We thank Dr. Arrington for his kind words about our article. In this letter we respond to his two main criticisms: that our use of the term bias in describing the situation facing the Democratic Party in 2006 is misleading and that using total vote to assess the results of the election is preferable to using average district vote, as we did in the article.
With respect to our conception of bias, we followed Gelman and King (Reference Gelman and King1990; Reference Gelman and King1994) in defining partisan bias as the expected percentage of seats over 50% a party obtains if it receives exactly half the average district vote. It is mathematically equivalent to note that there is partisan bias if the Democrats need more (or less) than half of the two-party vote in order to have an even chance at winning a majority of the seats in the House. In any case, this is one reason why we evaluate all our seats-votes curves at the 50% point.
The majority of Arrington's letter details a procedure for calculating the total vote obtained by the Democrats in the 2006 election. Arrington is correct that we did not account for runoff elections or uncontested races in our calculation of the total vote over time. As he notes, however, his estimates of the total vote deviate little from ours, which we presented in Figure 3. In addition we did account for both runoffs and unopposed races in computing the average district vote, which was the focus of the article. Endnote 4 explains how we treated uncontested races, while endnote 11 details how we treated runoff elections in the 2006 race.
Thus, the main point of contention is whether it is better to use total vote or average district to summarize election results. We do not deny that for some research questions it will be preferable to use the total vote. However the average district vote is well suited to estimating seat-votes curves, given that they depend on variation at the district level. And, as we noted in the article, average district vote is relatively immune to turnout variation and represents the national level of support for the party. It also can be viewed as more closely corresponding to national representation, because congressional districts have roughly equal population, and members of Congress represent not just voters but all Americans, including noncitizens, people under 18, people who are ineligible to vote, and people who simply choose not to vote (perhaps because there are no competitive elections in their districts). Another option would be to use population-weighted average district vote. In any case the fact that Democrats won only about 50% of the total vote in the elections leading up to 2006 does not change the fact that their average district vote was significantly larger than the share of seats they won. Our article explored the reasons for this, and how the Democrats were finally able to win at least 50% of the seats in the House.
