement
• Even after using multi-item scales (which should cut measurement error), respondents with low
placement knowledge still show low stability
• This breaks the “observational equivalence” between measurement error and real instability: the
instability is concentrated among people who lack “what goes with what” knowledge, indicating that
many citizens’ attitudes themselves are weak and shifting.
4. Placement knowledge matters more than general political knowledge
• In pooled regression models, an extra correct item of general knowledge modestly increases the
probability of crystallized attitudes, but an extra correct placement item is about 2.5–3 times more
influential on stability for multi-item scales and 7–9 times more influential for single items
• Within-respondent models (fixed effects) show that for the same person, issues on which they have
correct placement knowledge are more stable than issues where they lack it, ruling out many
alternative explanations (like general attentiveness).
5. Stability is strongest when people agree with their own party
• The stabilizing effect of placement knowledge is much larger when people both know their
party/candidate’s position and agree with it, and keep the same party/candidate over time
, • If respondents know where their party stands but disagree with it, their attitudes are much less
stable on most issues (with some exceptions like abortion, busing, marijuana where strong “easy
issue” attitudes can be stable even in disagreement).
6. How many citizens have stable attitudes?
• Using ANES 1972–2012 surveys, somewhat less than half of the US public, on average, can correctly
place both parties and candidates on policy questions; in the 1992–96 panel, only 19% correctly
place parties and candidates on all five items
• Combining this with the stability data, the authors estimate that in the US roughly 25–50% of citizens
have moderately stable attitudes (correlation ≥ 0.70) on key policy scales, depending on the panel
and issue set
• In Britain, the share with stable attitudes is higher, about 60% in the mid-1990s but falling to about
40% by the late 1990s as party differences shrink.
Takeaways
• Policy attitude instability in survey data is not mainly a measurement artifact; it is largely genuine
instability among citizens who do not know “what goes with what” – which issue positions go with
which party or ideological camp
• Stable attitudes are concentrated among people who know party/candidate positions and whose
own views line up with their preferred party, and this group is only about 20–40% of Americans on
many issues (somewhat more in Britain)
• For democratic accountability, this means that a sizable minority (but not a majority) of citizens have
well-anchored, stable policy views, while many others have weak and fluctuating opinions that can
shift over time.
B.Are Cultural and Economic Conservatism Positively Correlated? A Large-Scale Cross-National Test, Malka
et al. (2019)
The article asks whether cultural and economic conservatism usually “go together” in mass publics
worldwide and argues that they often do not, finding instead a common pattern where cultural right views
combine with economic left views (and vice versa).
Research question
• Are cultural and economic conservatism typically positively correlated within national mass publics
(i.e., do right-wing cultural attitudes go with right-wing economic attitudes)?
• How does this relationship vary across countries (e.g. post-Communist vs other, low vs high
development, traditional vs progressive) and across individuals (especially by political engagement)?
• Do psychological and social factors like needs for security and certainty and social class help explain
why cultural and economic attitudes sometimes align and sometimes oppose each other?
Main argument
• Standard political science sees little natural psychological reason for people to line up consistently
left or right across issues; when they do, it is usually because elites and media package issues into
right–left bundles that politically engaged citizens adopt.
• Political psychology’s “rigidity of the right” model instead claims a common psychological basis—
needs for security and certainty—should normally push people toward both cultural and economic
conservatism, so the two should usually be positively correlated.
• The authors argue that, once you look globally, neither story fully holds: needs for security and
certainty and social class often push people toward cultural right but economic left (protection in
, both domains), while elite right–left discourse pushes toward cultural right plus economic right, and
the balance between these forces varies strongly by context and engagement.
• As a result, attitude structure is often better described as a protection–freedom dimension (cultural
and economic protection vs cultural and economic freedom) than a single right–left dimension, and
the familiar right–left alignment is a special, context-dependent case.
Data, concepts, and methods
• Data: 229 national samples from the World Values Survey, covering 99 countries on all inhabited
continents, surveyed between 1989 and 2014
• Cultural attitudes:
- Sexual morality (index of abortion and homosexuality items)
- Immigration attitudes
- Women’s role in the workforce.
• Economic attitudes:
- Social welfare (index of views on income inequality and government responsibility for providing
for people)
- Preference for private vs government ownership of business/industry
• Political engagement: composite of political interest and the importance of politics in one’s life
• Needs for security and certainty: value-based scale (tradition, security, conformity vs self-direction,
stimulation) from the Schwartz value survey; analysed for 2 later waves
• Country-level variables: post-Communist status, Human Development Index (development), and
national traditionalism (average sexual morality traditionalism)
• Methods:
- Within-country bivariate correlations between cultural and economic attitude measures
- Multilevel (three-level) random-coefficient regressions: individuals nested in survey-years nested
in countries, modelling economic attitudes from cultural attitudes, demographics, engagement,
and country-level moderators; plus cross-level interaction tests
- Additional multilevel regressions by wave to see how needs for security and certainty and social
class relate separately to cultural vs economic attitudes.
Core empirical findings
1. Globally, cultural and economic conservatism are not usually positively correlated
• Across 6 cultural–economic pairings (3 cultural × 2 economic measures), the average within-country
correlation is small and negative in each case.
• For each pairing, significantly negative correlations across countries are more frequent than
significantly positive ones; for example, immigration conservatism vs business-ownership
conservatism is significantly negative in about 58% of countries and significantly positive in only
about 7%.
• Multilevel regressions confirm: pooled slopes relating cultural to economic conservatism are
generally near zero or small and negative; there is no evidence of a typical positive relationship in
the world as a whole.
2. Protection–freedom attitude organization is widespread
• It is more common for cultural right attitudes (e.g. traditional sexual morality, anti-immigration,
traditional gender roles) to go with economic left attitudes (support for welfare, more government
ownership) and for cultural left attitudes to go with economic right attitudes.
