Bill Skelly
Data, Consideration Scores, Prescriptive Analytics, Research
September 23, 2019
Sometimes it’s difficult to work at Causeway Solutions if you’re not well versed in Monty Python sketches. One of our favorites is the Spanish Inquisition and we believe it’s a great metaphor for how voters might make their decision leading up to Election Day.
In Alabama, voters are taking a long, hard look at Roy Moore. Roy Moore and Luther Strange. Roy Moore, Luther Strange, and President Trump. AMONGST THEIR CHOICES ARE…
And, quite frankly, we believe voters form independent likes and dislikes for each candidate separately BEFORE they make their final decision at the ballot box.
There might not be any more striking example of this independent candidate vetting that goes on in voters’ heads than the 2016 Presidential Election. Late last summer, we (in conjunction with our friends at TargetPoint Consulting and the RNC) identified 18 million Americans who were “rating” the Republican Party and President Trump less favorably than we believe they would have rated a different Republican nominee.
The more significant observation – and what we believed could ultimately bring the President over the finish line – was that they had not INCREASED their opinions of Hillary Clinton. While these voters might not select Donald Trump on a survey, they sure as heck weren’t going to walk into voting booths, close the curtains, and pull the lever for Clinton.
Recently, we’ve taken this concept further and begun testing these “consideration scores” in elections happening in the off year. Our goal with this post and others like it is not to make a prediction or stake our claim for a hidden “I told you so!” after the votes are cast, but to document our approach to this method, enabling us to review the value as we continue conducting additional studies.
Currently, polling for this race is all over the place.
Some attribute this to uncertainty in the turnout models; others attribute it to the lack of partisanship in Alabama for listed sample, which can result in inaccuracy.
We think it’s more complex than that.
From September 15th through the 19th we conducted an online research project of likely voters to evaluate the relative consideration for Roy Moore and Luther Strange and culminated the questionnaire with a head-to-head ballot. (Full Results)
Overall, we have the ballot at Roy Moore +18.4, which is about double the RCP average but in line with some of the individual polls that are out there. But that’s not where we think the story lies.
When asked to independently rate each candidate on a scale of 1-100 detailing how likely it is they would vote for each candidate, we observe a few notable things:
This probably comes as no surprise, but the Moore support is much deeper than that which Strange is currently experiencing. More so, in a race that could compete for the lowest of low-turnout scenarios, an 11-point gap in this type of support is almost crushing for Strange. This research was conducted before the President’s visit to Alabama, but there is significant enthusiasm that would have to come out of that one event in order to build a solid coalition behind Strange.
In my opinion, this is the difference between polls which show Moore at +18 and those which show him at+2, and the reason polling on this race swings wildly. And, as odd as it might sound, it’s possible BOTH sides of the surveys are right depending on how they are treating these voters.
These voters are a mixed bag, and it’s entirely possible that because they don’t have a strong allegiance to either candidate, they won’t show up and vote. They stay home, the likely voter screen dumps them out, and Moore squeaks out a 2-point victory.
If, on the other hand, these people do show up to vote, unless Luther Strange can make a last-minute argument to win them over, they will likely pull that lever for Moore – expanding his victory.
If I were on the Strange team, these are the people I’d be focused on almost exclusively… they’re interested in the race, they have equal views of each candidate, but for some reason they’ve chosen Moore over Strange in the toss-up.
All things equal, I believe the weak support of Strange is ultimately what dooms him in this race – Presidential support or not. He doesn’t have the enthusiasm behind him that Moore has, and those that do view him positively just aren’t fully committed to him. With only a few days left before the runoff, there are a lot of holes to fill in the ship.
The positive takeaway is that these numbers indicate it’s possible for Moore to consolidate the Strange vote, should he pull out a victory, in time for the General Election, meaning odds are high that Republicans will hold the seat.
We’ll see what tomorrow brings, but I think there’s more to survey research than just the topline numbers, and in a race like this – where polls are offering vastly different views, we need to accept that there may be truth in each of them. Survey research should be more prescriptive than predictive, and we believe this “consideration score” approach has the ability to give our clients a stronger road map for success.
Unlike Monty Python’s Spanish Inquisition, which NOBODY EXPECTS(!), I expect this will end in a Roy Moore landslide victory. At least that’s what the numbers say.