In a
recent diary I described my theory that "moral values" issues bring a lot of people to the poll box who normally don't attend church. I contended that church attendance is a moderating influence on people's politics and that a litany of racism, misogyny and homophobia motivated a lot of "Protestants" to get out and vote. I used Georgia as an example.
Now I have fuller numbers to back up my claim. My theory explains voting patterns nationally and in 35 states to within 5%. The mean differential between what numbers show and what my theory predicts among all 50 states is less than 3%.
Read below the fold for full analysis and results.
Introduction and Hypothesis
I am a Democrat-voting liberal with a love of Marxism and Existentialism. I am also a man of faith. My personal and highly unscientific experiences lead me to believe that faith, in general, is a positive experience. In extreme examples, it is used to oppress huge numbers of people. For most, though, faith is a source of emotional strength and moral fiber.
I have long believed that the practice of a faith-based religion is a moderating influence on one's politics. Regular church attendance or any other form of socializing with people of similar faith only heightens this effect. Ever since "moral values" took center stage in American politics, I have suspected that those most concerned with "moral values" are those who are not true participants in their faith. At the very least, such persons feel a low level of solidarity among others of their faith, and a common debate on "moral values" produces solidarity among such persons.
I did a bit of searching before deciding on a hypothesis with which to test this theory. I hypothesized that the sum of the percentage of people who are members of a Protestant church, the margin of error in that counting and the percentage of people who voted on "moral values" would be roughly equal to the percentage of people who considered themselves Protestant.
Method
Variables
P: The percentage of people who consider themselves Protestants, according to CNN's exit polls.
C: The percentage of people who consider themselves Catholic, according to CNN's exit polls.
PM: The percentage of people who are members of a Protestant church, according to the ARDA.
CM: The percentage of people who are members of a Catholic church, according to the ARDA.
MV: The percentage of people who reported to CNN that "moral values" were their #1 issue.
ME: The percentage of the populace not covered by the ARDA's study. More on this below.
Apparatuses
CNN provided my exit polling information. National results here.
The American Religious Data Archive provided my information about church membership rates. Their funding comes from the Lilly Endowment, and they have a charter to collect and distribute, every ten years, information on American religious activity. Front page is here..
Besides those websites, I used only my computer, which includes word processing, spreadsheet and calculator applications.
Procedure
I collected raw data from the ARDA and CNN websites. All statistics are derived from information found on one of those two websites. P and C are raw data, directly from CNN. PM is the sum of evangelical and mainstream Protestant church membership, as defined by the ARDA, divided by the total population of that state. In Idaho, Nevada, Utah and Wyoming, the four states with significant Mormon populations, the number of Latter Day Saints was also included in the calculation of PM; the ARDA does not count Mormons as Protestants, but Mormons consider themselves Protestants.
CM is similar to PM; it is the number of Roman Catholic adherents divided by the total population of the state. MV is raw data directly from CNN; for states where this figure is not available, the national figure of 22% is substituted. ME is always 11%. Approximately 34 million people, or 11% of the U.S. population, is known to be a member of a church which the ARDA did not include in its local numbers. For whatever reason, these churches did not participate in the study. Since that 11% is overwhelmingly Protestant, I have included them in my study. In actually, a large part of that 11% is traditionally African-American gospel churches, so the figure may in reality be higher or lower depending on the state. For simplicity's sake, I have assumed that 11% is a universal figure.
I added PM, ME and MV together for a hypothetical total of Protestants, then subtracted P from that total to derive a differential between my predicted figure and the actual figure. (Pdiff) I also added PM, ME, MV and CM then subtracted P and C to get a picture of Christianity as a whole. (Cdiff)
Results
Pdiff was between 5% and -5% in 27 states and nationally. Pdiff was between 10% and -10% in 37 states. In 35 states, either Pdiff or Cdiff was between 5% and -5%. Among all states, the mean Pdiff was 2.48%, and the median Pdiff was .5%. The mean Cdiff was -3.5% with a median of -3%. The standard deviation of Pdiff is 7.8.
The result of P - PM - ME ranged between 1% and 34%. The mean of this function was 19.32%. The result of C-CM ranged between -5% and 12%, with a mean of 4.32%.
The result of (P - PM - ME) / P ranged between .01 and .54, with a mean of .34. The result of (C-CM) / C ranged between -.16 and .7, with a mean of .23.
Possible Errors
ME is not a certain figure. While the national totals of people missed by the ARDA study is known, their regional distribution is not. I suspect that this variance in ME, which is not reflected in my study, skewed results in a few states. ARDA should endeavor to gather membership information from all religious institutions in 2010.
For some states, MV is not available. CNN, for some reason, did not publicize their "most important issue" results for that state.
Any study which is a review of multiple other studies with differing methodologies leaves itself open to a larger margin of error.
Conclusions
In general, it seems that my hypothesis was confirmed. The figures of most states conformed to my prediction of what they would be. In a study like this, it is my opinion that a differential of +/- 5% can be considered accurate. Below, I will provide explanations for the some of the more extreme results from my study.
Besides confirming my original hypothesis, some other issues were made clear. An average of 34% of people who call themselves Protestants neither are members of a church nor are explained by ME. Only 22% of Catholics consider themselves Catholic without being a member of a church. This could explain why Protestants seem to be more open to being swayed by "moral values" debates than Catholics are; Protestants are more in need of solidarity than their Roman Catholic cousins. Of course, there is a flip side to this statistic. Because of the way in which I conducted this study, those two percentages are a combined measure of Christians who aren't church members and the increased likelihood Christians will vote compared to non-Christians. Personally, I don't believe that Protestants are 34% more likely to vote than their demographic proportion would suggest, but that is just my opinion.
If my theory holds true, and I believe it does, then Pdiff and Cdiff can be used as measures of how liberal or conservative the Christian community in a given state is. If Pdiff, for example, is a negative percentage, then that means P > PM + ME + MV. This means that there are people who consider themselves Protestants in that state who neither voted on moral values nor are church members. Pdiff is the percentage of the state's population who fit that description. Alternatively, it could mean that ME is higher than 11% in that state or, in states where MV is not available, that the real value of MV is higher than 22%; either explanation would erode the measure of Pdiff somewhat. For positive values of Pdiff, there are multiple explanations. ME could be lower than 11%. In states where MV is not available, the real value of MV could be lower than 22%. Those two considerations aside, a positive Pdiff is the percentage of a state's population who both voted on moral values and are church members. Thus, a higher Pdiff implies a more conservative religious population.
Extreme Examples
Utah had the highest Pdiff of any state: 21%. Now, knowing Salt Lake and Provo as I unfortunately do, I have no problem believing that a high proportion of church-goers are homophobes, misogynists and racists. Also, since ME is comprised mainly of African-American denominations, I believe that ME in Utah is, in reality, nearly 0. (Utah is only 1% African-American)
Rhode Island had the second highest Pdiff, 16%. MV information for Rhode Island is not available, but I suspect that the real figure for MV in Rhode Island is lower than 22%. A high proportion of Catholics and of liberal sentiment in general causes me to believe this.
Pdiff for Oklahoma was also 16%. Given that they have one of the highest MV values and lowest CM values in the nation, I don't think it's a stretch to say that 16% of the state are church-going "moral values" voters. However, my knowledge of politics in that state is far from complete.
Similar explanations, I believe, can be found for other states with a Pdiff higher than 10% or lower than -10%.
Further Research
This area of research is wide open. There are all sorts of correlation studies that need to be done, from % of Evangelical Protestants and % of MV that voted for Bush, to % of CM and % of MV that didn't.
Appendixes
A state-by-state compilation of figures can be found here.
A spreadsheet of the figures can be found here.
So there you have it. The morons voting on "moral values" are NOT your typical church-goer. They are a minority of a minority. We already knew they didn't care about real morals, but apparently they don't even care about religion enough to join a church, either.