Ideological Cartography

(Co-written with Kevin Collins)

What will the field of candidates for the 2012 Republican presidential nomination look like? Presidential primaries tend to highlight divisions within parties, and the press has focused on moderates versus conservatives, social conservatives versus business conservatives, party insiders versus Tea Party insurgents, electable candidates versus long shots.

Some commentators, notably Nate Silver in his visualization of the potential Republican field, have depicted these dimensions as orthogonal. Specifically, he graphically depicts candidates ideological placement, status as insiders, and chances of winning.  However, while Silver uses Intrade values as a measure of candidate chances, his placement of candidates on the other two dimensions is based on his judgment alone. Based on his judgment, these three dimensions appear to be unrelated to one another. However, drawing on data from federal and state campaign finance records, we can more reliably estimate both candidate ideology and reliance on large donors, which we take as a proxy measure of insider status. Based on this analysis, we show that these dimensions are in fact strongly correlated.

The figure below summarizes the 2012 Republican presidential field. The upper panel plots the proportion of funds raised from donations of $500 or less (including unitemized contributions) against candidate ideology. As is evident, more conservative candidates, particularly those affiliated with the tea party, raise a greater proportion of their funds from small donors. The circle sizes are proportional to the Intrade share prices for the respective candidates as of June 13th. The circles are colored coded based on candidacy status. Those who have officially announced their candidacy are red, those who have not yet announced are purple, and those that have decided not to run are green.

The bottom panel overlays two kernel densities drawn from the 2010 Election cycle. In red is the ideal point distribution of Republican candidates. This gives a sense of how the presidential candidates locate with respect to the party as a whole. In gray is the distribution of Republican donor ideal points, each weighted by the total amount donated during the 2010 Election cycle. This characterizes the fundraising landscape with respect to ideology.

The Republican Field: The upper panel plots the proportion of funds raised from donations of $500 or less (including unitemized contributions) against candidate ideology. The circle sizes are proportional to the Intrade share prices for the respective candidates as of June 13th. The circles are colored coded based on candidacy status. Those who have officially announced their candidacy are red, those who have not yet announced are purple, and those that have decided not to run are green. The bottom panel overlays two kernel densities drawn from the 2010 Election cycle. In red is the ideal point distribution of Republican candidates. This gives a sense of how the presidential candidates locate with respect to the party as a whole. In gray is the distribution of Republican donor ideal points, each weighted by the total amount donated during the 2010 Election cycle.

The idea underlying measuring the ideology of candidates from campaign finance records is relatively straightforward. Contributors are assumed to prefer ideologically proximate candidates (i.e., those who share their views). A model then conditions on the ideology-based research conducted by the millions of political donors to provide estimates of candidate ideology.

When we compare the measures with those created by Nate Silver (link), reprinted below, we find that the rankings are correlated but exhibit a few large discrepancies. Silver ranks Ron Paul among the most moderate candidates, whereas the donors place him on the far right. (Paul may technically agree with Democrats on several key issues but has his own reasons.) He identifies Santorum as the most conservative candidate in the field, but Santorum’s donors place him nearer the center of the Republican Party. In both cases, the candidates’ DW-NOMINATE score reflects his CFscore–that is, roll call voting records consistently place Paul on the far right and place Santorum somewhere in the interior wing of the party.

Mapping candidates in this way highlights what we gain from quantitative measures of ideology. Cultural issues such as abortion and same-sex marriage happen to be strongly correlated (a pattern confirmed by looking at who donates to ballot campaigns), but they take a back seat to explaining how ideological preferences map onto political behavior and outcomes. In the end, traditional left-right economic issues best explain how a member votes and which candidates donors support.

As a point of reference, we include in the figure past Republican presidents. One of the advantages of measuring ideology from contribution records is that both candidates and contributors are typically active across multiple election cycles, which facilitates more reliable across-time comparisons (see the discussion here). If they were running today, both Bush’s would locate to the left of the mean ideal point of the Republican Party, which has become steadily more conservative over the last three decades, whereas Reagan would be slightly to its right.

Candidate positions are typically stable but are subject to change, especially for candidates seeking to reinvent themselves when competing for a new office. This certainly describes Romney predicament. Romney’s ideal point shifts dramatically over the course of his political career. Romney began his career with a score of 0.59 during his failed 1994 senate bid against Edward Kennedy, adopted a centrist position while running for governor, bringing his score to 0.48, and has currently settled near the center of his party with a score of 0.78.

These data come from the most recent available campaign finance records for each candidate, which do not include any current presidential exploratory or candidate committees but do include state records (for current and former governors) and older records, when the person in question has been out of office for a substantial period of time. These variations in source data could potentially bias the measures in a number of ways. First, if candidates have shifted their ideology since their last campaign record, their previous donation pattern may not reflect their current ideology. For example, Newt Gingrich’s fundraising as a Member of Congress may not perfectly reflect his ideology today.

Second, states have different campaign disclosure thresholds than do the federal government. While the Federal Elections Commission reports the individual records of all donations made by any individual who donates to a candidate over $200 in total, this disclosure threshold is lower in most states. In MN and IN it is $100, and in UT it is $50.  Since small dollar donors tend to be more ideologically extreme than large dollar, access seeking donors, if state candidates are receiving substantial amounts of donations in between their state and the federal disclosure thresholds, these amounts could make them appear more ideologically extreme. However, since the candidates for whom state records are used are distributed roughly evenly across the space, these differences in disclosure thresholds may not be a significant source of bias here.

Our measure of proportion raised from small donors is also in part a function of campaign finance laws. This matters for past candidates for president who fundraised subject to the $1,000 limit on individual contributions in place prior to the implementation of the BCRA during the 2004 Elections. This discrepancy make past president look as though they raised slightly more funds in small amounts than they actually did.

Variation in campaign finance laws matters much more for governors turned presidential candidates. Pawlenty, Daniels, Perry and Huntsman campaigned for governor under state laws that foster very different fundraising environments. (We rely solely federal contribution records for Romney and Palin). Minnesota’s contribution limits favor fundraising from small donors whereas limits in Indiana, Texas and Utah have no limits on the amount an individual can donate to a candidate, thus favoring fundraising large amounts from a smaller set of donors.# This highlights that Pawlenty has experience fundraising in small amounts but is not particularly informative when making comparisons to other candidates. Fortunately, this will no longer be a problem as future FEC filing deadlines approach.