The voters who didn't return were redder, but Jones's number doesn't move
About 640 of the voters who took our September poll didn't take this one, so we ran the same test we ran in SD-6: we took their September answers and weighted them the same way we weighted our October poll, including our likely-voter model. As likely voters, they're a more Republican group than our September sample as a whole. With leaners, they identify as Republicans by 14.3 points, compared to 12.8 for the full September sample, and they recall voting for Donald Trump by 13 points, compared to 8 points in September. They're also warmer to President Trump: his net approval with this group is -5, 4 points better than the -9 he had in our full September sample. They were a little less committed to voting, too: before the likely-voter adjustment, 84% said they were definitely voting in September, compared to 86% of the full September sample.
Even so, Tom Jones is at 48% with them, the same share he had in our September poll and a point below his 49% in October. What they lack is Malone support: James Andrew Malone is at 42% with this group, compared to 45% in September, with 9% undecided. Unlike in SD-6, their undecideds are redder than the undecideds in our full September sample: they back Josh Shapiro 39% to 20% over Stacy Garrity, a 19-point margin compared to 29 points in September, and 54% disapprove of President Trump while 31% approve, compared to 58% and 30%. Losing these voters explains part of why our October recall and our October undecideds look bluer, but it doesn't explain away Malone's gains, and Jones's share is flat no matter who we count.
State Navigate - Pennsylvania SD-36 poll briefing, PDF page 9.