So read the banner headline in the Chicago Tribune on November 3, 1948. A colossal blunder to say the least! How could a respected newspaper make a mistake like this?
A couple of reasons come to my mind. I’ll get to them shortly. Right now, though, hard core social Conservatives—you heard it here first—seem poised to do it again, only over a much more emotionally charged issue: abortion.
In 1948, ALL the polls gave Dewey a virtual landslide victory. Gallup had him winning by a margin of 49.5% to 44.5%. Crossley predicted the same. And, “Dewey 52.2% to Truman’s 37.1%,” declared Roper.
Also, the Chicago Tribune made two of its own blunders that had nothing to do with the above “polls.” First, it relied, almost exclusively, on the predictions of its own political analyst, Arthur Sears Henning. He had correctly called 4 of the past 5 elections.
Second, the Chicago Tribune had fallen prey to its own ego. It had begun to believe exaggerated reports of its own omniscience. A combination of the above polls and Arthur Sears Henning’s projection of a Dewey victory made the blunder inevitable.
As the world found out, Truman won. He beat Dewey in excess of two million votes. He carried 28 states with a total of 303 electoral votes. Yes, it was a landslide; only it was a Truman landslide.
Are polls that much of a joke? Absolutely NOT! For all the hoopla involved in criticizing them, inferential statistics boil down to nothing more than simple arithmetic.
But, there are mathematical rules involved in taking accurate polls. While this is not the place to detail all of them, understand that true randomness is THE cardinal rule of polling statistics. Here’s how it works.
Pollsters usually employ a 95% confidence interval. They choose a large enough sample size to comply with whatever margin of error they wish to include, most often around a ± 3 or 4 percent.
As an example, let’s say that Gallup conducts a poll to see how the country feels about a certain issue. The results of the poll indicate that people are in favor 51% to 42% with 7% undecided. The poll’s margin of error is ± 4% on a 95% confidence interval.
Simply put, these results mean that between 47 and 55% are in favor, while anywhere from 38 to 46% are NOT in favor AND that the poll’s predictions will be correct 95 out of 100 times.
Of course, the results could turn out differently, depending on which way those “undecideds” go. But, even this can be properly incorporated into the mathematical model.
This is precisely what the pollsters of 1948 did concerning the presidential race. The sample size was fine. The confidence interval was standard fare; and the margin of error was reasonable.
The only item that was off the mark was sample RANDOMNESS. Statistical samples must be random, meaning that every component within the group from which the sample is drawn must have an equal chance of being selected for analysis.
In 1948, the pollsters used (for the first time) the telephone in order to expand the sample size for their polls. The problem was that telephones were an expensive item back then. Republicans owning one outnumbered Democrats by a margin of three to one.
True randomness went out the window as a result. Such polls are worthless. Here is a well-written analysis of the subject.
Over the past three weeks, I’ve read articles from the Chicago Tribune, National Review Online, Slate.com, and Canada’s National Post. All of them involved analyses of a recent Gallup poll that seemed to show a shift away from “pro-choice” and towards “pro-life.”
In fact, the poll showed that 51 percent of us now identify with “pro-life,’ while only 42 percent identify with “pro-choice.” Or, perhaps NOT.
Adding 51 + 42 equals 93. So, what happened to the other 7 percent? Beats me! None of the columnists for the above publications went into this minor detail.
In fact, there was a conspicuous absence of technical methodology in ALL of those columns. But, the writers did make sure to fan the ever-present emotional flames with the usual array of incendiary nonsensical comments.
“Looks like we ‘pro-lifers’ aren’t the oddballs that we’re cracked up to be” came from Dennis Byrne of the Chicago Tribune.
As always, I refuse to make moral pronouncements concerning abortion. To me, it has always been a private matter between the women involved, their physicians, their families, and their consciences.
Neither am I a mathematician in the theoretical sense. However, I’ve always had a loving relationship with Mathematics’ loony kid sister, Arithmetic.
So, whenever people quote and pontificate from the Gospel according to “Polls,” the lovely Ms. Arithmetic and I go into a state of orgasmic ecstasy over the prospect of nailing a perfectly “good” hypothesis with a well-aimed nine-millimeter bullet of fact.
This poll, like the one back in 1948 showed all the signs of methodological perfection: a standard confidence interval, a realistic margin of error, and a solid sample size.
However, it CLEARLY identified 32 percent of the participants as self-described Republicans.
This wouldn’t have been a problem if NOT for the fact that only 20 percent of the voting public identified with the GOP (Compliments of Mr. Gallup’s other poll).
Randomness! Who needs it? Quoting the late Ronald Reagan, “Ask Mr. Dewey.” Skewing the numbers, intentionally or otherwise, always raises mathematical stupidity to ever-new heights.
The Limbaugh-Chaney wing of the GOP notwithstanding, it’s a big mistake to take this poll as anything near a shift in attitude for the bulk of American voters relative to abortion. The reason? Amy Farrows!
Amy Farrows is a friend of mine. She’s in her mid-forties with grandchildren. She’s as Liberal as they come, too. She becomes visibly ill at the mere sound of Rush Limbaugh’s voice. Mention Dick Chaney’s name around her and you take your life in your hands.
But, she’s apologetically “pro-life.” She’ll tell you, CLEARLY, that if she must mark a poll question as an either/or proposition, she’ll choose “PRO-LIFE” every time.
However, she COMPLETELY understands that abortion may well be the ONLY reasonable alternative from among several equally distasteful ones.
Amy does not approve of abortion as a means of birth control. Neither do most other Americans and the polls have consistently shown this. But, neither does Amy want it outlawed. And, as well, the polls have consistently shown this as a fact, also.
We Americans are an imperfect lot. We deal with complexity in the best way we can. Our feelings about many social issues are mixed, at times even contradictory.
Roughly, 68 percent of us believe that abortion is the taking of a human life. But, 74 percent of us say that it should be legal for reasons involving the health of both mother and fetus.
Those on the side of an absolute, unconditional right to “choose” and their counter parts on the other side: an absolute, unconditional outlawing of the practice, are minority participants in this controversy.
Whether hard-core social Conservatives and religious zealots like it or not, Roe v Wade is not going to be overturned to their satisfaction. On the other hand, hard-core social Liberals and equally zealous non-religious shouldn’t feel too snug, either.
The law is going to be modified, probably sooner than later, to include reasonable conditions attached to aborting a fetus. The bulk of us will welcome the move, too.
Joe Walther is a freelance writer and publisher of The True Facts. You may comment on his column by clicking here.