To substantiate this acceptance thesis, academics point to opinion surveys that show rising Catholic support for continued union and falling support for Irish unity. They link these profound changes in the distribution of Catholic constitutional preferences to the peace settlement reached in 1998. The Good Friday Agreement’s package of power-sharing, consociational governance, social and political rights, and cross-border institutions resolved the constitutional question by encouraging Catholics to abandon traditional Irish nationalism and develop a sub-state regional identity inside the UK. These commentators lament that Brexit interrupted the halcyon years of Catholic constitutional quietude by reviving popular interest in Irish unity. They look forward to better management of Irish-British relations, on trade and other matters, that will undo the external shock of Brexit and, most importantly, bring northern Catholics back to their constitutional senses and to union.
I show that the thesis of Catholic acceptance of partition in the years before Brexit is in all essentials egregiously wrong. Academics fabricated an outcome—substantive Catholic movement towards union and away from unity—and then conjured an explanation for their fabrication—Catholics’ experience and support of devolution caused a profound change in their constitutional beliefs. This consensus view remains unwavering, despite the feeble evidence marshalled in its favour. In fact, refuting the acceptance thesis is so easily accomplished that it invites speculation as to why so many scholars continue confidently to expound it.
Social Science Folly
Pro-union academic Adrian Guelke gives a classic statement of the acceptance thesis:
Prior to the 2016 [Brexit] referendum, it had seemed that the issue of the border had been settled once and for all and that Catholic/nationalist contentment with the functioning of the Good Friday Agreement had banished the grievance of partition. This was borne out by declining support for the option of a united Ireland in the polling of political attitudes in the province, despite the increase in the numbers of Catholics and those with a Catholic background as a percentage of the population in Northern Ireland. The trend of Catholic acceptance of the territorial status quo did not go unnoticed and it prompted the then-First Minister, Peter Robinson, to proclaim in a speech in November 2012: ‘The siege has lifted, the Troubles as we knew them are over, and the constitutional debate has been won.’ This was a high point in confidence in the Good Friday Agreement as a durable resolution of the Northern Ireland problem (Guelke, 2019, p. 397).
If Peter Robinson, the avowed tactical genius of the DUP, should proclaim that unionists have won the constitutional debate, who could doubt him?
Few did. Many scholars embrace the two major components of the acceptance thesis: Catholic constitutional beliefs profoundly change in favour of union, and the GFA is a major cause of that transformation. In a series of studies, Coakley plots opinion poll data over time to document what he regards as a significant alteration in northern Catholics’ attitudes on union and unity. He identifies the Agreement and its popular ratification as a decisive moment that entrenched the geopolitical status quo and marginalized belief in a united Ireland in both the north and the south (Coakley, 2005, 2007, 2015, 2017 & 2018). The Northern Ireland Peace Monitoring Report regularly points out the long-term trend of increasing Catholic support for union and decreasing Catholic support for unity (Nolan, 2012 & 2013; Gray et al., 2018 & 2023).[1] McCall and Nagle separately emphasize how the Agreement provides a new paradigm of transterritorial governance that realigns Catholic/nationalist identities and constitutional preferences towards acceptance of British sovereignty (McCall, 2001; Nagle, 2013). For Hayes and McAllister, the GFA’s devolution provisions are the key that helped attract Catholics to union over unity (Hayes & McAllister, 2013). Coulter and others argue that the Agreement establishes a constitutional tranquility in which Catholics relegate a united Ireland to the far-distant and fundamentally unrealizable realm of aspiration, effectively consenting to maintenance of the union. Brexit shatters this Catholic repose, much to the chagrin of the authors who look forward to reinstatement of the pre-Brexit calm (Coulter et al., 2021).
The annual Northern Ireland Life and Times (NILT) surveys dating to 1998 are the most comprehensive source of Catholic attitudes on the constitution. Advocates of the acceptance thesis routinely present NILT data to “substantiate” the extent of Catholics’ change of heart on partition. I use these same data to disconfirm the thesis of Catholic acceptance of union (ARK/NILT, n.d.).
In the analysis below, I employ the variable “religion” and the category “Catholic” because those are the terms in which the acceptance thesis is most commonly expressed. For many authors, the religion variable is useful in any study of long-term change because it facilitates comparison between survey and census data. That said, the acceptance thesis frequently uses “Catholic” as a proxy for people holding a nationalist political identity. Even though I present survey results only for Catholics, I’ve conducted a parallel analysis using the category “nationalist”, defined as all those respondents who say they generally think of themselves as nationalist as opposed to unionist or neither nationalist nor unionist. My analysis of nationalists produces similar results and the same interpretation as I report for Catholics. That is, it also strongly disproves the acceptance thesis.
Figure 1 shows Catholic support for unity (green line connecting square data points) and union (orange line with circular data points) from 1989 to 2024 using the measure of long-term constitutional preference. I draw the data for 1989 to 1996 from the Northern Ireland Social Attitudes (NISA) surveys. NISA, an extension of the British Social Attitudes surveys, is essentially a precursor of NILT. It repeatedly monitors the constitutional preferences of the northern electorate in the years prior to the GFA (Social and Community Planning Research, n.d.).
Figure 1 has three vertical black lines representing significant events that occurred during the data series. They will prove crucial to assessing the veracity of the acceptance thesis. The importance of the first line is evident—it signifies the signing and ratification of the Agreement in 1998. The last line of course marks the Brexit referendum in 2016. The significance of the middle line is less apparent: it indicates that in 2007 the NILT introduces a revised question measuring long-term constitutional preference. The original question, used in all the NISA and NILT surveys from 1989 to 2006, asks respondents: “Do you think the long-term policy for Northern Ireland should be for it to remain part of the United Kingdom or, to (re)unify with the rest of Ireland?” This question is balanced in that it gives respondents one opportunity to register their preference for union, and one opportunity to record their support for unity. The revised question, inserted in the 2007 survey and still in use, asks: “Do you think the long-term policy for Northern Ireland should be for it to remain part of the United Kingdom, with direct rule, to remain part of the United Kingdom, with devolved government or, to reunify with the rest of Ireland?” This question is unbalanced, giving respondents twice as many opportunities to say “union” as to say “unity.” In effect, the revised question is manifestly biased in a way that systematically overestimates support for union and underestimates support for unity. And this bias affects Catholics more than it affects Protestants or respondents with no religion (Burke, 2021). Any discussion of changes in the constitutional preferences of Catholics must take into account the impact of systematic measurement error from 2007 onwards.
Even an analysis as simple as that in Figure 1—setting out the basic distribution of Catholic constitutional preferences over time—uncovers fatal flaws in the thesis that the Agreement led to Catholic acceptance of union. The orange line shows that Catholic support for union is significantly lower in the ten years after the GFA than it was in the decade before. If anything, the data suggest that the Agreement turned Catholics away from union, which is a pattern that directly rebuts the thesis of Catholic acceptance.
To save their thesis, proponents might legitimately argue that the lack of Catholic acceptance of union immediately after 1998 reflects the many delays in implementing the Agreement, Britain’s controversial and unilateral decisions to suspend the devolved institutions on four occasions, and the long period without any Stormont government at all, which lasted from October 2022 to May 2007. Devolution is in constant crisis and did not have a fair opportunity to influence Catholic attitudes. Proponents could also point to Figure 1 to show that Catholic constitutional preferences begin to change decisively in favour of union in 2007, at the precise time the GFA really starts to work, thanks to the catalyst provided by the St Andrew’s Agreement and the restoration of devolution.
I should point out that I’m being very generous to the acceptance thesis, trying to come up with a plausible interpretation of it that is consistent with the data in Figure 1. I want to give the thesis every opportunity to be confirmed. The formulation of the thesis is in fact exceedingly vague and loose, rarely straying beyond the cryptic statement that Catholic constitutional preferences change in favour of union because of devolution. Its proponents never try to “save” the thesis for they don’t believe that it’s at risk. For many of them, the thesis is a statement of an unassailable truth.
Figure 1 confirms that something unusual happened circa 2007 that quickly reversed Catholic constitutional preferences. Between 2006 and 2007, Catholics’ support for union surges by 16.9 percentage points and their preference for unity plummets by 8.5 points, an absolute change of over 25 points in one year. This is the single biggest year-to-year change in the entire NISA/NILT series.[2] In 2008, for the first time, more Catholics favour union (43.4%) over unity (38.4%). And this reversal has lasting effects that endure until the unexpected intrusion of Brexit. As the figure shows, for the period from 2008 to 2016, the orange line indicating Catholic support for union is always above the green line showing support for unity. These years stand out as the only period in which more Catholics prefer maintenance of the union to Irish reunification. As I mentioned above, proponents of the acceptance thesis might point to the St Andrew’s Agreement and its promise of a more settled implementation of devolution as an explanation for Catholics’ changed attitudes during these years.
This seemingly logical but ultimately superficial argument cannot salvage the acceptance thesis. It’s true that Catholics were heavily in favour of the St Andrew’s Agreement, but there is no evidence to suggest that support for the new agreement changed their minds on the constitution. The 2006 NILT survey asks respondents how they would vote if a referendum were held on the St Andrew’s Agreement. Two-thirds of Catholics say they would vote “yes,” and only 2 percent say “no”. But almost one-third answer that they don’t know how they would vote. It’s inconceivable that—with so many Catholics undecided on its merits—the St Andrew’s Agreement could move Catholic opinion on the constitution to the degree shown in Figure 1. In any event, Catholics in favour of St Andrew’s are slightly less likely to support union (18.9%) than are those voting against the Agreement (25%). There is no comfort here for the thesis that Catholic support for the St Andrew’s Agreement produces acceptance of the union.Certainly, some years show the pattern predicted by the acceptance thesis: growing Catholic support for the Assembly moves along with growing support for union (2008-2010).
As we know, St Andrew’s declaration of a new dawn for devolved governance soon deteriorated into a series of bitter and prolonged disputes. The Executive did not meet for five months in 2008 because of an impasse over the devolution of policing and justice, which remained unresolved until the Hillsborough (Castle) Agreement in 2010. Loyalist protests over the flying of flags from public buildings erupted in December 2012, and the flags controversy contributed to the 2013 failure of the Haass talks, which tried to address not just flags but emblems, parades, protests and the legacy of the past. The 2014 Stormont House Agreement and 2015 Fresh Start Agreement—both in a sense continuations and extensions of the Haass process—also proved unable to solve the issues of how to deal with the past, welfare reform and other matters. In the wake of PSNI allegations of IRA involvement in the killing of Kevin McGuigan, DUP leader Peter Robinson refused to work with Sinn Féin, and stepped down as First Minister in September 2015. The DUP began an “in-out” or “hokey-cokey” approach to executive office, with a rotating group of its MLAs resigning their ministries and then retaking office only to resign again. In sum, this series of disagreements show that the decade after the 2006 St Andrew’s Agreement was as beset by devolution crises as was the decade before.[3] There is, then, nothing to recommend the argument that a period of effective devolution in the years after St Andrew’s accounts for what Figure 1 shows are the highest-ever rates of Catholic support for union.
Something else must be going on—something completely ignored by the acceptance thesis—that explains both the sudden transformation of Catholic constitutional attitudes in 2007 and the consolidation of the new belief pattern in the subsequent decade. That “something” is the revised and biased survey question measuring constitutional preference that the NILT begins using in 2007. Recall that the measure is beset by systematic error that results in overestimating support for union and underestimating support for unity. This explanation, rather than the acceptance thesis, accounts for the trends in Figure 1. Catholic support for union rises and Catholic support for unity falls in 2007, just as one would expect given the direction of bias in the revised measure. And the NILT’s continued use of the biased measure in each survey after 2007 explains why Catholics’ support for union remains high and their preference for unity low, until Brexit disrupted the distribution of Catholic constitutional attitudes.
Proponents of the acceptance thesis, with the partial exception of Coakley, never consider measurement error as a plausible alternative explanation for shifts in the constitutional beliefs of Catholics.[4] They mistake substantive change for change produced by measurement artefact. The sudden transformation in Catholic opinion in 2007 coinciding exactly with the introduction of a revised and obviously biased measure of opinion should have raised serious methodological concerns. But acceptance theorists, no less than the designers of NILT surveys, seem unperturbed by the problematic link between an altered measuring instrument and an altered statistical result.
Figure 2 rearranges the data used in Figure 1 to facilitate evaluating the relative worth of the two alternative explanations of Catholic acceptance and measurement artefact. It shows average Catholic support for unity and union over four periods of interest. These four periods are based on the vertical black lines of Figure 1 that mark notable events in the data series. The 1998 signing and popular ratification of the GFA is the demarcation between the first two periods. The introduction of the biased measure of constitutional preference in 2007 is the start of the third period, which I’ve labelled the pre-Brexit years. And the 2016 referendum on exiting Europe initiates the final, post-Brexit period. Figure 2 emphasizes how anomalous is the pre-Brexit period: 2007 to 2015 are the only years in which average Catholic support for union exceeds that for unity. The distributions of constitutional opinions in the other three periods are remarkably similar in their pro-unity profile: with Catholic preference for a united Ireland averaging near 50 percent, and some 20 to 30 percent of Catholics on average backing union.
The pattern of Catholic constitutional beliefs across the various periods appears at first to be an enigmatic feature of the data. Recall that the biased question on constitutional preference—which inflates support for union and deflates support for unity—is in continuous use from 2007 onwards. It therefore corrupts the measurement of opinion in both the pre-Brexit and post-Brexit years. Why then do these two periods show fundamentally different distributions of Catholic beliefs on the constitution? And why does the post-Brexit distribution, when the biased measure is in use, so closely resemble the pre- and post-GFA periods, when an unbiased measure is in place? The answer to both questions lies in the impact of Brexit on constitutional beliefs. Many analysts show that Brexit increases support for Irish unity and decreases support for continued union (Connolly & Doyle, 2019; Garry et al., 2021; Hayward, 2021; Burke, 2023; European Movement Ireland, 2025). Certainly, Catholics’ preference for unity was trending upwards and their support for union trending downwards in the three years prior to 2016. But Brexit consolidated and accelerated these trends, opening an increasingly large gap between the high number of Catholics supporting unity and the low numbers favouring union in the years after 2016 (see Figure 1). This Brexit effect in a sense undoes or compensates for some of the systematic error in the measure of constitutional preference. That is, the anti-union, pro-unity influence of Brexit offsets the pro-union, anti-unity bias of the measure. The post-Brexit distribution then can be seen as a snapshot of Catholic opinion with some of the measurement error removed.[5] That it parallels Catholic constitutional beliefs in the years straddling the GFA underlines the persistence and stability of Catholic preference for unity over union, and illustrates the poverty of the acceptance thesis. Catholic support for partition is never a normal condition of politics in the north.
There is no plausible interpretation of Figures 1 and 2 that lends any credence to the thesis of Catholic acceptance of British rule. But proponents of the thesis would have you believe otherwise. They exhibit a kind of magical empiricism, which allows them to hold on to the integrity of their thesis even though there is clear statistical evidence refuting it. For acceptance theorists, the only time that Catholics’ constitutional beliefs really count is when they show an overall preference for union, as in the pre-Brexit years from 2007 to 2015. They ignore or otherwise marginalize Catholic opinion in all other periods, when Catholics show strong support for Irish unity over continued union. In the years immediately before and after the GFA, they dismiss Catholic preference for a united Ireland as merely aspirational. In the post-Brexit years since 2016, they similarly discount Catholics’ emphatic wish for unity as either ephemeral, a belief that will dissipate once the external disturbance caused by Brexit settles down, or as superficial because Catholics will jettison their preference for a united Ireland when they begin to understand the costs and consequences of unity (Coulter et al., 2021). These ad hoc and futile attempts to rescue the acceptance thesis are self-serving, and smack of imperial conceit. Acceptance theorists are in effect saying that if Catholic opinions do not align with the constitutional tastes of the pro-union academics studying them, then those opinions must be specious. Such magical empiricism has no place in the analysis of social surveys.
The absurdity of the acceptance thesis becomes even more apparent once we appreciate fully that measurement bias is a plausible alternative explanation for changes in Catholics’ constitutional beliefs after 2006. Advocates of the thesis assert that the period from 2007 to 2015—the only pro-union period in the 35-year data series—represents the one true expression of Catholic preference. Yet this is the very period in which the survey data are most contaminated by systematic measurement error boosting support for union and sinking support for unity. The biased measuring instrument creates an artificial Catholic plurality favouring union that acceptance theorists misconstrue as real. They must remain oblivious to the impact of measurement error because it thoroughly subverts their thesis. They prefer a fanciful explanation of substantive change in Catholic opinion over the conspicuous explanation of artefactual change induced by measurement bias.
Up to this point, I’ve tested the acceptance thesis using the single variable of long-term constitutional preference. I’ve examined Catholic beliefs over time, from 1989 to 2024, and supplemented this basic data with contextual and methodological information on notable events occurring during the NISA and NILT series of surveys. This elementary analysis fatally undermines the proposition of Catholic approval of union. But, as part of my never-ending quest to be as fair as possible to the acceptance thesis, I’ll conduct one more test that adds other variables to the mix. I’ll examine attitudes towards the GFA and Stormont governance to test directly the hypothesis that Catholics’ support of the devolution settlement leads them to accept union over unity. This analysis buries (finally) the acceptance thesis, with no prospect of resurrection.
Figure 3 gives the proportion of Catholics, Protestants and respondents with no religion who say they voted “yes” to support the GFA in the May 1998 referendum, and who say they would vote “yes” today if another referendum were held. It documents what everyone knew at the time of the Agreement, that more Catholics than Protestants were in favour of the provisions of the settlement. This pattern holds to 2005, the final year in which the NILT asks the question about voting in a GFA referendum. The level of Catholic support for the GFA is high not only when compared to other religious groups, it’s also absolutely high: the number of Catholics voting to accept the Agreement averages 91 percent from 1998 to 2005.
The NILT does not again regularly examine overall assessments of the Agreement until 2019. Figure 4 shows the number in each religious group who say that the GFA remains the best basis for governing the north, as opposed to saying it is no longer a good basis for governing, or it has never been a good basis. As in Figure 3, Catholics show the most support for the Agreement. Compared to Protestants and the no religion group, they are more likely to say that the GFA remains the best basis for governing. And again, like in Figure 3, the absolute level of Catholic support of the GFA is high, averaging 78 percent over the period. The stability of Catholics’ strong approval of the overall settlement—almost as sturdy in 2024 as in 1998—is noteworthy.[6] The acceptance thesis, then, is off to a promising start. But the crucial question is: does this impressive level of support for the Agreement push Catholics to accept union? Here, the thesis founders.
Figure 5 is a direct test of the acceptance thesis that, because Catholics strongly approve of the Agreement, they have come to see partition as the normal constitutional arrangement for the north. The black lines with triangular data points are an alternative representation of the data in the previous two figures, which show Catholics’ overall support for the GFA across two disparate periods. Figure 5 tracks that support alongside Catholic preference for union. If the acceptance thesis is accurate, the data should show that the strong and steady Catholic support for the devolution settlement is associated with increases in Catholic preference for union. The data show no such pattern. From 1998 to 2005, Catholics’ overwhelming willingness to vote “yes” in any referendum on the Agreement does not change their constitutional preference. Very high Catholic support for the GFA, never falling below 87 percent over the period, persists in the presence of low Catholic approval of continued union, averaging only 20 percent. In particular, there is no evidence for the proposed explanation that strongly positive Catholic evaluations of the new governance arrangements lead them to accept union. Data from 2019 to 2024 also disconfirm the acceptance thesis. Heavy Catholic majorities in favour of the GFA (averaging 78%) go along with falling levels of support for continued union (averaging only 23%). In fact, the observed pattern is exactly the opposite of what the acceptance thesis predicts: while Catholic approval of the GFA as the best basis for governing the north remains robust in the years 2019 to 2024, Catholic preference for union is trending sharply downward.[7]
There is a sizable gap in Figure 5, spanning the years 2006 to 2018, in which the NILT surveys did not ask respondents how they might vote if another GFA referendum were held today, or whether they believed the Agreement remains the best basis for governing the north.[8] During the gap years, though, the NILT does ask questions on select aspects of the Agreement. Let’s take a look at one of those questions that the NILT asks in enough years to allow meaningful tracking over time.
Figure 6 overlays the data in Figure 5 with information on Catholics’ assessment of the effectiveness of the Assembly during some of the gap years. The blue line connecting the diamond-shaped data points shows the number of Catholics who say the Assembly achieved “a lot”, as opposed to those saying it achieved “a little”, “nothing at all”, “too early to tell” how much the Assembly achieved, or “don’t know”. The NILT asks this question in every survey year from 2007 to 2015 (there was no survey in 2011 because of lack of funding). Crucially, the measure of Assembly achievement distinguishes Catholics who have positive opinions about devolution from those who do not. It is therefore an ideal candidate to assess the accuracy of the acceptance thesis. Unfortunately for the thesis, favourable Catholic assessments of the Assembly are not consistently related to Catholic warmth towards union. Certainly, some years show the pattern predicted by the acceptance thesis: growing Catholic support for the union.
Assembly moves along with growing support for union (2008-2010). In other years, though, the data directly contradict the thesis of Catholic acceptance of union: increases in support for the Assembly are associated with declines in the number of Catholics preferring union (2010-2012, 2014-2015). In still other years, the data appear incomprehensible to the thesis: falling Catholic support for the Assembly goes with rising Catholic preference for union (2007-2008, 2012-2013).[9] Finally, the very few Catholics who assess the Assembly positively, averaging just 14.5 percent over the period, cannot possibly explain the huge and persistent jump in the number of Catholics preferring union in the decade after 2006. The data in Figure 6 provide no relief to advocates of the acceptance thesis.[10]
Conclusion
My analysis soundly refutes the thesis that Catholics’ experience or support of devolution induces them to prefer continued union with Britain and therefore accept partition as the normal constitutional condition. It argues that the illusion of Catholic support for union in the decade after 2006 is explained by systematic error in the measure of long-term constitutional preference, which artificially escalates Catholics’ support for union and diminishes their preference for unity. It shows that Catholics simultaneously hold two views that undermine the acceptance thesis. First, they strongly believe that GFA-mandated political institutions are the preferable form of British rule in the north, certainly much better than the obvious alternative of direct rule from Westminster. The remarkably high levels of Catholic support for the overall devolution settlement persist under very different conditions. Support is almost universal in the early days of the Agreement, no doubt a time of promise and hope; and it remains exceedingly strong today, after more than two decades of dysfunction and ineffectiveness in devolved governance. Second, and most importantly, Catholics continue decisively to favour a united Ireland over maintenance of British sovereignty. Contrary to the acceptance thesis, Catholics’ endorsement and experience of devolution has not pushed them to support continued union.
In sum, Catholics can distinguish the best form of practical British governance in the here and now from the united Ireland they fundamentally want. This pair of attitudes is entirely understandable and legitimate. But advocates of the acceptance thesis do not in fact accept the constantly expressed pro-unity beliefs of Catholics. In a word, they do not accept Catholics as nationalists or republicans. And they seem particularly unsettled by talk of a border poll, which moves Catholics’ preference for Irish unity onto the daily agenda of practical politics.
The sophistication and subtlety of Catholic political-constitutional beliefs compares favourably to the rigidity and denialism characterizing the views of acceptance theorists. As long as there is British rule, Catholics prefer a form of governance that gives them some institutional role in decision-making. And they are discerning enough to recognize when these governing arrangements are operating poorly. But they do not want British rule at all, and welcome the political, constitutional and demographic developments portending its demise.
Notes
[1] Gray et al. also point out that this long-term trend started to be interrupted in 2013.
[2] The other big swings in Catholics’ constitutional preference seem to be related to significant political developments. The 1994 change in Catholic opinion—a sharp increase in the numbers favouring unity and steep fall in the numbers supporting union—probably reflects the Downing Street Declaration of December 1993 and the IRA’s announcement of a cessation of armed struggle in August 1994. A similar swing in 2001 may be linked to the British government unilaterally suspending devolved institutions twice in six weeks. The 2012 downturn in support for union could be prompted by the flags controversy that erupted at the end of the year, while the NILT interviews were being conducted. All these swings in Catholic opinion were short-lived, with the status quo ante restored in a year or two. The 2007 swing was more enduring, upending Catholic constitutional preferences for almost a decade.
[3] These crises of course continued after 2016. Martin McGuinness resigned as Deputy First Minister in January 2017 in protest against DUP obstruction, collapsing Stormont for three years. In February 2022, Paul Given resigned as First Minister as part of the DUP’s campaign against the Brexit protocol. The devolved institutions were not restored for a further two years.
[4] Coakley notes that the bias in the revised measure of constitutional preference will have an impact on observed opinions. In his conclusions, however, he fails sufficiently to incorporate measurement bias as a significant alternative explanation for the changes he records. Coakley is also more careful than are other acceptance theorists in that he constantly comments on the fragility of trends on constitutional preference, which can be altered by social and political developments (Coakley, 2005, 2007, 2015, 2017 & 2018).
[5] Of course, measurement error continues to affect the estimates of long-term support for union and unity in the post-Brexit period. We would expect that an unbiased measure of Catholic constitutional beliefs for this period would show lower support for union and higher support for unity than does the biased measure of long-term preference. The data confirm this expectation. Recall that the measure of long-term preference after 2006 is biased in a pro-union direction because it gives respondents two chances to say they favour union and only one chance to express a preference for unity. The NILT surveys contain another measure of constitutional beliefs that is balanced or unbiased. The question asking respondents how they would vote if there were a referendum tomorrow on the future of the north gives respondents one opportunity to support union and one to support unity. The NILT includes this question in every survey since 2019. Comparing both measures across the relevant survey years shows, as expected, that the unbiased measure (referendum vote) consistently yields lower estimates of Catholic support for union and higher estimates of Catholic support for unity than does the biased measure (long-term preference). Some of these differences are small, some are large, but the pattern is consistent over time.
[6] Catholic approval of particular aspects of the GFA varies over the years. Support for a power-sharing executive and north-south bodies remains fairly strong, but support wanes for the south dropping its territorial claim to the north, the majority consent principle and the Assembly (Hayes & McAllister, 2013).
[7] Figure 5 examines aggregate trends, comparing the overall number of Catholics who support the GFA to the overall number of Catholics who prefer union. The individual-level data confirm the aggregate trends in the figure. They show that there is no consistent evidence confirming the thesis that Catholics who support the GFA are more likely to prefer union over unity than are Catholics who do not support the Agreement.
[8] In 2007, the NILT asks whether the GFA was a good thing for the north, a bad thing, or it did not make much difference. I do not include this question in my analysis in the paper because it appears in just one survey, and it is sufficiently different from previous surveys asking about voting in a GFA referendum that concerns about comparability arise. Putting these concerns aside, including the question in the analysis only strengthens the case against the acceptance thesis.
[9] Catholic dissatisfaction with the operation of the Assembly in these years did not lead to any significant increase in the number of Catholics preferring union with direct rule over union with devolved government.
[10] The individual-level data confirm the aggregate trends in Figure 6, much as we found with the previous figure (see endnote 7). Generally, Catholics with positive opinions of the Assembly’s achievements are not more likely to support union over unity than are Catholics who view the Assembly negatively. That is, the individual-level data also refute the acceptance thesis. I’ve examined two other measures of opinions of devolution during the gap years (2007-2014), whether Catholics believe the Assembly is giving ordinary people more say in how the north is governed, and whether they are satisfied with the way MLAs are doing their jobs. The analysis of these measures also disproves the acceptance thesis: there is no consistent survey evidence at the aggregate or individual level showing that favourable assessments of Assembly governance or satisfaction with MLA job performance move Catholics to prefer union to unity.
References
ARK/NILT. (no date). ARK. Northern Ireland Life and Times Surveys [computer files]. Retrieved from.
Burke, M. (2021). “Up with Union, Down with Unity: Measuring Constitutional Preference.” The Pensive Quill. 29 May. Retrieved from.
Burke, M. (2023). “The Crisis of Democratic Legitimacy in the North.” The Pensive Quill. 18 July. Retrieved from.
Coakley, J. (2005). “Society and Political Culture.” In Politics in the Republic of Ireland. 4th ed. Ed. J. Coakley and M. Gallagher, 36-71. London and New York: Routledge, in association with PSAI Press.
Coakley, J. (2007). “National identity in Northern Ireland: stability or change?” Nations and Nationalism 13:4 (October): 573–597.
Coakley, J. (2015). “Does Ulster still say ‘no’? Public opinion and the future of Northern Ireland.” In The Act of Voting: Identities, Institutions and Locale. Ed. J.A. Elkink and D.M. Farrell, 25-55. London: Routledge.
Coakley, J. (2017). “Resolving international border disputes: The Irish experience.” Cooperation and Conflict 52:3 (September): 377–398.
Coakley, J. (2018). “Catholics in Northern Ireland: Changing Political Attitudes, 1968–2018.” In The Contested Identities of Ulster Catholics. Ed. T. P. Burgess, 21-37. Cham, Switzerland: Palgrave Macmillan.
Connolly, E., and J. Doyle. (2019). “Brexit and the changing international and domestic perspectives of sovereignty over Northern Ireland.” Irish Studies in International Affairs 30, pp. 217–233.
Coulter, C., N. Gilmartin, K. Hayward and P. Shirlow. (2021). Northern Ireland a generation after Good Friday. Manchester: Manchester University Press.
European Movement Ireland. (2025). EU POLL 2025: Ireland and Northern Ireland. May. Retrieved from.
Garry, J., B. O’Leary, K. McNicholl and J. Pow. (2021). “The future of Northern Ireland: border anxietiesand support for Irish reunification under varieties of UKexit.” Regional Studies 55:9 (September): 1517- 1527.
Gray, A.M., J. Hamilton, G. Kelly, B. Lynn, M. Melaugh and G. Robinson. (2018). Northern Ireland
Peace Monitoring Report. Number 5. October. Community Relations Council. Retrieved from.
Gray, A.M., J. Hamilton, G. Hetherington, G. Kelly, B. Lynn, P. Devine, J. Topping and R. Martin. (2023). Northern Ireland Peace Monitoring Report. Number 6. November. Community Relations Council. Retrieved from.
Guelke, A. (2019). “Northern Ireland, Brexit, and the Interpretation of Self-Determination.” Nationalism and Ethnic Politics 25:4 (October-December): 383-399.
Hayes, B.C., and I. McAllister. (2013). Conflict to Peace: Politics and Society in Northern Ireland over Half a Century. Manchester: Manchester University Press.
Hayward, K. (2021). “The precarious position of Northern Ireland.” June. UK in a Changing Europe. Retrieved from.
McCall, C. (2001). “The Production of Space and the Realignment of Identity in Northern Ireland.” Regional and Federal Studies 11:2 (Summer): 1-24.
Nagle, J. (2013). “From Secessionist Mobilization to Sub-state Nationalism? Assessing the Impact of Consociationalism and Devolution on Irish Nationalism in Northern Ireland.” Regional and Federal Studies 23:4 (December): 461-477.
Nolan, P. (2012). Northern Ireland Peace Monitoring Report. Number One. February. Community Relations Council. Retrieved from.
Nolan, P. (2013). Northern Ireland Peace Monitoring Report. Number Two. Community Relations Council. Retrieved from.
Social and Community Planning Research. (no date). Northern Ireland Social Attitudes Survey. 1989-1996. Datasets [computer files] downloaded from the UK Data Service. Retrieved from.
[1] Gray et al. also point out that this long-term trend started to be interrupted in 2013.
[2] The other big swings in Catholics’ constitutional preference seem to be related to significant political developments. The 1994 change in Catholic opinion—a sharp increase in the numbers favouring unity and steep fall in the numbers supporting union—probably reflects the Downing Street Declaration of December 1993 and the IRA’s announcement of a cessation of armed struggle in August 1994. A similar swing in 2001 may be linked to the British government unilaterally suspending devolved institutions twice in six weeks. The 2012 downturn in support for union could be prompted by the flags controversy that erupted at the end of the year, while the NILT interviews were being conducted. All these swings in Catholic opinion were short-lived, with the status quo ante restored in a year or two. The 2007 swing was more enduring, upending Catholic constitutional preferences for almost a decade.
[3] These crises of course continued after 2016. Martin McGuinness resigned as Deputy First Minister in January 2017 in protest against DUP obstruction, collapsing Stormont for three years. In February 2022, Paul Given resigned as First Minister as part of the DUP’s campaign against the Brexit protocol. The devolved institutions were not restored for a further two years.
[4] Coakley notes that the bias in the revised measure of constitutional preference will have an impact on observed opinions. In his conclusions, however, he fails sufficiently to incorporate measurement bias as a significant alternative explanation for the changes he records. Coakley is also more careful than are other acceptance theorists in that he constantly comments on the fragility of trends on constitutional preference, which can be altered by social and political developments (Coakley, 2005, 2007, 2015, 2017 & 2018).
[5] Of course, measurement error continues to affect the estimates of long-term support for union and unity in the post-Brexit period. We would expect that an unbiased measure of Catholic constitutional beliefs for this period would show lower support for union and higher support for unity than does the biased measure of long-term preference. The data confirm this expectation. Recall that the measure of long-term preference after 2006 is biased in a pro-union direction because it gives respondents two chances to say they favour union and only one chance to express a preference for unity. The NILT surveys contain another measure of constitutional beliefs that is balanced or unbiased. The question asking respondents how they would vote if there were a referendum tomorrow on the future of the north gives respondents one opportunity to support union and one to support unity. The NILT includes this question in every survey since 2019. Comparing both measures across the relevant survey years shows, as expected, that the unbiased measure (referendum vote) consistently yields lower estimates of Catholic support for union and higher estimates of Catholic support for unity than does the biased measure (long-term preference). Some of these differences are small, some are large, but the pattern is consistent over time.
[6] Catholic approval of particular aspects of the GFA varies over the years. Support for a power-sharing executive and north-south bodies remains fairly strong, but support wanes for the south dropping its territorial claim to the north, the majority consent principle and the Assembly (Hayes & McAllister, 2013).
[7] Figure 5 examines aggregate trends, comparing the overall number of Catholics who support the GFA to the overall number of Catholics who prefer union. The individual-level data confirm the aggregate trends in the figure. They show that there is no consistent evidence confirming the thesis that Catholics who support the GFA are more likely to prefer union over unity than are Catholics who do not support the Agreement.
[8] In 2007, the NILT asks whether the GFA was a good thing for the north, a bad thing, or it did not make much difference. I do not include this question in my analysis in the paper because it appears in just one survey, and it is sufficiently different from previous surveys asking about voting in a GFA referendum that concerns about comparability arise. Putting these concerns aside, including the question in the analysis only strengthens the case against the acceptance thesis.
[9] Catholic dissatisfaction with the operation of the Assembly in these years did not lead to any significant increase in the number of Catholics preferring union with direct rule over union with devolved government.
[10] The individual-level data confirm the aggregate trends in Figure 6, much as we found with the previous figure (see endnote 7). Generally, Catholics with positive opinions of the Assembly’s achievements are not more likely to support union over unity than are Catholics who view the Assembly negatively. That is, the individual-level data also refute the acceptance thesis. I’ve examined two other measures of opinions of devolution during the gap years (2007-2014), whether Catholics believe the Assembly is giving ordinary people more say in how the north is governed, and whether they are satisfied with the way MLAs are doing their jobs. The analysis of these measures also disproves the acceptance thesis: there is no consistent survey evidence at the aggregate or individual level showing that favourable assessments of Assembly governance or satisfaction with MLA job performance move Catholics to prefer union to unity.
References
ARK/NILT. (no date). ARK. Northern Ireland Life and Times Surveys [computer files]. Retrieved from.
Burke, M. (2021). “Up with Union, Down with Unity: Measuring Constitutional Preference.” The Pensive Quill. 29 May. Retrieved from.
Burke, M. (2023). “The Crisis of Democratic Legitimacy in the North.” The Pensive Quill. 18 July. Retrieved from.
Coakley, J. (2005). “Society and Political Culture.” In Politics in the Republic of Ireland. 4th ed. Ed. J. Coakley and M. Gallagher, 36-71. London and New York: Routledge, in association with PSAI Press.
Coakley, J. (2007). “National identity in Northern Ireland: stability or change?” Nations and Nationalism 13:4 (October): 573–597.
Coakley, J. (2015). “Does Ulster still say ‘no’? Public opinion and the future of Northern Ireland.” In The Act of Voting: Identities, Institutions and Locale. Ed. J.A. Elkink and D.M. Farrell, 25-55. London: Routledge.
Coakley, J. (2017). “Resolving international border disputes: The Irish experience.” Cooperation and Conflict 52:3 (September): 377–398.
Coakley, J. (2018). “Catholics in Northern Ireland: Changing Political Attitudes, 1968–2018.” In The Contested Identities of Ulster Catholics. Ed. T. P. Burgess, 21-37. Cham, Switzerland: Palgrave Macmillan.
Connolly, E., and J. Doyle. (2019). “Brexit and the changing international and domestic perspectives of sovereignty over Northern Ireland.” Irish Studies in International Affairs 30, pp. 217–233.
Coulter, C., N. Gilmartin, K. Hayward and P. Shirlow. (2021). Northern Ireland a generation after Good Friday. Manchester: Manchester University Press.
European Movement Ireland. (2025). EU POLL 2025: Ireland and Northern Ireland. May. Retrieved from.
Garry, J., B. O’Leary, K. McNicholl and J. Pow. (2021). “The future of Northern Ireland: border anxietiesand support for Irish reunification under varieties of UKexit.” Regional Studies 55:9 (September): 1517- 1527.
Gray, A.M., J. Hamilton, G. Kelly, B. Lynn, M. Melaugh and G. Robinson. (2018). Northern Ireland
Peace Monitoring Report. Number 5. October. Community Relations Council. Retrieved from.
Gray, A.M., J. Hamilton, G. Hetherington, G. Kelly, B. Lynn, P. Devine, J. Topping and R. Martin. (2023). Northern Ireland Peace Monitoring Report. Number 6. November. Community Relations Council. Retrieved from.
Guelke, A. (2019). “Northern Ireland, Brexit, and the Interpretation of Self-Determination.” Nationalism and Ethnic Politics 25:4 (October-December): 383-399.
Hayes, B.C., and I. McAllister. (2013). Conflict to Peace: Politics and Society in Northern Ireland over Half a Century. Manchester: Manchester University Press.
Hayward, K. (2021). “The precarious position of Northern Ireland.” June. UK in a Changing Europe. Retrieved from.
McCall, C. (2001). “The Production of Space and the Realignment of Identity in Northern Ireland.” Regional and Federal Studies 11:2 (Summer): 1-24.
Nagle, J. (2013). “From Secessionist Mobilization to Sub-state Nationalism? Assessing the Impact of Consociationalism and Devolution on Irish Nationalism in Northern Ireland.” Regional and Federal Studies 23:4 (December): 461-477.
Nolan, P. (2012). Northern Ireland Peace Monitoring Report. Number One. February. Community Relations Council. Retrieved from.
Nolan, P. (2013). Northern Ireland Peace Monitoring Report. Number Two. Community Relations Council. Retrieved from.
Social and Community Planning Research. (no date). Northern Ireland Social Attitudes Survey. 1989-1996. Datasets [computer files] downloaded from the UK Data Service. Retrieved from.
⏩ Mike Burke has lectured in Politics and Public Administration in Canada for over 30 years.
No comments