one of the big things is likelihood to vote; another is response rates.

how often is your 65+ sample disproportionately skewed with people over 80, who are more likely to stay on the line and answer questions? and, at what age does the increased likelihood of seniors to vote start reversing?

you could end up with a large sample of 80+ in the 65+ demographic and conclude they're more likely to vote, when they in fact can barely leave the house at all.

and, opinions and worldviews are likely to be dramatically different nowadays between a fully functional 69 year old and a totally geriatric 87 year old.