Cheat-Seeking Missles

Tuesday, April 05, 2005

Mining The Pope Poll Data

Double Toothpicks has a great post today about an AP/Ipsos poll that says 60%-plus majorities think the new pope should liberalize the church. Steve mined the data and found only 23% of respondents were Catholic -- leading to his good point that Catholics should be very offended that AP seems to think Protestants, Mormons, non-believers and Muslims should have a say in what is essentially a Catholic theological matter.

I mined the data myself and found some other interesting points:
  • Party affiliations are skewed way left: 36% Republican, 51% Democrat, 12% Independent
  • The age of the sample was skewed young, leading to more liberal results. The largest group was 18-34 (31%), and the percent dropped progressively with each age grouping until the over 65's were just 15%. The biggest segment of the popluation, Baby Boomers, was next to last in representation in the survey (24%).
  • More than half (51%) think John Paul was one of the greatest popes of all time, but they were comfortable saying his core theology was bad. You can't have it both ways. If they're so eager to change the Catholic theology he held so conservatively, why did only 1% rate him "below average" and 1% rate him "one of the worst popes ever?"
  • The big question: "Should priests marry?" was weighted in its wording. It gave a choice between whether the next pope should "change church policies to allow priests to mary" or "maintain traditional policies requiring priests to be celebate." The words "traditional" and "celibate" are loaded; more balanced results would have been gained by making one question "change church policies and let priests marry" versus "keep church policies so priests don't marry."
Pollsters can do a lot of mischief with surveys, and the blogosphere has proved again and again that you should never take them at face value. Polling isn't an exact science, and you don't have complete control over how your samples end up, but MSM polls have established a wretched tradition of sampling the segments of the population that are most likely to yield the results they want to publish.