There has been extensive analysis and speculation regarding the 2011 census results. There has also been some criticism of those who have attempted to correlate the religious community background data with politics. In Northern Ireland religious community background is almost 100% correlated with political party preference. The recent Spotlight poll questioned over 1,000 residents of Northern Ireland on numerous political issues and analyzed the responses by age, gender, level of education, religious community background and other variables. There were 368 individual from the Protestant community background who were willing to give their 1st preference party vote and only 4 choose a nationalist party (all SDLP). Of the 328 individuals from the Catholic community background who were willing to give their 1st party preference only 3 choose a unionist party. Cross community voting preference is only 1%. In view of these findings, the census provides the most reliable data when attempting to determine current and future election results within a constituency.

The only difficulty in using the census data, especially with religious community background, is that the data is incomplete. In the 2001 census, there were 80,670 people who did not answer the question on religion, 4.8% of the population. Of that 4.8%, 3.6% did not answer any part of the question, 0.6% checked off the box indicating they had a religion but did not identify the religion, and 0.6% indicated their religion was “None” but did not indicate whether they were raised in a religion. The census office estimated that there were an additional 81,627 people who did not return census forms. The census office did a follow up survey of 10,000 households to confirm the accuracy of the census. Approximately 500 households (~2,000 people ) would have been people who never responded to the initial census and the data from those 2,000 people was used to extrapolate to the entire estimated 81,627 who never returned census forms. Thus, there was no religion background data for over 9% of the population.

Fortunately, there are other sources of religion data which are more complete than the census. This would be the annual school census, which includes all children enrolled in any school in Northern Ireland. The school census for 2001 was completed in October 2001, 6 months after the 2001 census. The data for primary school children is the most accurate to use in comparison with the 2001 census, since primary school attendance is compulsory. The only students missing from the primary school data are students in special schools and this data was included in later years and is similar to data for other primary schools. The data that is available covers 98% of primary school age children. The religion data for primary school students in 2001 showed the following:

Catholic    Protestant   NonChristian           None-Not Stated   Total

85,048          75,644          684                    9,177                  170,553

For the data from the 2001 census I used the religious information for the 6 to 11 age cohort (they would be age 6 ½  to 11 ½ at the time of the school census ) and the majority of the age 5 cohort . I also did an alternative analysis with the 6 to 11 age cohort with the majority of the age 12 cohort instead of age 5 though it differs slightly since religion data changes little from one age cohort to the next.

Catholic    Protestant   NonChristian   None-Not Stated   Total

84,575          76,911         567               8,500                170,553

+473              -1,267            117               677

The school census records more Catholics, Non Christian and None and fewer Protestants than the 2001 census. Most of the excess None are in integrated schools and state schools. They are probably of secular background (i.e.: the parents have a tenuous religious connection and put down that religion for themselves and their children in the census but did not wish the child to be considered as having a religion for school purposes).  How to analyze these 677? I did so based on where they attended school and almost 90% would be Protestant. It could also be argued that they should be 75% Protestant and 25% Catholic based on the religion brought up in data ( for the None group) in the census. It could also be argued that some or many are of Catholic background and the parents do not wish to reveal their Catholic background where they would be in a small minority in an overwhelmingly Protestant school. The adjusted data, removing the excess None, shows the following difference compared to the 2001 census.

Catholic    Protestant   NonChristian   None-Not Stated

554                 -674             120                    0

For the secondary school data the school census data is:

Catholic    Protestant   NonChristian   None-Not Stated   Total

80,034          68,015           386               7,068                155,503

For the census I used the age 12 to 16 cohorts and some of the age 17 cohort.

Catholic    Protestant   NonChristian   None-Not Stated   Total

79,358            70,260           452           5,434                155,503

The difference between the school census and 2001 census is:

Catholic    Protestant   NonChristian   None-Not Stated

+676             -2,245           -66                    +1,634

After adjusting for the excess None (assuming mainly Protestant).

Catholic    Protestant   NonChristian

+872             -811               -62

The 872 extra Catholics in the secondary school census compared to only 554 in the primary school census may be due to more Catholics undercounted in secondary school. It is also possible that it is due to another factor. Some older students drop out of secondary school, so the school census does not record the school dropouts. If there is a higher dropout rate among Protestant students compared to Catholic students then this could account for the higher number of Catholics compared to the primary school census (where school attendance is compulsory). Even if the true number similar the primary school number (554) the secondary school census also confirms an undercount of the Catholic school age population.

It appears that the primary school and secondary school census enumerates a Catholic undercount compared to the 2001 census. This number would range from 1,100 (assuming a higher dropout rate in secondary school for Protestant students) to 1,426 (assuming no differential in dropout rate in secondary school). Of course, all these students have younger siblings who are not yet in school and older siblings who have already finished school. They also have parents, grandparents, cousins, uncles and aunts who are not in school. Since the school census only records 20% of the population, the undercount of the Catholic population for all of Northern Ireland would be 5 times as high-at least 5,500. The lowest estimate would be 4,000, if I used the 6 to 11 age cohort and some of the age 12 cohort and extrapolated the undercount only using the primary school census. The highest underestimate would be 10,000. The 10,000 estimate would come from using the higher secondary school number (872 for Catholics) and assuming that 40% of the excess None are of Catholic origin. I tend to prefer a lower number, perhaps 5,000.

The 2001 school census confirms a small undercount of Catholics in the 2001 census. Why did the 2001 census record fewer Catholics than the school census? There were 80,670 people that did not record their religion on the census form. The census office used other markers (religion of other household members, age and knowledge of the Irish language) to estimate the religion of these 80,670. They found that their religion would be no different than the percentages of those who did answer the religion question. Those census respondents who did not answer the religion question are not the source of any discrepancy. There were an estimated 81,627 people who did not return census forms. The data for these people was estimated from households in the follow up census survey who were found to have been non respondents in the original census. This is where the source of error probably originated. The census office extrapolated from this sample (up to 2,000 people) to the entire 81,627. The census office found that the non-responders were 47.1% Catholic, 31.8% Protestant and 20.6% None or refused to reveal their religion (the latter much higher than the 13% in the census). It is clear that the non-responders were more likely to be Catholic and less likely to be Protestant than those who responded to the census (the Protestant percentage was over 50% in the census). Of course, many non-respondents did so simply because they could not be bothered to fill out the initial census form. However, the higher Catholic percentage and the lower Protestant percentage in the non-responders indicate that there was small Catholic boycott in the 2001 census.  Of course, this sample of non-responders only includes those who were willing to cooperate with the census takers who came to their home. Those who refused to cooperate likely included a disproportionate number of people who were unwilling to cooperate with the census takers for political reasons. These people would include dissident republicans and their sympathizers. The census office is unable to account for those who refuse to cooperate and this would affect their extrapolation of the non-responder data to all the non-responders. Another possible source of error is the large percentage of those unwilling to reveal their religion to the census workers in those non-respondents in the follow up census. The census office analyzed that 20.6% of None and Not Stated as if they were similar to the 13% of None and Not Stated in the census. They assumed that 13% were actually None and only 7.6% Not Stated. This would lead to an overestimate of the None and Protestant section of these people since the None group is 54% of Protestant background, 27% of No religious background and only 17% of Catholic background. If those who refused to reveal their religion where instead analyzed according to the percentages of those who did reveal their religion it would result in 3,000 more Catholics and 1,500 fewer Protestant and 1,500 fewer None in the final tally of the 81,607.

The 2001 school census confirms a small undercount of the Catholic population in the 2001 census. It is approximately in the range of 5,000 people, though it could be higher. This 5,000 undercount is only 0.3% of the entire population. The school census also confirms that the 2001 census has very accurate data for the religious background of the entire Northern Ireland population. The age religion data for the 2011 census will be released later this year. It will be possible to compare the 2011 school census data with the data from the census. There maym or may not, be an undercount or overcount of the Catholic population.