“Cognitive Madisonians” and Congressional Approval

Declining approval of Congress is a popular topic these days (note: low Congressional approval is always popular, just more so recently).  The importance of this issue was aptly described in 1974 by Arthur Miller: “a democratic political system cannot survive for long without the support of a majority of its citizens” (Miller, 1974, 951).  When, in December of last year, a Gallup poll found Congressional disapproval at an all time low, Josh had an excellent post on the matter which got some serious attention (from, among others, Ezra Klein at the Washington Post).  In that post, Josh argued that disapproval is “built into the institution’s DNA” citing published research which has shown that one of the determinants of Congressional disapproval is the passage of legislation.  Josh noted that “By simply doing its job Congress can alienate large parts of its constituency.”  But one part in particular caught my attention:

it’s frustrating voters don’t understand and appreciate the process better. It’s difficult to meld 350 million voices into a bill. Reasonable people can and will disagree. Having an appreciation for the difficulties legislators face may be a constructive step toward restoring faith in our governing institutions

This is a sentiment many academics echo (in my experience).  For example, Mike Sances (guest posting over at the Monkey Cage) recently asked “if [declining trust] about deeply held feelings of alienation, then maybe what is needed is a massive educational campaign about the virtues of American government.”

Josh and Sacnes’ suggestions share, at their core, a belief that the public’s trust of our governing institutions is tied to a basic understanding of how government functions (its constitutional design, virtues and inherent tendencies).  One particular function that Josh hints at in his post is the institutional constraints on active government.  Regarding checks and balances, as just one example, Madison noted in Federalist no. 51 that one solution for “faction” is to “divide the legislature into different branches; and to render them, by different modes of election and different principles of action, as little connected with each other as the nature of their common functions and their common dependence on the society will admit.”

So the question is: Do people with an appreciation for limited government and/or more political information have higher approval of Congress?  To assess this matter I used data from the ANES time-series.

I thought we could test the first question–about limited government and trust–by looking at the behavior of split-ticket voters (those who vote for candidates for public office of rival parties), or what researchers have colloquially labeled “cognitive Madisonians.”  At the minimum, these individuals are “ok” with the idea of split-party control.  At the maximum, they outright prefer constraints on one-party rule.  For this data I’m using a dummy variable that records if the respondent voted for candidates of rival parties for the House and Senate (coded 1) or whether they voted a straight party ticket (coded 0).  I’m also including a variable of self-reported political information, where higher values indicate that the respondent has a high degree of information about politics.  This allows us to assess whether a general understanding of U.S. politics enhances support for Congress (independent of one’s support for institutional constraints).

There are, of course, a range of factors that are likely to affect Congressional approval other than whether one splits their ballot or has high political information.  As John Sides has repeatedly pointed out (see here, for example), the state of the economy is one of the more significant predictors.  To capture this effect I’m using a question asking respondents whether they thought the economy got worse over the prior year (note: Sides’ prior analysis was macro-level, so this offers an additional piece of evidence).  The main trust of Sances’ guest post (cited earlier) was that when a respondent’s political party is in power, trust among that individual increases.  For this factor I’m using a dummy variable recording whether the respondent’s party had unified control of both chambers.   This was coded according to party control in the prior two-year periods and the respondent’s political identification.  I also included an additive index of both parties’ thermometer rating in order to gauge support for political parties in the aggregate.  Finally, I included a series of demographic controls—labeled Age, Education and Family Income—and institutional controls—such as the presence of quasi-divided government (Quasi) and divided government (Divided).

Here are the results from a logit model of congressional approval (1 approve, 0 disapprove) with standard errors clustered by year:

The positive and significant effect on Split-Ticket lends support to the hypothesis that cognitive Madisonians (those who approve of split-party control and, presumably, support greater checks on single-party rule) have higher levels of Congressional approval.  The insignificant effects on Quasi and Divided suggest that it’s not institutional constraints per se, rather the effect is driven by an implicit or explicit support for such constraints.  I also find that those with greater self-reported information about politics (labeled Info) have higher levels of Congressional approval.  However, the effect Education is not significant.  The marginal effects of both factors are such that split-ticket voters are 5% more likely to approve of Congress while a change in political information from ‘average’ to ‘very high’ increases the likelihood of approval by 11%.  In sum, those who have some level of appreciation for institutional constraints (in this case, checks and balances) have a higher approval of Congress as do those those with greater political information.  Thus, trust in governmental institutions is linked to an understanding and appreciation for how government functions (particularly a recognition that government is slow and contentious, by design).

Some other factors are significant predictors of Congressional approval as well.  The variable on economic conditions is negative and significant (labeled Economy).  Because this variable is reverse coded, this negative value confirms John Sides’ claim that when the economic conditions are favorable, Americans approve of Congress.  The variable Party Power and is positive and significant, confirming Sances’ claim that when a respondent’s party is in power that individual is likely to hold a favorable view of Congress.  The model also adds that favorable views of both parties adds to Congressional approval (labeled Parties).  Finally, Age is statistically significant and negative, indicating that older (perhaps less idealistic) Americans have less support for Congress as an institution.

So what’s the solution to lower congressional approval?  Force Americans to read the Federalist papers!