The polling company Behaviour & Attitudes carried out an Exit Poll for RTE on the day of the abortion referendum in 2018. Questions asked of the 3779 respondents included two questions on religious belief and practice, permitting analysis to be carried out on the relationship between the respondents’ voting patterns and their religion.
Results of this analysis were posted on the RTE website at the time, and are still accessible online. However, the way these results are presented on the RTE website is not that helpful, if one wants to ask direct questions like “How did Catholics vote” or “How did regular church-goers vote?” The method of presentation chosen on the RTE website is to use column percentages, which actually answer a different question. For example, if you go to the No column of these tables under the heading “How voted”, you will find 88% listed in the “Catholic” row – meaning that 88% of the No voters were Catholics. This might sound comfortingly high, from the standpoint of the Catholic hierarchy, but it actually conveys very little on its own. This percentage is bound to be high, because Catholics also formed a large percentage of the total sample.
What should have been provided in these tables were the row percentages, not the column percentages. That is, instead of telling us that x% of No voters were Catholic, it should have told us that y% of Catholics voted No. That gives a much clearer picture of what happened in this referendum. Before writing this article, I contacted both organisations with a request for the information in the way I wanted it, but neither of them responded.
But I do have a fall-back position. I can use the information that was given on the website to estimate the row percentages I require. But it is a bit clumsy, and also a bit inaccurate, because the percentages displayed on the website have all been rounded to the nearest integer. Thus, when I see that 74% of the 3779 respondents are Catholics, I can calculate that there are 74% of 3779 = 2796 Catholics in the survey, and when I see that 88% of 1157 No voters were Catholic, I can calculate that 88% of 1157 = 1018 Catholics voted No. I can then deduce that 2796-1018=1778 Catholics voted Yes. That is where the data in row 1 of my Table 1 came from.
These calculations should be reasonably close to the correct figures, but they are unlikely to be exact. Nevertheless, my tables should be sufficiently accurate that the conclusions I draw are valid.
Table 1: 2018 Exit Poll: Voting behaviour by religious affiliation
|Denomination||Voted No||Voted Yes|
Table 1 shows the voting patterns by religious denomination. Some religious minorities have been omitted, because their small numbers in the sample did not admit of analysis. Different Protestant denominations are combined in this table. Most denomination titles in Table 1 are self-explanatory, but I should explain that “Not religious” is my abbreviation for “I’m not religious, but I do consider myself a spiritual person”.
The first thing that stands out in Table 1 is that, contrary to the impression created by the 88% listed under No on the RTE website, a big majority of Catholics (63.6% vs 36.4%) actually voted Yes in the referendum. Not that different from how the Protestants voted, in fact. The second thing that stands out is that the non-religious subgroups voted overwhelmingly for abortion and they make up 18% of the total sample.
Table 2: 2018 Exit Poll: Voting behaviour and regularity of church attendance
|Church attendance||Voted No||Voted Yes|
|Several times a week||127(84.1%)||24(15.9%)|
|Once a week||578(58.8%)||405(41.2%)|
|Once a month||150(28.4%)||379(71.6%)|
|Few times a year||197(19.3%)||823(80.7%)|
Table 2 is also very informative. To me, it is more informative than Table 1, because it classifies respondents in a way that distinguishes those who are serious about their religion (attend church once or several times a week) from those who are not. There is an extraordinary decline in those voting No (i.e. in those opposing the legalisation of abortion) as we move down this table.
I assume that the picture in Table 2 would remain the same if restricted to Catholics only, but I have no way of checking this without access to the actual poll data. On this assumption, we can conclude that nominal Catholics largely ignored Church teaching and voted for abortion in much the same way as other societal subgroups.
Jim Stack MSc PhD is a retired Mathematician/Statistician, and is writing here in a personal capacity. Most of his research output is listed on
The two tables he presents here also featured in a blog which was posted on the Iona Institute website in June 2020