In most countries secularism is positively correlated with support for welfare, but does welfare make people more secular? Anthony Gill of the University of Washington says yes; in 2004, he and grad student Erik Lundsgaarde published a paper arguing that welfare provides a substitute for church attendance, making people less likely to attend church.
The full theory goes as follows: in the 19th century, the power of Christian churches came from their ability to provide social services such as charity, education, and health care. As the state started providing the same services without requiring or expecting church attendance, it became less economic for people to attend church, and less economic for church leaders to focus on welfare activities.
This theory has a lot of holes in it, but the study has some empirical backing. There’s a statistically significant relationship between a Christian country’s welfare spending as a percentage of GDP and the percentage of people in it who report attending church weekly, even when controlling for such variables as education and whether the country is Catholic or not. The weakness of the study comes not from its lack of data, but from flaws in how the variables are defined, failure to look for alternative explanations, and problems with individual case studies.
First, the study doesn’t explicitly say how welfare spending is measured. This is significant because it right off the bat fails to control for key factors. Most importantly, the most expensive part of the welfare state is social security, whose cost increases with the old age dependency ratio. But more religious states have higher population growth rates, leading to younger demographics and lower social security costs.
It’s possible to get around that by looking at states that buck the trend and are both relatively religious and relatively old. The best case study here is Poland, which is simultaneously the most religious nation in Europe and one of the oldest. Additional examples include Spain, Portugal, and to some extent Italy. The only one of the four that appears in the scattergram plotting church attendance and welfare spending is Spain, which is considerably more religious than the regression line predicts.
In addition, even when one controls for old age pensions, not all governments spend welfare the same way. The USA prefers targeted tax breaks, making its welfare system appear stingier than it actually is. In addition, some benefits can be distributed either as welfare or as spending on health care and education, which the study doesn’t account for. A good example in the US would be free lunches in schools, a welfare service that adds to the education budget.
Second, the omission of education spending is crucial. A church often thrives by having its own set of parochial schools. The standard British joke about catechism is that religious education only secularizes people, though the more common sensical effect is the opposite, namely that greater availability of parochial schools will make the population more religious. Education spending is correlated to welfare spending via the mediating variable of economic liberalism or socialism. As such, Gill and Lundsgaarde commit a grave sin of omission by overlooking it.
Likewise, a more direct political mediating variable could account for much of the correlation. In a followup paper, Gill notes that the correlation between welfare and religosity holds within US states, too. But within the US, both welfare and secularism fall under the rubric of liberal politics, contrasted with the welfare-busting and religiosity of conservative politics.
This in fact holds true in Europe and Latin America, which comprise all countries in the study but two, the US and Australia. Throughout Europe and Latin America, even more so than in the US, there is a strong tradition of anti-clerical liberalism. It’s likely that all Gill’s motivating example of Uruguay shows is that Uruguay has a long history of domination by the left-liberal Colorado Party.
Third, the main measure used for religiosity, reported church attendance, is deeply flawed. The USA’s real church attendance rate is half its reported rate. The church attendance variable tracks not how many people attend church, but how many would like pollsters to believe that they attend church. This variable has some value, but is overall less important than data based on actual church attendance.
The other figure used, the percentage of people who declare themselves nonreligious, is flawed as well. There are two dimensions to religious affiliation – one’s choice of religion, which tracks culture, and one’s position along the religious-secular spectrum. More plural areas, especially those with strong connections between religion and culture, will have a lower percentage of people calling themselves nonreligious than less plural areas.
Fourth, many of the assertions in the study admit too many inexplicable case study exceptions. Ireland and the Philippines’ unusually high levels of religiosity are attributable to the role the Catholic Church played in pro-independence and anti-Marcos politics respectively; I presume Poland could be similarly explained away, were it in the study. But other exceptions require seriously modifying the theory.
For example, the study would predict an increase in American church attendance rates after the welfare reforms of the 1990s. The American study only finds a slightly less significant correlation between welfare and religion in 1995; meanwhile, there was a measurable increase in church attendance in the two months following the 9/11 attacks.
For another example, the case study of Britain goes in almost the opposite direction as the one the study predicts. Britain hasn’t had a serious welfare system since Thatcher’s economic reforms. And yet, in the 1990s, religious belief crashed, and while children of secular parents always grew up to be secular, children of religious parents had only a 50% chance of growing up to be religious. Levels of belief crashed even among Muslims, who Britain forces a religious identity on in many respects.
And fifth, there are alternative explanations that the study should look at but doesn’t. First, it’s legitimate to ask why support for welfare correlates so nicely with secularism in Western politics. It could be an ideological accident that modern liberalism is secular and pro-welfare and modern conservatism is religious and anti-welfare; after all, in turn-of-the-18th-century Britain, it was the Tories who were more supportive of extensive Poor Laws and the Whigs who favored a libertarian economic policy.
Or, equally well, it could be the realpolitik version of what the study is trying to say: welfare is a substitute for religion. As such, religious organizations are likely to ally themselves with political groups that oppose welfare. It holds to some extent for modern conservatives, though by no means for all. In 1900, the US populists were both pro-religion and pro-welfare, and would only embrace prosperity theology in the 1960s and 70s.
A good way of gauging such political explanations is seeing if the same trends hold for non-Western countries. Muslim organizations provide the same welfare Christian ones do; in fact, one of the main power sources of Islamist movements is their strong performance in disaster relief. Of course, Islamism has an entirely different dynamic to it – its main promise isn’t charity but change – but it’s useful to examine this dynamic and see how it can apply to the West. How relevant is the promise to change the morally uncertain status quo to the rise of American Dominionism?
I should stress that except perhaps for the problematic definitions of the variables, this study is not shoddy. A data set comparing religiosity and welfare is always useful. The study’s downfall is in using the data to confirm a theory that has no other evidence to it. Although the study seems to satisfy the falsification criterion in that Gill intended for it to highlight the failure of the theory, in fact it does not falsify the statement “welfare does not cause a decline in religiosity.” All it does is superficially confirm the statement that welfare does in fact cause religiosity to fall.
Of the many different angles the study could take, the one about a direct effect of welfare on religiosity is one of the most obvious two, which is probably why Gill went with it. The other, that religious groups lobby against welfare, is more empirically plausible than the converse direction of causation, but does not fit well into Gill’s theory. But more indirect links, for example with education or political liberalism as a mediating variable, look far more fruitful. The study’s ultimate downfall is not so much that it is wrong as that it is woefully incomplete, concentrating on perhaps the least enlightening theory available.