Voters in last week's midterm election appear to have voted more liberally on ballot measures than on the candidates -- even in more conservative states and on more conservative measures.

We looked at 125 of the measures, propositions and constitutional amendments that appeared on ballots last week to gauge how voter opinions on them compared to the candidates higher up on the ballots. States approved ballot measures that would expand the minimum wage or legalize marijuana even as they voted overwhelmingly to elect Republicans to the Senate and the state House, as numerous reports have indicated. We wanted to figure out the extent of that apparent conflict.

The 125 measures were on the ballot in 40 states. We gave each measure a rating -- conservative, neutral or liberal -- based on its content. Most of them were neutral, which included bond measures and anything that seemed as though it could go either way or received strong bipartisan support. For each state, we also figured out the partisan split in the highest statewide office on the ballot. (In Missouri and Washington, we averaged House ratings for each party.)

The ballot measures broke down like this.

Then we compared the margin of support for the politically charged measures with the margin of support for the candidate under consideration. So Alaskans picked Dan Sullivan (R) to replace Mark Begich (D) in the Senate by 3.2 points and backed marijuana legalization, a liberal proposal under our definition, by 4.6. So, Alaskans voted 7.8 points more liberally on marijuana legalization.

On the 16 conservative ballot measures, voters were 4.7 points more liberal than their candidate voting. On the 26 liberal measures, they were 15.2 points more liberal. (The medians were 12.5 and 18.6, respectively.)


You'll notice that the graph uses "more Democratic" and "more Republican." We're comparing apples and oranges a bit here, because we are comparing liberal vs. conservative politics with Democrats vs. Republicans. As the example of marijuana probably proves, those categories don't always overlap cleanly. Not to mention that a different set of considerations comes into play when one is evaluating candidates: who the person is, his or her background, and even partisanship. That almost certainly colors those decisions.

One last note. Contrary to expectations, most of the initiatives -- 87 of the 125 -- received more than half of the vote. This doesn't mean that those 87 passed; some, often spending-related measures, needed a two-thirds majority. Neutral ballot propositions were more popular overall, receiving 15 net points of support, on average. Conservative measures averaged at plus-12. And liberal propositions were at plus-11. But because they did, in fact, often pass in more Republican states, it made a bigger difference in our above calculations.