22 March 2017
Did all-mail ballot elections increase voter turnout in Oregon?
That’s the official truth as told by former Oregon Secretary of State Phil Keisling and others, and is based on the turnout of registered voters statistic. Writing in the Washington Monthly in January, Keisling said:
Turnout in Oregon, my home state, used to be buoyed by demographics: through the early 1980s, income and education levels were above the national average, and 93 percent of the population was white.
But by 2000, Oregon’s timber industry had collapsed and our population had grown steadily poorer and more diverse relative to the rest of the country. (Today, Oregon’s per capita income is 91 percent of the national average, and only thirteen states have a larger Hispanic share of population.) Meanwhile, from 1980 to 1998, turnout of registered voters fell from about 77 percent to 71 percent in presidential elections, and from about 77 percent to 59 percent in midterms. That decline looked set to continue.
But in 1998, Oregon voters approved a universal vote by mail system. In the 2000 presidential election, voter turnout leaped to 80 percent and has since climbed as high as 86 percent, while midterm turnout rebounded to an average of 71 percent. This decade of higher turnout happened even as Oregon lost its status as a presidential battleground and was seeing far fewer competitive statewide elections.
That sounds impressive — but the turnout of registered voters is the wrong statistic for analyzing voter turnout. Experts in the field, among them Michael McDonald of the United States Elections Project, employ a different statistic, the turnout of the voting eligible population (VEP). The VEP is the voting age population (VAP =>18 years of age) minus foreign nationals, prison inmates, and other disqualified classes. In Oregon, the VEP is 93 percent of the VAP. Registered voters comprise 85 percent of the VEP.
Why VEP turnout is the more illuminating statistic can be illustrated by the imaginary state of Fremont, which has a population of 1.4 million, a VEP of one million, and just 100,000 registered voters. In a general election, 90,000 votes are cast. The registered voter turnout is 90 percent, good enough for national bragging rights, but the VEP turnout is nine percent, low enough to conclude that democracy is in pretty sad shape in Fremont.
Voter registration rolls, incidentally, tend to cycle from inflated with deadwood to deflated by periodic purges of inactive and allegedly ineligible voters.
Which brings us back to Oregon. Did going to all-mail ballot elections there increase VEP turnout? The answers are No for midterm elections, and Maybe for Presidential elections.
Applying the Student’s T and Wilcoxon tests confirm’s one’s eyeball evaluation of turnout in the midterm elections: there’s no significant difference between the pre and post all-mail ballot eras. For Presidential elections the all-mail ballot era produces a 3.2 percent higher mean VEP turnout. This is not significant at the 95 percent level, but the difference does exist. Can we then conclude that switching to all-mail ballot elections boosted VEP turnout in Presidential elections?
No.
The switch to mail ballots was not the only variable affecting turnout. In 2004, there were eight ballot measures on Oregon’s general election ballot. One concerned taxes. Another addressed timber production. Medical marijuana dispensaries was the subject of another. And one ballot measure defined marriage. Interest in these issues could have boosted turnout regardless of the voting methods, as could have unusually interesting or divisive contests for elective office. Exit polls and other post election surveys might not be of much help in teasing out the mail ballot effect because the sampling error of those surveys may exceed the 3.2 percent difference in the pre and post all-mail ballot eras.