by Steve Gill, Political Editor of The Tennessee Star
Vanderbilt University has just released a new political poll that has anybody with even a remedial understanding of politics scratching their heads about the methodology and results. With Christmas approaching it looks like Santa will need to add a new category on his Naughty or Nice checklist to accommodate the Vanderbilt political science experts: Nitwits!
First, the poll targeted “registered” rather than “likely” voters, which always guarantees a less informed and involved pool of responses. Those who are “registered,” simply because they are automatically registered when they get a driver’s license but don’t actually vote, are not the folks who spend much time getting informed about the candidates or political issues; nor do they consume much news.
Second, the Republican-Democrat composition of the poll gives Republicans a small 7 point margin over Democrats (34-27). Really? In a state that has super majorities of Republicans in the State House and Senate; a 7-2 Republican majority in the Congressional delegation; a 26 point Donald Trump margin of victory over Hillary Clinton; a Bill Lee margin over Karl Dean in 2018 by 22 points; and a Marsha Blackburn victory over Phil Bredesen by 11 points? Yet, Vanderbilt produces a poll constructed on a “plus seven” Republican base in the “red state” and thinks it should be taken seriously?
Third, the poll was conducted over an eighteen (18) day period from November 19 – December 6. Polls are intended to be a snapshot of opinion, not a mini-series. No credible pollster would be in the field for nearly three weeks when so many news stories and issues occur that can impact the results within a mere week. That’s why most serious polls are completed in 3-4 days. Vandy should at least be flagged for delay of game!
Finally, let’s take a quick glance at the track record of previous Vanderbilt polls. In October 2016, just a few weeks before the Presidential campaign concluded, Vanderbilt claimed Trump was leading Clinton by an 11 point margin (44-33%). Trump won by 26. In May, 2012 Vanderbilt claimed Mitt Romney had a tiny one point lead over incumbent President Barack Obama in Tennessee; Romney carried the state by over 20 points.
And in 2018, Vanderbilt said Bredesen had a one point lead over Blackburn in their poll just two weeks before Election Day. She won by eleven. That same poll said Lee lead Dean by 11 points; Dean lost by 21. At least Vanderbilt is consistent…they miss the actual election results by double digits virtually every single time!
The new poll claims that two thirds of Tennesseans support expansion of Medicaid (or more accurately “Obamacare”, which would have produced dramatically different results if that name had been used in the poll question). Bill Lee campaigned in opposition to expanding Obamacare while Karl Dean advocated expansion. Bill Lee won by 21 points with actual voters.
Similarly, Vanderbilt says Marsha Blackburn (that would be Senator-Elect Marsha Blackburn for those at Vanderbilt who remain in denial about the November election results) has a 45-50 approval-disapproval gap with Tennesseans despite winning a month ago by a 55-44 margin — carrying all but three of Tennessee’s 95 counties. (There’s that ten point error gap that Vanderbilt consistently produces, and that 2018 election map is identical to the county vote split that Trump enjoyed over Clinton two years ago.)
Meanwhile, Vanderbilt says Bredesen is much more popular than Blackburn, with a 54-34 favorable gap of twenty points. (In fact, according to the poll data, Bredesen’s 54% approval ranks him almost equal to Bill Lee at 57%, though Lee won by twenty points while Bredesen was getting trounced.)
The only thing more amazing than Vanderbilt’s consistent ability to produce flawed poll data is the fact that so many media outlets in Tennessee continue to treat their results as legitimate “news” and credible “information.” Perhaps it’s just a marriage made in heaven: #fakenews and #junkpolling.
– – –
Photo “Polling Station” by Marco Verch CC 2.0.