Monday, March 22, 2010

Lies, Damned Lies, and Statistics

Well before I left but also before I had time to respond.  It was just pointing to another blog that is claiming the statistical data for atheists being underrepresented in prisons is a lie.  The site is Thoughtful Faith and the post is called Atheists in prison.

He starts by looking at Scottish prisons.

A higher percentage of Scottish inmates claim to have no religion in comparison to the general population. According to the Scottish Government’s Statistical Bulletin, Prison Statistics Scotland 2007-8, Table 8, 34.1% of the prison population has no religion. This compares with 19% of the general Scottish population answering that they had no religion, according to the ICM research faith survey conducted for the BBC in November 2005.

You can find the prison numbers on page 21.  Scotland has a total of 7,154 males and females in prison.  2,442 list none as their religion.  The faith survey he links uses an even smaller number of people, 1,019.  Of the 1,019 surveyed 227 listed no religion.  I am actually looking at the 2001 Census, which surveys everyone.  It gives a more accurate number of 27.55% listing as no religion.  Still the number of atheists in jail in Scotland is more than the population by 6.58%.  I am fairly sure that 6% would fall within standard deviation.  Also that just because one small subset in one area is equal does not mean that it is equal everywhere.

This is a bell curve:

When dealing with statistics one has to understand what a bell curve is and what it means.  Notice the majority of responses will fall into the middle range around the mean, which is usually simplified to average.  You will have outliers on both ends of the spectrum.  So having a small prison system that lies outside the mean is to be expected.

He also uses stats from England and Wales.

A higher percentage of English and Welsh inmates claim to have no religion in comparison to the general population. According to the March 2000 report, Religion in Prisons, 31.9% of inmates claimed to have “no religion”, of whom 0.2% who specifically answered that they were “atheists” and 0.1% who answered that they were “agnostic”. The national census, 15.5% of people in the general population answered that they had “no religion” and 7.3% gave no answer at all. Therefore between 15.5% and 22.8% of the general population were atheists, in comparison with  31.9% of inmates.

First off, if they don't claim to be an atheist you can't just relabel them as atheists.  So atheists consisted of 0.2% of the prison population.  By his own links, he shows that the census didn't even distinguish between atheists, agnostics and other non religious people.  So there is no way to compare the numbers here, without just making stuff up.

Finally he gets to America, which is what the original subject of atheists in prisons is about (apples, oranges, what's the difference).

The United States keeps no official statistics on religious beliefs of inmates. The claim that atheists were under-represented in prisions was seemingly started, by Rod Swift, who wrote it on his website, and publicized the claim through the internet and sceptical magazines. He claims that he received an email from an employee of the Federal Bureau of Prisons, Denise Golumbaski. According to this email, 0.2% of those surveyed specifically answered they were “atheist” and 19.8% give no answer. This compares with 0.5% of the US population at the time who identified as atheist, and 4 to 6% (according to Gallop) who gave no answer.

Rod Swift was one of the two sources I originally quoted.  The other source I found was from the Skeptic files.  It verifies the findings of Rod Swift over a longer period of time than just a single year.

In the end, there are many factors that make up why people are in prison and claiming any one is a bit silly.  I actually agree with his last statement, well except for the first and last part.

Although all three surveys indicate that people of “no religion” are overrepresented in prison, there are many complex factors which could be linked to criminal behaviour – including levels of education, age, economic well being. It may well be, for example, that atheists tend to be younger and more active – and therefore more likely to display criminal behaviour for reasons which have little or nothing to do with the atheist worldview. Nevertheless the claim that atheists are under-represented in prison is contradicted by the data in England, Wales, the US and also in Scotland.

I leave you with Norway via Russell's Teapot:

Edit:  Apparently FreeThoughtPedia thinks they own the rights to Russell's Teapot just because they were one of the many hosting the image.  So I have grabbed it from another source.


Anonymous said...

Hi mate. Thanks for replying.

First up. Good work on the Scottish 2001 census data. I was surprised to see "no religion" is actually higher than the more recent BBC poll. Perhaps this has more to do with the way the question was asked (which is discussed in your link).

We can calculate out the standard deviation. With a sample size of 7154, the standard error (the standard deviation in the mean) is less than 1/sqrt(7154) = 1.1%. So a difference of less than one percent is not significant, but six or seven percent is.

BeamStalk said...

I haven't done statistics in years so I couldn't remember standard deviation. I think we agree that there are more factors than just religious choice or non-choice goes into why a person is in prison though.

Anonymous said...

Definitely. These things are pretty complex.