October 23, 2006

On Waves and Stability - Part II

My last post looked at evidence of an "anti-Republican wave" in the form of consistent Democratic advantages on the so-called generic House vote. Today I want to consider another indicator of that potential wave, the various measures of an "enthusiasm gap" between Democrats and Republicans. The generic ballot provides a blunt measure of the vote preferences of registered or likely voters, but it tells us about another decision many will make, whether they will vote at all.

Predicting the likely turnout would be helpful, because the last two off-year "wave" elections had higher than average voter turnout. As the table below shows, voter turnout as a percentage of the eligible population was 39.8% in 1982 (when Democrats gained 27 seats) and 38.6% in 1994 (when the Democrats lost 54).

Most of the analysis I have seen of the 1994 election reached the same conclusion as this report by Curtis Gans of the Committee for the Study of the American Electorate:

[In 1994] Republican turnout was up in every region of the country, while Democratic turnout was down in every region of the country except the Middle Atlantic States and the Far West, where the party recorded exceedingly modest gains.

Will turnout work in favor of the Democrats as it did in 1994 and (presumably) in 1982 as well?** The widely reported "enthusiasm gap" detected on many recent national surveys suggests that may be a possibility. As the table below shows, Democrats have been more likely than Republicans to report feeling "more enthusiastic than usual about voting this year." While Republicans had expressed more enthusiasm earlier in the year, the gap has widened significantly (14 to 18 percentage points) in recent weeks.

An analysis (subscription only) by Gallup's Jeff Jones in April indicated that this pattern is a "departure from recent history" in two ways:

This is the first time in a midterm election year since Gallup began asking the question in 1994 that a significantly higher percentage of Democrats than Republicans claimed to be more enthusiastic about voting.

Additionally, this is only the second time that Gallup has found a higher percentage of Democrats in the "more enthusiastic" than in the "less enthusiastic" column. The other time was at the beginning of this year.

The following table shows the results on these questions among registered voters for Gallup's final pre-election survey in 1994, 1998 and 2002:

Jones questioned the relationship between enthusiasm and turnout with a rather complicated point in his analysis. While demonstrating that the party that gains seats on an off-year election typically has "a better net enthusiasm score than the losing party," Jones argues that an enthusiasm gap "expands the party's strength at the more basic level of party identification, rather than increasing turnout rates among that base." His evidence is that party identification among likely voters in Gallup's final pre-election surveys has been "fairly stable" from 1994 to 2002, "regardless of the party's enthusiasm about the election."

The analysts at the Pew Research Center were less hesitant about the connection between enthusiasm and turnout in reporting on their most recent survey results. Looking at their historical data (in a report well worth reading in full), they suggested the possibility of a "higher than normal" turnout in 2006 given the "record-high levels of Democratic enthusiasm:"

Turnout in the 2006 midterm election may well be higher than normal, given the level of interest expressed by voters. Today, 51% of voters say they have given a lot of thought to this November's election, up from 45% at this point in 2002 and 42% in early October of 1998. Even in 1994 a recent high in midterm election turnout just 44% of voters had thought a lot about the election in early October.

So regardless of the underlying mechanism, we have more evidence of the "anti-Republican wave" that Mark Mellman talked about. I have expressed skepticism in the past about the ability of opinion surveys to precisely predict levels of voter turnout, but the data above certainly argue that some sort of Democratic turnout advantage is likely this year. And as with the generic vote, these measures are worth watching closely on the final round of national surveys for signs of any last minute shifts.

Of course, the size of the wave -- as measured by the the generic vote and "enthusiasm gap" -- is just half of the puzzle. As Mellman said, the real question is what happens when this apparent anti-Republican wave crashes against "a very stable political structure." I'll take that up in the next post.

[Continues with Part III]

**I assume that similar analysis exists somewhere in the political science literature for the 1982 elections showing Democratic regions turning out at a higher rate than Republican regions, but I have been unable to locate anything in the public domain online. I did find a 1982 New York Times interview (via Times Select) with Ann Lewis, then political director of the Democratic National Committee, in which she claimed that "voter turnout among Democratic constituencies was high" in that election.  But she did not offer any data at the time to back up the assertion.