What Stokes calls “nominal” conservative Protestants, who attend church less than twice a month, have higher divorce rates even than the nonreligious. And you can find similar patterns with other indicators — out-of-wedlock births, for instance, are rarer among religious-engaged evangelical Christians, but nominal evangelicals are a very different story.

It isn’t hard to see why this might be. In the Christian penumbra, certain religious expectations could endure (a bias toward early marriage, for instance) without support networks for people struggling to live up to them. Or specific moral ideas could still have purchase without being embedded in a plausible life script. (For instance, residual pro-life sentiment could increase out-of-wedlock births.) Or religious impulses could survive in dark forms rather than positive ones — leaving structures of hypocrisy intact and ratifying social hierarchies, without inculcating virtue, charity or responsibility.

And it isn’t hard to see places in American life where these patterns could be at work. Among those working-class whites whose identification with Christianity is mostly a form of identity politics, for instance. Or among second-generation Hispanic immigrants who have drifted from their ancestral Catholicism. Or in African-American communities where the church is respected as an institution without attracting many young men on Sunday morning.