Almost The Entire Difference In The Generic Ballot Polls Can Be Explained In Their Samples

At this point in the run – up to Labor Day Weekend, the narrative of a GOP wave has flipped (at least among the MSM and punditry) to more of a puddle. You can also take the unhinged rantings of say Stan Greenburg and his twisted logic individual House race polls showing the party doing worse than their 2020 numbers are a good thing (it’s funny, just check his Twitter). But, that narrative seems to be driven entirely by a left leaning set of polls who’s samples are among the worst I have ever seen.

I want to highlight the RCP average as of 8.31.22. Note the three polls which show a Democratic result better than 2020 and the GOP leaning polls below them.

On the surface these results show a close race and a divided electorate. But a deeper dive into each poll – especially the Democratic leaning ones – show the entire difference can almost entirely be attributed to their samples.

Let’s start with the most egregious survey sample of the bunch – Economist/Yougov. It should be noted in 2020 their final tracking survey had Democrats up ten in the generic ballot. They won by three. This also could be attributed to their sample which, calling horrendous, is an understatement. Yougov’s sample is 39D/35I/26R. Put another way, this electorate is worse than 2008 for the GOP. Heck, the poll would look even worse if the GOP was not leading among Independents in the survey.

Politico/Morning Consult is not much better. Their sample is 39D/27I/34R. It should come as no surprise ever since the beginning of last year these polls have also always shown Democrats ahead.

Perhaps the outlier in this trio is Quinnipiac. The survey has a history of leaning left despite not spitting out terrible samples on the surface. That said, even they forecast a 30D/36I/26R/8U sample. In every sample, Republicans are more motivated which will be interesting to see when or if these surveys switch over to Likely Voter models.

The GOP leaning surveys, Emerson, Trafalgar, Rasmussen, CBS, NBC News and NewNation all have DRAMATICALLY different samples. Emerson has a sample of 37D/27I/35R which is a D leaning electorate but not by too much. Both Trafalgar and Rasmussen hide their samples but both consistently lean to the right. CBS has a divided sample of 37D/27I/36R. NBC has a sample of 39D/21I/40R and Newsnation has the most GOP leaning sample of 34D/29I/37R.

Take these samples all together and you find the obvious. Pollsters with incredibly D leaning samples have Democrats ahead while pollsters which split or narrow GOP leaning samples have Republicans ahead. In a nut – shell, the reason the RCP average has closed at least is simply because of the partisan composition of their samples. This means unlike 2010, the polls are not going to point us in the direction of a wave for the GOP.

So, instead, we need to look to the issues and in this CBS’s survey is enlightening and believable. Majorities of voters trust each side on major issues. On the economy, Republicans are solidly ahead while on healthcare, abortion and climate Democrats lead. But what is instructive is what voters prioritize and in this case the economy and inflation remain tops. Now, this cannot top absurd samples but even so it shows why in a D = 13 sample in Yougov the GOP is only down eight and leading among Independents.

I have written about this many times before but it looks like we are set to hit the period of the election where it is a choice between the horse – race the media loves (breathlessly reporting on survey results without any analysis of the survey itself) and the anecdotal evidence coming to us. The horse race numbers based on polls show a mixed bag (determined purely by sample composition). The anecdotal evidence points to a red November (at least in the House).

First off, anecdotally, the evidence of shrinking payrolls, persistent inflation, concerns over costs, all benefit the GOP. Virtually every argument about inflation is that it is not going up but rather “not as much.” Meanwhile companies are seeing shrinking profits due to clogged supply chains and budget crunched shoppers. Combine this with more and more people living on debt and workers seeing wage increases wiped away by inflation and Republicans are ahead in this category.

Even the mother of anecdotal evidence around elections – special elections – favors the GOP. But wait. Isn’t the GOP underperforming in special elections after Dobbs? Well, no actually, not in every single one. In MN – 1, the district might have voted for Trump by nine points in 2020, but it also voted Republican at the Congressional level by +3. This time it was +4. In NY – 19, Biden carried the district by 2 points but the Democrat Congressional candidate won it by 11. This time it was +2. Then there is the following point best illustrated by Decision Desk.

“Democrats have been gaining strength in higher educated areas this entire century, but it wasn’t until the 2016 Presidential Election that educational polarization began to accelerate. Now six years later, Democratic vote share is exceptionally strongly correlated with educational attainment in heavily White areas, including the districts where the four recent Special Elections were held. Broken down at a county-wide level, Democrats dominated the higher education areas while Republicans dominated the lower education areas in these four special elections. But Democrats are not only winning highly educated areas by a modest amount through these post-Dobbs special elections, they are winning them by more than ever before. While Democrats have shifted highly educated counties in their direction between the 2020 Presidential Election and the 2022 House Special election, Republicans have still gained in less educated counties.”

“Fast Stat: Democratic Special Election House candidates outperformed Joe Biden by ~6% on average in four recent special elections after Roe v. Wade was overturned. In the 3 counties (Tompkins NY, Olmsted MN, and Lancaster NE) with 40%+ Bachelor’s Degree attainment, Democrats outperformed their 2020 results by 8%. In the other 48 counties, Democrats only outperformed their 2020 results by 1%. Not only did highly educated counties vote for Democrats by unprecedented margins, but voters from these areas turned out at incredible rates for the party. Democrats have lagged behind Republicans in voter enthusiasm for most of this election cycle, but the Dobbs decision may have turned this trend on the head. Now Democrats are trouncing Republicans, driven by astronomical turnout within these highly educated counties.”

But here is the rub……

“The four districts that held special elections are not representative of the nation. All of the districts are heavily white (average 85% vs. 75% for the nation as a whole) and consist of only small to mid-sized towns and cities (Lincoln, NE is the largest single metro area primarily within any of these districts, at only 151st in the nation). Given that white voters in small to mid-sized towns and cities are trending towards the Democratic Party (especially in disproportionately well-educated areas), it shouldn’t be a surprise that a significant event such as Roe v. Wade could trigger more rapid partisan realignment through educational polarization. Republicans are gaining the most in areas with a high minority population, which they could not showcase in these special elections. The Democratic electorate in these districts is much more white and well-educated than the party’s national voter base. These white and well-educated voters are much more likely to be energized by the Supreme Court’s Dobbs v. Jackson decision, and Democratic turnout in more diverse and less-educated communities may not increase as much. Finally, it is essential to remember that even elections in these districts between these candidates may look significantly different in November. These special elections averaged 33% turnout with respect to the 2020 presidential election, a far lower turnout than expected in the midterm elections this November. Given that highly educated voters tend to be more politically active than the average American, it is unsurprising that more highly educated groups of voters and the party they support (Democrats) dominated these low turnout elections.”


“Democrats have already been gaining in the exceptionally educated portions of America to the point where the party is almost completely maxed out in these districts. After the 2016 Presidential Election, Democrats controlled 27 of the 38 house districts where more than 45% of adults had a bachelor’s degree or higher. Now, after the 2020 General Election, Democrats control 36 of these 38 districts (only wins by Van Taylor in TX-03 and Ann Wagner in MO-02 were able to prevent a complete shutout for the GOP). And most of these incumbents won’t even be in competitive races this November; all 36 are running in districts to the left of the nation as a whole, and only Sharice Davids (KS-03) and Tom Malinowski (NJ-07) have less than a 90% chance of winning according to our latest House forecast. Even if voters in these districts turnout for Democrats at an absurd rate and produce absurd margins for the party, just as these voters have in the four recent special elections, they will have a negligible effect on the overall control of the house.”

Could not have said it better.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s