retain Jacksonian and anti-Jacksonian party labels, I have relied on party coding in the ICPSR
#9822 data set. Conflicts between the ICPSR coding for the 24
th
and Poole and Rosenthal
coding for later Congresses were cross-checked and corrected with the current Biographical
Directory of the United States Congress (http://bioguide.congress.gov). Third party members are
excluded from the event history and pooled logit analyses so that the party dummy variable is a
true dichotomous measure.
Slave Percentage. Slave populations and percentages are among the few district-based measures
available for the 1830s House. For southern districts, I have coded census-derived slave
percentage data from Parsons, Beach, and Hermann (1978) and Parsons, Beach, and Dubin
(1986). The coded percentages from the Parsons data combine free blacks and slaves in the
south.
Region. Regional variables (north/south divisions and coding for northwest and border
members) follow the ICPSR state code regional divisions, with the exception of Missouri, which
is coded as a border state in the analysis.
Antislavery Vote. Using electoral data provided by Craig Goodman, and following his general
approach of using antislavery vote as a measure of constituency interest (Goodman 2002), I have
created an indicator variable for the presence of antislavery vote in each northern House district.
This variable reflects district votes for parties associated with antislavery positions in the
preceding House election. The Liberty party and Anti-Masonic parties are the two parties whose
votes are coded for this time period; other antislavery parties such as the Free Soil party do not
emerge until after the period of this study.
Electoral margin. The electoral margin variable represents the difference between the
incumbent’s vote percentage in the election to the current House and that of his opposite-party
28
Convention