Public Opinion, Ideology & Polling

Classic definition (V.O. Key): “opinions held by private persons which governments find it prudent to heed.”
Focus is on collective views, not single individuals.
Governments track it to:
- Decide which bills, regulations, or programs to pursue.
- Anticipate voters’ reactions and craft re-election strategies.

Aggregate (collective) opinion
- Remarkably stable; shifts are slow and incremental.
Individual opinion
- Highly volatile—can change with the day’s news, moods, or recent conversations.
Paradox: individually volatile data can aggregate into a smooth, predictable line.

Shapes the “shortcuts” (heuristics) voters use: attitudes, ideologies, partisanship.
Serves as feedback for officials seeking re-election or legitimacy.

Definition: a manner of thinking about a specific person, policy, or event in the political world.
Features
- Loosely structured; most people haven’t deeply processed every stance.
- Inconsistent: People often hold contradictory views (e.g., favoring tax cuts and higher spending).
- Mutable: New info can shift an attitude; over time they may calcify.
- Confirmation bias reinforces existing beliefs.
- Sophistication gap: Most Americans are low-information voters—consume little, often one-sided news.

Ideology = “elaborately organized set of political attitudes.”
Benefits for voters:
- Provides an interpretive framework → promotes internal consistency.
- Simplifies new information processing—acts like a “worldview”.

Modern Conservatism
- Govt should uphold social/moral order.
- Prefers less economic regulation.
- Derives from a shift away from “traditional conservatism” (royalist, pro-elite control).
Modern Liberalism
- Govt as tool to produce equality of opportunity (e.g., K-12, food security).
- Descended from classical liberalism which distrusted govt power.
Libertarianism
- Maximum personal & economic liberty; “govt only protects safety and borders.”
- Spectrum from “small-govt conservative” to near-anarchic positions.
Democratic Socialism
- Expanded social services (healthcare, housing, college) through democratic means.
- Seeks equal opportunity and more equal outcomes.
Communism (rare in U.S.)
- Collective/public ownership of major productive assets to prevent worker exploitation.
- In practice, most self-declared regimes say they are merely “pursuing communism.”
Fascism (historical/edge):
- Total control by a leader/party; nationalist, often racial lens; regulates economy & private life.

Simple 1-D left↔right spectrum fails to place libertarians, populists, etc.
Political-science 2-axis grid:
- Y-axis = degree of social/cultural regulation.
- X-axis = degree of economic regulation.
- Produces 4 quadrants → Liberal, Conservative, Libertarian, “Authoritarian/Populist” (state-heavy in both realms).

$\approx 25\%$ call themselves liberal, $\approx 35\%$ conservative, rest moderate.
Lines are steady; liberal ID edged up ≈8 pts in 30 yrs.

Columbia School (1940s-50s): Community & personal influence dominate vote choice.
Michigan School (1960s): Party ID is a strong psychological attachment funneling votes.
Kinder & Kalmo (2000s):
- Ideology often follows party, not vice-versa.
- For average/low-information voters, self-label “conservative/liberal” in Feb ≠ guarantee same label in 1-2 yrs.
- Party ID, however, is highly stable year-to-year.

Definition: attitude toward / identification with a party label (D or R, etc.).
The single best predictor of vote choice.
Not identical to ideology (e.g., conservative Democrats, liberal Republicans exist).
Stability: PID rarely changes; big shocks (Great Recession, Tea Party, Trump era) only nudge lines.

$81\%$ of self-declared independents lean D or R.
Among public:
- $17\%$ lean Democrat, $13\%$ lean Republican, only $7\%$ true “no-lean” independents.

Women, non-whites, college grads, urban, non-religious → lean Democratic.
Men, whites, non-college, rural, white evangelicals/Mormons, older → lean Republican.

1800s: anecdote & “straw polls” (jelly-beans, fairs) → unscientific.
1930s-40s: George Gallup pioneers modern scientific polling rooted in probability theory.

Population (N): entire group you care about.
Sample (n): small subset actually interviewed.
Probability theory: Properly drawn sample → infer population traits.
Random Sampling (gold standard):
- Everyone in population has ≈equal chance to be selected.
- Phone era: random-digit dialing (RDD) → mix $70\%-75\%$ cell / $25\%-30\%$ landline.
Sample Size & Margin of Error (MoE):
- Typical national poll $n\approx1050$ → $\text{MoE}\approx\pm3\%$ – $5\%$ at $95\%$ confidence.
- Larger $n$ ⇒ smaller MoE $\big(\text{MoE}\propto\frac{1}{\sqrt n}\big)$ .
Internet polling: growing but still biased by digital divide & tech literacy.

Report: A = $47\%$ , B = $43\%$ , C = $10\%$ , MoE $\pm5\%$ .
- Adjusted ranges: A $[42,52]$ , B $[38,48]$ , C $[5,15]$ .
- Overlaps ⇒ race is statistical tie (within the margin).

Selection bias: sample systematically excludes/overrepresents groups (e.g., Fox News text poll, 90 % Romney).
Sampling error: unavoidable variance due to using a sample; quantified by MoE.

Heavy horserace coverage encourages “bandwagon” attitudes.
Exit polls no longer released in real-time to avoid influencing later voters.
Always ask:
1. Does this poll align with other recent polls?
2. Does the article’s headline actually match what the numbers (and MoE) say?

Delegate model of representation → officials should follow voters’ preferences, hence polling invaluable.
Campaigns use favorability polls to decide whether to run and craft messages.
Risk: over-emphasis on numbers can crowd out substantive policy discussion.

Distinguish attitudes, ideology, partisanship—building-block hierarchy.
Know agents of political socialization.
Contrast classical vs. modern ideologies (liberal & conservative).
Understand polling methodology: population vs. sample, random sampling, MoE, selection bias.
Be able to read MoE into a reported percentage.
Recognize demographic patterns in PID & ideological self-ID.
Remember: party ID is the most stable, predictive factor in U.S. voting behavior.