A few days ago BlueSouth commented:
"The writer of the other kos diary asked the question, "why split the data up by area code?"
My thought was just that it is easier since you are dealing with IVR, but I know you pull your lists from voter records, so why not congressional districts? The other writer seemed upset by it for some reason, to me it really doesn't matter, since a cross tab is a cross
tab, but it has made me curious."
Good question. We split it by area code because it is easy. We can and do make geographical cross tabs in other ways, like Congressional districts, especially if a client wanted it.
I personally like using area codes because it splits the Triangle, Triad and Charlotte into distinguishable regions, where congressional districts do not. Also, with only 6 regions each cross tab is somewhat statistically significant, but with 13 regions, for example, the significance would decrease significantly.
Subscribe to:
Post Comments (Atom)
2 comments:
The significance is a great point.
Your polls are usually 600 people for a primary poll correct? What size would they need to be to split it up by congressional district and be as significant?
Our polls range between 400-800 people. Let's say its 600. Over six regions, each region would have roughly 100 people. It would take a poll of 1300 people to create congressional district crosstabs of about 100 people each.
Post a Comment