Voting by the Numbers
By Phil Erwin
The latest FOX News national poll is reported to “show a tightening” among the remaining candidates for the GOP Presidential nomination.
Well, that much is true. The largest gap between candidates in this poll is 24 percentage points, where two months ago it was 35. The drop from 9 candidates to 3 meant “tightening” was likely – but not inevitable. Donald Trump could have gained 30 points; it appears he only gained 5.
But do poll numbers actually tell us anything? This one represents spur-of-the-moment answers from about 1000 people… out of some 240 million potential voters.
Just how accurately do you think those 1000 folks actually represent the rest of us?
Polls are nothing more than guesswork as to how a handful of people answering “Who do you like?” questions illustrates what everybody else thinks. As if their sentiments, and yours, don’t change in a heartbeat based on the latest news – or your kids screaming, or your bills mounting, or that lunchtime pepperoni pizza churning in your gut.
Pollsters say that’s what the “margin of error” is for – to account for some variability in respondents’ moods, or the random (meaning unpredictable) variations in who happened to be available (and willing! ) to answer questions when the telephone rang.
But how is that “margin” determined? What makes it 5 percent in some polls, 3 percent in others? Two percentage points out of a sample of 1000 doesn’t seem to matter much; but 2 percent of 240 million is more than a room full of voters. Why should we believe the margins are accurate, any more than are the reported “results” of the polls?
But the reporting of polls is what polling is all about – The news media gotta have something to report on, else they’d be staring at the camera and twiddling their thumbs. Not much to hold an audience’s interest. But Guess Who’s Winning Today? headlines get your attention. Give ‘em the poll results, and they’ll dazzle you with their brilliant reading of public sentiment, as demonstrated by those 1000 lucky folks.
The latest FOX poll shows Trump at 41%, Cruz at 38%, with John Kasich at 17%. The margin of error is 3%. And a FOX reporter correctly stated that Trump and Cruz are “within the margin of error.”
So far, so good. But then she commits what, to a statistician, would be a cardinal sin: She draws hard-and-fast conclusions from numbers that don’t even suggest a statistically probable conclusion. (It doesn’t matter who she is, they all make similar mistakes.)
“Trump remains in the top spot,” she opines, “while Cruz has 38%.”
Wrong. They are separated by 3 percentage points, with a 3% margin of error, which means they are in a statistical tie. In fact, it’s possible that it is Trump who is at 38%, and Cruz at 41. That’s what “margin of error” means.
This poll indicates there may be no front-runner. It merely suggests that Trump is more likely the front-runner than Cruz.
Doesn’t make for a good headline, does it? So that’s not what they report. They probably don’t even know that they’re mis-reporting the poll’s meaning.
Yet, in stating that Trump “remains in the top spot,” this reporter is affecting the dynamics of the election. People favor the front-runner, because they want to vote for a winner. And she is convincing people – arguably, most of her listeners – that Donald Trump is on track to win the GOP nomination. In fact, that is precisely what I heard reported on news analysis shows; and it is absolutely not what this poll shows.
The correct way to report this poll is: “Donald Trump still appears more likely to win the GOP nomination, but his margin for victory is shrinking, and the end result is not at all certain. In fact, Trump may just now be giving up front-runner status to Ted Cruz.”
Not pithy. But accurate. And a very, very different impact on the listening public.
You can’t just trust what you hear, even from those who are trying to deliver the news to you “straight.” You have to pay attention.
Don’t accept polls, or their inevitable “analysis,” without question. Forget taking them with a grain of salt – keep the salt shaker handy; and …
Pay attention!
____________________________________________
Phil Erwin is an author, IT administrator and registered Independent living in Newbury Park. He sometimes wishes he could support Democrat ideals, but he has a visceral hatred for Lies and Damn Lies, and is none too fond of Statistics. If his writing depresses you, he recommends you visit Chip Bok’s site for a more lighthearted perspective.
Get free Citizensjournal.us BULLETINS. Please patronize our advertisers to keep us publishing and/or DONATE
