How to test election results between two election years?

#1
Hi All, I would like to test two general election results between two dates whether there is a significant change in any party's vote percentage (or vote number)? There is a change in registered voter numbers between election years also. How can I be sure that also sample size (or population size here maybe) is not a problem for comparison?
 

Karabiner

TS Contributor
#2
What is meant by significant here? A test of statistical significance would
consider the actual voters in both years as two samples, but that doesn't seem
to make much sense. Samples of actual voters, from what populations of actual
voters?
In addition, a frequentist test would assume as Null hypothesis that the difference
of a party's vote percentage between the years was exactely 0.00000 % between the
populations from which the two samples were randomly drawn, and that idea
does not seem to make much sense either. In addition, sample sizes in elections
usually are so huge that this Null hypothesis is certainly rejected. Do you perhaps
mean by significant something like important, large, relevant?

With kind regards

Karabiner
 
Last edited:
#3
Thank you for your reply. So here the question in simple terms is whether a political party statistically significantly increases its share of votes relative to the others on the basis of previous election results. Voters registered the previous term was 199k, the current term is 203k by the way.