One of the elegances of sociology is found in the unintended consequences of our actions. In my studies of attitudes towards abortion, I found a nice example of such unintended consequences regarding the Catholic church. But, I doubt that the findings are warranted by the analyses.
The theoretical background of the article by Cook, Jelen, and Wilcox (1993) is rather straightforward. It is well known that the Catholic church opposes against the practice of induced abortion. Therefor, it is expected that individual members of the Catholic church will be influenced by this doctrine, and will object against induced abortion more often than, for instance, non church members. On the other hand, people not objecting against abortion, or even having an explicit pro-choice stance on the issue, might feel threatened by the presence of a strong Catholic church in their presence, and express their pro-choice attitudes more strongly. From this the expectation is derived that in regions with high proportions of Catholic church members the non-Catholics will be more permissive towards abortion than in regions with low proportions of Catholics.
The authors test these two assertions using data from an exit-poll, collected in 42 American States. Using regression analyses, they find that indeed individual Catholics have a stronger pro-life stance than non-Catholics. On the contextual level they find the expected opposite result: when controlled for individual Catholicism, people living in a state with a high proportion of Catholics tend to be more permissive towards towards abortion. Apparently, according to the authors, the Catholic church is very well capable of instilling their pro-life stance on abortion in its members. And, again according to the authors, the presence of a strong Catholic church in a state mobilizes the non-members to express an strengthen their pro-choice stance.
However interesting the expectation and supposed finding about the unintended consequences of the presence of strong Catholic church might be, I doubt these findings are warranted by the analyses. Allow me to be a little bit technical. The analyses basically consist of two parameters (plus several controls): individual and contextual level Catholicism. Both variables are added to a multiplicative regression model simultaneously, which has led to the findings as described above. However, the authors seem to interpret the model as if a cross-level interaction effect between individual and contextual catholicism had been estimated as well, which was not the case. Their interpretation of the findings that the effect of contextual Catholicism only instills pro-choice attitudes amongst the non-members would only be warranted by such an interaction term.
Instead, I would think that a proper interpretation of their findings would be that, when controlling for individual Catholicism, the presence of a high proportion of Catholics leads to stronger pro-choice attitudes for everyone. Even, on average, for the Catholics themselves. Of course this is not explicitly tested, requiring an interaction term added to the model as argued above, but it could be interpreted as that people tend to fight for what they think is important when their opinion is contested by the presence of others with a different opinion, and what in a different context would remain salient.
How’s that for an unintended consequence of membership of the Catholic church?
Cook, Elizabeth Adell, Jelen, Ted G., Wilcox, Clyde (1993). Catholicism and Abortion Attitudes in the American States: A Contextual Analysis Journal for the Scientific Study of Religion, 32 (3), 223-230