The Predictive Power of Presidential Job Approval on Election Outcomes, according to 1978-2012 Exit Polling

Presidential Job Approval based on exit polling, 1978-2012

It's been posited by some that a president's job approval rating is a pretty decent predictor of eventual election outcomes. But study's that have looked at this before tend to base their findings on pre-election political surveys on who says they will vote, rather than exit poll data that surveys actual voters.

A look at what exit polls have reported on past presidential job approval shows that while attitudes towards the President's job performance are occasionally on-the-mark with regards to the national popular vote, they're not always. For example, voters are frequently willing to vote AGAINST a President or his party's nominee whose job performance they APPROVE of.

More than that, exit poll findings regarding a president's job approval are far more predictive of actual results when an incumbent is seeking re-election, as opposed to the incumbent party's nominee seeking election.

Job performance also tends to be less prognosticative of actual results in midterm elections, as opposed to Presidential, while the President's APPROVAL rating is more likely to reflect his proportion of the vote than his DISAPPROVAL rating is to reflect his opponent's vote share.
The chart below documents the President's job approval rating in every presidential and midterm election since 1978. From 2004-2012, exit poll data from CNN is relied upon. From 1978-2002, exit poll data is provided by Samuel J. Best and Brian S. Krueger's Exit Polls, Surveying the American Electorate, 1972-2010, with the exception of 1988 data, which was retrieved here. Unfortunately, exit pollsters did not ask voters their opinion regarding presidential job approval in 1980, 1984, 1992, and 1996. Thus, in 4 of the 18 elections examined in the chart below, the President's job approval rating is based on the final Gallup Poll taken prior to that election, not exit poll data.

* indicates the President's job approval rating for this year is based on Gallup's final pre-election poll. All other findings come from exit poll results. Red indicates a Republican election victory, blue indicates a Democratic election victory. "H" means "House." "S" means "Senate." House and Senate votes tallies are provided by USHouse.gov

As the chart indicates, excluding midterm elections, voters have frequently voted against Presidents they approve of. A solid 54% majority approved of the job President Barack Obama was doing as they headed to the polls on Nov 6, 2012, but the President only won 51.0% of the vote.

The same phenomenon occurred the last time an incumbent President sought re-election, though to a lesser degree. George W. Bush carried 50.6% of the popular vote, though 53% of voters that day approved of the job he was doing.

Al Gore could only have hoped his vote total reflected the job approval rating of his boss in 2000. Voters that day gave Gore 48.3% of the vote, despite overwhelmingly approving of the job Bill Clinton was doing in his final days as President (58/42%, to be exact). Again in 1996, Bill Clinton fell just barely shy of a popular vote majority, attracting 49.2% of the vote, despite having the approval of 54% of Americans.

Obviously, this occurrence doesn't take place in every election, but it has happened frequently enough to conclude that a President's approval rating cannot always be expected to match his or his party's nominees' final vote tally.

The chart above also shows that exit poll findings on presidential job performance are more similar to actual results when an incumbent is seeking re-election, as opposed to when a member of the incumbent's party is seeking the presidency.  The approval rating for President's seeking re-election has been within 5 points of his actual vote share in every election since 1980. But in 2000, when the incumbent was term limited, Gore ran 10 points behind Clinton's job approval rating (48% vs. 58%).

In a similar situation in 1988, George H.W. Bush's vote share perfectly matched President Reagan's 53% job approval rating, though Dukakis captured all 40% of voters that DISAPPROVED of Reagan's performance, as well as the rest that neither approved or disapproved of him.

Furthermore, the above chart illustrates the fact that a President's job approval rating correlates with actual results much more so in Presidential election years, as opposed to midterm elections. The reasoning for this seems obvious - in midterms, though voters may express their satisfaction or dissatisfaction with the President with their vote, they're actually voting for or against local candidates. But whatever the reasoning, exit poll findings on voter opinion of the President have often diverged wildly from midterm results.

Consider the 2002 midterms. Congress had just approved the Iraq War resolution, America was still on guard just one year following the September 11th terrorist attacks, voters were in near complete agreement with the Republican Party's increasingly hawkish foreign policy stance, and President George W. Bush was wildly popular, with 2 out of 3 voters approving of his job performance. Yet even then, the Republican party only won a 5 and 4 point popular vote victory in the House and Senate, and needless to say, came no where close to 66% of the popular vote (George Bush's approval rating among exit poll respondents).

In 1998, voters gave Bill Clinton a strong 57/43% job approval rating in the wake of the Lewinsky scandal, while essentially splitting their vote between Republicans and Democrats in the Senate and House. In 1986, President Reagan's party experienced a landslide 10 point popular vote loss in the House of Representatives, despite Reagan sporting an enviable 63/37% approval/disapproval rating.

What, does all this tell us about the 2014 midterm elections? Little to nothing.

As 2002, 1998, and 1986 have taught us, there can be very little correlation between actual midterm election results and the President's job approval rating, much less his job approval rating 19 months out from election day (by the way, Obama currently sits at an average 48/47%, per pollster.com). But look at it like this: in November 2010, the Republicans experienced impressive 7 and 5 point national popular vote wins in the House & Senate respectively, while the President's job approval/disapproval rating was 44/55%; a result that could probably be expected given Obama's unpopularity. But the two midterms where the President's approval/disapproval rating was nearly tied (1978 & 1982) produced very different results. In '82, Reagan was just above water with a 52/48% rating, yet Republicans still experienced major popular vote losses in the House and Senate. Jimmy Carter, on the other hand, was just under water in 1978 (48/52%), but Democrats still managed fair popular vote wins in Congress.

In other words, good luck trying to parse election results from Obama's job approval rating next November. He could be popular and experience congressional losses (like in '86 and '90), or unpopular and experience gains (like in '78). Of course, this fact won't stop me from 'parsing' anyway.