Ahead of today’s midterm, many are banking on a massive wave of democrat support to drown the Trump Administration. I believe we should temper our expectations.
Prior to today’s midterm, I dove into the latest CNN poll, a supposed barometer of how things will go. I’ve pulled back the weeds and swept away the grime to reveal:
- the CNN poll has been poorly weighted, and therefore the top-line figure is unreliable;
- the 2018 midterms are most certainly a referendum on Donald Trump; and
- there will be no “Blue Wave”; Trump’s white identity politics equally match the “Resistance”.
CNN’s latest poll puts the Democrats way ahead of the Republicans in what is called the “Generic Congressional Vote”. Pollsters asked respondents, “If the elections for Congress were being held today, which party’s candidate would you vote for in your Congressional district?” This question is used to divine the final midterm outcome. Typically when one party scores well on this question, they take House and Senate seats away from their opponents. The CNN poll indicates the Democrats have a +13% advantage over Republicans. Based on the last 20 years of midterm elections, this suggests the Democrats will pick up something like 35 seats. If the CNN poll is accurate, the Democrats will take back the House of Representatives and probably the Senate as well.
I don’t think this poll is reliable and I’ll explain why. A close examination of the poll also dispels the myth there will be a “Blue Wave” in 2018, and shows that the “Resistance” has been equally met by the anxiety Trump has stimulated through his use of white identity politics.
The first thing you want to know when analysing any poll is: what is its composition (sample weighting)? i.e.; when pollsters refine their data from the 1,000 random people who picked up the phone to have a chat, how many Republicans, how many Democrats, and how many unaffiliated voters (Independents) are they including? Obviously, if the top line figure of the poll (in this case +13% Democrats) is based on a sample that is disproportionately “registered Democrats”, or people over the age of 65, or people who live in cities, the result will “less reliable”. This is because that type of sample doesn’t properly reflect the electorate, let alone the voters likely to come out and vote on Election Day. “Refining” aggregate polling data in this way is called “weighting the sample”. The best polls disclose their weighting. This CNN poll does not. All it says is, “The entire sample was weighted to reflect national Census figures for gender, race, age, education, region of country, and telephone usage.” While this is entirely unhelpful, it’s not the end of the story. There are other ways to get a sense of how reliable the sample weighting might be…
Also on The Big Smoke
- The midterms: Judgement day for the democrats
- In all fairness, bipartisan discussion is dead (and we killed it)
A handy way to see how accurate the weighting of the CNN poll might be is to look at the Trump favourability/unfavourability results. CNN indicate results of 41% approve/57% disapprove for their “likely voter model”. However, when we compare this figure to the RCP (RealClearPolitics) average of 43% approve/53.2% disapprove, we can clearly see the CNN poll is weighted with a disproportionate number of people predisposed against Donald Trump. The deviation is ~6.5% points. Interestingly, if we correct CNN’s top-line figure by the deviation on Trump approval, we get an adjusted +6.5 for Democrat the “Generic Congressional Vote” (+13% – 6.5% = +6.5%). This new (adjusted) figure is almost the same as the RCP Generic Congressional Vote average of +7.3. This was a long (yet empirically verifiable) way of saying the CNN poll has not been weighted properly.
That being said there is a lot of great information to be gleaned from this poll!
The most important findings of doing a “deep dive” into this poll are, as mentioned:
- the 2018 midterms are most certainly a referendum on Donald Trump; and
- there will be no “Blue Wave”.
Let me show you why I can boldly make these claims…
A Referendum on Trump
The poll asked all its respondents, “Will your vote for a congressional candidate be made in order to send a message about Donald Trump?”
Message that I support Trump 28%
Message that I oppose Trump 42%
My vote isn’t to send a message 28%
Compare this to when the same question was asked about Obama in 2014 (54% said their vote wasn’t to send a message), Obama in 2010 (54% again said their vote wasn’t to send a message), and George W Bush in 2006 (42% said their vote wasn’t to send a message). The vast majority of those polled by CNN said they are voting in 2018 to send a message about Donald Trump. Far more people responded this way, this time, than any other time this question has been asked. The 2018 midterms are absolutely a referendum on Trump.
There’s no Blue Wave
The “Blue Wave” is the notion that a large number of Democratic voters (the party colour is blue) will come out to vote for their candidates in this election (in a wave). Ostensibly, because Trump is so repugnant – and the Resistance has been a meaningful protest since Trump’s inauguration in 2017 — Democratic voters are fired-up and can’t wait to let their voice be heard at the ballot box. The CNN poll indicated this isn’t going to happen. In fact, what is more likely is a “purple wave” – a historically large number of both Democratic and Republican voters coming out to vote in these midterms. I’ll stick my neck out and say these midterms will see the largest number of voters on record (early voting stats give the game away).
In a poll, a useful proxy for likely voter turnout is “enthusiasm to vote”. CNN reports that overall enthusiasm to vote is very high: 64%. Compare this to a historic turnout for midterm elections of about 40% of the electorate. Critically, however, both Democrats and Republicans have the same degree of “high enthusiasm” to vote: 72%. If we really were going to see a Blue Wave, it would appear in these results. What you’d need to see would be a higher level of enthusiasm from Democrats relative to Republicans. While the Democrats are certainly fired-up by a disdain for Trump to vote in the midterms, Republicans are equally highly motivated (probably due to the Kavanaugh Supreme Court nominating affair) to pull the lever.
The other dimension meant to drive a “Blue Wave” was the “Resistance”. The Resistance is the umbrella term for all those standing against Trump’s bigotry, sexism, misogyny and xenophobia – predominantly women and people of colour. Unfortunately, the CNN poll does not reflect a “Resistance push”. 63% of women are highly motivated to come out and vote (compared to the average of 64%), 60% of non-white respondents are enthusiastic to vote, and only 56% of millennials (those aged 18-34) appear keen to make their way out of doors on voting day. If the Resistance was going to fuel a Blue Wave the CNN poll ought to reflect this by showing high results in enthusiasm for younger, female and minority respondents. This is not what the data shows. Instead, we see that men are very enthusiastic to come out and vote (65%), as are white voters (66%). All in all, the poll shows that Trump has energised white men to the same extent that the Resistance and Democrats have invigorated women, young voters, and people of colour. Trump’s (and Fox News’) effective use of white identity politics – the aggrievement of whites in reaction to decades of affirmative action, illegal immigration and Democratic claim that all Republicans are essentially racist – has energised white male voters living outside the major cities in a way no president ever has before.
In the end, it’s a wash, not a wave, and its hue will be purple – certainly not blue.