Talk:2024 United States presidential election in Pennsylvania

Polls to exclude

After reading TNR article on ideological pollsters trying to game the averages, it seems like we should also exclude unreliable pollsters like Rasmussen, ActiVote and Trafalgar. Anyone else have thoughts here? Superb Owl (talk) 23:37, 24 October 2024 (UTC)

While TNR is a reliable source, it's also an acknowledged left-leaning opinionated source. I don't think we should be in the business of including or excluding formal polling data based on some sources complaining that they're biased; virtually all polling involves some degree of bias. The whole point of providing a broad list of polling sources here is to let the reader come to their own conclusions from the data available, not for us to be guiding them to a conclusion. cheers. anastrophe, an editor he is. 05:25, 25 October 2024 (UTC)

My issue is that WP:RS beyond TNS (eg 538) have found Rasmussen to be biased to the point of also being unreliable. Why would we include polls that WP:RS find to be unreliable? Wikipedia does not let anyone upload a poll that they did, so what I am wondering is where is the line of reliability on polling and is it any different from WP:RS guidelines? (538 also rates Trafalgar as 0.7 stars out of 3 fwiw) Superb Owl (talk) 20:20, 25 October 2024 (UTC)

The problem again is one of selection. Nate Silver's Silver Bulletin (he of course being the originator of 538) gives a graph of 'reliability' for polling sources by their own data regressions. It also shows Trafalgar and Rasmussen as biased right, but similarly shows Public Policy Polling and Morning Consult biased left. All of the polling sources are biased one direction or the other to some degree, with only one/two sources at 'zero' bias per Silver. All of these polls are used in these "2024 United States presidential election in" articles (ten of them). Different reliable sources come to different conclusions - the least biased per Silver are different from the least biased per 538. But are we going to throw out only the right leaning polls? Who makes the decisions on which RS's data and interpretations are the most "accurate" (as if there's any true accuracy in public opinion polling)? How narrowly do we distinguish what is "accurate" and by which sources?

All I see is a giant can of worms, and one not worth opening with only a little more than a week left until the election. cheers. anastrophe, an editor he is. 21:24, 25 October 2024 (UTC)

(Here's a direct link to Silver's analysis of tendency to lean towards one or the other: https://www.natesilver.net/p/which-polls-are-biased-toward-harris?utm_source=publication-search ) cheers. anastrophe, an editor he is. 21:31, 25 October 2024 (UTC)

Silver analyzes much more than bias but also transparency, reliability and other best-practices. I think we should exclude any polls that receive less than 1 star on Silver's rating and do not see how doing so 11 days before an election has any bearing on the merits of cleaning up Wikipedia from unreliable sources. There are 259 pollsters who have a 1-star rating or better. Plenty of conservative pollsters fall into that camp. Superb Owl (talk) 21:37, 25 October 2024 (UTC)

Silver's results don't have stars. If you insist that this is a valuable exercise, then Public Policy Polling and Morning Consult must also be culled from this - and the nine other similar articles - as they are similarly biased but to the left. There's no reliable sources policy I'm aware of specific to opinion polling itself. Note that 538 is not listed at WP:RSP, so thus far we have a single source - heavily biased left, that you are suggesting is basis for removing right-biased polling. I see nowhere in your arguments an interest in a neutral approach of removing least-reliable polls regardless of which way they lean; perhaps I'm mistaken. cheers. anastrophe, an editor he is. 22:04, 25 October 2024 (UTC)

Here's a possibly related - or unrelated - thing I just noticed in the poll results in the article. Next to some polling organization names is an (R) or a (D). There are ten (D)'s and 34 (R)'s - but nowhere on the page that I can find are those designations specifically defined or explained. No 'key', no nothing. I'm curious how and why they're there; there's no apparent pattern or consistency to them. Perhaps before anything else, those designations should be removed, since they are unidentified. cheers. anastrophe, an editor he is. 22:28, 25 October 2024 (UTC)

once again, we are talking about different things entirely. You seem obsessed with political leaning, I am only concerned about reliability. Does anyone else have any thoughts? Superb Owl (talk) 03:52, 26 October 2024 (UTC)

Results by congressional district

Is there a reason why results by congressional district aren't shown in detail when they're an option on the map? 2A02:C7C:EE9B:B100:955F:76F7:B017:C81E (talk) 11:45, 24 November 2024 (UTC)