Infertility is killing off the secular world, a number of writers have observed, including Phillip Longman, whose 1994 book The Empty Cradle I reviewed last year. [1] In the former Soviet empire, where atheism reigned as state policy for generations, the United Nations forecasts extreme declines in population by 2050, ranging from 22% for the Russian Federation to nearly 50% for the Ukraine. Secular western Europe will lose 4% to 12% of its population, while the population of the churchgoing United States continues to grow. Is secularism at fault? The numbers do not suggest otherwise.

 Humankind cannot abide the terror of mortality without the promise of immortality, I have argue in the past. [2] In the absence of religion human society sinks into depressive torpor. Secular society therefore is an oxymoron, for the death of religion leads quickly enough to the death of society itself.

These are impressionistic rather than rigorous arguments, however. Explaining the causes of population change has frustrated statisticians for years. Many factors influence fertility, including urbanization and literacy. Subsistence farmers view children as cheap labor and a source of wealth; peoples who remember a high infant mortality rate have more children; uneducated people do not plan the size of families, and impoverished people cannot afford the means to do so.

Having found no academic research that specifically measures the impact of religious belief on fertility controlling for these factors, I have done some calculations of my own using a cross section of data for 174 countries. My analysis, preliminary as it is, supports the conclusion that religious belief strongly influences fertility after controlling for wealth and education. There are lies, damned lies, and statistics, of course, and results of this kind should be viewed with caution. Still, this analysis passes the first cut of tests for rigor.

A visual comparison of population growth rates and degree of religious belief shows strong correlation. The World Christian Database ( reports the percentage of individuals declaring themselves “atheists” or “non-religious” in more than 200 countries, as well as economic and demographic data. [3] Using the 2005 population projections published in February by the United Nations Economic and Social Council, I compared a number of measures of population growth to the data for religious belief. The results are shown below.

All the countries with high population growth rates (vertical scale) have an extremely low percentage of non-religious people (horizontal scale), while all the countries with extremely low population growth rates have a high percentage of non-religious people. There are of course some countries (e.g., France and the UK) with population growth rates above zero despite a very large proportion of non-religious. Very high fertility of immigrant populations, though, helps explain why the French and British numbers deviate from the trend. Although a sample of 83 countries permits a great deal of differentiation, the overlap of cultures due to immigration necessarily will lead to some anomalies. [4]

More important is that a scatter-plot of population growth vs percentage of non-religious people, of course, is a naive comparison, ignoring other factors that influence fertility. One could (and demographers do) spend a lifetime fitting different pieces of data into the jigsaw puzzle to explain the variation in population growth across countries. This effort is a sketch rather than a finished picture, of course, and I have been able to test only a few variables.

By far the strongest predictor of population growth rates is adult literacy. That is not surprising, as illiterate people are likely to let nature take its course without any consideration for the implications of family size. Nonetheless, religious belief (measured by the log of the percentage of non-religious in the population) remains a strong predictor even when adult literacy is introduced as a control variable (at the nearly 100% confidence level). Wealth, that is, per capita GDP, shows no significance in the equation.

t StatConfidence
Log of %
GDP per capita0.0000.000-0.0493.9%
Adult literacy-0.0190.005-4.130100.0%

The equation

explains 68% of the variation in population growth across countries. More important than the degree of explanatory power (r squared ) is the significance of each predictor, that is, the probability that the true coefficient is not zero. Both religious belief and literacy show significance at close to the 100% confidence level. The results change little if the 50, or 100, or 150 largest countries are included in the sample.

Underlying the demographic crisis of the industrial world, I believe, is a spiritual crisis. If the above analysis has any merit, the issue is not wealth, but rather the desire of men to continue to inhabit this planet. Secular ideologies – socialism, positivism, and so forth – promised a world free of bigotry and hatred, and an unending vista of peace and prosperity. Humankind, however, has vomited up these ideologies. Secular Europe and radical Islam in that sense represent two sides of the same coin: both have rejected the secular order, the latter through open battle, and the former through fatal resignation.

Demographic analysis can help strip secularism of its progressive mask and reveal the death’s-head underneath. The analysis shown above may be the work of an amateur, but it will serve a good purpose if it provokes the professionals to do a more thorough job.


[1] Faith, fertility and American dominance, Sep 8, 2004.

[2] See Why Europe chooses extinction, Apr 8, 2003.

[3] Generously, the World Christian Database allows some free downloads although the full dataset requires a paid subscription.

[4] A case could be made that a certain threshold of secularism must be reached in order to influence population growth. Using the logarithm of the percentage of non-religious approximates this threshold effect, but this is a way of linearizing a relationship that evidently is not linear to begin with. A LOGIT model might permit better specification than linear regression.

Leave a comment

Your email address will not be published. Required fields are marked *