Friday, January 19, 2007

Atheists in Prison

There seems to be a general belief among atheists that the amount of atheists in the US prison population is proportionally much smaller than the amount of atheists in the general US population. I myself am an atheist and have read this interpretation of the statistics on many atheistic websites. I never really found those accounts convincing, since they seemed to declare that anyone categorized as “No Religion” in the general population was an atheist. During my previous semester at school I had to complete a project for my statistics class. For my project I decided to compare the religious affiliations of inmates in the US prison population to the religious affiliations of the overall US population. My goal for the project was to discover any existing correlations between a particular religious affiliation among the US prison population and the overall US population. I will summarize the results here, but you may also download the power point I made for the class if you prefer here (it contains graphics that may make it easier for you to interpret the data):

You may also find a copy of this document in Microsoft Word format here:

Immediately following is a description of how I obtained and analyzed the data, potential biases, and a description of the major error made by those who claim that there is a significant difference between atheists among the two different populations. The overall results are found at the bottom.

Data Gathering:

For the overall U.S. Population a received my data from the 2001 American Religious Identity Survey (ARIS).

The prison data came from prison survey data from the Federal Bureau of Prison compiled in 1997.

Potential Biases:

Already, there are some potential biases among the data that I must note. The first is that the two surveys are from different years. This could be a problem since we have good reasons to believe that the religious affiliation among the U.S. Population did not remain static between 1997 and 2001. A comparison between the 1990 ARIS and the 2001 ARIS will show that there was a downward trend in religiosity in the US between the two studies. However, we cannot assume that this trend was linear.

Another potential bias results from the sex of most prison inmates. Since most prison inmates are male, and the prison data doesn't list its data according to sex (or whether or not both sexes were studied at all), we must remember that this could affect our conclusions.

Additionally, each study categorized religious groups differently, particularly among Christian denominations (it is highly likely that the data for the “Protestant” group is not at all reliable) and those that either refused to answer or did not have a religious affiliation.

Lastly, as with any survey, respondents may not always be truthful.


For the following data, I have only shown the groups present in both surveys.

2001 ARIS data ( ):

76.6% - Christian

19.1% - No Religion/Refused to Answer

1.3% - Jewish

0.7% - Other/Not Specified

0.5% - Muslim/Islamic

0.5% - Buddhist

0.4% - Atheist

0.3% - Hindu

0.05% - Native American

0.04% - Bahai

0.027% - Sikh

0.026% - Scientology

0.01% - Santeria

0.005% - Rastafarian

Because of the way the prison data was grouped, I chose to combine the “No Religion” and “Refused to Answer” groups into one group. However, I separated the “Atheist” group (Atheists were listed as a subgroup which helped to make up the No Religion group) from the “No Religion” group as it was found in the 2001 ARIS study because of the way that the prison data grouped it's findings. This is the area where those who claim that there is a significant difference between atheists among the two populations make their mistake. Among these claims that I had found, they tended to assume that all of the respondents who claimed that they had no religious affiliation were automatically atheists, however this is not the case. Just because someone does not affiliate themselves with a particular religion does not mean that they do not have a belief in a god. Further, and more importantly, the prison survey data (shown below) asks inmates to explicitly identify themselves as “atheist”. Any comparison of the prison data must be with data which asks an identical (or at least very similar) question. You cannot use survey data which asks questions such as “Do you believe in a god?” or “Are you religious?” and then assume that the conclusions based on these comparisons are valid. There may be people who technically fit the definition of the word atheist, however we cannot consider that data because whether or not they fit into the definition of what an atheist is, and whether or not they refer to themselves explicitly as an “atheist” are two different things.

2001 ARIS – Christians:

34.7% - Protestant

24.4% - Catholic

11.2% - Other/Non Specified

2.1% - Pentecostal

1.3% - Mormon

1.2% - Church of Christ

0.6% - Jehovah's Witness

0.34% - Adventist

0.31% - Orthodox

The data for the “Protestant” and “Other” groups are likely to be unreliable when forming conclusions because of the grouping differences among each survey (For the most part, I had to form the Protestant and Other groups myself by combining denominations which I myself thought belonged in those groups, making them very unreliable.). Mostly likely, the Protestant group should be larger.

1997 Prison Data ( ):

31.4% - Catholic

28% - Protestant

19.7% - None/Refused/Other

7.6% - Muslim/Islamic

2.5% - Native American

1.5% - Rastafarian

1.4% - Jewish

1.39% - Church of Christ

1.1% - Pentecostal

1.1% - Moorish

0.9% - Buddhist

0.7% - Jehovah's Witness

0.6% - Adventist

0.4% - Orthodox

0.3% - Mormon

0.2% - Scientology

0.16% - Atheist

0.12% - Hindu

0.12% - Santeria

0.01% - Sikh

0.009% - Bahai

0.007% - Krishna

If you follow the link to the website (found at the top of the prison data) and explore that particular page, you will find two different versions of this same data. Each version groups the data slightly differently but still uses the same numbers. From these two different groupings I was able to get the preceding data.

Which Religions Were Studied:

The following religious affiliations were present in both sets of data and compared via confidence intervals with a 95% confidence level:

Christian (including various denominations)

None/Refused to Answer



Native American






Santeria, Bahai, & Sikh

Among Christian denominations, the following were studied:

Catholic, Protestant, Church of Christ, Pentecostal, Mormon, Jehovah's Witness, Adventist, Orthodox

Which religions had a proportionally equal amount of prison inmates?


Prison: 1.37% - 1.48%

General: 1.28% - 1.45%


Prison: 0.068% - 0.184%

General: 0% - 0.09%


Prison: 0% - 0.068%

General: 0% - 0.123%


Prison: 0% - 0.073%

General: 0% - 0.11%

Which religions had a proportionally lower amount of prison inmates?


Prison: 64.15% - 64.26%

General: 76.56% - 76.73%


Prison: 0.0699% - 0.19%

General: 0.29% - 0.45%


Prison: 0.11% - 0.23%

General: 0.35% - 0.52%

Note: While the amount of atheists among the prison population is proportionally lower than the amount recorded among the general population, the discrepancy is not nearly as significant as previously claimed. Furthermore, we do not have the correlational strength between these two variables and cannot assume any kind of a causal connection. If we were to have correlational data that showed a strong relationship between these two variables then we could conclude that they were somehow related, although this would not alone tell us what was causing the relationship. A better way of discovering whether or not a theistic believe leads to greater or less propensity to commit crime would be to ask questions such as “Does God exist?” or “Is there a hell?” as opposed to asking people what label they give themselves.

Which religions had a proportionally greater amount of prison inmates?

No Religion/No Answer:

Prison: 19.68% - 19.799%

General: 19.11% - 19.28%

Note: Due to the differences in the way the two surveys categorized their findings, these comparisons are unreliable.


Prison: 7.64% - 7.76%

General: 0.45% - 0.61%

Native American:

Prison: 2.53% -2.64%

General: 0% - 0.13%


Prison: 1.54% - 1.65%

General: 0% - 0.093%


Prison: 0.146% - 0.262%

General: 0% - 0.1%


Prison: 0.89% - 1.005%

General: 0.44% - 0.6%

What about Christian denominations?:

Among Christians, the following displayed an equal proportional amount:


Prison: 0.34% - 0.46%

General: 0.23% - 0.39%

Jehovah's Witness:

Prison: 0.66% - 0.77%

General: 0.56% - 0.72%

The following displayed a lower proportional amount:


Prison: 28.04% - 28.16%

General: 34.77% - 34.94%

Note: This is the group that is most likely influenced by bias for the General US Population, although it is likely higher.


Prison: 1.12% - 1.23%

General: 2.04% - 2.2%


Prison: 0.26% - 0.38%

General: 1.26% - 1.43%

The following displayed a greater proportional amount:


Prison: 31.37% - 31.49%

General: 24.44% - 24.6%

Church of Christ:

Prison: 1.34% - 1.46%

General: 1.12% - 1.29%


Prison: 0.61% - 0.72%

General: 0.26% - 0.43%


Religious affiliations displaying an equal proportion among prison inmates:

Jewish, Santeria, Bahai, & Sikh

Orthodox, Jehovah's Witness

Displaying a lower proportion:

Christian(total), Hindu, & Atheist

Protestant, Pentecostal, Mormon

Displaying a greater proportion:

No Religion/No Answer, Muslim/Islamic, Native American, Rastafarian, Scientology, Buddhism

Catholic, Church of Christ, Adventist

Note that:

Data from the “Protestant” and “No Religion” categories could be unreliable and therefore conclusions may be invalid.

Since the studies were done four years apart, religious affiliations may have changed during this time, invalidating conclusions.

We don't know the correlational strength.

Conclusions don't explain why certain religions are disproportionate. Some that are disproportionate aren't significantly so.

Disproportionalities may be caused by: an independent factor, one causes the other, both reinforce each other.