Religion vs. ACT Scores in America

While I was doing some leisure reading and enjoying Taos, NM over Spring Break, a research question came to mind.  How religious is each state in America?  Along with that, how intelligent is each state on average?  The reason this came to mind is that through my own personal experience, I noticed I was able to break through the boundaries of my thinking back when I was religious and become, in my own opinion, much more intelligent.

Pretty quickly I found some data that Gallup collected on March 27, 2012 regarding how religious individuals felt.  Nationally, about 40% of Americans are very religious, in that they attend church services and believe religion to be a very important part of their lives.  Only 28% of Americans see themselves as moderately religious, where religion isn't that important but they attend services or they may feel it is important but don't attend services often.  The growing number is those Americans who feel they are not religious, as 32% fit into this category [see Gallup poll].

It turns out Mississippi is the most religious state, with Utah and Alabama coming in second and third respectively.  The least religious states in America (which also happen to be the most non-religious states) are Vermont, New Hampshire, and Maine.  Although Mississippi and Alabama are also the top two least non-religious states, there are at least 17 states that have fewer non-religious people than Utah.  I found that interesting.

The average ACT scores of graduating students is hardly a measure of the overall intelligence of a state, but it is one measure.  The ACT website had data on 2012 graduates from all 50 states, including the composite average and averages in Math, English, Reading, and Science.  I've put both the religious and ACT data in an Excel file that you can obtain here.

It turns out Mississippi, Tennessee, and Kentucky have the lowest three average math ACT scores, while Massachusetts, New Jersey, and Connecticut have the highest three math ACT scores.

I fit a total of fifteen different simple linear regressions relating my response variable of either Percent of Very Religious (y1), Percent of Moderately Religious (y2), or Percent of Not Religious (y3), to one of five independent variables Composite ACT (x1), English (x2), Math (x3), Reading (x4), and Science (x5).

Here were the t values for the coefficients that go with each independent variable and the coefficient of determination for each regression:

Most likely, you're not a math person, and you're wondering what the hell this is telling you.  This is telling you that the average Math ACT score in a state is a very significant predictor of whether a state is Very Religious or Not Religious.  Here are some graphs that may help you see what is going on. 

The larger the percentage of "Very Religious" people in the state, the lower the average Math ACT score is in that state.

The larger the percentage of "Not Religious" in a state, the larger the average ACT Math score is in that state.  

Please note: this in no way shows or proves causation!  But isn't the correlation interesting?  There are many other variables at play here, no doubt, but nobody can argue that there is not a correlation between how religious a state is and how well they do on their Math ACT.  

This was the most fun I had with statistics since I've been alive.