Fig 1.
Directed Acyclic Graph (DAG) denoting the assumed causal structure of the data.
For simplicity, all potential confounders from Table S3 have been grouped together in the ‘Baseline confounders’ node. The node ‘U’ represents potential unmeasured confounding. Note that data on ‘political beliefs’ are only available in the G1 offspring generation.
Table 1.
Descriptive statistics for the religious/spiritual beliefs and behaviours (RSBB) exposures in the G0 mother (n = 14,216), G0 partner (n = 10,916) and G1 offspring (n = 14,524) samples. Note that the percentages of missing data are calculated separately from the observed data.
Table 2.
Descriptive statistics for the climate beliefs and total number of pro-environmental actions performed in the G0 mother (n = 14,216), G0 partner (n = 10,916) and G1 offspring (n = 14,524) samples. Note that the percentages of missing data are calculated separately from the observed data.
Fig 2.
Results of the mothers ordinal regression models with ‘belief that the climate is changing’ as the outcome for four religious exposures (belief [n = 2,569], identity [n = 2,542], attendance [n = 2,542], and latent classes [n = 2,575]; models are separated by dashed horizontal lines).
Odds ratios above 1 indicate an increased belief in climate change. See Table S9 in S1 Text for full results.
Fig 3.
Results of the mothers ordinal regression models with ‘believes that humans are to blame for climate change’ as the outcome for four religious exposures (belief [n = 2,561], identity [n = 2,535], attendance [n = 2,535], and latent classes [n = 2,567]; models are separated by dashed horizontal lines).
Odds ratios above 1 indicate an increased belief that humans are to blame for climate change. See Table S11 in S1 Text for full results.
Fig 4.
Results of the mothers linear regression models with ‘total number of actions performed due to climate change’ as the outcome for four religious exposures (belief [n = 2,218], identity [n = 2,195], attendance [n = 2,197], and latent classes [n = 2,224]; models are separated by dashed horizontal lines).
Values above 0 indicate an increased number of pro-environmental actions performed. See Table S13 in S1 Text for full results.
Table 3.
Summary of key results. A dash (“-“) indicates little-to-no association between the religious exposure and climate outcome. Neg = Negative association (e.g., among G0 mothers a Christian identity was negatively associated with belief that the climate is changing); Pos = Positive association (e.g., among G0 partners regular religious attendance was positively associated with concern regarding climate change); U-shaped = ‘U’-shaped association (e.g., among G0 mothers the association between the religiosity latent classes and concern over climate change was U-shaped, with “agnostic” participants less concerned than “atheist” participants, but no difference between “highly religious” and “atheist”); J-shaped = ‘J’-shaped association (e.g., among G0 mothers the association between the religiosity latent classes and number of pro-environmental behaviours was J-shaped, with “agnostic” and “moderately religious” participants engaging in fewer actions than “atheists”, but with “highly religious” engaging in more actions than “atheists”). For all religious exposures the baseline/reference categories reflect lower/no religiosity, so a positive relationship (for instance) between a religious exposure and climate outcome indicates that higher levels of religiosity are associated with greater climate awareness, concern or behaviours.