Now I am not the best person around to consider statistics, and indeed my wife over in Reading is the person doing the course. Unfortunately she has had a busy day, and is in need of her sleep (statistical relationship 1 for a specific time of day). She tells me that the diagram (which I have redrawn - this is the sort of thing that appeals to me) shows moderate linear regression. If 1 then it would be completely strong, and if 0 then no relationship.
She says there is a phrase in her notes that is worth remembering, but will find who said it only after sleep. It goes something like:
None of the models are right; some of them are useful.
The line above takes an average through the various positions that more or less deviate from it on the variables. There is a definite weakening of the relationship (away from 1) by the Roman Catholics. Without them, R2 would rise to 0.81 (which is stronger - nearer 1). Visually, the flatter the curve, the nearer to 1, and the more responsive the relationship between the two variables on the x and y axes.
What this graph does not show and yet tempts the conclusion is that as you learn you become less biblically literalist. That is reading the graph back up the line, which you can do, but is to forget that this only concerns the percentages of postgraduate education. There is an initial starting point along the y axis, the 25.94 in this case, and then the linear regression that follows as one moves down the percentages of the postgraduate educated and sees a relationship rising of biblical literalism.
The importance of the denominations cannot be overlooked. They are a variable in this, and why Razib goes on to make his explanation based around them. It works like this (my explanation, not his). Unitarian Universalism preaches a message that is relatively attractive to postgraduates. The message involves a discouraging of biblical literalism. The same is so, but less so, with The Episcopal Church. At the other end, the Church of God in Christ , the message is less attractive to postgraduates and involves a strong dose of encouraging biblical criticism. What the Roman Catholic Church does is impose another dogma that relatively discourages postgraduates and yet involves itself an absence of biblical literalism. The actuality, thinks Razib, is that the population is heavily Latino and still has expectations of authority and an absence of above higher education.
This is the point, of course: the actual situation. My above explanation implies rational and free choice, which sets up relationships between cultural and institutional collectivities. You end up mapping denominations. It is, in the end, qualitative at final explanation because of the cultural nature of the phenomena. What would be the position in the UK? I venture to suggest that the education level of Unitarians is reasonably high - many teachers - but many a professor is more likely to be an Anglican. All sorts of multi-variate analysis could be made: self-understood social status, for example.
Given that the flatter the line the stronger the relationship, here follows a purely imaginative set of lines that follow a more general set of rules about education and literalism.
This shows that the impact of science education is less on biblical literalism than social science education, but that modern theology has an even bigger impact on removing literalistic beliefs. Remember each line would have the denominations scattered all around in deviation from it.
And also, as a commenter has pointed out, and as my sixth form and undergraduate student life showed with supply and demand diagrams, these diagrams are not going to show straight lines regarding the relationships between the variables involved, but curves. In other words, these are marginal relationships that trade off education (here postgraduate) and biblical literalism: indeed one can do indifference curves at each denomination (and on and on).
So very very roughly speaking, you are more likely to be a biblical literalist if you are thick.