For a previous post on primary votes, see here.

In this post, I’ll discuss findings from the 2016 primary and caucus results. Only comparable primary and caucus (e.g., no Dem caucus and GOP primary in a state) results are used, though Oklahoma (which had an open Dem and closed GOP primary) is kept in the data.

These correlations are between states, not counties. Looking at counties would obviously give different (and probably more helpful) results, such as a strongly positive correlation between Rubio and Kasich vote shares within each state (as is obvious by looking at the results in Massachusetts, where Trump was seen as the True Republican candidate, Cruz as some right-wing extremist, and Rubio and Kasich RINOs).

*Cruz primary vote as a percentage of total primary vote (Dem and GOP) is strongly negatively correlated with Clinton primary vote as a percentage of total primary vote between states. This was the strongest correlation found between the total primary vote shares of two candidates in the 2016 presidential primaries.

*Trump primary vote as a percentage of total primary vote (Dem and GOP) is strongly negatively correlated with Sanders primary vote as a percentage of total primary vote between states. This was the second-strongest correlation found between the total primary vote shares of two candidates in the 2016 presidential primaries.

*Cruz primary vote as a percentage of total primary vote (Dem and GOP) is negatively correlated with Sanders primary vote as a percentage of total primary vote between states.

*Trump primary vote as a percentage of total primary vote (Dem and GOP) is only slightly negatively correlated with Clinton primary vote as a percentage of total primary vote between states.

*Trump primary vote as a percentage of total primary vote (Dem and GOP) is positively correlated with Cruz primary vote as a percentage of total primary vote. In fact, it seems higher levels of Trump primary vote as a percentage of total primary vote (Dem and GOP) necessitated a higher percentage of Cruz primary vote as a percentage of total primary vote (Dem and GOP) (though not visa versa -cf., Utah).

*Kasich primary vote as a percentage of total primary vote (Dem and GOP) seems to be only slightly positively correlated with Trump primary vote as a percentage of total primary vote (Dem and GOP) between states.

*Kasich primary vote as a percentage of total primary vote (Dem and GOP) seems to be uncorrelated (if anything, slightly negatively correlated) with Cruz primary vote as a percentage of total primary vote (Dem and GOP) between states.

*Kasich primary vote as a percentage of total primary vote (Dem and GOP) seems to be strongly negatively correlated with Rubio primary vote as a percentage of total primary vote (Dem and GOP) between states. No doubt, this is due to them being close substitutes for each other, even though Rubio is substantially more conservative and neoconservative than Kasich. This was the third-strongest correlation found between the total primary vote shares of two candidates in the 2016 presidential primaries.

*Kasich primary vote as a percentage of total primary vote (Dem and GOP) seems to be positively correlated with Sanders primary vote as a percentage of total primary vote (Dem and GOP), but negatively correlated with Clinton’s. Kasich is the only GOP candidate running whose primary vote as a percentage of total is positively correlated with that of either Democratic candidate for President between states.

*There was no correlation between Rubio and Trump primary vote as a percentage of total primary vote (Dem and GOP) between states.

I also sorted the candidates by perceived (and, where data exists, actual) ideology. Sanders is to the left of Clinton who is to the left of Kasich who is to the left of Christie who is to the left of Bush who is to the left of Fiorina who is to the left of Rubio who is to the left of Trump who is to the left of Carson who is to the left of Cruz. The decision to sort Trump to the right of Rubio and Cruz to the right of Carson is substantiated by the exit polls in the Virginia open primary, as well as Rubio winning the GOP primaries in such “bastions of conservatism” as Minnesota, the District of Columbia, and Puerto Rico. The candidate the ideological median primary/caucus voter voted for is shown below for each state where this can be reasonably calculated. Nevada and Maine’s Democratic Party rules for delegate selection are, to put it lightly, byzantine, and no popular vote can be reasonably calculated from their results.

medianprimaryvoter

Yellow is Rubio (DW-NOMINATE Joint Scaling: .579), dark yellow is Trump, light blue is Kasich (DW-NOMINATE Joint Scaling: .313), blue is Clinton (DW-NOMINATE Joint Scaling: -.381), and dark Blue is the Bern (DW-NOMINATE Joint Scaling: -.523).

North Carolina’s result was due to Democrat overregistration there. Ohio’s result was due purely to the favorite son effect. I’m not sure what went on in South Dakota, as Trump did very well there in the general election, especially in areas Hillary Clinton did well in during the primary, curiously enough. As one can see, the New Hampshire result portended it going blue; the Iowa result portended it going red. Had their primaries been open, South Dakota’s and Pennsylvania’s primary votes would probably have had a somewhat different composition.

The below are the primary results, from New Hampshire (the first Trump victory; I didn’t bother with Iowa due to the presence of too many candidates with known ideology scores, like Paul and Santorum, and the fact the ultimate nominee of the Republican Party didn’t win it) to Indiana, the last contested Republican primary. The contests are presented in the chronological order the Republicans had them. This is due to the convenient fact the dynamics of the Democratic contests, unlike those of the Republican ones, did not substantially change from February to June.

primaryresultsnhtoin

GOP only:
primaryresultsnhtoingoponly

Data here.

Compare the 2012 Republican race:

primaryresultsnhtoin2012

After Indiana (the 2012 equivalent were the Wisconsin [Romney v. Santorum] and Virginia [Romney v. Paul] primaries), pundits complained about lack of party unity for Trump:

All I can say is, Mitt Romney won Pennsylvania with no real opponents with 58% of the Republican vote. Trump won the same state with two viable opponents, one of the strongest exponents of modern conservative Republicanism in the Senate and one of the strongest Republican candidates with the college-educated and a popular and accomplished current Republican governor of a neighboring key swing state, with a mere 56.6% of the Republican vote. Same with New York, which Mitt won uncontested with 62.4% of the Republican primary vote and Trump won facing the same opponents as in Pennsylvania with a mere 59.2% of the Republican primary vote. Even more the case with Rhode Island, which Mitt won uncontested with 63.0% of the Republican vote, which Trump won (running against the same opponents as in Pennsylvania and New York) with a mere 63.8%. The fact is, by the New York primary, the Republican Party was solidly unified behind the Only Man who Could Even Remotely Save the Nation, the Greatest GOP Nominee Since Reagan, the man who can and may well Make America Great Again, the businessman whose endorsement Rmoney appreciated and craved during the 2012 Republican primary season, The Donald.

The most unusual Romney overperformance over Trump in the uncontested phase of the primaries is New Jersey, Trump’s best state during the primaries, which, however, is a blue state with no real friendship for social conservative Santorum and a decent deal of elite Acela Corridor pro-Kasich #NeverTrumpers. The most unusual Trump overperformance over Romney in the uncontested phase of the primaries is in South Dakota, probably due to the Santorum social conservative vote. Both New Jersey and South Dakota were closed primaries. Ultimately, Trump ended up overperforming Mitt Romney both in South Dakota and New Jersey in the general election.

Advertisements