March 2, 2016

The Path to Office

Candidates have to win the approval of many veto-players to take office.

  • General electorate
  • Primary electorate
  • Elites
  • Candidate's self-respect
  • Optimal strategy at each stage is to appeal to median voter
  • But if candidates have limited ability or inclination to move ideologically, candidates have to compromise, accomodate binding constraints

Primary Electorates

"[T]he most common explanation for the failure to observe Downsian convergence [is] the extremism of primary electorates" - Gary Jacobson (2012)

  • Old literature sees primary electorates as extreme relative to primary nonvoters (many papers, 1978-1985)
  • Geer (1988) uses exit polls to show that primary voters are less extreme than general election voters
  • Using validated 1980 turnout from the ANES Norrander (1989) finds little ideological difference
  • With validated CCES data, Hill (2015) uses IRT to show that primary voters are more extreme

Survey Data: ANES Face-to-Face

  • Conducted in-person by professional interviewer

  • Sampling frame is active addresses from USPS Delivery Sequence File in 126 sampled tracts

  • 2,054 completed pre-election interviews, 2,006 respondents gave a full name

  • Matched to 3 vendors, 2 of which gave ANES primary turnout

  • Mix of vendor's proprietary matching (trust but verify) and address-then-name matching

  • 1,624 of 2,006 matched to primary vote-vendors. Weighted: 86%

Methodology

  • Identified 310 interesting ANES questions, assigned to categories, recoded to numeric
  • PCA Scaling: ideological scale with domestic policy and economy items; SES scale
  • Also performed t-test on each question for each party
  • Generate \(310 \times 2 = 620\) \(p\)-values
  • Test hypotheses using the Benjamini & Hochberg (1995) procedure, controlling the FDR at \(.05\)-level

A Two-Factor Theory of Primary Turnout

Extreme Moderate
Rich High Medium
Poor Medium Low



  • High SES people vote more
  • Extremists vote more

Distribution of Partisans

Cell Population:
Dem Rep
Liberal Moderate Moderate Conservative
Rich 22 19 30 34
Poor 28 31 20 16
Primary Turnout by Cell:
Dem Rep
Liberal Moderate Moderate Conservative
Rich 22 18 30 57
Poor 14 18 25 35

Composition of Primary Electorate

  • 49% of Democratic Primary Voters are "extremists" vs. 64% of Republicans
  • 54% of Democratic Primary Voters are above-median SES vs. 63% of Republicans
  • Ideology explains R Primary turnout (~12.5%) but not D turnout (~0%).
  • SES also more predictive of turnout for R's.

Benjamini-Hochberg

% of Questions with Differences by Category

Ideological Self-Placement

Issues & Party Approval

What's Next?

For primaries:

  • Compare to voterfile estimates using MRP, where possible
  • Apply same methodology to CCES
  • Compare to 2016 ANES

Non-Presidential elections broadly:

  • Who votes in local elections?
  • What are the implications for Oliver's (2012) home ownership hypothesis?
  • Do low-turnout elections always advantage conservatives?