Increased religiosity in residents of states in the U.S. strongly predicted a higher teen birth rate, with r = 0.73 (p<0.0005). Religiosity correlated negatively with median household income, with r = -0.66, and income correlated negatively with teen birth rate, with r = -0.63. But the correlation between religiosity and teen birth rate remained highly significant when income was controlled for via partial correlation: the partial correlation between religiosity and teen birth rate, controlling for income, was 0.53 (p<0.0005). Abortion rate correlated negatively with religiosity, with r=-0.45, p=0.002. However, the partial correlation between teen birth rate and religiosity remained high and significant when controlling for abortion rate (partial correlation=0.68, p<0.0005) and when controlling for both abortion rate and income (partial correlation=0.54, p=0.001).
What does this all mean? Well, the authors sum it up in their abstract's conclusion in this way:
With data aggregated at the state level, conservative religious beliefs strongly predict U.S. teen birth rates, in a relationship that does not appear to be the result of confounding by income or abortion rates. One possible explanation for this relationship is that teens in more religious communities may be less likely to use contraception.What this means is that there are no strong outliers in the dataset, and this still is a positive correlation even if you take abortion rates and income into account. They also seem to have chosen to remain cautious with their list of explanations, and don't make the jump between this correlation and the religious right's opposition to actual sex eduction in schools (i.e., they don't make the jump to say that abstinence-only education, strongly supported by those with high levels of religiosity, doesn't work).
It would be interesting to see what sort of explanation abstinence-only supporters would give for this study.