Monthly Archives: March 2016

What are the odds that Donald Trump will win the general election?

A friend asked me today whether I thought there were “any chance” that Donald Trump could win the general election election. The short answer is “yes.”

The long answer is this:

Since the 1950s there have been a few key variables that have done a pretty good job of predicting the outcome of U.S. presidential elections: 1) domestic economic growth during the election year, and 2) incumbent president approval rating. The incumbent party’s tends to win when those things are good; the challenging party’s candidate tends to win when those things are bad. Observe the following correlational graph between domestic economic growth and incumbent party’s share of the vote (source):

economic graph

Using these basic forecasting models as a baseline, we can plug in President Obama’s current average job approval rating of 49% and 2016 GDP forecasts are somewhere in the range of 2%. This would give the Democratic candidate (very likely Hillary Clinton at this point) about an 88% chance of winning, leaving the presumptive GOP nominee Donald Trump with about a 12% chance.

That being said, these forecast models make some big assumptions. One assumption is that the national political/electoral system is roughly in the same condition as has characterized the system since the WW2 era in terms of partisan constituencies, the relative equilibrium of the Republican and Democratic parties at the national level, etc. There is some evidence that this system may be giving way to a new system, though, although it’s unclear at this point the extent to which this has occurred (characteristics of the new system include the emergence of the internet, hyper-polarization, etc.). The spectacular inability of the Republican “establishment” to exert any control over this year’s nominating system is another indicator that the electoral system may be undergoing a sizeable shift. The predictive model described above may not be as reliable in an environment that differs substantially from the elections upon which it was based.

Another big assumption is that both parties nominate generally qualified and competent candidates and that they will run about evenly-matched campaigns in terms of resources, strategy, etc. I’ll leave it readers to determine the extent to which “qualified and competent” applies to the current Republican front-runner. To the extent to which one of the two parties nominates a candidate who doesn’t meet this criteria, it may cause the predictive model to over-estimate that candidate’s chances to win the general election.

All in all, my “hunch” at this point is that Donald Trump has somewhere between a 20%-40% chance of winning the general election should he indeed eventually win the nomination. This is based on the evidence that 1) given the prevailing political and economic conditions, any generic Democrat would have an advantage over any generic Republican this year, 2) Donald Trump’s rhetoric and campaign style will likely demobilize a significant portion of the Republican base, and 3) there is a good deal of uncertainty as to the extent to which we are entering a “new party system” and the consequent extent to which the patterns found in previous elections can predict outcomes in current and future elections.

Under normal circumstances, I would say with a good deal of confidence that “all other things being equal, the political science models point to an edge for the Democratic candidate this year.” This year, however, the political science models have done a very poor job of predicting the rise of Trump and the collapse of the power of the GOP establishment. I therefore preface my “hunch” of a 20%-40% odds for a Trump general election victory with a great deal of caution and humility.

A quick analysis of the Kentucky 54th legislative district special election

This week Republican Daniel Elliot won the special election for the Kentucky 54th state legislative seat which comprises Boyle and Casey counties. According to the Advocate-Messenger report, he received 58.4% of the vote compared to the 41.6% received by Democratic challenger Bill Noelker. Noelker evenly narrowed out Elliot in Boyle County with 50.5% of that county’s vote, while Elliot won a clear majority (78.4%) of the Casey County vote.

This morning a friend asked a question about the precinct turnout patterns that prompted us to look at the relationship between turnout and party registration in each precinct. Here’s a quick summary of what we came up with:

Democrat/Republican registration ratio:

  • Boyle County: 1.47
  • Casey County: 0.23

Noelker/Elliot voting ratio:

  • Boyle County: 1.02
  • Casey County: 0.27

This suggests that NoelkerĀ underperformed significantly in Boyle County relative to party registration and overperformed slightly in Casey County relative to party registration. Also, the precinct-by-precinct D-R ratio and Noelker-Elliot voting ratios correlations are 0.42 in Casey County and 0.19 in Boyle County which means that partisan registration ratios were more predictive of voting patterns in Casey County than in Boyle County.

This admittedly back-of-the-envelope analysis suggests that Elliot won by turning out Republican voters in Boyle County (or that Noelker was less successful at turning out Democratic voters in Boyle County) and that each candidate did pretty well among their respective party bases in Casey county. It’s also possible that mobilization efforts on the part of campaigns mattered less and that Republican voters in Boyle County were simply more enthusiastic to show up to vote on Tuesday than were Democratic voters. It’s not possible just from the turnout statistics to know definitely one way or the other.

In the nearly six years that I’ve lived in Kentucky, I’ve observed that Republicans have been trending more and more successful at the state and local level in the state of Kentucky. Over the past several decades “ticket splitting” has been declining, meaning that voters have become more and more consistent in voting patterns between national and state/local elections. Kentucky has resisted that trend for a long time: continuing to elect Democrats at the local level while electing Republicans at the national level. The federal and state elections of 2012, 2014, and 2015 have generally trended more and more Republican at the local level here in Kentucky. My hunch is that Daniel Elliot’s victory this week is part of the broader on-going trend of the decrease in “ticket splitting” among Kentucky voters who are becoming more consistent in their Republican preferences at the local as well as state and national levels.