Accepting Bad Data Uncritically
Humans seem to have an inbuilt propensity to accept 'evidence' that supports their existing beliefs with critically evaluating it. Then, somehow, other people take the comments at face value and the 'evidence' is slowly accepted without any real investigation into it's credibility. Today, I noticed an article on the Australian Christian Lobby website entitled 'The Twighlight of Atheism', which is a reprint of the article which appeared in Science and Theology News by Matt Donnelly. The summary line of this article is that 'Statistics show that atheism is growing in Europe, but declining elsewhere'.

The crux of the article is therefore, the following paragraph.

Is atheism dying, though? Among members of the prestigious American National Academy of Sciences, 93 percent classify themselves as unbelievers. Influential scientists, such as Richard Dawkins, Peter Atkins and E.O. Wilson, have argued that more than a century of discoveries in disciplines from biology to astronomy render belief in God obsolete. Moreover, evidence is emerging that shows a solid link between high rates of atheism and societal health. High levels of atheism are strongly correlated with low rates of homicide, poverty, infant mortality and illiteracy, according to Pitzer College sociologist Phil Zuckerman, writing in the forthcoming Cambridge Companion to Atheism. Zuckerman also indicates that high levels of atheism are correlated with high levels of educational attainment, per capita income and gender equality.

I was immediately reminded of the hopelessly flawed and biased 'research' done by Gregory Paul, so I did a little more research into Phil Zuckerman's contribution to the Cabridge Companion to Atheism. Thanks to the wonders of modern technology, Phil's chapter is available online, so we can investigate where this supposed 'strong link' is coming from.

Note suprisingly, Zuckerman uses Paul's research without question (and indeed, Gregory Paul, our freelance paleontologist and artist has release other research on the prevalence of non-theism) as support for the correlation between atheism and societal health. As noted over at Verum Serum, Paul cherry picked countries and indicators in order to reach his conclusion (such as avoiding theft, assault and other societal negatives.

But Zuckerman's chapter has even more credibility problems than just relying on Paul's work (which in itself shows how uncritical Zuckerman is in accepting something that agrees with his beliefs). As this chapter was released earlier this year, Zuckerman makes the priceless comment that

nations marked by high levels of organic atheism – such as Sweden, the Netherlands, and France -- are among the healthiest, wealthiest, most educated, and most free societies on earth.

I guess he feels that 60 cars being burnt every night, high unemployment figures and major cultural issues 'healthy'.

Zuckerman goes on to use UN's Human Development Report (HDR) and compare it to his list of non-theistic adherents (I'll come back to this list in a minute), saying

According to the 2004 Report, the five highest ranked nations in terms of total human development were Norway, Sweden, Australia, Canada, and the Netherlands. All five of these countries are characterized by notably high degrees of organic atheism. Furthermore, of the top 25 nations ranked on the “Human Development Index,” all but one country ( Ireland) are top-ranking non-belief nations, containing some of the highest percentages of organic atheism on earth. Conversely, of those countries ranked at the bottom of the “Human Development Index” -- the bottom 50 -- all are countries lacking any statistically significant percentages of atheism.

If you check out his list, Vietnam is second proportionally, yet Zuckerman conveniently ignores that Vietnam is 112th on the HDR, The Czech Republic is 6th on Zuckerman's list but 32nd on the HDR, Estonia is 10th on Zuckerman's list, but 36th on the HDR and so on and so forth. Someone probably has time to do a proper plot relating the two datasets, but Zuckerman certainly hasn't. I doubt that the plot will have any sort of good fit. All in all, there are disparities that would indicate that atheism is not a strong factor. Even if it was, it would only be showing correlation, not causation. It is completely reasonable to assume that a society with more freedom is going to have more 'organic' (Non-forced) atheism as opposed to showing that atheism causes freedom. What is really missing is a trend analysis of countries as their atheistic population changes.

Finally, dealing the Zuckerman's list itself. I have noticed several dodgey practices. Firstly, the list is ordered by the maximum percentage if a range of percentages exist from literature. Why? Surely the more statistically valid thing to do would be to use the average values?

Secondly and much more importantly, the values Zuckerman is using are not values for Atheism. He has combined values for atheism, agnosticism and 'non-believer in god' whatever those terms mean (They aren't defined). Yet, in his summary of research we can glean just how plastic these definitions can be -

According to Norris and Inglehart (2004), 64% of those in Sweden do not believe in God. According to Bondeson (2003), 74% of Swedes said that they did not believe in “a personal God.” According to Greeley (2003), 46% of Swedes do not believe in God, although only 17% self-identify as “atheist.” According to Froese (2001), 69% of Swedes are either atheist or agnostic. According to Gustafsoon and Pettersson (2000), 82% of Swedes do not believe in a “personal God.” According to Davie (1999), 85% of Swedes do not believe in God.

Zuckerman goes on to talk about Japan

According to Norris and Inglehart (2004), 65% of those in Japan do not believe in God. According to Demerath (2001:138), 64% do not believe in God and 55% do not believe in Buddha, however a very strong majority have engaged in some form or Shinto, Buddhist, or Japanese folk/cultural ritual, such visiting a shrine or temple on the previous New Year’s Day. According to the 1999 Gallup International Poll, nearly 29% of the Japanese chose “none” as their religion. According to Johnstone (1993:323), 84% of the Japanese claim no personal religion, but most follow “the customs of Japanese traditional religion.”

These two lists make me suspicious that many of the survey's don't always deal well with panteistic concepts like new age beliefs.

Based on Zuckerman's 'careful assessment' (Did he 'carefully assess' Paul's work?) what can we conclude? Not much. Which is why it is troubling to see other people, such as Matt Donnelly, putting forth Zuckerman's work as credible.

Three final notes. Firstly, the use of 'poverty rates' should always be taken with caution. The UN generally defines poverty as less than half the median income. This definition favors a socialist system, rather than reflecting meeting the real needs of individuals. It is fairly well known that secular humanists/atheists commonly push for socialist systems.

Secondly, many of the indicators used by the UN ignore abortion in their mortality rates. Such an exclusion represents the definition of the atheist, not the theist, further biasing results.

And finally, a better study would investigate trends over many more factors, such as specific religion's (not just lumping them all together), government types and freedom. Until then, this sort of research and pronouncement is essentially useless. That won't stop the zealous from trumpeting it as excellent because it agrees with their beliefs though.
- posted by Alan Grey @ Thursday, December 15, 2005