top of page


Two years after the abortion referendum, I found myself thinking quite a lot about how that referendum result came to be.

A big part of the explanation was, of course, the relentless pro-abortion bias in media coverage of the issue, stretching over most of the thirty-five years of the Eighth Amendment. There is statistical evidence that even the national broadcaster was guilty of egregious bias. In general, in both print and broadcast media, we heard an awful lot from women hurt by the Eighth Amendment, and very little from women hurt by abortion. In fact, speaking as a statistician, I found the reliance on anecdotal evidence, rather than proper scientific evidence, by the political, media and medical establishments, to be very strange. Having spent much of the last twenty years acting as statistical advisor/author on a range of medical studies, for publication in peer-reviewed medical journals, and having been put through a rigorous review process every single time, for me it was very strange indeed to see momentous decisions on abortion justified, even by doctors, based entirely on individual stories, or on vague talk of “risks”.

However, I had also come to the conclusion that part of the explanation for the abortion vote must be the decline of Irish religious belief and practice. If we were still a believing people, it would have taken a lot more, in the shape of hard evidence, to make us consider even minor changes to our abortion laws.

One feature of media coverage was the tendency of journalists to associate only Catholicism with a particular position on abortion. There was an unstated, perhaps unconscious, assumption that other subgroups in society would make rational decisions on abortion, but that Catholics would be pre-disposed to a particular position. One striking example of this was a Prime Time debate during the referendum campaign, when David McCullough began his interrogation of Rónán Mullen with a question about his religious beliefs, but did not pose the same question to the other participants at all. It was particularly unfair to Mr Mullen, given that RTE policy on the night was to severely ration the amount of programme time allocated to each speaker. Mr Mullen was forced to use up precious airtime defending his faith rather than defending the Eighth Amendment.

Given this background, I was particularly interested to see, in the recent Amárach / Iona Institute survey of 1000 randomly-selected adults, that 43% of respondents reported that they “did not pray”. It seemed clear to me that, if someone does not pray in difficult times such as this pandemic, then that person is highly likely to be, practically speaking, if not philosophically, atheist or agnostic. I was struck by the size of this subgroup in the survey; I would have expected it to be about 20%. But, from a statistical viewpoint, the larger number was a bonus, because it made it possible to split up this 43% into smaller subgroups. Also, precisely because it is a large subgroup, it can have a large electoral effect if it votes differently from other subgroups, and is worth studying on that account alone.

I contacted Iona for more information about this “do not pray” subgroup, and David Quinn kindly provided me with a breakdown of the figures. The Iona survey was not about abortion, or any other political or social issues, but it did include useable data for my purposes – for age, gender, social class and location of respondents. I also have access to data from the RTE Abortion Exit Poll, so I have two good sources of information to help me address the question posed in the title of this article. I am going to refer to this subgroup (the 43% who do not pray) as ‘non-religious’ in what follows.

What can we learn about non-religious from the Iona survey?

  1. 45% of the males are non-religious, 42% of the females.

  2. Younger respondents are more likely to be non-religious: they comprise 54% of the 18-24 age group and 60% of the 25-34 age group and 50% of the 35-44 age groups, but only 43% and 26% of the 45-54 and 55+ age groups.

  3. They comprise 47% of the ABC1 socioeconomic group and 40% of the C2DEF group.

  4. They comprise 46% of the Dublin respondents, 43% of Leinster, 44% of Munster and 39% of Connacht/Ulster respondents.

The interesting thing about these findings is that, for the most part – gender being the exception- the subgroups where the non-religious are over-represented correspond with the subgroups which had higher Yes votes in the abortion referendum, according to the 2018 RTE Exit Poll. That survey, of 3779 randomly-selected voters, found as follows:

  1. 65.9% of the males voted Yes, 72.1% of the females

  2. 87.6% of the 18-24 age group voted Yes, 84.6% of the 25-34 age group, 72.8% of the 35-49 group, 63.7% of the 50-64 group and 41.3% of the 65+ group. These are not all the same age-groups as in the Iona survey, but the pattern is very similar – the percentages are highest in the younger age groups and decline sharply in the older age groups.

  3. 76.7% of ABC1 group voted Yes, 63.1% of C2DE and 52.5% of F.

  4. 79.8% of Dubliners voted Yes, 67.2% of Leinster, 66.3% of Munster and 62.0% of Connacht/Ulster.

The stand-out result here is surely that, in the two age groups in the Iona survey where the non-religious are in a majority, the Yes vote in the referendum was well in excess of 80% in both. It is also interesting that Dublin, where the Yes percentage vote was nearly 80%, and the ABC1 socioeconomic group, where the Yes vote was 77%, both had higher percentages of non-religious than their counterparts.

Additional information from the RTE Exit Poll

The RTE Exit poll actually had Atheist (5% of respondents) and Agnostic (2%) categories, and a third category “I am not religious, but I consider myself a spiritual person” (11%). These three together amount to 18% of respondents, a far cry from the 43% who “don’t pray” in the Iona survey. My preference is for the Iona figure of 43%, even though I have already acknowledged that I was surprised by its magnitude. I suspect that e.g. many cradle Catholics will tick the Catholic box in surveys, but have in effect abandoned their faith and should be including themselves in one of the three aforementioned groups instead. The “don’t pray” category, therefore, seems a more accurate barometer of religious belief. In the same vein, I did not just accept the Denomination categories in the Exit Poll which respondents said they belonged to (Table 1 below) but I also report results for frequency of attendance at religious services (Table 2 below).

Minority religions mostly did not have enough respondents in the Exit Poll to be useful for analysis, and are excluded from Table 1 below. I constructed this table myself, it was not on the website in this form. To do so, I had to use the rounded percentages provided on the RTE website, so the numbers displayed in Table 1 for the different categories may be slightly inaccurate. The same applies to Table 2 which follows later.

Table 1: Voting behaviour by religious affiliation

Denomination No Yes

Catholic 1018 (36.4%) 1778(63.6%)

Atheist 12 (6.4%) 177 (93.6%)

Not religious 35 (8.4%) 381 (91.6%)

Agnostic 12 (15.8%) 64 (84.2%)

Protestant 35 (31.0%) 78 (69.0%)

Comment on Table 1:

The percentages displayed in the table are row percentages. Thus, 36.4% of Catholic respondents voted No in the referendum, but just 6.4% of Atheist respondents etc. For comparison, in the Exit Poll, 30.6% of all respondents voted No. (In the end, a third of voters voted No). According to Table 1, the percentage of Catholics voting No is higher than the overall No percentage in the Poll, but not excessively higher. The percentage of Protestants voting No is very much in line with the overall No percentage. The startling results in Table 1, however, are for the three non-religious categories: in all three, the percentage No vote is way below the overall Poll No vote.

There was another question in the RTE poll, on regularity of attendance at church. 4% of respondents attended church several times a week, and 26% once a week. 29% of respondents answered “never/hardly ever” and 27% “a few times a year”. These categories seem to me to better reflect actual religious beliefs of respondents than the Denomination categories of Table 1.

Table 2: 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%)

Never 116 (10.6%) 980 (89.4%)

The evidence in Table 2 is clear. Regular churchgoers voted No in very large numbers. Infrequent attendance at church, in contrast, is associated with very high numbers voting Yes.


The Iona survey was not designed for present purposes, but does provide evidence that non-believers are a substantial presence in our society, and are found disproportionately in certain demographic subgroups – younger adults, ABC1 socioeconomic group, and Dublin-based. These subgroups voted disproportionately for abortion in 2018, according to the RTE Exit Poll. Analysis of the Exit Poll data suggests that, in fact, unbelief negatively affected the size of the No vote in the referendum more than strong belief positively affected the size of the No vote. All conclusions are tentative, in so far as the RTE poll may have underestimated the numbers of unbelievers, and the Iona poll was not about the abortion referendum at all. But it appears that we can answer the question posed in the article title in the affirmative. Unbelief, at least on the issue of abortion, seems to be associated with voting behaviour as much as, if not more than, belief.

Jim Stack MSc PhD, a retired Mathematics Lecturer, has acted as statistical adviser/author on more than forty published, peer-reviewed medical studies (listed on the website PubMed ). He writes here in a personal capacity.

bottom of page