The creation of a reasonably correct data set for both House and presidential election results was fairly involved. While I would like to state that the results are absolutely correct, I still occasionally find an error. Furthermore, as will be noted in the following sections, there can be disputes about what a so-called correct score is for some elections. First, the creation of the House data set will be discussed, and then the presidential data set will be discussed.

HOUSE ELECTION RESULTS

With regard to House results, I relied on Michael J. Dubin's (1998) United States Congressional Elections, 1788–1997. This work allowed the correction of several kinds of error that show up in Inter-University Consortium for Political and Social Research (ICPSR) data files.

One kind of so-called error is particularly noteworthy. In both California and New York, cross-endorsement of candidates has occurred, and continues to exist in New York. In California a candidate might run with the endorsement of the Democratic and Republican parties. In these cases Dubin (1998) records the votes cast on each party endorsement line of a candidate. I was able to verify the actual party affiliation (not endorsements) of candidates by checking their affiliations in the prior Congress, using either results for prior elections or by consulting the Biographical Directory of the United States Congress (http://bioguide.congress.gov/biosearch/biosearch.asp). In the ICPSR data set, many of these districts have no recorded votes, and these districts end up missing in analyses of vote percentages.