Former New York Times data wonk Nate Silver has little patience with journalists who casually tell readers that Question A in a poll is linked to Question B, so therefore Conclusion C is obviously true.
Well, no, said Silver, the two aren't linked just because Joe Authoritative Journalist thinks they are.
Silver was exercised by a recent story in Politico on the Virginia governors race between longtime Clinton fund-raiser Terry McAuliffe and Attorney General Ken Cuccinelli, according to Talking Points Memo.
"'Recently for example, I saw an article in Politico that claimed the government shutdown was responsible for the Democrat Terry McAuliffe's expanding lead in the Virginia gubernatorial race,' Silver said at an Online News Association conference.
"'The article cited one question in the poll that the government wasn't popular and another showing McAuliffe well ahead, and asserted there was a link between the two. There was no evidence however, of a causal link that anyone had switched their vote because of the shutdown,'" Talking Points Memo said.
Silver's point is absolutely correct - maybe responses to the two questions bear some correlation (which is NOT causation), but the fact a Politico reporter, or any other journo, says it's so doesn't make it so.
Now if there was some concrete evidence - say a question that asked Virginia respondents if the government shutdown in Washington affected their decision about who to vote for in the gubernatorial contest - it might be that there is a link.
"It was a fine theory, but instead it was stated as a fact when there was no proof of it whatsoever," is how Silver put it, according to Talking Points Memo.
Silver's comments about the Politico story bring to mind his observations previously recorded in this space concerning the possible impact of the government shutdown of 2013 on the 2014 congressional elections.
According to an NBC/Wall Street Journal survey, the public blamed the Republicans more than the Democrats for the shutdown by a 22-point margin.
The survey also said Republican public approval was at its lowest-ever ebb.
The unstated but clearly implied conclusion of the survey was that the shutdown would damage Republican prospects in 2014:
"And one year until next fall’s midterm elections, American voters prefer a Democratic-controlled Congress to a Republican-controlled one by eight percentage points (47 percent to 39 percent), up from the Democrats’ three-point advantage last month (46 percent to 43 percent)."
But, as was noted in that October 13 Beltway Confidential post, Silver has long questioned whether the government shutdowns of 1995 or 1996 had any discernible effect on the 1996 election for either Democrats or Republicans.
And Silver is highly skeptical that any of the "big" political stories so often promoted by the traditional and digital media months ahead of elections actually have much at all to do with the outcomes.
Blue Collar Perspective's Charles Hawkins makes a related point, using the NBC/Wall Street Journal survey as the hook.
Hawkins noted that the survey's sample was heavily weighted in favor of Democrats, by a full 11 percent. Then he asked an interesting question:
"What if we leveled polling, with even numbers questioned in each party? Let’s remove 5.5 percent from Democrats, add 5.5 percent to the Republicans, and adjust responses accordingly.
"Q4 asked for Approve/Disapprove comments about how Obama’s doing his job. 48 percent disapproved, 47 percent approved. Re-adjusted: 53.5 percent disapprove, 41.5 percent approve.
"Q8 asked about congressional election preferences for Nov 2014. 47 percent said Democrat, 39 percent said Republican. Re-adjusted: 44.5 percent want Republicans, 41.5 percent Democrats
"Q10 asked if gov’t should do more to solve problems, or less, leaving it to individuals and businesses to solve them. 52 percent said gov’t is the answer, 44 percent said businesses. Re-adjusted: 49.5 percent favor businesses and individuals, 46.5 percent favor government
"Q16 asked if Obama’s being a strong leader, or pushing his own political agenda ahead of what’s good for the country. 46 percent said ‘leader’, 51 percent ‘putting agenda before country’. Re-adjusted: 40.5 percent strong leader, 56.5 percent Obama’s putting agenda before country."
In other words, using the same sample employed by NBC/Wall Street Journal but reweighting it to reflect a balance of Democrats and Republicans produces dramatically different results, virtually the opposite of what the NBC/Wall Street Journal authors concluded.
All of which is to say, yet again, readers should be wary whenever journalists start having fun with numbers. And that is a judgement I would apply to myself, as I suspect the estimable Mr. Silver would to himself.