Statistical Mischief and "The Catholic Vote"
Much has already been written about “the Catholic vote” in this year’s presidential election. Brace yourself for more—much of it thinly disguised politicking by partisans.
Why the focus on Catholics?
Because Catholics represent the largest religious denomination in the nation and Catholic voters are disproportionately represented in many of the critical “swing” states a presidential candidate needs to win.
But contrary to much of what you might read, because of the very size and diversity of the Catholic population, it would be an error to consider this large group of voters as somehow constituting a “voting bloc”—a bloc somehow broadly receptive to political appeals aimed at Catholics-as-Catholics.
However, studies have shown, that a subset of “Catholic voters”—those who attend mass on at least a weekly basis—are far more likely to say their faith and values are significant factors in how they vote. This is the group of “Catholic voters” who actually help determine election outcomes.
In recent presidential elections this group of voters have become increasingly Republican in their presidential voting. According to CNN election day exit polls, in 2012 Mitt Romney received 57% percent of their support compared to just 42% for President Obama.
Which brings us to the latest bit of statistical mischief gaining traction in this year’s bizarre presidential election.
The wellspring of this mischief is a recent Pew survey, “Evangelicals Rally to Trump, Religious ‘Nones’ Back Clinton.” Pew has a longstanding and well-deserved reputation for its comprehensive national surveys on a wide range of topics and the mischief is not theirs, rather it is the result of others abusing one small piece of their survey.
Although the overall findings of Pew Research Center’s survey on the voting patterns of religious voters were unsurprising—and tracked patterns in 2012—much has been said about the poll’s findings regarding a change in voting habits of weekly church-attending Catholics. And herein lies the statistical misunderstanding which has morphed into mischief.
According to the poll, weekly church-attending Catholics have shifted from the Republican presidential candidate to the Democrat presidential candidate by a swing of 22 points since a comparable poll in 2012. In the June 2016 poll, Clinton had a 19-point (57-38) advantage among weekly church-attending Catholics, whereas Mitt Romney had a 3-point (48-45) lead among this subgroup in the comparable June 2012 Pew survey.
These findings are being presented by some as the Clinton/Kaine ticket bringing Catholic voters back to the Democrats, but it is worth taking a closer look at the data before jumping to conclusions.
Anyone citing the recent Pew poll numbers as strong evidence of a seismic shift in “faithful Catholic” voting is on very shaky ground.
In the Pew survey, “weekly church-attending, Catholic, registered voters” has a sample size of only 138 people—giving it a whopping +/- 10% margin of error. But even more significant is that the demographics of that sample also changed dramatically from the comparable 2012 survey.
In the 2012 survey “White, non-Hispanic” respondents comprised 78 percent of the sample; in 2016 they were only 61 percent. Given that Hispanics, Blacks, and Asians overwhelmingly lean Democratic, the increase in their portion of the sample from 22 percent to 39 percent (a 77 percent increase!) would have caused a large shift in Pew’s findings, regardless of the candidates. Clinton supporters are anxious to point to this shift as specific evidence of faithful Catholics rejecting Trump, when, in fact, the huge demographic sampling change would have negatively impacted any Republican candidate.
Clinton partisans are using a dubious statistic the same way a drunk uses a lamp post: more for support than illumination. More thoughtful observers should not make the same mistake.