Anticipation for the 2018 midterm elections has already started to grow. While vote choice patterns across different groups prove similar between midterm and presidential years, turnout rates–and resulting electorate makeups– do not. I wanted to check the midterm-presidential gap in turnout rates within major demographic groups, and whether that has changed over time.
The best available data to measure these trends can be found at the U.S. Elections Project website, operated by Professor Michael McDonald. The data contains turnout rates from the Current Population Survey for every midterm and presidential election from 1986 to 2014 and broken up by four different racial, age, and educational attainment subgroups. Importantly, McDonald has a version of this turnout data in which he applies a vote overreport bias correction. Given how survey respondents notoriously overreport past voting in elections, this correction is very useful and thus I use this version of turnout in the figures below.
For all racial groups, the gap in turnout for midterm and presidential elections was the narrowest during the first half of this time period in the late 1980s and the 1990s. The gap steadily grows thereafter: while more people across all racial groups turn out to vote in presidential elections, turnout rates in midterm years stagnate if not decline. The pattern is most pronounced among blacks, likely generated by Barack Obama’s presence on the ballot in the 2008 and 2012 presidential elections, which explains the surge in black turnout at those times but lack of a corresponding increase during surrounding midterm years.
The one other key note here is the recent context: turnout in the last midterm election in 2014 dropped by a considerable amount for all racial groups. That was especially the case among Hispanics–with a record low of 21.1 percent turnout during this time frame–and other races–also with a turnout nadir for this period–of which Asians make up a big part.
For all of the age groups in the above graph, the midterm-presidential year turnout gap generally gets to its smallest level around the 1990s. Since then, the gap has grown for the three of the four youngest age groups here, a trend that follows a similar dynamic observed in turnout rates by race: for age groups 18-29, 30-44, and 45-59, turnout rates in presidential elections grow, while midterm turnout stay about constant. For Americans age 60 or more, however, turnout in midterms does not stagnate in the same way in the latter half of this time span, and instead steadily grows from 1998 to 2010. This explains why older age groups have increasingly taken up a larger share of midterm electorates over the last few decades. Yet older voters were not immune to the turnout drop in 2014, as all age groups experienced this turnout rate decline, especially Americans age 45-59.
The midterm-presidential election turnout gap among groups of different educational attainment also grows a bit in the second half of this time period. However, the trend of an increasing gap is not as strong as it is among other demographic subgroups, as it appears more stable for Americans of different educational levels. The gap tends to be larger among those with some college education or a college degree, followed by those with post-graduate education. Americans with less than a high school education, on the other hand, have the smallest gap of all four groups here: their turnout habits differ much less between midterm and presidential years compared to Americans of other educational levels. And once again, all groups experience a turnout drop in 2014, particularly among people at the two highest educational levels.
3/16/17 update: I had this piece re-printed at the Decision Desk, check it out here.