A
antunesaa
Guest
I know my share of microeconometrics and using the data from the PEW forum 2008 survey the results I mentioned are quite robust (and the result is quite robust in the economics literature). These are publicly available data for a large cross-section of US inhabitants (26k subjects in the sample, which is representative of US residents in 2008, except Hawai) and if you want I can email you Stata scripts and logs of ordinal probit models atesting just that. Studies using cross-country regressions to find individual-level relationships are usually wrong, and it is not surprising that you have found them in a journal of evolutionary psychology; I doubt the editors know enough microeconometrics to identify this problem, which in the economics literature has been identified long ago. What you have to do is to use individual-level covariates to explain individual-level measures of well-being; these covariates account for differences of race, sex, marital status, income, etc. Finally, you add state- or country-level fixed effects to account for unknown common factors within the geographic unit.Actually, on a social level, the exact reverse is true. A study published in Evolutionary Psychology, July 2009 (vol.7(3), 399-434) called “The Chronic Dependence of Popular Religiosity upon Dysfunctional Psychological Conditions” tested whether religion is necessary to the health and prosperity of a society, and whether higher degrees of nonreligiosity are more likely to be associated with better social and economic conditions. Seventeen first world nations (per capita income of at least $23,000) were tested.
Criteria for religiosity included belief in God, biblical literalism,evolution, heaven, and hell; frequency of prayer and church attendance, and the number of unbelievers.
Criteria for dysfunction included per capita number of homicides, incarcerations, suicides, abortions, infant mortality, life expectancy, infant mortality, life expectancy, marriage duration, divorce rate, alcohol use, corruption, income and income disparity, poverty and unemployment levels.
The USA came in first in both lists. Overall, there was a marked correlation between how religious a society was, and how dysfunctional it is, with the lowest rates found among the least religious of countries. You can find a summary of this study where I did, in Bugliosi’s book Divinity of Doubt (which, by the way, I cannot endorse wholeheartedly, since its quality and knowledge level are very uneven. This study, however, was endorsed by an impressive list of health organizations).
If you do this, the results are overwhelming. Relative to Catholics (which is the largest single-name religious group in the sample, with roughly 22% of the total), people declaring themselves “atheist” or “agnostic” (1.5% and 2.6% of the sample) have significantly lower scores in a measure of personal happiness (check questions q1 and q1a in the methodological notes of the survey). People declaring “nothing in particular” (10%) also have a lower propensity to happiness.
In marginal terms, relative to catholics, atheists have more 22 percentage points of probability of being in the worst category (very dissatisfied with life), while for agnostics the figure is 16 pp and for the other group it is 9 pp. A typical catholic is 3.2 times more likely to be in the better category (very satisfied with life) than an otherwise identical fellow atheist. For agnostics the factor is 2.3 and for the other group it is 1.6.
All these numbers are statistically significant. (Of course, all this is after controlling for income, number of children, sex, ethnic background, marital status, etc. I used US states fixed effects.)
Again, I can provide you a Stata script with sources and the execution log but I think this particular point (for whatever reasons) is pretty much settled in the microeconometric literature, and is found in other countries as well. The reason why the cross-country regression is wrong (or at least your interpretation is) is that it compares, say, Denmark with the US, but it might be the case that there is something (mood?) not captured by the country-level covariates used in the regressions that makes everyone happier in Denmark, for instance. But, within Denmark, religious people are still happier, because the finding at the individual level is robust.