Editorial: Why polls are misleading, and it's in Trumps Favor

[Zeo]

This is a direct data interpretation from a September 2019 poll by New York Times/Sienna college of likely voters, and is solely my opinion.

I am doing something called cross reference polling interpretation, and it does lean toward to who ever wins the most categories wins, but that's it's intent.  Voters only have one choice in the ballot box, but these polls have several questions.  You can't just say the majority of this poll says T+1, when you as them to vote 3 times, based on criteria you give them.  And then weigh them on statistics(which how they do it isn't 100% clear.)

Meanwhile I just simply take each Unique category, and add all the percentages together and average them out.

	Trump	Harris	Percentile of responses	Spread	Outside of a 5% margin of error
Under 30	37	58	13	21	TRUE
30-44	39	45	22	6	TRUE
45-64	49	47	33	2	FALSE
65+	51	44	28	7	TRUE
men	54	40	not stated	14	TRUE
women	37	54	not stated	17	TRUE
no college	56	38	25* see note 1	18	TRUE
4 year college degree	35	61	25* see note 1	26	TRUE
black	15	79	11	64	TRUE
white	54	43	67	11	TRUE
Asian/Other(as per poll question)	43	46		3	FALSE
Hispanic	51	39	10	12	TRUE
Final result raw	43.42%	49.50%		6.08%	TRUE
final result without double counting Hispanics (toggling the "Hispanic" row since they are also considered white or black)
Final result adjusted	43.73%	50.75%		7.73%	TRUE

While I have slight hesitation calling Penn, hence the election, for Kamela, it's only because this is only one poll.  I am doing so anyway.   I can't see the cross voting going far enough to eat that approximately +6-7 advantage even with the NYT lean.  It's over the adjusted normal polling error edge too(normal is 3%, but no year has been normal with Trump), so it would  take a  somewhat significant shift in fortune toward Trump to change, and it's getting pretty damn late for that.  Kamela has gained pretty much the entire cycle, and baring a major fuck up, I don't see that changing.

However it's just a demographics poll of likely voters Sept 19th New York/sienna, with their aggregate being trump 1% ahead. While I disagree with a trump +1 result (keep in mind the NYT polls are considered D+ 1 at least, so them showing a T+1 on this kinda data is them trying very hard to be fair, and not repeat 2016.), it's important to remember, as I noted in  my foot notes, issues are just as important, and I found zero good data on that,


Kamela has also picked up like 1-2 points since then(this is the most recent full data set I could find.). Also with Hispanics being counted both in and out of their racial group, this kinda makes a big difference. Which is why we have a "Latino" vote, however to do that correctly, they should be 100% their own race. ( I call this part specifically bad polling.)


Also again I see intent to not underestimate Trump, again, despite the sampling being heavily in his favor by default, as evidenced by the amount of whites polled, and As noted in my disclaimers as well, Hispanics are counted twice (see point 2), as well as education being counted several hodgepodge ways. See point 1.


Despite the high rank of this poll, I find this "deskew" for Trump somewhat aggressive, and am leaving it, because it proves my point even better then if I tried to fix it. While I'm not unqualified to do so, and know how, I am no where near as such as Nate Silver or Professor Litchman, and both explain how to do skew and deskew correctly if interested.


*Disclaimers


1. "with college" education  voting data shouldn't be presented ONLY as combined with racial data. An associates degree is important to separate from race, as is not all races are equally represented, and a more then significant amount of people with an associate SHOULD have a BA, save Foreign language gen ed requirements, esp among millennial voters. This is especially true when they have data on what percentile of respondents have some college!


2. Race and heritage are different, I do not approve of the double categorization of Hispanics, or Asians just being part of other.


3. I am an independent, but I am definitely against trump, however I hope my methods show I'm not massaging the numbers.

4. I wanted to includes stances on abortion, immigration, and lgbt as the big topics, but can't find a reliable poll on those to factor in for this election, that doesn't combine  these with party affiliations(which are useless to me THIS election) . Its not R vs D, its literally referendum on Trump!


https://projects.fivethirtyeight.com/2024-election-forecast/
https://www.natesilver.net/p/nate-silver-2024-president-election-polls-model

Edited for grammer and punctuation (Capital letters and a coma replaced with a period).