Chapter 17 Exercises

These exercises move you from reading about public opinion to working with public-opinion data. The first three are required for the Democracy Audit project; the rest are optional but recommended.

Exercise 1 — Pull a Pew Research study and read it like a methodologist

Objective: Read a complete public-opinion study from a non-partisan source and identify what it does and does not tell you.

Go to pewresearch.org/politics-policy and find a study published within the past 12 months on a topic that interests you (current options often include immigration attitudes, presidential approval, religious-political alignment, generational politics, attitudes toward institutions, foreign-policy views).

Open the methodology section (Pew puts it at the bottom of every study; some are linked separately as "How we did this"). Identify and write down:

  1. The sample frame (who could be selected — registered voters, all adults, panel members?).
  2. The sample size (n) and the margin of error.
  3. The mode (online panel, phone, mixed).
  4. The weighting variables.
  5. The field dates.
  6. The sponsor of the study.

Then look at the actual results. For one finding — pick the most attention-grabbing statistic in the study — write a paragraph that does the following:

  • States the finding accurately, including the MoE.
  • Identifies one or two demographic sub-groups where the finding diverges from the headline.
  • Identifies what question Pew did not ask that would have helped you understand the finding better.
  • Identifies one piece of context (a related question from a different source, a historical trend) that frames the finding.

Submit a 400–600 word memo of this analysis. The point is not to find Pew doing something wrong — Pew is generally careful — but to practice the discipline of reading polling at the level of the underlying methodology rather than the headline.

Exercise 2 — Analyze ANES Time Series Data on a long-running question

Objective: Get hands on the gold-standard academic dataset and look at change over time.

Go to electionstudies.org and find the ANES Time Series Cumulative Data File (a single file containing harmonized variables from every ANES survey since 1948). You do not need to download or run statistical software for this exercise; the ANES website provides simple cross-tabs and trend tools.

Pick one of these long-running questions:

  • "Do you trust the government in Washington to do what is right?" (VCF0604).
  • Party identification (VCF0301).
  • The seven-point liberal-conservative self-placement (VCF0803).
  • Approval of the way the government is handling its job, by demographic group.

For your selected variable, produce:

  1. A trend chart of the population-level percentages over time, from 1958 (or earliest available) to most recent ANES.
  2. The same trend, broken out by partisanship (Democrat, Independent, Republican).
  3. A 200-word interpretation: what does the trend show? When did inflection points occur? What political-historical events do those inflections correspond to? Where does the partisan gap widen or narrow most dramatically?

Submit the charts and interpretation as a single document. If you can produce the charts cleanly with ANES's tools, do so; if not, sketch them in any tool you prefer (spreadsheet, Python, R, or hand). The interpretation matters more than the visualization quality.

This exercise is the closest thing in this textbook to "how political-science research is actually done." Time to learn the muscle memory.

Exercise 3 — Compute a perception-gap measurement on yourself and your peers

Objective: Replicate the More in Common perception-gap research at micro-scale.

You will need at least 6 peers (more is better) — ideally a mix of self-identified Democrats, Republicans, and Independents.

For each participant, ask the following questions:

Set A — your own positions: - "What percentage of police use of force in the U.S. do you believe is unjustified?" (Number 0–100.) - "What percentage of immigrants currently in the U.S. would you support being granted a path to legal status if they had no criminal record and had been here 5+ years?" (Number 0–100.) - "What percentage of new gun purchases should require a background check?" (Number 0–100.)

Set B — your guess at the other party's positions: For each question above, ask: "What percentage do you believe the average voter of the OTHER party would answer?"

Now compare:

  • For each respondent, compute the gap between (their guess of the other party's position) and (the actual mean of the other-party respondents in your sample).
  • Aggregate the gap by respondent partisanship.

You should find — at least roughly, with a small sample — that members of each party systematically estimate the other party as more extreme than the actual median of that party. The size of the gap will vary, but the direction is usually consistent.

Write a 300-word reflection: was the gap larger or smaller than you expected? Were the gaps symmetric across parties or asymmetric? What does this tell you about the information environment your peers and you live in?

Exercise 4 — Steel-man both sides of a current ideological caricature

Objective: Practice the steel-manning discipline that Chapter 17 (and the Balance Guide) center on.

Pick a current political controversy where you have a clear position (your own ideological caricature of the other side is fine starting material). Examples: gun-control legislation, immigration enforcement intensity, abortion regulation, climate policy aggressiveness, university DEI programs, transgender athletes in women's sports, the role of religion in public schools.

For your chosen issue, write two essays of approximately 300 words each:

  • Essay A: The strongest version of the position you currently hold. Cite the empirical evidence, the philosophical case, the values invoked. No straw-manning of the other side.
  • Essay B: The strongest version of the position you currently do not hold. Cite the empirical evidence, the philosophical case, the values invoked. Take the position seriously. Imagine the smartest, most thoughtful person who genuinely holds this position; write what they would write.

Submit both. The grading is not on which conclusion you reach but on the quality of each steel-man — whether you took the other side as seriously as you took your own.

This exercise will be repeated through the textbook in different forms. The ability to articulate the strongest version of a view you do not hold is the precondition for any useful political conversation.

Exercise 5 — Democracy Audit: your district's demographics vs. its representative's positions

Objective: Produce the public-opinion component of your district-level Democracy Audit deliverable.

For your home congressional district (use Census Bureau 2020 redistricting data; tools at census.gov):

  1. List the demographic profile: population by race/ethnicity, education, age, urbanicity, household income, religion (use Pew's regional religious-landscape data as a proxy where district-level religion data are unavailable).
  2. List your representative (use govtrack.us or congress.gov). Pull their party, voting record on five recent contested votes, and any publicly available statement on each.
  3. For each of the five votes, find a recent national poll on the underlying issue (Pew, Gallup, or any other AAPOR-transparent source). Note the headline result.
  4. Write a 400-word analysis: where does the representative's voting record align with the national-polling majority? Where does it diverge? Where does the district's demographic profile suggest local public opinion may differ from national-polling majorities?

You will reference this analysis throughout Part III. Save it.

Exercise 6 — Find the moral foundation in a recent controversy

Objective: Apply Haidt's moral-foundations framework to political language.

Pick a recent political controversy with substantial public debate (a Supreme Court decision, a piece of legislation, a presidential action, a major social-media flashpoint).

Find one op-ed or longform essay arguing each side of the controversy. (For balance, pick outlets across the partisan spectrum: e.g., one from the New York Times or The Atlantic; one from National Review or The American Conservative.)

For each essay, identify which of Haidt's six moral foundations the author appeals to, and where:

  • Care / harm
  • Fairness / cheating
  • Liberty / oppression
  • Loyalty / betrayal
  • Authority / subversion
  • Sanctity / degradation

Quote the specific sentences that invoke each foundation. Note which foundations the author does not invoke that could have been used to make a similar case.

Write a 300-word comparison. What does this reveal about how each side is framing the disagreement? Could each side make a stronger case by also invoking the foundations they did not? Where do the two arguments actually disagree, and where are they appealing to different moral concerns past each other?

Exercise 7 — Identify the heuristic you use most

Objective: Self-knowledge.

Reflect on a recent political decision you made (a vote, a donation, a candidate evaluation, a policy support). Write a 200-word analysis identifying:

  • Which heuristics from Section 17.7 you used (party label, endorsement, demographic cue, performance evaluation).
  • Which information you did not gather but could have.
  • Whether, on reflection, the heuristic produced a decision aligned with your considered preferences or whether deeper investigation might have changed the decision.

This is not an indictment of heuristics. Heuristics are rational economies. The reflection is to make the cognitive shortcut visible to yourself, so you can decide which heuristics serve you well and which do not.

Exercise 8 — Test the salience-asymmetry hypothesis

Objective: Distinguish broad-and-shallow support from narrow-and-deep support.

Pick a single issue with high polled majority support but limited legislative success (universal background checks, Dreamers path to citizenship, infrastructure spending, prescription-drug-price negotiation, raising the federal minimum wage, marijuana legalization).

Find at least two pieces of evidence on each of the following:

  1. Polling support. Pew, Gallup, or other AAPOR-transparent source.
  2. Salience among supporters. Issue-importance polling — when respondents are asked to rank the most important issue facing the country, how many list this one in their top three?
  3. Salience among opponents. Same question, opponents only.
  4. Organizational backing. What organizations support and oppose? What are their relative resources (membership, lobbying spend, PAC spending — use FEC and OpenSecrets data)?

Write a 400-word memo: based on these four pieces of evidence, why does this issue not produce legislation despite majority support?

The point is to internalize the gap between polling support and political force — the gap that explains a great deal of the apparent dysfunction in American policymaking.

Exercise 9 — Track presidential approval against an event window

Objective: See how (and how little) presidential approval moves with specific events.

Pick a recent presidential administration and a specific high-salience event during it: an executive order, a major legislative achievement or failure, a foreign-policy crisis, a scandal, a Supreme Court appointment. Examples: the Dobbs decision (June 2022, Biden); the U.S. withdrawal from Afghanistan (August 2021, Biden); the first impeachment (December 2019, Trump); the ACA passage (March 2010, Obama); the announcement of Operation Warp Speed (May 2020, Trump); the Inflation Reduction Act signing (August 2022, Biden).

Using Gallup, RealClearPolitics aggregates, or 538-archive aggregates (any one will work; the patterns are robust), pull the approval numbers for the four weeks before and the eight weeks after the event. Plot or tabulate the trend.

For each of the following sub-groups, do the same: in-party respondents only; out-party respondents only; Independents only.

Write a 300-word interpretation:

  1. Did the event move overall approval? By how much, in what direction, for how long?
  2. Did the event move the in-party, out-party, or Independents disproportionately?
  3. Does the size of the movement match what the event's media coverage suggested it should?

You will likely find — for almost any event — that the movement is smaller than the coverage implied. Approval moves slowly. The 10-point band most administrations live within is the strong attractor; events typically push 2–4 points and the effect dissipates within 6–10 weeks.

This is a good exercise for understanding why presidential approval is often a more useful indicator of underlying partisan structure than of current events.

Exercise 10 — Re-examine your own information diet

Objective: Identify the structural features of your information environment that may produce perception gaps in you.

Over a 7-day period, log every political news source you encounter, by name. Include: what TV programs you watched (or had on); what podcasts you listened to; what news websites you visited; what social-media accounts you actively engaged with on political topics; what conversations you had where political news came up.

At the end of the week, classify each source on two dimensions:

  1. Partisan lean (left-coded, right-coded, center, mixed) — using AllSides or Ad Fontes Media as a rough guide if you don't know.
  2. Mode (legacy mainstream, partisan opinion, social-media-driven, podcast/longform, in-person conversation).

Tabulate the time-weighted distribution of your information sources across these categories. Compute, roughly: what percentage of your information diet came from sources lean-aligned with your own partisan identification? What percentage from sources lean-opposed? What percentage from center/mixed sources?

Write a 250-word reflection: what does your information diet look like? Does it look like the kind of diet associated with smaller perception gaps (mixed-source, legacy-mainstream, in-person) or with larger ones (homogeneous-partisan, social-media-driven)? What change in your diet, if any, would you want to make based on the perception-gap research?

This exercise is for self-knowledge, not for indictment. The point is awareness of the structural features of your own information environment, not guilt about consuming the content you choose.