Tuesday, November 23, 2004

2 is still greater than zero, right? Not in Nagourney and Elder's world

Yes, we're still on Adam Nagourney & Janet Elder's New York Times front page story "Americans Show Clear Concerns on Bush Agenda" (http://www.nytimes.com/2004/11/23/national/23poll.html?oref=login) and the more I look at the polling questions and responses and compare it to the summary in the paper, the more concerns I have. (Or worries, as the poll might characterize it.)

Let's analyze two paragraphs from Nagourney & Elder's front page story and then compare their summary to the polling data:

That said there is little question that Americans have grown increasingly unhappy with the influence of popular culture on daily life, and that was a significant dynamic in this election. Nearly two-thirds of respondents said that Hollywood was lowering the standard of popular culture. And 70 percent said that all popular culture -- music, movies, and television -- was lowering moral standards in America.
The poll also found, though, that Americans were evenly divided on whether television, movies and books were including too many gay themes and characters.

This is an interesting section and I'll try to keep it simple.

We're looking at three questions in the polling for the basis of the excerpt above:

"55. How worried are you that popular culture -- that is, television, movies, and music -- is lowering the the moral standards in this country?"

40% say lowering; 30% say somewhat;, 17% say not too worried; 12% say not worried at all; 1% gave no answer or didn't know.

"56. What kind of impact would you say Hollywood is having on the standards of popular culture -- is Hollywood lowering the standards, raising the standards or not having much impact on the standards of popular culture?"

62% said lowering; 6% said raising; 29% said "not having much impact"; 4% didn't know or gave no answer.

"57. Do you think popular entertainment media -- such as movies, television, and books -- are including too many gay themes and characters these days, not enough, or about the right amount?"

42% said too many; 6% said not enough; 44% said the right amount; 7% said they didn't know or had no answer.

Let's deal with the "two thirds . . . [who] said that Hollywood was lowering the standard of popular culture" first.

That's the 62% who responded to question 56. That's simple enough. And they're asked what they think the effect is, please note, they're not asked about their concerns or their worries. Nagourney and Elder report this figure correctly.

But let's now look at the 70% who "said that all popular culture -- music, movies and television -- was lowering moral standards in America." That's not what they were asked in the question. It wasn't "do you believe that ___"; it was "are you worried that it might be."

I mean, I'm worried that I might be gaining weight. Am I gaining weight? Do I think I'm gaining weight? Asking about a worry or a concern, isn't the same as asking do you think _____. (No, I don't think I'm gaining weight. Yes, I do worry about gaining weight.)

Here's another example, I'm worried I may end up with cancer. Do I think I have cancer? (No, I don't think I presently have cancer. Yes, I worry about getting it.) Those are two different questions and there's no basis for the statement to appear in print as it did.

But let's look at the "70%" figure. To arrive at that (see question 55 above and note the bold type) you have to add the categories "very" worried and "somewhat" worried (40% and 30%). Are "somewhat" and "very" close enough to be added together (to form a total for a worry, not to form a total for a statement of belief)? Let's say they are and turn to the statement about an even division of people as to whether television, movies, and books were including too many gay themes and characters. The journalists are using the answers to "too many" (42%) and "right amount" (44%). Where is the 6% that felt that there weren't enough gay themes and characters in popular culture? It's okay to combine "very" and "somewhat" in one instance, but you can't combine "right amount" with "not enough"? If you combine those two (44%; 6%) you get fifty percent. And since Nogourney and Elder are comparing "whether television, movies and books were including too many gay themes and characters" the 6% who say there's not enough characters and themes need to be factored in.

42% are saying that there are too many gay characters and themes. How many are saying that there are not too many? 50% (7% didn't know or had no answer). That's what you get when you add the group that feels there is just the right amount and the group that feels there are not enough. Too many vs. not too many gives you 42% and 50% which isn't an even divide.

Here's where someone might want to bring in the margin of error, but there's no caution of margin of error in Nagourney and Elder's other discussion of totals (such as the seventy percent).

Does the poll indicate that we are evenly divided on the amount of gay characters and themes?
Even with the margin of error giving 3% to the ones who think there are too many -- for a total of 45% -- and taking away 3% from the ones who do not think there are too many -- leaving 47% -- we've still got two percent more respondents who say that there are not too many gay themes and gay characters in popular culture; that's a difference outside the poll's range of error. And this, according to Nagourney & Elder, is an even divide?

Not in any statistical class I took. That's a two percent lead for the groups who feel that there are not too many gay themes and characters represented in popular culture.

That's "slightly more" people feel there are not enough gay characters and themes in popular culture (books, movies and television) -- that is not "evenly divided."

And remember, the 3% is a plus or a minus. For the discussion above, we added to the group that felt there were too many and subtracted from the group who felt there weren't too many. So let's flip that. 3% taken away from the ones who said "too many" drops them to 39%; 3% added to those who felt there weren't too many brings the total to 53%. So 39% of Americans may feel that there are "too many" gay themes and characters in popular culture and 53% do not feel that there are "too many." That's a fourteen percent difference. How is that "evenly divided"?

The printed results of the poll reveal 8% more respondents saying that there are not too many gay themes and characters. Factoring the margin of error in full force towards the "too many" crowd still gives the not too many respondents a 2% lead. Factoring the margin of error in full force towards the not to many gives them a lead of 14%. And Nogourney and Elder want to say "evenly divided"?

There are problems with this poll but there are also problems with the way Nogourney and Elder have summarized the results. A few posts back, I jokingly stated that math wasn't the Times strong point, Nogourney & Elder prove the joke true in at least this one instance with their summary report of the poll.

I'll also add that popular culture (in the questions) always includes TV and movies. But it doesn't always include books and music. (Check questions 55-57 above.)