Polls go in waves as enthusiasm goes up and down. Harris peaked after the convention, it makes sense that the numbers are starting to normalize more than a month later.
But FTA:
“I urge journalists and researchers to dive into FiveThirtyEight and see how the red wave pollsters have flooded the zone again. MT, PA, NC were initial targets but now it’s all 7 battleground states.”
Fine, let’s look at PA since that’s the “must win” swing state:
Note that you are repeating the same polls multiple times in that listing. 538 lists the same poll multiple times based on the different results from it (likely voters vs register voters) and head to head vs full field often giving 2 to 4 results per poll. These are not separate polls. The NYT only did one poll of PA recently, don’t assume those are 4 at the same time
As an aside, some of the pollsters have gotten more partisan this cycle. That recent TIPP poll there was the most egregious where a previously reputable pollster just decided to just assume that philadelphia was going to have 1/10 of the normal turnout in their likely voter screening (look at the unusually massive difference in their likely voters vs registered voters). This is despite asking how likely they were to vote and people in philadelphia respondeded with normal numbers, not anything anywhere close to 1/10th. It wasn’t a mistake either, they replied saying they were the ones who did the likely voting screen and there were no errors, but didn’t really offer much of an explication of why they basically assumed philadelphia wouldn’t vote
OK, here we go, so of the 22 PA polls currently up on 538:
Not red wave: 6
The New York Times/Siena College
InsiderAdvantage
Emerson College
Quinnipiac University
YouGov/Center for Working Class Politics
Just Bad polling: 3
Redfield & Wilton Strategies
Research Co.
Hunt Research
OnMessage
Red Wave: 2
TIPP Insights - flooded the channel with 4 polls, but the polls split 50/50 with Harris showing a larger margin on her two than Trump did on his two.
Fabrizio, Lee & Associates/McLaughlin & Ass. - McLaughlin being Trump’s preferred pollster.
Fabrizio, Lee & Associates/GBAO - While Fabrizio is a Red Wave pollster, GBAO is not. Both polls show Trump up so it’s likely GBAO is just a beard here.
So of the 22 polls, 7 are run by clearly red biased pollsters. I wouldn’t call 31.8% a “flood”, it’s about in line with MAGA support in the general population.
6 are run by just questionable sources, I wasn’t able to determine bias other than “bad at their job” bias. Results seem evenly split, 2 Harris, 2 Trump, 2 tied.
Which leaves 9/22 run by unbiased, quality sources.
Ok, so I’m just scrolling through the most recent polls and picking out numbers based on this. I’m just going to grab the most recent numbers from Silver Bulletin. I’ll grab the latest, most recent number for a given pollster, Harris & Trump only.
Not red wave: 6
The New York Times/Siena College
Harris/ Trump:
44%/ 49% (but also, I’m not sure whats going on with the repeat entries on this poll…)
InsiderAdvantage
Not in Nate Silvers database, cant find in 538’s either, different name?
Emerson College
Harris/ Trump:
50.2%/ 48.6%
Fabrizio, Lee & Associates/GBAO - While Fabrizio is a Red Wave pollster, GBAO is not.
The Red Wave polls and the Not Rave polls are in good agreement. These polls are all with each others MOE, and would fail a t-test.
Harris:
t-statistic: -0.405
p-value: 0.706
Trump:
t-statistic: -0.245
p-value: 0.818
Both p-values are significantly higher than 0.05, indicating that there is no statistically significant difference between the two groups in terms of Harris and Trump percentages.
The article is wrong. Unless it meant to say that Times/ Sienna is a “Red Wave” pollster, this an article targeted towards Blue MAGA to give them something they want to hear.
Polls go in waves as enthusiasm goes up and down. Harris peaked after the convention, it makes sense that the numbers are starting to normalize more than a month later.
But FTA:
“I urge journalists and researchers to dive into FiveThirtyEight and see how the red wave pollsters have flooded the zone again. MT, PA, NC were initial targets but now it’s all 7 battleground states.”
Fine, let’s look at PA since that’s the “must win” swing state:
https://projects.fivethirtyeight.com/polls/president-general/2024/pennsylvania/
Oct. 7-10
857LV
The New York Times/Siena College
Harris 49%
Trump 45%
Stein 1%
Oliver 0%
Oct. 7-10
857LV
The New York Times/Siena College
Harris 50%
Trump 47%
Oct. 7-10
857RV
The New York Times/Siena College
Harris 49%
Trump 45%
Stein 2%
Oliver 1%
Oct. 7-10
857RV
The New York Times/Siena College
Harris 50%
Trump 47%
Oct. 8-9
707LV
Redfield & Wilton Strategies
Harris 46%
Trump 48%
Oliver 1%
Stein 0%
Oct. 7-9
803LV
TIPP Insights
Harris 48%
Trump 49%
West 1%
Stein 0%
Oct. 7-9
1,079RV
TIPP Insights
Harris 49%
Trump 45%
West 1%
Stein 1%
Oct. 7-9
803LV
TIPP Insights
Harris 48%
Trump 49%
Oct. 7-9
1,079RV
TIPP Insights
Harris 49%
Trump 45%
Oct. 6-9
800LV
Fabrizio, Lee & Associates/McLaughlin & Associates
Harris 48%
Trump 49%
Oct. 7-8
800LV
InsiderAdvantage
Harris 47%
Trump 49%
Oct. 5-8
1,000LV
Emerson College
Harris 49%
Trump 50%
Sept. 28-Oct. 8
600RV
Fabrizio, Lee & Associates/GBAO
Harris 46%
Trump 47%
Sept. 28-Oct. 8
600RV
Fabrizio, Lee & Associates/GBAO
Harris 45%
Trump 46%
Stein 2%
Oliver 1%
West 0%
Kennedy 0%
Oct. 5-7
450LV
Research Co.
Harris 50%
Trump 49%
Oct. 3-7
1,412LV
Quinnipiac University
Harris 49%
Trump 47%
Oct. 3-7
1,412LV
Quinnipiac University
Harris 49%
Trump 46%
Stein 1%
Oliver 1%
Oct. 2-7
1,037LV
Hunt Research
Harris 47%
Trump 47%
Stein 1%
Oliver 1%
Oct. 2-7
1,037LV
Hunt Research
Harris 47%
Trump 48%
Sept. 27-Oct. 2
5,686LV
Redfield & Wilton Strategies
Harris 48%
Trump 47%
Oliver 1%
Stein 0%
Sept. 24-Oct. 2
1,000RV
YouGov/Center for Working Class Politics
Harris 47%
Trump 45%
Sept. 24-Oct. 2
500LV
OnMessage
Harris 46%
Trump 46%
(Whew! That’s a lot of formatting! Swinging back after Googling each company)
It’s simpler to just say they are trying to flood the zone with bullshit 🤣
Note that you are repeating the same polls multiple times in that listing. 538 lists the same poll multiple times based on the different results from it (likely voters vs register voters) and head to head vs full field often giving 2 to 4 results per poll. These are not separate polls. The NYT only did one poll of PA recently, don’t assume those are 4 at the same time
As an aside, some of the pollsters have gotten more partisan this cycle. That recent TIPP poll there was the most egregious where a previously reputable pollster just decided to just assume that philadelphia was going to have 1/10 of the normal turnout in their likely voter screening (look at the unusually massive difference in their likely voters vs registered voters). This is despite asking how likely they were to vote and people in philadelphia respondeded with normal numbers, not anything anywhere close to 1/10th. It wasn’t a mistake either, they replied saying they were the ones who did the likely voting screen and there were no errors, but didn’t really offer much of an explication of why they basically assumed philadelphia wouldn’t vote
So what’s your conclusion?
I gave it quick scan and referenced based on your summaries ( thanks for that btw).
To me this all seems very much “in place” for polling data, as in, noisy, but pointing in the same direction.
I always use 5% as my rule of thumb to give the Pubs as a handicap. If a Dem is leading by 5% or more, they are actually leading.
Working on the conclusion, got distracted by SNL. :)
Following! (:
OK, here we go, so of the 22 PA polls currently up on 538:
Not red wave: 6
The New York Times/Siena College
InsiderAdvantage
Emerson College
Quinnipiac University
YouGov/Center for Working Class Politics
Just Bad polling: 3
Redfield & Wilton Strategies
Research Co.
Hunt Research
OnMessage
Red Wave: 2
TIPP Insights - flooded the channel with 4 polls, but the polls split 50/50 with Harris showing a larger margin on her two than Trump did on his two.
Fabrizio, Lee & Associates/McLaughlin & Ass. - McLaughlin being Trump’s preferred pollster.
Fabrizio, Lee & Associates/GBAO - While Fabrizio is a Red Wave pollster, GBAO is not. Both polls show Trump up so it’s likely GBAO is just a beard here.
So of the 22 polls, 7 are run by clearly red biased pollsters. I wouldn’t call 31.8% a “flood”, it’s about in line with MAGA support in the general population.
6 are run by just questionable sources, I wasn’t able to determine bias other than “bad at their job” bias. Results seem evenly split, 2 Harris, 2 Trump, 2 tied.
Which leaves 9/22 run by unbiased, quality sources.
Ok, so I’m just scrolling through the most recent polls and picking out numbers based on this. I’m just going to grab the most recent numbers from Silver Bulletin. I’ll grab the latest, most recent number for a given pollster, Harris & Trump only.
Not red wave: 6
The New York Times/Siena College
Harris/ Trump:
44%/ 49% (but also, I’m not sure whats going on with the repeat entries on this poll…)
InsiderAdvantage
Not in Nate Silvers database, cant find in 538’s either, different name?
Emerson College
Harris/ Trump:
50.2%/ 48.6%
Fabrizio, Lee & Associates/GBAO - While Fabrizio is a Red Wave pollster, GBAO is not.
Harris/ Trump:
45%/ 50%
Quinnipiac University
Harris/ Trump:
47%/ 49%
YouGov/Center for Working Class Politics
Harris/ Trump:
51%/ 48%
Just Bad polling: 3
Redfield & Wilton Strategies
Harris/ Trump:
46%/ 48%
Research Co.
Harris/ Trump:
49%/ 45%
Hunt research:
47%/ 47%
OnMessage:
Can’t find in either database.
Red Wave (2):
TIPP Insights
Harris/ Trump:
48%/ 49%
McLaughlin
Harris/ Trump:
48%/ 49%
Doing the quick stats…
Not Red Wave Polls:
Just Bad Polling:
Red Wave Polls:
The Red Wave polls and the Not Rave polls are in good agreement. These polls are all with each others MOE, and would fail a t-test.
Harris:
Trump:
Both p-values are significantly higher than 0.05, indicating that there is no statistically significant difference between the two groups in terms of Harris and Trump percentages.
The article is wrong. Unless it meant to say that Times/ Sienna is a “Red Wave” pollster, this an article targeted towards Blue MAGA to give them something they want to hear.