The devout and the secular pulled apart

Cultural-conservatism index (z-scored on the pooled GSS sample; higher = more culturally conservative) by church-attendance level, across five eras, 1972–2024.

Frequent (2–3×/month to >weekly) Occasional Seldom (never to < yearly)

Takeaway: the frequent attenders drift gently down off their conservative perch, but the seldom-attenders fall away far faster — the gap widens because the secular keep getting more liberal.

Index z-scored on the full pooled sample; attendance collapsed to seldom (GSS attend 0–1), occasional (2–4), frequent (5–8). Cells with Kish effective N below 50 are suppressed; effective N per cell ranges from 2,278 to 4,796. Source: GSS 1972–2024, weighted by weight_use.

A residual became a quarter of the country

Share reporting no religious affiliation by birth cohort (bars), with the era trend overlaid (line). Each younger cohort secularizes further; the nones now reach 37 percent of the youngest.

“No religion” share, by birth cohort Era trend (survey period)

Takeaway: every younger cohort is more secular than the last, and the period trend climbs in step — the unaffiliated rose from one-in-thirteen to better than one-in-four.

Cohort = survey year minus age (approximate birth year); the 1990s cohort pools births 1990–2006. Nones = GSS relig == None. Source: GSS 1972–2024, weighted by weight_use; effective N per cohort cell from 1,620 (1990s) to 14,570 (pre-1940).

The god gap bites culture, not economics

Weighted correlation of two religiosity measures with cultural and with economic conservatism. Religiosity tracks the culture-war axis far more than the spending axis.

vs cultural conservatism vs economic conservatism

Takeaway: on culture the bars stand tall for both the religiosity index and attendance alone; on economics they barely clear the floor — the god gap is a culture-war phenomenon.

Religiosity index = z-mean of attendance, belief in God, and reversed fundamentalism and strength-of-affiliation (HIGH = more devout). Cultural and economic indices z-scored on the pooled sample. n = 68,004 (religiosity×cultural), 39,085 (religiosity×economic). Source: GSS 1972–2024, weighted by weight_use.