Measuring turnout effects of vote by mail in California

I just got sent this recent article by Mark DiCamillo (Field Poll) published in the California Journal of Politics and Policy. This looks like a good journal with an excellent of editors.  I don’t know how the attached article made it by the proofreaders, however.  The data that are reported are fine, but I think the conclusions the author reaches are not really supported at all by the data.

DiCamillo writes, on pg. 1

The recent rapid expansion in mail ballot voting coincides with the rebound in voter turnout in California’s three most recent presidential elections, after an almost continuous 32-year decline. Table 1 shows the trend in voting participation in California presidential elections since 1976 and compares this to the number of mail ballot votes cast in these elections. Californians’ accelerated use of mail ballot voting appears to have had a beneficial effect on voter turnout in recent elections. Evidence for this can be found in the higher turnout rates of mail ballot voters compared to other registered voters (see Table 2).

It’s important that we look at the evidence for these claims.  DiCamillo makes, in my opinion, two errors:

  1. In the first figure, he plots presidential turnout rates in California since 1976 along with the number of by mail ballots.  He uses this to argue that the growth in turnout is caused (I can see no other way to interpret “has a beneficial effect on”) in part by voting by mail.
    The problem is that this attributes all or a significant part of the change in turnout to the increased use of a particular mode of balloting.  I am almost certain, based on substantial research in political science, that virtually all of that increase is due to the changing nature of California elections and the California political environment.
    If voting by mail were the cause–and the increased use not an effect–then why did it take so long for that effect to be seen (no excuse absentee voting has been around since 1978)?  If voting by mail causes increased turnout, how to explain the negative correlation from 1976-1996?
  2. DiCamillo’s second chart compares the turnout rate among those issued mail ballots to all registered voters in 2008, and again uses this to argue that voting by mail leads to higher turnout.
    It’s hard to know where to start with this claim.
    – It is impossible to conclude anything from a single year.
    – Excluding those on permanent absentee rolls, those requesting absentee ballots have already indicated a strong preference to actually vote. You can’t compare these voters to those who are just registered to vote who have made no positive action toward casting a ballot.
  3. – The data in Table 4 explain why the turnout differences exist.  It may be partially due to by-mail balloting, but it is much more likely due to the kind of citzens who opt for that mode: older, better educated, higher income.

This wouldn’t concern me much except that the reason I was sent the article is I am working with a post-election review commission in another state that is considering an aggressive move to voting by mail.  Misleading conclusions about the impact of voting by mail on turnout don’t help election administrators make good policy decisions.

The last point about the demographic profile of by absentee voters–I’ve made that point in at least four published articles at this point, as has Adam Berinsky, Michael Traugott, Michael Hanmer, Priscilla Southwell, Jeffrey Karp–almost everyone who has written on this topic.