This is a topic I’ve been hoping to someday write extensively on, and the truth is I’m not quite ready to do so. But I have an observation that is so startling and so much in line with my thinking on this issue that I thought I’d share it as a way of introducing the topic, as I continue to think about it and collect data.
There are a lot of things right and wrong with our election system, but I’m going to propose that a particular problem, if solved, would make many of the other problems go away.
People are not in the habit of voting, and therefore, don’t get into the habit of voting. Many think (correctly, actually) that some elections matter more than others, and they use that as a handy excuse to not bother to vote. We see this clearly in the off and on pattern of voting during presidential vs. “off” years. See? We even call them “off years.” Take a year off! Don’t bother to vote!
If there was more equivalence between election years — if most of them were considered important for some reason — then more of the years would be higher turnout years, and those years would be feeders, potentially, for later years, so over time the overall turnout would go up and up, rather than up-down-up-down. In other words, our current system involves alternating between taking steps forward and taking steps backwards, rather than forward progression.
But this was not always true. This is a bit complex, but I’ll give you a provisional explanation. In the old days, such as the 19th century, state elections were much more important than they are today, to national politics. If you were running for President, it mattered a lot which party was in control in each state, and the composition of the senate was controlled by state politics, not voting (legislatures sent US Senators to Washington, not voters). That meant that state elections were very much part of the national election process. This was even more important, and complex, when two other thing were true. First, there were more than two parties, even if most states really had only two going at once. Second, the parties were not perfectly aligned with platform. So, for instance, prior to the Civil War, southern Democrats were mainly pro-slavery, while northern Democrats were split on slavery and their main interest was something else. For several decades a century, also, we tend to have a party turnover, where one party eats itself (and is ultimately finished off by other forces) while a different party takes over. (This may be happening now.) During those periods, the two party system is obviously a three party system for a while.
Between the greater heterogeneity in political orientation of parties, with that variance structured by state and region, and the importance of state elections to the presidential election process and the Senate, people watched the state elections outside of their own states, and within the states, much more closely.
This all probably made each election year more interesting to everybody than it is now, but a second factor, much more important but potentiated by the aforementioned factors, also pertained. Today, most Governor’s races are done in a four year cycle, with over half (36) being during the so-called “midterm” year (two years offset from the POTUS election) and most (but not all) of the rest being during the POTUS cycle.
This guarantees that almost all governor-level election activity is done on a two year cycle nationally. The POTUS election and the governor’s elections are therefore in sync with each other, as well as with Congress. This is because the US Senate elections are, effectively, every two years as they are never held on “off years” and the US House is voted on every two years, again, in sync with POTUS.
For this reason, every two years there is something to do, every two years there is almost nothing to do. And, those two year periods, synced as they are with POTUS, are divided into two parts, POTUS years and midterm years.
For this reason, a double-digit percentage of the voters get three years off and a similar number get two years off, by their own way of thinking about what is important. So hardly anybody goes and votes one, two, or three out of four years.
This pattern of being almost perfectly in sync with POTUS and Congress is new. Many states, in the 19th century, had three year governor terms, or two. There have been states with four year terms but on odd years (at present I think there is only one of these).
Given a) the greater importance, as a national story, of state elections, and b) the nearly chaotic pattern of elections, where there was alway something going on, meant that the difference between one year and another was not so great and this, because of the effect I propose above, meant an overall higher turnout.
Starting some time after 1900, state by state (and probably in cities as well, for mayor, etc.) elections lined up in the two year pattern, and terms for governor changed to four years across the board. As this happened, US citizen participation in elections fell.
The overall pattern shows a huge increase in voter turnout from the early days to the Federal era. That is not the subject of discussion here. Subsequently we experience a high rate during presidential election and a lower but still high rate during midterm years. The midterm years during this period are higher than some presidential years now, and represent an 85% or so fraction of the presidential years. Then we see a drop, from around 1890 to the 1930s, followed by a new pattern with two features. First, the overall participation is way down, with midterm elections being well below 50%, and second, the drop during midterms is greater, with midterms being about 66% of presidential years.
I think both of these patterns can be explained, at least in part, by the normalization of elections to be on a 4-year/2-year cycle synced nationally.
There, I connect details of timing of election with a major overarching pattern, using a beautiful hypothesis. What remains is testing the hypothesis in several cases where we can see the details, to see if it works out as expected. I have one case for you. The following graph shows Minnesota primary turnout over time for presidential and midterm years.
This is the observation I stumbled on last night that prompted me to write this post. Here we see the large binary oscillation between POTUS and midterm years, as expected, from the early 1960s to the present. We aslso see that over this time there is a steady decrease in participation. The break point, 1960, when the pattern changes, is the year Minnesota went form a two year to a four year governor term. Boom.
We also see a diminishing in the up and down cycle since 1994, and some interesting anomalies. That is for discussion another time. At this point, we can say that with respect to this hypothesis — syncing elections ruined turnout — that the center holds.
One solution to this problem would be to unsync the elections. I can think of several ways to do this.
1) Keep the two year term for the US House, but give half of them one three year term once, so about half the house races are happening every year.
2) Move around the Senators so there are senatorial races every year.
3) Move all gubernatorial races to one of the other odd year that is not a POTUS or Midterm year.
I predict that if we did all of that, participation in our elections would spring to 80%. Well, not spring. Move. It would take a decade, but it would get there.