Carl Klarner writes:

I’m currently doing work on state legislative elections that uses Democratic success as the dependent variable. I do these analyses with either the percent of the two-party vote for the Democrat as Y, or a dichotomous measure of a Democratic victory as Y.

The problem is, I don’t know how to handle free-for-all multi-member districts. The range in the number of positions from 2 to 11 or so places. If ordered logistic regression was used, the problem would be that the number of possible Democratic victories would vary by the number of positions in the district. Is there a way to do an ordered logistic regression where you can specify the overall maximum number a case can obtain? Additionally, can you think of a way to model this but keeping the continuous measure of Democratic success so that information on how successful isn’t thrown away?

My reply:

First, I recommend using vote share rather than win/loss as an outcome. Using win/loss throws away data. And, from the standpoint of vote intention, the difference between 51% and 53% is the same as the difference between 49% and 51%.

For the multimember district issue, my quick suggestion would be to take D/(D+R), where
D = total votes for all the Democratic candidates in the district
R = total votes for all the Republican candidates in the district.
I could see that there will be some settings where this won’t work, but I’d think it would be a good start.