Fake Precision

Here’s another adventure in what may become a regular feature, “Debunk the Statistical
Conclusions”. This time it’s a survey about Catholic voting preferences.

Keep in mind that the margin for error is +/- 5%.

In the most recent poll, Catholic registered voters favored Kerry by a 51 percent to 45 percent margin.

Adjusting for the error margin, Kerry potentially has 46% to 56% of the Catholic vote and Bush has 40% to 50%. I wouldn’t assume Kerry really has the lead until either the margin for error drops or Kerry’s lowest percentage beats Bush’s highest. The sample size may not have been large enough to be predictive. Furthermore, we’re not given the sample breakdown. There may be a selection bias.

Combined data from Gallup’s two most recent polls, conducted July 19-21 and July 30-Aug. 1, show that Catholic registered voters, who attend church weekly, support Bush over Kerry by a 52 percent to 42 percent margin. This group represents about one-third of all Catholic registered voters.

Among Catholic registered voters, who attend church on a semi-regular basis, that is, nearly every week or monthly, Kerry leads Bush by 50 percent to 45 percent. This group represents 27 percent of all Catholic registered voters.

Among Catholic registered voters who attend church on an infrequent basis, Kerry has a 57 percent to 39 percent lead. This is the largest group of Catholics, representing 38 percent of all Catholic registered voters.

What we’re actually looking at are 37% to 47% vs. 47% to 57%, 45% to 55% vs. 40% to 50%, and 52% to 62% vs. 34% to 44% for Kerry and Bush, respectively. The only group with a clear preference is the marginal Catholics. The strict Catholics are just shy of definitive and the semi-regular Catholics are a toss-up.

Now let’s look at the population breakup. From the strict Catholics, Kerry gets (rounding to the nearest whole percent) 13% to 16% and Bush gets 16% to 20%. From semi-regular mass attenders, Kerry gets 12% to 15% and Bush gets 11% to 14%. From marginal Catholics, Kerry gets 20% to 24% and Bush gets 13% to 17%.

Total it all up (and rounding to nearest whole percent) and Kerry has 45% to 55% and Bush has 40% to 51%. The survey gives 51% to 45%. If I were to take the middle of the ranges, I’d get 50% to 45%. That’s close enough for government work.

Ok, so I’ve bored you to tears with math. Am I getting to a point? Yes, it’s that Catholic voters aren’t going to hand either candidate a landslide victory. Don’t believe me? Here’s one more piece of math. Catholics make up about 25% of this country’s population. That means that, based on this survey, Kerry can expect about a 14% contribution to the popular vote and Bush can expect about 11%. Kerry and Bush would need another 37% and 40%, respectively if the popular vote decided the election, which it doesn’t. Doesn’t look like such a significant margin now, does it? I’m not saying it doesn’t look a little better for Kerry, but he’s not likely to break out the champagne over this.

One last thing…

Historically, Catholics voted for the Democratic presidential candidate in large part. However, in the past three decades, Catholics, who comprise 25 percent of the U.S. population, have become a key swing voting group. They proved their influence on election results when they broke with their historical voting pattern to support the winning Republican candidates in 1972, 1980, and 1984.

I’m quite curious to see just how strongly Catholics backed candidates in those elections.

(The basis for this analysis “Fake Precision” on Fallacy Files)

[An observant reader pointed out that the sample breakdown doesn’t add up to 100%. To make that happen, the “about one third” must be interpreted as 35%. I also accidentally swapped Bush and Kerry’s numbers in a couple places. I’ve adjusted my numbers and interpretations to reflect that. – Funky]