29
Appendix 4. Variable Coding for Empirical Analyses on Senate Elections
Variables relating to Candidate Ideology (see equation 3)
Variable
NES Pooled Senate
Election Study Variable ID
Description/Coding
Candidate
ideology
Respondent’s Liberal Con-
servative Placement of candi-
date (v.556 for Democratic
candidates; v.557 for Repub-
lican candidates)
Candidate’s position computed as the mean location as-
cribed to the candidate, averaged over all respondents
who were willing to locate the candidate.
State ideology
Respondent’s Liberal-
Conservative Self-Placement:
v547
Mean position over all state respondents who were will-
ing to self-place, averaged over all state respondents in-
terviewed in the year of the senate election.
Partisan
divergence
Respondent Lib-Cons Self-
Placement: v547
Respondent Party Identifica-
tion: v497
Difference between the State ideology (defined above)
and the mean ideological self-placement of all respon-
dents who self-identified with the candidate’s party. For
these computations respondent were coded as Democratic
Identifiers if V497 = 0,1,2 and as Republican identifiers if
V497 = 4,5,6
Partisan size
Respondent Party Identifica-
tion: v497
Proportion of state respondents who self-identified with
the candidate’s party
Variables Relating to Democratic Candidate’s Vote Margin (see equation 4)
Variable
NES Pooled Senate
Election Study Variable ID
Description/Coding
Democrat’s vote
share
Party of Candidate Voted for:
v321
1 if respondent reports voting for the Democratic Candi-
date, 0 if Republican Candidate. Votes for third party
candidates and abstain coded as missing.
Democrat’s
proximity ad-
vantage
Resp. Liberal-Conservative
Placement: v547
Resp. L-C Placement of Dem
Candidate: v556
Resp. L-C Placement of Rep.
Candidate: v557
Squared distance to Republican candidate minus squared
distance to Democratic candidate (see equation 1). Esti-
mates use the state/year average of v556 & v557, respec-
tively, in this calculation.
Convention