Christopher Newport University’s Wason Center has new numbers this morning showing embattled Democratic gubernatorial nominee Ralph Northam sliding in the polls, with Republican Ed Gillespie dramatically gaining four points and in a statistical dead heat with Northam at 48-44.
Yet once again, there are some real question marks in this poll. 28% of those polled signaled they were graduate level educations or better (the Virginia average is 15%). 43% of the respondents indicated they made salaries of over $100K or better (23% over $150K). Over half the poll was conducted via cell phones.
What’s slightly more concerning? The poll was munged according to race, sex, age, and region — but not on other demographics such as income, ideology, or party affiliation… partly, one assumes, because CNU Wason did not know how or where to munge the data in that regard.
Even with Gillespie within a statistical tie, the thumbs are all over the scales in a poll that still skews towards limousine liberals.
One other interesting tidbit in this poll. Gillespie seems to be tacking on a percentage point per week, while Northam has effectively flatlined. Should the trend continue, Gillespie crosses up at or just about on Election Day.
Of course, one can’t be too hard on CNU Wason for their polling methodology. We are in the middle of a political realignment, independents are a tricky bunch nowadays, cell phones are mucking up results, and to make matter slightly more complicated? Virginians are voting in an off-election year.
Yet there are still some very basic amendments — leaving off undecideds, focusing on 2013 voters, asking voter intensity questions, munging the data for income and education levels — that ought to be taken into consideration.
The real twist in all of this is that while the data infallibly records the sentiments of those polled, presenting them back to the public as an accurate reflection of the electorate is murky at best. Certainly the election might very well demonstrate Northam with a 4-7 point lead, but at some point we get into territory where the numbers feel good while the results as to how we arrive at such numbers are all over the map.
The solution? Adopt the methodologies of the major campaigns. Yes, it is expensive… but when millions of campaign dollars are at risk, accuracy in polling is critical. For those representing hubs of information, the investment is worth considering.
UPDATE: Quick not to appear too-terribly-biased, Fenit Nirappil over at the Washington Post observes via Twitter:
Last week’s poll had Northam up seven, but our resident pollster tells us the changes are within realm of sampling error https://t.co/qqSemCUzJz
— Fenit Nirappil (@FenitN) October 17, 2017
…which for those counting at home, means Northam +13 according to the WaPo was really, really off.
Yet more to the point, such an observation really doesn’t dive into the realities that the 30% of the Virginia electorate isn’t graduate-level educated or better, or that 46% of Virginians probably don’t make $100K or more (average salary for graduate level incomes is slightly north of $100K).
Each and every mismatch when leveled up? Advantage to Gillespie.
