Take a look at an article in the Washington Post today reporting the results of a poll about attitudes regarding war in Iraq. Do you see the same virtual disconnect I do between the headline and text, on one hand, as opposed to the actual data?
The headline:
Doubts Temper War Support
Gender, Age and Politics Fuel Gaps in Opinion on Attacking
Lead paragraph:
Surveys conducted since the Sept. 11, 2001, terrorist attacks have consistently shown that a majority of Americans favor military strikes against Iraq. But this general agreement that force should be used is neither absolute, unconditional nor uniformly shared by key voting groups, an analysis of recent Washington Post-ABC News surveys suggests.
Before turning to the disconnect, I wonder when, or even if, the results of any poll on any major national issue has ever found a “general agreement” that was “absolute,” “unconditional,” or “uniformly shared by key voting groups.” Does their absence justify a finding of “doubts”?
Alas, the online article is not accompanied by the very revealing bar graph that is in the hard copy. The article is accompanied by something called “complete poll data,” but those seem to be from a different poll from the one the article is about, or at least from the results reported in the bar graph in the paper edition. Too bad. Thus you’re going to have find a copy of today’s paper or take my word for what it says and shows.
Here’s the question the pollees were asked:
The Bush administration says it will move soon to disarm Iraq and remove President Saddam Hussein from power, by war if necessary, working with countries that are willing to assist, even without the support of the United Nations. Overall, do you support or oppose this policy? Do you support it with reservations or without reservations?
There follows a bar graph showing “Percent who support war.” The groups (“key voting groups”?) and their support are as follows:
Republican men 89%
Republican women 82%
All men 67%
35-44 year olds 67%
Whites 63%
Southerners 62%
Midwesterners 61%
18-34 year olds 60%
All Americans 59%
Independent men 59%
Westerners 58%
45-54 year olds 57%
55-64 year olds 55%
All women 51%
Northeasterners 51%
Independent women51%
Age 65 or over 49%
Democratic men 42%
Nonwhites 35%
Democratic women 34%
Thus, out of 20 categories reported, only 4 oppose going to war even without the UN. If you were writing the article about these numbers, would your headline be “Doubts Temper War Support”?
Not surprisingly, I suppose, the major division seems to be between red and blue, Republicans and Democrats. Assuming an equal number of men and women in each party (I know, but wait a minute), 85.5% of Republicans support war compared to 38% of the Democrats. In real life, the disparity is no doubt greater because more than half of the Republicans are men and more than half of the Democrats are women.
If there had been polls 150 years ago, I wonder if they would have shown this great a difference between the North and South over slavery.
Fun with algebra!
Conveniently, an identical majority of both “all women” and “independent women” support the war (51%). This allows us to conclude that a 51% majority of the aggregate set of “Republican women” and “Democratic women” support the war. Since we know a 82% majority of “Republican women” support the war a 34% minority of “Democratic women” support the war, we can determine the proportion of “Democratic women” and “Republican women” used in compiling the poll. (This process is not so easy for men because it would require assumptions as to the number of “independent men” not required when analyzing women.)
After a little black box algebra, we determine that among women who stated a party preference, 65% are Democrats and 35% are Republicans, nearly 2 to 1.
That’s the algebra. Draw your own subjective conclusions.