I wrote a few weeks ago about the commonly observed phenomenon that the president's party tends to lose House seats in mid-term elections, but I didn't really get into why this happens. What's going on there? Is it pretty common for people to be enamored of a presidential candidate, but then turn against him and his party two years later when he's failed to deliver on his promises?
Not really. Lynn Vavreck had a nice post out in April explaining that mid-term seat loss isn't really about swing voters changing their minds between elections. Rather, it's about people staying home. Presidential campaigns, with their high levels of spending and media attention and turnout efforts, draw a great many peripheral voters into the electorate. These are the people who put the winning presidential candidate over the top. In a mid-term, which sees substantially less attention and spending, many of these voters stay home. Turnout in mid-term elections tends to be around 20 percentage points lower than in presidential ones.
If your state requires voters to register weeks or even months before Election Day, if it requires photo identification for voters, or if the polls are open for shorter hours, fewer people will participate.
Can this turnout be manipulated? A bit, but not easily. National party groups may try to boost turnout in mid-terms, but they rarely have much impact. Part of the problem is that there just aren't that many competitive House or Senate elections across the country. In a presidential campaign, everyone's exposed to the election news to some extent, but the campaigning in a mid-term year is very focused on a few competitive areas. And, perhaps, more importantly, the president's name—the most polarizing name in the country—doesn't appear on the ballot.
What else can influence turnout? There's quite a bit of evidence suggesting weather has an impact. Steven Rogers has some recent research showing that presidential coattails (a presidential candidate's ability to help congressional candidates of his party) are shortened when there's more rain on election day. That is, rain makes it costlier for a person to vote, so fewer of these peripheral, non-habitual voters show up. This is consistent with some research by Brad Gomez, Thomas Hansford, and George Krause that examined county-level rainfall records and found that an inch of rain on Election Day tends to depress turnout by about 0.8 percentage points. They additionally found that poor weather seems to affect Democratic voters more, giving Republican candidates an advantage. The link between precipitation and participation is well established.
Logan Woods, a recent graduate student at the University of Illinois, extended this research to primary elections, and found some very tentative evidence of a precipitation effect there as well. But there's an important wrinkle here: The people less likely to turn out due to adverse weather are the more ideologically moderate. So bad weather tends to make the primary electorate more ideologically extreme, which might well mean that bad weather produces more polarized nominees. This is a chilling thought for the future given forecasts for more extreme weather.
Of course, turnout is also influenced by institutional rules. If your state requires voters to register weeks or even months before Election Day, if it requires photo identification for voters, or if the polls are open for shorter hours, fewer people will participate. And it doesn't help that we, unlike many other countries, hold our elections on a weekday, rather than on a weekend or a national voting holiday. Indeed, it's somewhat perverse that we hold more elections than almost any other democracy but make it relatively hard to participate in them.