|
 |
 |
|
MICHIGAN |
| |
|
|
|
| POLITICAL REF'S AVG: |
| Trump +1.7 |
| Harris 47.2, Trump 48.9 |
|
| EXIT POLL & 2024 TURNOUT ASSUMPTIONS |
| POLLSTERS: Trump +1.0 |
| Harris 47.4, Trump 48.4 |
|
| If a poll is missing, it does not release cross tabs, has too small a sample (<500) or averaged 3 or more off the result in the last two presidential elections. |
| FULL POLL RESULTS AND EXPLANATIONS OF ADJUSTMENTS |
| Poll |
Date |
Sample + Party Distribution |
Result |
Pollster Avg |
Ref's Avg |
| INSIDER ADVANTAGE |
10/8-10/9 |
800LV |
Harris 46, Trump 48 |
x |
|
| Political Ref's Take |
Awaiting release of the crosstabs. Until then, excluded. |
Excluded |
|
| EMERSON |
10/5-10/8 |
950LV, Party ID: 38D/25I/37R. Gender ID: 46/54 M/F. Education ID: 36/64 BA+/NoBA. Age ID: 11/29/20/39. 18-30-50-65+. Racial ID: 78/13/3/2/5 W/B/L/A/O. |
Harris 49, Trump 49 |
x |
|
| Political Ref's Take |
Poll Bias Score 27, insignificant bias, included in my average. A few too many white ID offset by slightly pro-Harris age group and slightly too few GOP. Included in my aveage. |
Harris 49, Trump 49 |
|
x |
| WALL STREET JOURNAL |
9/28-10/8 |
600RV, Party ID: 45D/12I/42R. Gender ID: 47/51 M/F. Education ID: 37/63 BA+/NoBA. Racial ID: 77/13/3/2/5 W/B/L/A/O. Age ID: 26/23/24/25 18-35-50-65+ |
Harris 49, Trump 47 |
x |
|
| Political Ref's Take |
Poll Bias Score 110 towards Dems, severe bias, will adjust. Too many Dems, slightly too many college and post-grad degrees, too few white ID and heavily disproportionate pro-Harris age groups. The Dem biases were not sufficiently offset by somewhat too few women/many men. After adjustment, Harris gets 48.5 to Trump 47.9. See all assumptions and calculations here. |
Harris 48.5, Trump 47.9 |
|
x |
| QUINNIPIAC UNIV |
10/3-10/7 |
1007LV, Party ID: 31D/30I/32R. Inferred Gender ID: 46/54 M/F. Education ID: 36/64. Racial ID: 81/12/7 W/B/O. Age ID: 19/22/30/30 18-35-50-65+. |
Harris 47, Trump 50 |
x |
|
| Political Ref's Take |
Poll Bias Score -6, no bias, included in my average. In a surprise, I had no problem making sense of the Quinnipiac internal data. When the pollster includes all racial vote data, meaning vote ratios including Latinos and Asians, the data adds up. I can infer the education ID when all racial vote data is provided. When it's not I have to start assuming things. Just speculating, but perhaps QU wanted a certain result in Pennsylvania and made the data fit raising the need to hide Latino and Asian data? The profoundly pro-Trump gender gap in Michigan made it futile to try and get a desired result. I do not trust Quinnipiac becuase their final results often grossly favor Dems and they lack sufficient transparency. |
Harris 47, Trump 50 |
|
x |
| MIRS / MI NEWS SOURCE |
9/30 |
709LV, Party ID: 40D/22I/38R. Gender ID: 46/54 M/F. Inferred Education ID: 55/45 BA+/NoBA. Inferred Racial ID: 80/13/7 W/B/Other. Age ID: 14/22/35/29 18-30-45-65+. |
Harris 48.0, Trump 49.0 |
x |
|
| Political Ref's Take |
Poll Bias Score 302 to Dems. Will adjust. The bias comes almost completely from a severe education bias that is not balanced by any other bias. I had to infer the actual education ID, but the numbers show a larger lead than 1 for Trump, much more in line with a 3-point lead. Additionally, the numbers show between a 7 to 15-point lead for Trump among Independents, and Republicans and Democrats are likely to have equal turnout. This adjusted result is what the numbers support. Adjusted poll bias score is 36, or insignificant bias. See calculations here. |
Harris 47.0, Trump 50.1 |
|
x |
| TRAFALGAR |
9/28-9/30 |
1086LV, Party ID: 36D/33I/32R. Gender ID: 48/52 M/F. Age ID: 10/13/14/28/35 18-30-40-50-65+. Racial ID: 75/15/4/2/5 W/B/L/A/Other. |
Harris 44.7, Trump 46.9 |
x |
|
| Political Ref's Take |
Poll Bias Score -35, insignificant bias, included in my average. |
Harris 44.7, Trump 46.9 |
|
x |
| NYT / SIENNA |
9/21-9/26 |
688LV, Party ID: Party ID: 30D/35I/27R. Gender ID: 48/51 M/F. Age ID: 14/22/33/28 18-30-45-65+. Education ID: 34/66 BA+/NoBA. 70/8/4/1/3 W/B/L/A/O. |
Harris 48, Trump 47 |
x |
|
| Political Ref's Take |
Poll Bias Score -19 to GOP, no significant bias. Too few GOP and too few white ID sufficiently offset by too few woman and too few college degrees and post-grads. Included in my average. |
Harris 48, Trump 47 |
|
x |
| ATLAS INTEL |
9/20-9/25 |
918LV, Party ID: 29D/38I/33R. Gender ID: 47/52 M/F. Age ID: 19/23/36/22 18-30-45-65+. Education ID: 40/60 BA+/NoBA. Racial ID: 73/15/6/2/4 W/B/L/A/O. |
Harris 47, Trump 51 |
X |
|
| Political Ref's Take |
Poll Bias Score 119 towards Dems. Far too few white ID and somewhat too many black ID plus somewhat too many college and post-grads not sufficiently offset by too many Republicans slightly too few women. Will adjust all biased categories and average with the original topline, Harris gets 46.7 to Trump 50.7. See calculations here. |
Harris 46.7, Trump 50.7 |
|
x |
| RASMUSSEN REPORTS |
9/19-9/22 |
1086LV, Party ID: 32D/30I/38R, Gender ID: 47/53 M/F. Racial ID: 65/21/14 W/B/O. Age ID: 10/40/26/24 18-30-40-50-65+ |
Harris 48, Trump 48 |
X |
|
| Political Ref's Take |
Poll Bias Score 107 towards Dems. Severe bias, will adjust all categories and average with original topline. The biggest bias in this poll is that far too little white ID and far too much black ID are included, not sufficiently offset by too few Dems, slightly too few women/many men, and slightly disproportionately GOP leaning age groups. After adjustment, Harris gets 46.3 to Trump 49.5. See calculations here. |
Harris 46.3, Trump 49.5 |
|
x |
| OUTDATED OR REPLACED POLLS. NO LONGER INCLUDED IN THE AVERAGES. |
| USA TODAY / SUFFOLK |
9/16- 9/19 |
500LV, Party ID: 35/29/35 D/I/R. Gender ID: 48/52 M/F. Age ID: 6/13/25/28/28 18>25>35>50>65+. Racial ID: 79/12/3/1 W/B/L/A. Education ID: 36/44 BA+/No BA. |
Harris 48, Trump 45 |
x |
|
| Political Ref's Take |
Poll bias score of -18, or no significant bias. Included. |
Harris 48, Trump 45 |
|
x |
| THE HILL / EMERSON |
9/15- 9/18 |
875LV, party ID 38/26/37 D/I/R. Gender ID: 46/54 M/F. Education ID: 36/64 BA+/No BA. Age ID: 11/21/37/31 18>30>44>65+. Racial ID: 78/13/3/2 W/B/L/A |
Harris 49, Trump 47 |
x |
|
| Political Ref's Take |
Poll bias score, 35 to Dems. No signifcant bias. Included |
Harris 49, Trump 47 |
|
x |
| MARIST |
9/12- 9/17 |
1138LV, Party ID: 35D/32I/32R. Gender ID: 51/49 M/F. Education ID: 65/35 BA+/No BA. Racial ID: 76/12/3/9 W/B/L/Other. Age ID: 23/15/62 18>35>45+. |
Harris 52, Trump 47 |
x |
|
| Political Ref's Take |
Poll bias score -31 to GOP. No significant bias, included. They took too many Dems and not enough GOP but offset by taking too few women and too many men. Balances out. Included. |
Harris 52, Trump 47 |
|
x |
| QUINNIPIAC |
9/12-9/16 |
905LV, Party ID: 30D/34I/28R, Gender ID: 37/62 M/F. Education ID: 40/60 BA+/No BA. Racial ID: 67/33 White/Non-White. Inferred Age ID: 17/21/33/28 18>35>50>65+ |
Harris 51, Trump 46 |
x |
|
| Political Ref's Take |
Poll bias score of 310 toward Dems, a severe bias that must be adjusted in three areas. Recall that I only adjust until the poll balances, so we are getting the poll to a non-biased place. The poll assumed a 62/37 F/M split, 60/40 No BA/BA+ education split and 67/33 White/Non-White split. Correcting for all three and averaging the results together gave Harris 49.7 to Trump 47.6. Final adjusted bias was 20 to the Dems, or no significant bias. Included after adjustment. |
Harris 49.7, Trump 47.6 |
|
|
| INSIDER ADVANTAGE |
9/11-9/12 |
800LV, Party ID: 34D/32I/34R. Gender ID: 53/47 F/M. Racial ID: 80/12/2/1 White/Black/ Latino/Asian. |
Harris 48.0, Trump 48.6 |
x |
|
| Political Ref's Take |
This poll has a bias score of -14 toward the GOP, which is a very slight bias and is within acceptable range. See full calculations here. No adjustment necessary. |
Harris 48.0, Trump 48.6 |
|
x |
| MIRS/MI NEWS SOURCE |
9/11 |
580LV, Party ID: 38D/24I/38R. Gender ID: 53/47 F/M. Racial ID: 81/12/0/0/7 White/Black/ Latino/Asian/Other. |
Harris 48.0, Trump 48.0 |
x |
|
| Political Ref's Take |
This poll has a bias score of -30 toward the GOP, which is a slight bias and is within acceptable range. See full calculations here. No adjustment necessary. |
Harris 48.0, Trump 48.0 |
|
x |
| CBS NEWS / YOUGOV |
9/3-9/6 |
1086LV |
Harris 50, Trump 49 |
x |
|
| Political Ref's Take |
Party Vote: D's 99-1 Harris, I's 52-47 Trump, R's 96-3 Trump; Education Vote: White Degree 53-46 Harris, White No Degree 60-40 Trump. Applying party vote to more likely party ID, Harris 49 to Trump 49.3 |
Harris 49.0, Trump 49.3 |
|
|
| TRAFALGAR GROUP |
8/28-8/30 |
1089LV, 36D/33I/32R, 75% White, 15% Black, 4% Hispanic, 2% Asian, 5% Other |
Harris 46.6, Trump 47.0 |
x |
|
| Political Ref's Take |
This sample is less white by 6% and more black by 3% than 2020. While significant, it is reasonable considering there is a black candidate in the race. Included. |
Harris 46.6, Trump 47.0 |
|
x |
| THE HILL / EMERSON |
8/25-8/28 |
800LV, 38D/26I/37R, Education: 64% Non-college, 36% College degree |
Harris 50, Trump 47 |
x |
|
| Political Ref's Take |
The voter turnout model is reasonable. No adjustment necessary |
Harris 50, Trump 47 |
|
x |
| EPIC-MRA |
8/23-8/26 |
600LV, 30D/26I/30R |
Harris 46, Trump 47 |
x |
|
| Political Ref's Take |
The voter turnout model is reasonable. No adjustment necessary |
Harris 46, Trump 47 |
|
x |
| AMERICAN GREATNESS / TIPP |
8/20-8/22 |
741LV, 35D/29I/33R |
Harris 48, Trump 46 |
x |
|
| Political Ref's Take |
The voter turnout model is reasonable. No adjustment necessary |
Harris 48, Trump 46 |
|
x |
| RASMUSSEN REPORTS |
8/13-8/17 |
1093LV, 36D/30I/34R |
Harris 48, Trump 47 |
x |
|
| Political Ref's Take |
The voter turnout model is reasonable. No adjustment necessary |
Harris 48, Trump 47 |
|
x |
| NYT SIENNA |
8/5-8/8 |
619LV, 35D/31I/29R |
Harris 50, Trump 46 |
x |
|
| Political Ref's Take |
The sample includes too few Republicans, and it is obvious. Will adjust to the more likely 37D/24I/38R. Applying the party vote of D's 94-5 Harris, I's 49-43 Trump and R's 92-8 Trump to the more likely voter turnout, Harris receives 48 to Trump's 48.6. |
Harris 48.0, Trump 48.6 |
|
x |
| FULL POLL RESULTS AND EXPLANATIONS OF ADJUSTMENTS |
| Poll |
Date |
Sample + Party Distribution |
Result |
Pollster Avg |
Ref's Avg |
|
| CURRENT TURNOUT ASSUMPTION: 38D/23I/38R. The Democrats will NOT replicate their 2020 turnout. The odds of achieving this without pandemic conditions are very low. Nevertheless, I'm assuming they are able to do it. |
| |
| 2020 EXIT POLL - 38D/23I/38R without leaners (PDF); Demo - White 81%, Black 12%, Hispanic/Latino 3%, Asian 1%; Age - 18-29(12%), 30-44(21%), 45-64(37%), 65<(30%). Education: 64% Non-college degree, 36% College Degree. Education by Race: White Degree 32%, White No Degree 35%, Non-White Degree 10%, Non-White No Degree 24% |
| |
To see results of 2020 michigan polls below, click here
Best 2016 michigan Poll: Emerson College
Polls that got michigan right: UPI/CVoter, Reuters/Ipsos, CCES/YouGov, Data Orbital, NBC/Marist, CNN/OpRes, CBS/YouGov, Monmouth
Polls that got michigan wrong: Survey Monkey, Saguaro Strategies |
|
