I am new to this list, and new to the field of EI and am hoping I can
get some advice.
I am a history graduate student at the University of Canterbury, NZ, and
I am looking at gender and voting over a 50 year period, 1893-1954,
during which there were 22 elections.
The official election results for this period included the number of men
and the number of women who cast a vote for each of the 80 odd
electorates for each election.
I am keen to use EzI to test relationships between gender and the party
vote and between gender and the vote for and against the prohbition of
alcohol.
Since I know gender turnout I have used this to test EzI, using
x=proportion of the registered voters who are women
t=overall turnout (number of votes/number of registered voters)
n=number of registered voters
to estimate
BetaB=female turnout (number of votes cast by women/women registered as
votes).
and BetaW=male turnout (number of votes cast by men/men registered as
voters)
The software seems to work okay, and indeed is very easy to use, but
gives results that systematically overestimate female turnout and
underestimate male turnout.
I tried using co-variates, but was warned I needed to adjust the
priors. (I got the message "Warning: Including covariates without
priors works but is not generally recommended").
I am uncertain how to determine priors, other than just taking a random
stab. Are there any guidelines? Can any one point me towards
publically available examples that show how others have set their
priors?
Any help would be much appreciated.
Thank you
Linda Moore
History Department
University of Canterbury
Christchurch
New Zealand
lmm72(a)student.canterbury.ac.nz
-
ei mailing list served by Harvard-MIT Data Center
List Address: ei(a)latte.harvard.edu
Subscribe/Unsubscribe: http://lists.hmdc.harvard.edu/listis.cgi?info=ei