What does primary turnout tell us about the general election. “Nothing,” states Harry Enten of fivethirtyeight.com. Respectfully — I’m a big fan of Enten and the website — I disagree. I’ve been noting starting Feb. 3, that Republican turnout has been higher than Democratic turnout in just about every state primary and caucus, and that Republican turnout is up sharply compared to 2012 or 2008 and Democratic turnout have been sharply down from 2008. And I think that that has some significance for the general election.
Of course, as Enten points out, year-to-year comparisons of turnout are often misleading. Different sets of states have held primaries in every election cycle starting with 1972, when primaries became the predominant means of selecting national convention delegates, so total turnout figures for one year aren’t commensurate with those for another.
When Texas doesn’t have a primary in one cycle and then has one in the next, or when it has a primary for one party and not the other, that makes a big difference in total primary turnout, which has no significance for November. In any particular year one party’s nomination may be uncontested (Republican in 1984 and 2004, Democratic in 1996 and 2012) or only mildly contested (Republican in 1972 and 1992) while the other party’s nomination is up for grabs. Or one party’s nomination fight may be settled much earlier than the other’s. Turnout naturally tends to decline when the nomination is not at stake.
Example: In 2000 Bill Bradley withdrew after Iowa and New Hampshire and left Al Gore essentially unopposed; the fact that Republican turnout was larger than Democratic turnout that year had no significance for the general election.
Also, state contests can affect turnout. Some states, like North Carolina, hold presidential, congressional and state primaries on the same day; non-presidential contests may boost turnout. Democratic turnout in Illinois this year was higher than Republican; one reason is party registration, but another may be the hot Democratic primary for prosecutor in Cook County, where 57 percent of Democratic votes were cast. So comparisons of the two parties’ turnout in a particular year or a party’s turnout as compared to past years ought to be made with caution.
Even more so when comparing, over time, turnout during the period 1968-88 of split ticket voting, when Republican won five of six presidential elections and Democrats maintained majorities in the U.S. House of Representatives and in most Americans’ state legislatures, and the period 1996-2016 of straight ticket voting, when there has been an increasing convergence of voting in presidential, congressional and state elections.
Enten, seeking as good analysts usually do the largest possible n (number of examples), looks back on primary turnout going back to the 1976 cycle. But Democratic turnout before 1988 was much higher than Republican turnout in the South — between 65 and 81 percent of Southern primary voting was in Democratic contests between 1976 and 1988 — even though in only one election (1976) did the Democratic nominee carry more than a handful of Southern states. (I define the South as the 11 states of the Confederacy plus West Virginia, Kentucky and Oklahoma.) Southern primary turnout was not just insignificant to the November result, it was hugely misleading.
That was also true, to a lesser extent, outside the South and particularly in states with party registration. Party registration tends to lock into place the party alignments of the past and it incentivizes voters in one-party localities or districts to register in the dominant party even if they vote regularly for the other in general elections. Thus landslide losers Barry Goldwater ran ahead of Republican party registration in New York City and George McGovern ran ahead of Democratic party registration in Upstate New York. Kentucky and (until last year) Oklahoma still had more registered Democrats than Republicans, even though they voted 61 percent and 67 percent for Mitt Romney in 2012; New Hampshire still has more registered Republicans than registered Democrats even though it has voted Democratic in the last three presidential elections.
But if more people turn out for the Republican than the Democratic primary in a state like Oklahoma, as happened this year but not in 2008, that may have some significance for the general election.
My argument is that comparisons of primary turnout have some significance for the general election in a period of party stability and straight-ticket voting, when large majorities of voters tend to support the same party in election after election. They give clues or hints as to general election turnout which, with the two parties closely divided, can make the difference between victory and defeat.
Of course there are very few clean comparisons to be made. Democrats had no primary contests at all in 1996 (Bill Clinton never even had to formally announce he was running for reelection) or 2012, and Republicans had no such contests in 2004. The only serious contest in the 2000 Democratic race, as noted, was over after Iowa and New Hampshire. The only years in which both parties have had contested nominations were in 2008 and this year. To get another example, you have to go back to 1988 — a year when Southern primary turnout was nearly two-to-one Democratic but George Bush carried every Southern state except West Virginia.
So what do I see in turnout so far that may have significance for the general election. On the Democratic side, we see much lower turnout than 2008 by blacks and Latinos — and a near-disappearance of Democratic primary voters labeling themselves as “moderate” or (as very few do) “conservative.”
I see Hillary Clinton’s narrower margins among Northern than Southern blacks as problematic—perhaps signs of lower turnout or decreasing Democratic percentages or both — for her. I see Sanders’ 80-plus percent wins among under-30 voters as a sign that Clinton will have trouble duplicating Millennial enthusiasm and turnout in 2008.
On the Republican side, I see increased turnout (including significant increases in evangelical turnout) as very interesting, coupled with the fact that Republican debates have drawn double to triple the audience as the most-viewed Republican debate in the past.
Some of this — it is unknowable how much — is due to interest in and curiosity about Donald Trump. And the potential for higher Republican turnout in November may be more than counterbalanced by either Trump’s nomination or by his rejection at the Republican convention. I think that the low Democratic turnout, together with the dire picture of the nation presented by the two Democratic candidates in the eighth year of a Democratic administration, is bad news for Democrats. Not the politics of joy!
So, problems for Democrats and advantages for Republicans, which they might very well squander.
That said, I agree with Harry Enten that same-year comparisons of the two parties’ turnout have less significance once one party’s nomination seems settled and the other remain up for grabs. You can make the argument that there was never any doubt that Hillary Clinton would be nominated, or that it became clear she would be after March 1, or after March 15.
Factors pointing in the other direction are that Sanders has carried nine states, come very close in four others (Iowa, Massachusetts, Missouri, Illinois) and has run about even outside the South. Sanders keeps getting a lot of votes, though delegate-counters point out he has no chance to win.
But then, delegate-counters pointed out at this stage in 2008 that Hillary Clinton had very little chance (though more than Sanders does now) to win. What is clear from the results so far is that the huge boom in Democratic primary voting in 2008 has not carried forward, whereas Republican turnout is considerably more robust than it was in either 2008 or 2012. I think those things have some significance — though it’s not entirely clear precisely what — for the general election.
