Public attitudes to referendums on Irish unification in Northern Ireland: evidence from an online consultation

ABSTRACT

Debate on Irish unification has increased in recent years, yet public attitudes on the processes that might lead to it are underexplored. This article examines the results of an online public consultation carried out in Northern Ireland in the summer of 2020 on how any future referendums on the unification question would best be conducted. It shows that hopes and fears on this question exist across society in Northern Ireland. It also reveals that public views are yet to crystallise on many of the design features of a referendum. There are some exceptions to this: there is a widespread feeling that the Brexit referendum process should not be replicated; and there are signs of divergent views emerging between communities on the franchise, referendum threshold, and use of citizens’ assemblies. Nevertheless, we conclude that the lack of hardened views on most questions of process points to an opportunity for policymakers and impartial observers to foster agreement on how a referendum process should be conducted.

KEYWORDS:

• Northern Ireland
• direct democracy
• Brexit
• Belfast/Good Friday Agreement

The framework for possible referendums on Northern Ireland's constitutional future is laid out by the Belfast/Good Friday Agreement of 1998. The Agreement gives the UK government’s Secretary of State for Northern Ireland the power to call such a vote in Northern Ireland at any time, and requires them to do so if a majority for a united Ireland appears to them ‘likely’ (Belfast/Good Friday Agreement, 1998, Constitutional Issues Section, Annex A). For many years, the possibility of such a vote appeared remote, but discussion of the matter has recently increased, and polling suggests some shift in public attitudes in favour of unification (Working Group on Unification Referendums on the Island of Ireland, 2021, pp. 50–51). Research has examined whether people want unification (e.g. Ashcroft, 2019; Lucid Talk, 2021; Northern Ireland Life and Times [NILT], 2020; Shirlow, 2021), what factors might shape their choice (Garry, O’Leary, Coakley, Pow, & Whitten, 2020; NILT, 2019), and what form of united Ireland they would prefer if it came about (Garry et al., 2020). By contrast, public views on processes of deciding the constitutional question – including how any future referendum process would best be conducted – have remained underexplored.

That gap in research might appear unsurprising: public interest in the processes, as distinct from the outcomes, of politics is generally assumed to be low. Indeed, research has repeatedly found that citizens’ preferences about political processes are strongly outcome-contingent (e.g. Armingeon & Guthmann, 2014; Bol, Blais, Gillard, Lopez, & Pilet, 2018; Landwehr & Harms, 2020; Werner, 2020). In Northern Ireland’s context, however, the omission is problematic. Any referendum on the constitutional question would be contentious and potentially destabilising. To minimise these dangers, the process itself would need to be designed and conducted well, in order to maximise clarity. Yet the 1998 Agreement is silent on many aspects of any referendum, so crucial decisions on process remain unmade. Public expectations about the process would be crucial to perceptions of fairness and legitimacy. Whether clear public expectations already exist on these matters, what they are, and whether they vary across society therefore matters.

To help address this gap, we analyse here the results of an online public consultation conducted in Northern Ireland in the summer of 2020, which sought to elicit people’s views and feelings in relation to a possible referendum. The consultation was carried out for the Working Group on Unification Referendums on the Island of Ireland, which was formed by the UCL Constitution Unit in 2019 and which released its final report in May 2021 (Working Group on Unification Referendums on the Island of Ireland, 2021).¹ Without taking a collective view on the desirability of different outcomes, the Working Group examined all aspects of a possible referendum process, including how referendums might be sequenced, how they might be configured in relation to other steps in a decision-making process, and how they would best be conducted.

We begin by reviewing existing research on public attitudes towards unification referendums and setting out four unanswered questions that we seek insights on through our study. Then we describe our methodology, in terms of consultation design, dissemination, and analysis. We also set out who responded to the consultation, and consider implications for how those responses should be interpreted. We present the consultation results first in terms of attitudes towards a referendum and a united Ireland in broad terms, and second in relation to views on specific aspects of the referendum process. We conclude by reflecting on the implications of our findings for how the process of designing any future referendum on the constitutional question should be approached.

Existing research

We are not the first to investigate public views on the prospect of a referendum on Northern Ireland’s constitutional status. Existing research takes both quantitative and qualitative forms. Some of that research is academic, but relevant work has also been done through political and civic channels.

Quantitative studies, drawing on opinion polls and surveys, typically focus on the basic questions around support for or opposition to unification itself and whether a referendum should be held. Until around 2013, almost all such studies found support for unification to be below 30%. Since then, results have become much more varied. While some studies continue to suggest little or no change (e.g. Shirlow, 2021), a few have placed support for unification at or close to 50% once undecided respondents are excluded (e.g. Ashcroft, 2019; see Working Group on Unification Referendums on the Island of Ireland, 2021, pp. 49–53). The Northern Ireland Life and Times (NILT) survey, carried out annually, has asked more specific questions, including on how attitudes on the constitutional question are affected by Brexit and other factors (NILT, 2020). Some studies also look at support for the idea of holding a referendum. Lucid Talk (2021) found majority backing for such a vote ‘in the next five years’. None of these studies, however, has delved into attitudes on referendum process.

Deliberative methods have recently been deployed to dig deeper into views towards a possible united Ireland, and some of these have also generated evidence on attitudes towards the process. A small-scale citizens’ assembly conducted by researchers in 2018 explored the impact of Brexit on attitudes towards unification. Participants shared their fears that a referendum would cause division and lead to violence, and they favoured clear information during any referendum and an inclusive and rational debate (Garry, O’Leary, McNicholl, & Pow, 2021). An assembly held in the Republic explored views on process in more depth. It found that, before deliberation began, most participants wanted the form of a united Ireland to be specified before rather than after a referendum; at the end of the assembly, that majority was overwhelming: 42 out of 50 participants (Garry, O’Leary, Gillespie, & Gjoni, 2021, p. 16). Multiple participants contrasted this approach with that taken to Brexit (Garry, O’Leary, Gillespie, & Gjoni, 2021, p. 19). These are valuable insights. Equivalent questions have not, however, been explored in the North.

A sentiment analysis of unionists and nationalists in a town in Northern Ireland has found that, while unionists are clear they do not want Irish unity, some are willing to talk about the issue, despite its previously ‘taboo’ status (Dornschneider and Todd, 2020, pp. 17). Elsewhere, an Oireachtas report by Senator Mark Daly (2019) drew on contributions from unionist politicians and civic activists as well as focus groups with members of unionist organisations in Northern Ireland. It identified seven fears among unionists concerning a united Ireland: identity loss; Republican triumphalism; retribution towards former members of the Royal Ulster Constabulary (RUC), the British military, and the prison service for their roles during the Troubles; confiscation of land; violence; return to the EU; and concerns about healthcare, economy and welfare (Daly, 2019, p. 8). The report thus focused primarily on attitudes towards unification itself. But it did also repeatedly highlight one concern about the process of a referendum: that a referendum held before the detail of the proposed change has been worked out is undesirable (Daly, 2019, p. 3). As above, this point was explicitly connected to experiences around Brexit.

Thus, while existing work focuses primarily on attitudes to unification itself, it also provides some information on views about the referendum process. We see, in particular, a view that voters should be presented with a clear choice, the form of united Ireland on offer having been worked out in advance. And voters – at least some of them – see that as a lesson learnt from Brexit.

Yet the evidence that we have to date remains scant, particularly in Northern Ireland. We have four outstanding questions:

• First, to what extent are people in Northern Ireland thinking about the process by which a decision on the constitutional future might be made, rather than just about the possible outcomes? Who wants to talk about this issue, and who does not?

• Second, what are people’s broad attitudes towards this process – specifically, to a possible referendum? What hopes or fears do they have in relation to it?

• Third, to what extent do people have views in relation to specific aspects of the design of such a referendum? Do people already have clear ideas about the form that a referendum should or should not take – such that any deviation from that form might delegitimate the process in their ideas? Or do they not have such views?

• Fourth, insofar as people do have such views, what are they?

We turn now to describing the evidence through which we seek answers to these questions and the ways in which we have analysed it.

Methodology

The evidence presented here comes from an open consultation conducted for the Working Group on Unification Referendums on the Island of Ireland. Through the consultation, the Working Group hoped to access as wide a range of opinion from Northern Ireland’s different communities as possible. In this section, we address four issues: the choice of an open consultation as our research tool, and its implications; the content of the survey questionnaire; how the consultation was publicised; and how the responses were analysed.

Open consultation

The evidence that we draw on comes from an open consultation. That is, participation in the consultation was open to anyone who wished to take part: no form of representative sampling was employed. The choice of this methodology was dictated largely by the fact that the consultation was conducted for the Working Group on Unification Referendums on the Island of Ireland. The legitimacy of the Working Group required that any member of the public who wished to express their thoughts to the group was able to do so. The use of this approach clearly means that the sample of respondents is not representative of the population of Northern Ireland: rather, it consists of people who became aware of the consultation and chose to participate in it. That places limits on how the results can be interpreted: as discussed in the summary of responses below, those who chose to respond were strongly skewed towards particular parts of the population.

On the other hand, the open consultation approach also has strengths. First, it provides evidence in relation to the first of our questions above: on who wants to engage with this subject. Second, it shows what people who are engaged are thinking. Given that the topic of how a decision on Northern Ireland’s constitutional future should be made (as distinct from what the decision should be) is relatively esoteric, a research process that gathers the thoughts of those who are already interested in the subject – and may be influential on future debates should they become more mainstream – is instructive. Third, given the paucity of existing evidence, this approach is a valuable first step in research: by canvassing for ideas in an open-ended way, it enables themes to be identified that might subsequently be probed further through survey questions or deliberative methods.

These strengths – particularly the first – can best be realised if the main factor driving participation in the consultation is interest in the topic rather than awareness of the consultation. The processes through which we publicised the consultation are therefore important, and we discuss them below, after outlining the consultation’s content.

Consultation questionnaire

With the first of our research questions being addressed by the use of an open consultation approach, we sought to address the remaining three questions through the design of the questionnaire. Specifically, we sought to enable respondents to express their general thoughts and feelings regarding the prospect of a referendum on the unification question; and to gather information on what thoughts, if any, respondents had about the specific features of any such referendum. The design process began by reviewing the features that the Working Group had identified as needing attention. We then engaged with civil society representatives in Northern Ireland, discussing our purposes and piloting questions. This helped ensure that the consultation was easy to understand and likely to yield meaningful responses. Once the consultation questionnaire was finalised, we placed it online.

The online consultation form began by explaining the purpose of the consultation and how the responses would be used. Respondents gave consent by clicking through to the main consultation page. Because the survey was fully anonymous, the study was exempt from UCL ethics approval, but all appropriate ethical standards were nevertheless carefully followed.

In line with our second research question, the questionnaire, after the introduction, began with very broad questions allowing respondents to express their thoughts about a referendum freely:

Q1: What do you feel when you hear talk about such a referendum? What are your hopes? What are your fears?

Q2: Would you want answers to any questions ahead of a referendum, to help you decide how to vote? If so, what questions would you want answers to?

Q3: What (if anything) do you think might help to overcome your fears? Or, what would help to fulfil your hopes?

Our pilot exercise suggested a clear answer to our third research question: very few people have thought in any detail about the specific features of a referendum. That, in turn, would make answering the fourth question difficult: there would be a risk that respondents would find the process of completing the questionnaire uncomfortable, and that the results would lack meaning. We concluded that it was therefore important to provide background information on some of the particular issues, which we provided largely in the form of a series of background questions (BQ1–5), shown in Box 1.

Box 1. Background questions set out in the consultation questionnaire.

1. How would the Secretary of State decide when to call a referendum? How should the Secretary of State judge whether a majority of voters in Northern Ireland are likely to support a united Ireland? For example, would he or she rely on opinion polls or would there need to be evidence from election results or some other source? (BQ1)

2. When would a referendum happen? Would a referendum take place before discussions on the form of a united Ireland? Would it take place after that? Might referendums be desirable at both of these stages? Should a referendum (or referendums) take place in Northern Ireland and the Republic of Ireland on the same day, or on different days? (BQ2)

3. How would plans for a united Ireland be worked out? For example, how would it be decided whether Northern Ireland would retain its own devolved government (the Assembly held in Stormont) within a united Ireland or whether powers would transfer to Dublin? How would future arrangements for things like healthcare, pensions, and policing be decided? Would such matters be decided, say, within the current Republic of Ireland, or by people across the whole island of Ireland? What roles, if any, would be played by the UK or Irish governments or the Northern Ireland Executive? What would be the role of political parties or of organisations in civil society? What would be the role of members of the public in this process – for example, through participation in a citizens’ assembly? (BQ3)

4. What about the option to stay in the UK? Would there also be a process for deciding whether, if voters opted to stay in the UK, Northern Ireland’s relationship with the rest of the UK would be reformed? If so, what would this process look like? (BQ4)

5. How would the referendum be run? Referendum procedures and regulations are very different in the UK and the Republic of Ireland. Considering this, how would the referendum campaigns be regulated? How much could campaigners spend and who could make donations? How would reliable information about the options be made available to voters, and how would any misinformation be tackled? Who would be entitled to vote in referendums both north and south? (BQ5)

In order not to overwhelm respondents with questions that they might not previously have thought about, we did not ask for responses to the background questions one by one. Instead, we asked three broader questions:

Q4: Do you have views on how any of these issues should be resolved? If you do, please indicate clearly in your response which aspect or aspects you are referring to.

Q5: You may feel you do not have enough information on these questions in order to express a view on them. If so, what further information would you want?

Q6: Have we missed anything?

Finally, in the last part of the questionnaire, we asked questions about demographics and identity, based on those asked by the annual NILT survey (NILT, 2020).

How we used the background questions has important implications for the interpretation of the results. The fact that those questions highlighted some issues but not others is likely to have affected the frequency with which topics were mentioned by respondents: those covered by the background questions may have gained extra prominence. And the approach of not seeking responses to the background questions one by one may have reduced the number of respondents offering views on each. We consider these implications further when examining the responses below.

Administering and publicising the consultation

The consultation ran from 22 July to 2 September 2020. Our aim was to reach across society and the various communities as much as possible, so we promoted the consultation as widely as possible: through social media, the mainstream media (the consultation was featured in articles in major newspapers and radio outlets in both Northern Ireland and the Republic), and blogs, and through the newsletters, email lists, and Facebook pages of civic organisations. We should note that outlets with traditionally nationalist audiences covered the consultation, as did BBC Radio Ulster, but outlets with traditionally unionist audiences did not. This may have affected who saw the survey announcement – a point that we return to below. Two weeks into the consultation, it was apparent that some groups were underrepresented: in particular, women, Protestants, unionists, those identifying as British, and those educated to A Level or lower. To address this, we identified and contacted an additional 55 civic organisations, many of which agreed to help with publicity.

Analysis

After the consultation closed, we developed a coding framework for each of the main questions asked. Three people coded the responses. For each question, they began by coding separately. They then examined each other’s work, conducted a series of blind coding rounds, and gradually developed final categories. These categories reflected answers to the key questions in the survey, as well as other issues that were raised multiple times. The codes had two layers. The first was the overarching theme or issue the response was situated within: for example, ‘fear of a united Ireland’. The second captured the more specific point being raised: for example, ‘loss of healthcare’. After initial analysis, we assessed the coding scheme as a whole and identified some gaps. We renamed, split, merged, and in some cases added new codes based on an agreed scheme. One coder added the new codes to the survey responses, while a second person checked them.

We then analysed the responses, looking at patterns both overall and within particular demographic and identity groups. As we elaborate in the following section, one key feature of the responses was that they came disproportionately from nationalists. As a result, the aggregate patterns across all respondents considered collectively have little meaning: they are in no sense representative of opinion across Northern Ireland’s population. In the sections that follow, we therefore report only breakdowns by respondents identifying as unionist, nationalist, or neither. Given the self-selected nature of our sample, caution is needed in interpreting even these patterns. But they do give a flavour of the thinking of those members of each group who chose to engage.

During the early analysis, the coders noticed repetition in a small number of responses. We found 39 responses that appeared to contain coordinated messages, with respondents raising similar themes framed in almost identical language. A further 27 responses had somewhat similar messages or language. These responses tended to express nationalist views or concerns. While they may have had some impact on the results, they constituted only 3–5% of the total sample (depending on whether the additional 27 are included), so they did not strongly affect the distribution of opinion. The issues raised were not uncommon, and there is no reason to doubt that these were genuine people expressing genuine views. Thus, while noting the patterns, we have not removed these responses from our analysis.

As we have indicated, caution should be exercised in attaching meaning to the precise numbers of respondents expressing particular views, for two reasons: the self-selected sample; and the impact of the questionnaire framing on responses. We do, however, report such numbers (as percentages of respondents from each group), to enable readers to make their own judgements. The figures show the issues that members of each community raised with us and the hierarchy of salience among them. Alongside these numbers, we also quote the words of respondents directly, to provide illustrations of the precise points that they wanted to make. We quote responses exactly as they were submitted, without editing for spelling or grammar.

Summary of responses

In total, we received 1,377 responses, including 803 from respondents who said they lived in Northern Ireland. Northern Ireland residents were our primary focus – gathering a cross-section of views in the Republic of Ireland and Great Britain as well would have required a wider promotional exercise on our part, which was beyond the scope of the exercise. The following analysis therefore focuses on responses from Northern Ireland. We did examine the responses from elsewhere; this did not identify any major views or themes that were not voiced by respondents from Northern Ireland.

Table 1 breaks down the Northern Ireland respondents across demographic categories and groups within society. As is evident, they skew strongly towards some groups. Most notably, and unsurprisingly given the subject, many more nationalists and people identifying as Irish responded than did unionists or people identifying as British. As noted above, this might partly reflect the news outlets that carried information about the consultation, but it is likely to reflect deeper patterns too. That is, it provides an important indication of the state of debate on these matters: while many nationalists are keen to consider the prospect of a referendum on the unification question, most unionists are very wary. Despite this, significant numbers identifying as British and/or unionist, or identifying as neither nationalist nor unionist, did respond, and so the consultation gives valuable insights into thinking in these communities too.

Table 1. Demographic breakdown of consultation responses.

Demographic category	% of all respondents	Demographic category	% of all respondents
Gender		Education
Male	69.1%	Degree level or higher	67.5%
Female	29.5%	Diploma or equivalent	12.3%
No information	1.4%	A-Level or equivalent	8.3%
		GSCE or equivalent	7.3%
Age		No qualifications	2.9%
17 or younger	1.0%	No information	1.6%
18–24	4.9%
25–34	15.6%	Location
35–44	20.2%	A big city	28.6%
45–54	21.3%	Suburbs/outskirts of a big city	16.2%
55–64	20.4%	A small city or town	29.0%
65–74	9.1%	A country village	12.5%
75+	3.6%	A farm/home in the country	13.6%
Ethnicity		National identity
White	94.8%	British	14.9%
Other	0.7%	Irish	66.0%
No information	4.5%	Northern Irish	13.0%
		Ulster	0.6%
Religiosity		Other	5.2%
Yes	59.5%
No	38.7%	Political identity
No information	1.%	Unionist	17.8%
		Nationalist	62.0%
Religion of upbringing		Neither nationalist nor unionist	18.9%
Catholic	42.2%	No information	1.2%
Protestant – all denominations	16.1%
Other	2.5%
No information	39.2%

As noted above, we also observed a strong gender imbalance early in the consultation period and sought to address it by contacting organisations working specifically with women. That a strong imbalance remained may reflect wider patterns of political discourse (see Potter, 2020). Respondents also skewed strongly towards those with a higher level of formal education, perhaps reflecting patterns in society, or as a result of the survey having been produced and disseminated by a team based in a university.

An issue arose regarding place names. The consultation asked respondents to select their location from a prescribed list, which included ‘Northern Ireland’ and ‘Republic of Ireland’. Some selected ‘elsewhere’ and then entered ‘the North’ or words that implied they lived in the six counties that constitute Northern Ireland (23 respondents). Some simply wrote ‘Ireland’ (25 respondents) which made it unclear which jurisdiction they were resident in. In almost all of these cases, respondents identified themselves as nationalists. Given the ambiguity, these responses were not added to the final analysis. Doing so would not have significantly altered the results.

Responses 1: hopes and fears

Our broad opening questions (Q1–3) elicited a wide range of perspectives. Respondents expressed hopes and fears about many different things: about what would happen if there was a referendum or if there was not; about ways in which particular actors might, in the eyes of the respondent, behave helpfully or harmfully in the context of a referendum. In addition, many respondents did not express hopes or fears at all. In particular, a majority of unionist respondents and some respondents who identified as neither unionist nor nationalist said that they were either completely opposed to a referendum or that it was premature to discuss the matter. Given these patterns, it would not be meaningful to present aggregate numbers of those who expressed hopefulness or fearfulness in the round. It is notable that, though we did not ask specifically for respondents’ views on a potential united Ireland – Q1 was about hopes and fears concerning a referendum – many nonetheless shared them.

Below, we highlight key general themes that emerged from respondents’ answers: about their hopes and fears, concerning a united Ireland and concerning a referendum (responses to Q1); and about how to fulfil their hopes or address their fears (responses to Q3). We report the percentage of respondents from each community who offered thoughts coded within a particular theme. We focus on themes that were mentioned by around 10% or more of respondents from at least one of the communities in Northern Ireland, plus some themes that, though mentioned less often, were relatively prominent in the small group of unionist respondents. Given the nature of the sample, we do not report very small numbers of below 4%.

Hopes and fears for a united Ireland

In response to Q1, some respondents shared their hopeful expectations for a united Ireland. Table 2 outlines the most prominent themes they voiced. It shows that aspirations for better community cohesion in a united Ireland were the most prominent positive expectations. One nationalist said: ‘I have a hope and aspiration of a new inclusive and prosperous Ireland of equals. A country that is welcoming and representative of everyone’. Smaller numbers also had hopes for a positive economic impact of unification. One respondent who identified as neither nationalist nor unionist made that point alongside several others:

The possibility of a referendum excites me as I think a united island inside the EU could have enormous potential in many areas: the environment, health and education, business and inward investment, potential to influence peace and wellbeing on the world stage and within the EU, and more.

Re-joining the EU in the event of unification was mentioned by significant numbers of respondents who shared hopes about unification. One unionist said: ‘I would also be pleased if there was an immediate EU membership for all NI citizens of such a referendum passed’.

Table 2. Positive expectations for a united Ireland.

Issue	Nationalist	Unionist	Neither
Hopes for better community cohesion in a united Ireland	25%	<4%	13%
Hopes for a positive economic impact of unification	12%	<4%	7%
Hopes for re-joining the EU in the event of unification	6%	<4%	11%

Some respondents also shared fears and concerns regarding a united Ireland, as summarised in Table 3. Significant numbers of unionist respondents and smaller numbers in the other groups shared fears and concerns about the future of unionists and unionist identity. Some voiced fears of ‘ethnic cleansing’, a loss of British identity, or discrimination in a united Ireland. One unionist said:

[I am] extremely fearful for the future if this takes place and scared for my life due to high level of support between Republican political parties and paramilitary groups. […] I would be fearful that certain cultural groups could only commemorate behind closed doors and secretly as they would be fearful of physical and emotional attacks. I would be forced to live in a state I have no wish to be a part of and feel I would not be welcome in.

Significant numbers of unionists and small numbers of other respondents expressed fears of an economic downturn in the case of unification. One nationalist said: ‘The economic impact is the greatest fear, as it is the lack of jobs that will prove to be a strong influence on appetite for violence’. The third prominent theme involving negative expectations for a united Ireland was the fear of losing the existing standard of healthcare in a united Ireland. A respondent who identified as neither nationalist nor unionist said: ‘I feel excited. Hopes: peace, unity, and stability on the island. My only fear is the loss of the NHS’. One nationalist said: ‘My fears would be, we would loose our NHS service and also the economics’.

Table 3. Negative expectations for a united Ireland.

Issue	Nationalist	Unionist	Neither
Concerns/fears about the future of unionists and unionist identity	< 4%	19%	7%
Concerns/fears about economic downturn in case of unification	< 4%	15%	< 4%
Concerns/fears about losing the existing standard of healthcare in a united Ireland	<4%	8%	5%

Hopes and fears for the referendum process

We did not identify particular hopes as to the referendum process itself that were mentioned by many respondents. That may in itself be instructive: supporters of unification focused on the outcome, but did not often offer positive reflections on the process for getting there.

Many respondents, by contrast, shared concerns and fears about the referendum process. As indicated in Table 4, concerns that unification referendums would be divisive and could further polarise society in Northern Ireland were shared prominently by respondents from all communities. One nationalist said: ‘I fear that some parties will use the referendum campaign to stoke fear, division and sectarianism in order to influence the electorate’. Fears of violence were prominently raised in response to the question about hopes and fears (Q1), as shown in Table 4, but were also frequently mentioned throughout the survey (across all questions, by 19% of nationalists, 31% of unionists and 35% of respondents who identified as neither). For example, one respondent who identified as neither nationalist nor unionist said: ‘I feel anxious when I hear it discussed, because I know it’ll cause violence, but I also feel hopeful for a brighter future’.

Table 4. Fears and concerns about the referendum process.

Issue	Nationalist	Unionist	Neither
Concerns/fears that a referendum would be divisive	10%	23%	29%
Concerns/fears that violence would break out due to a referendum	15%	21%	27%
Concerns that a clear roadmap for unification should be produced before a referendum	6%	4%	10%

Fulfilling hopes and overcoming fears

In response to Q3, about how their hopes could be fulfilled and their fears overcome, respondents proposed a range of measures, as summarised in Table 5. The highest numbers of nationalists and respondents who identified as neither nationalist nor unionist called for open discussion to draw up plans for unification. One nationalist said: ‘indepth discussion and agreement from all concerned political, civil and financial sectors on both parts of the island’. The second most frequently shared theme was cross-community engagement throughout the referendum process. A unionist said: ‘An honest dialogue, north and south. Serious thinking, and some compromise shown by republicans in NI that we are different and will remain so in UI’. Respondents also stressed the need for a clear roadmap for unification to be drawn up before any referendum. A respondent who identified as neither unionist nor nationalist said: ‘A coherent plan of action for post referendum. The referendum must clearly state what the people are being asked to vote for’. Many respondents also said that collaboration between the British and Irish governments throughout the referendum process would be necessary. One who identified as neither nationalist nor unionist said: ‘I believe that both governments would need to work together on a draft framework before the campaign’.

Table 5. Calls for action about the referendum process.

Issue	Nationalist	Unionist	Neither
Calls for an open discussion to draw up a roadmap for unification	20%	<4%	11%
Calls for cross-community engagement throughout the referendum process	15%	4%	16%
Calls to produce a clear roadmap for unification before a referendum	13%	8%	13%
Calls for campaigning and deliberation on a prospective united Ireland based on impartial information	9%	6%	13%
Calls for collaboration across governments during the referendum process	8%	5%	12%
Views that the unionist community should be protected in a united Ireland	<4%	9%	5%
Views that all communities should be protected in a united Ireland.	5%	7%	7%

Some respondents called for campaigning and deliberation to be based on impartial information. Indeed, this request featured in responses to various consultation questions – see also the more detailed discussion in the following section. A unionist said: ‘A fully informed detailed list of everything that would change if there was an all Ireland and this to be made public by the media and I think they could never cover everyone’s questions’.

A notable further theme was the view that the unionist community should be protected in a united Ireland. This was expressed particularly, but not exclusively, by unionist respondents. A nationalist said:

I would hope that it is handled correctly so as to allay the fears of the unionist community so that they could realise that a new Ireland would also be their home and that their culture and identity would not be under threat.

Another respondent who identified as neither nationalist nor unionist said: ‘A guarantee that the British Identity in Northern Ireland and our way of life would stay the same’.

Responses 2: views on specific referendum features

In this section, we outline attitudes to key design features of a referendum itself. While certain design features were mentioned frequently, there were no particularly prominent views as to how specific technical design aspects of a referendum should be handled. Significant numbers did, however, mention previous constitutional referendums as illustrations of how they thought unification referendums should or should not be designed. Respondents sometimes cited previous referendums as examples to follow. In other cases, they highlighted problems in past referendums that they believed should be avoided. We begin by examining the role of past examples in shaping respondents’ thinking, before turning to specific design features in themselves.

How past referendums shaped respondents’ thinking

As illustrated in Table 6, the Brexit referendum of 2016 was the past referendum that featured most prominently among the responses. That was true in all three groups, but particularly for respondents identifying as neither nationalist nor unionist. The Scottish independence referendum of 2014 and the referendums on the Belfast/Good Friday Agreement in 1998 were also cited by significant numbers of respondents.

Table 6. Total mentions of previous referendums in the UK or Ireland.

Issue	Nationalists	Unionists	Neither
Brexit referendum (2016)	18%	12%	24%
Scottish independence referendum (2014)	11%	5%	<4%
Belfast/Good Friday Agreement referendums (1998)	<4%	<4%	7%

Other referendums, however, received scant attention. Most strikingly, only one respondent cited any Irish constitutional referendums besides the Agreement vote of 1998.

Although it was mentioned in varied contexts, the Brexit referendum was most often cited as an example that should not be followed. Table 7 summarises the commonest themes mentioned in relation to this vote. Significant numbers of respondents from all communities stated that any unification referendums should offer a clear plan for a united Ireland, developed ahead of the vote. Respondents contrasted that with what they saw as the unclear concept of Brexit presented in the 2016 referendum. One unionist said:

Fears: Brexit was supported because there was no detail. Irish unity sounds nice until you put some detail on it, then hardline republicans will join hardline unionists in rejecting it, so we might have the same vague ideas dominating the discussion and find that we could have chaos as bad as Brexit, but with guns in the background.

Small numbers of respondents from all communities said that campaigning in the Brexit referendum had been based on misinformation, and argued that this should be avoided. One nationalist said:

Table 7. Calls to learn lessons from the 2016 Brexit referendum.

Issue	Nationalists	Unionists	Neither
Calls for a clear plan	9%	7%	15%
Calls to prevent the use of misinformation	4%	<4%	<4%

I think the referendum debate should be had in a calm and respectful manner, which allows people to sound off on their fears without being ridiculed, but also calls out scaremongering for what it is. It should be an informed, facts based discussion. Essentially the opposite of the Brexit campaign is what I want.

The Scottish independence referendum, by contrast, was most often cited as an example to follow. Some respondents invoked it to support their argument for lowering the voting age. For example, one nationalist said: ‘I believe that 16/17 year olds should have voting rights on this occasion, like in the Scottish referendum’. Others commended the plans for a potential independent Scotland drawn up ahead of the Scottish vote and called for the same to be done ahead of any unification referendums in Ireland. One respondent who identified as neither nationalist nor unionist said: ‘Scotland have made major plans if they were ever to vote for independence and the same should be done in Ireland’.

When the Belfast/Good Friday Agreement referendum of 1998 was cited as an example to follow, respondents were often calling for voters to have clear information on the options on the ballot in advance of the vote: they noted how information about the Agreement was provided to households in 1998. One nationalist said: ‘Produce a booklet addressing the benefits of reunification in a clear concise manner and deliver to every household i.e. like the Good Friday Agreement’.

Aspects of the referendum process: overview

The aspects of any referendum process that were mentioned most often by respondents to the consultation are summarised in Table 8. These include design features of a referendum in itself – such as the franchise and the threshold that must be reached for the proposition put to voters to pass. They also include features of the wider process, including the evidence that the Secretary of State would use to decide whether a referendum should be called, and the possible use of citizens’ assemblies as part of the process of designing the options on the ballot paper.

Table 8. Proportion of all respondents mentioning particular aspects of the process.

Issue	Nationalists	Unionists	Neither
Evidence for the Secretary of State	38%	14%	32%
Sequence of referendums	28%	8%	17%
Use of citizens’ assemblies during process	28%	<4%	17%
Information and misinformation	27%	10%	23%
Referendum franchise	18%	10%	13%
Referendum threshold	13%	10%	12%
Intergovernmental cooperation during process	8%	6%	12%

Most of the aspects of the process listed in Table 8 were included in background questions within the consultation questionnaire, with one exception: the referendum threshold. We did not mention the threshold at all in the questionnaire, but respondents brought it up themselves in response to a number of questions. Also notable is the fact that many of the points raised in our background questions are absent from Table 8 and were, indeed, mentioned by very few respondents. We now examine each of the items in Table 8 in turn.

Evidence for the Secretary of State

We included a background question on how the Secretary of State should decide whether a majority of voters in Northern Ireland are ‘likely’ to support a united Ireland (BQ1). This highlighted two possible sources of evidence – opinion polls and election results – and floated the possibility of ‘some other source’ without mentioning what it could be.

This was the most frequently mentioned aspect of the referendum process among respondents from all three groups. Furthermore, majorities of those sharing their views on this matter from each group supported the use of election results as evidence. One unionist said: ‘SoS should call a Border Poll once Nationalist Parties have an overall vote share more than 50%’. The second most often mentioned source was evidence from opinion polls. This was again supported by high numbers of nationalists and respondents who identified as neither nationalist nor unionist, but only small numbers of unionists mentioned it. One respondent who identified as neither unionist nor nationalist said:

The Secretary of state should commission an independent and internationally adjudicated indicative poll to confirm the voting intentions of the majority of Northern Ireland for remaining part of UK, or transitioning to Irish Unity.

The use of census data was supported by some nationalist respondents but by only small numbers of respondents from other communities. One nationalist said: ‘I feel that census data is particularly useful in determining the political background of all citizens’. It should be noted, though, that the census, which is conducted only once every ten years, does not ask about respondents’ political background. It does include a question on religion, but the link between this and political perspective is far from perfect.

Some nationalists and respondents who identified as neither nationalist nor unionist, as well as small numbers of unionists, explicitly mentioned the use of mixed evidence. For example, one unionist said: ‘A series of opinion polls, north and south over a 5 year period. Results of election results also considered along with a series of votes in the Assembly (run alongside the opinion polls)’.

Sequence of referendums in Northern Ireland and the Republic of Ireland

The sequencing of referendums – that is, whether referendums north and south would happen on the same day or on different days – was the second most frequently mentioned aspect of the process. The majority of respondents who engaged with this question were in favour of simultaneous ballots. One nationalist said: ‘The referendum must be carried out on the same day north and south otherwise exit polls might influence voters who had yet to cast their vote’. Only small numbers of respondents from any community favoured holding referendums separately. For example, one unionist said: ‘A referendum should take place in ROI first. Only then can the electorate in NI see if a referendum here is needed at all’.

Our questionnaire included a background question on this issue (BQ2). This was part of a set of questions about referendum timing, which also covered whether a vote would take place before or after (or potentially before and after) discussions on the form of a united Ireland. That issue did also come up frequently in responses, but was more often framed in terms of the information that would be available to voters ahead of the poll. We discuss this below.

Citizens’ assemblies

Respondents brought up citizens’ assemblies in relation to planning for a united Ireland in response to multiple consultation questions. The differences between the three groups were especially large here. Over a quarter of nationalist respondents and a sixth of those who identified as neither nationalist nor unionist thought that citizens’ assemblies would be useful for discussing terms of unification prior to a referendum. By contrast, only a small number of unionists proposed the use of such assemblies. This appears to mirror patterns in the wider discourse. One nationalist said: ‘The Irish government must take the lead in establishing an all-Ireland forum to plan for Irish unity and the future of this country’. A respondent who identified as neither nationalist nor unionist said: ‘A citizens assembly should be set up to answer all the questions for 12 months before the vote’.

It would seem that, in a relatively short space of time, holding a citizens’ assembly has become associated with one particular side of the constitutional debate. The use of such assemblies was one of the themes raised in some of the coordinated responses we discussed earlier, but there is no reason to doubt that the views expressed were genuine. Given that citizens’ assemblies are designed to bring all parts of a community together in dialogue, the skew in support for an assembly on the constitutional issue could raise difficulties, which we explore in the discussion section below.

We mentioned citizens’ assemblies in our background questions, but this was just a small part of a much wider question on how plans for a united Ireland might be worked out. That other aspects – notably, the roles of political parties and of civil society – evidently had much less traction is striking.

Information and misinformation

Respondents from all groups shared fears of manipulation of referendum campaigns by politicians, media, social media, or external forces. Fears were variously expressed of ‘media bias’, ‘bot manipulation’, ‘misinformation’, and campaigners’ use of ‘social media to manipulate public opinion’. These fears led to proposals both to provide high-quality information and to counter misinformation.

Many respondents from each group said that impartial information should be provided during a referendum campaign. One nationalist said: ‘I think the more information that is available (factual) on both sides of the argument would enable people to make an informed decision’. In this context, Brexit was often brought up as an example not to follow. One respondent identifying as neither nationalist nor unionist said:

Factual information with tight legal rules about what claims can be made during campaigning – Unlike the Brexit referendum where no one was held to account for outright lies told whilst campaigning. To be treated in same way as an election, not a referendum, where the telling of lies in a campaign could mean that result was overturned.

Small numbers of respondents in all communities referenced the 1998 referendum on the Belfast/Good Friday Agreement as an example of good practice when providing impartial information.

Significant numbers of respondents – particularly nationalists – called for an independent body to oversee referendum campaigns and tackle misinformation. One identifying as neither nationalist nor unionist said:

Fact and evidence based research and scenarios. An independent cross country (UK, Ireland, N. Ireland) body to determine a charter for how this process can be managed, regulated and holding political leaders to account for participating fairly in the process.

Small numbers of respondents from all communities advocated sanctions for spreading misinformation during a campaign. One unionist said: ‘Misinformation issued by any person should be punished severely – through financial penalty and judicial sanction’.

The referendum franchise

We included a background question on who should be entitled to vote in referendums both north and south (BQ5). Respondents who mentioned this issue offered a variety of views. The theme that was highlighted most often – by nationalists and respondents who identified as neither nationalist nor unionist, as well as by small numbers of unionists – was lowering the voting age to 16. One nationalist said: ‘ … young people over 16 should also be given a vote as this whole thing would be about their future’.

Smaller numbers of respondents, predominantly unionists, mentioned criteria relating to citizenship and/or residency. One respondent who identified as neither nationalist nor unionist said: ‘I think the people allowed to vote should be Irish or UK citizens/passport holders, and not merely resident in the countries’. A unionist said: ‘Any British or Irish Citizen over 18 registered to vote and resident in NI or ROI on the day of the vote should be allowed’.

Referendum threshold

As noted above, we did not mention referendum thresholds in our background questions. That was because the 1998 Agreement is clear that a simple majority (50% + 1) in Northern Ireland is what would be required for it to consent to unification in a referendum, and the Working Group agreed early in its deliberations that this requirement should be adhered to (Working Group on Unification Referendums on the Island of Ireland, 2021, p. 196). Likewise, for any referendum on this matter in the Republic of Ireland, the constitutionally entrenched threshold of 50% + 1 would apply. Nevertheless, respondents raised the issue of the threshold throughout the survey. The majority of nationalists who did so, and small numbers of respondents who identified as neither nationalist nor unionist, favoured a simple majority threshold. One who identified as neither nationalist nor unionist said:

Re Simple majority – a weighted majority is unavoidably undemocratic and means that some votes weigh more than others. A slim majority either way would be damaging, but this should be avoided by using good data to decide when to call a referendum (i.e. election results).

Among respondents expressing a view, most unionists and those who identified as neither nationalist nor unionist, as well as small numbers of nationalists, favoured a supermajority threshold. Some said simply that the threshold should be ‘more than 50%’ in favour of unification; others proposed a threshold varying between 60 and 75%. One unionist said this was needed ‘to avoid a split country’.

Intergovernmental cooperation

The final theme mentioned by significant numbers of respondents – particularly among those who identified as neither nationalist nor unionist – was that of cooperation between the British and Irish governments. Some also mentioned the Northern Ireland Executive. For example, one respondent who identified as neither nationalist nor unionist said: ‘I believe that both governments would need to work together on a draft framework before the campaign’. A unionist said, ‘A very very detailed agreement between at least the Irish government and Westminster would be good, especially about money’.

At the same time, other respondents highlighted the role of one or other government. Notably, some nationalists – but no other respondents – said that the design of a united Ireland would be up to the government in Dublin only, and that the UK government should remain neutral in any referendum campaign.

Discussion

We identified four questions at the beginning of our analysis. The first asked to what extent people in Northern Ireland are thinking about the process by which a decision on the constitutional future might be made, rather than just the possible outcomes, and also who wants to talk about this issue, and who does not.

The second part of that question is answered by who chose to participate in our consultation. Of all respondents in Northern Ireland, 62% identified as nationalist – compared to 19% of respondents professing that identity in the 2020 NILT survey (NILT, 2020). This overwhelming skew in engagement towards nationalists is unsurprising – a referendum is the only route to their desired outcome. Perhaps more strikingly, response rates among unionists and those identifying as neither nationalist nor unionist were broadly the same as each other: there were slightly more ‘neithers’ than unionists among our respondents, as there were in the NILT sample. That may reflect a balance of two factors: unionists, being most opposed to Irish unification, might be expected to be most reluctant to discuss a referendum that could lead to such unification; conversely, many who identify as neither unionist nor nationalist focus less on the constitutional question at all.

We gain evidence on the first part of the question by delving into the content of the consultation responses. We found that most respondents from all communities were thinking much more about unification itself than about the processes through which a decision about unification might be made. The hopes and fears that people expressed when we asked very broad opening questions were mostly about unification, even when the question was about a referendum. This suggests that even people who chose to respond to our consultation, who are likely to be unusually engaged with the issues raised, were not thinking much about the process through which the constitutional issue might be decided.

That leads directly to our second question: what are people’s broad attitudes towards this process – specifically, to a possible referendum? Most people – even most respondents to this consultation – do not really have many thoughts about the process as yet, beyond seeing it as a route to an outcome that they either want or hope to avoid. In so far as they do have unprompted thoughts about the process, these are mostly fears about the divisive nature of such a vote, and about the potential for violence.

Our third and fourth questions asked to what extent people have views in relation to specific aspects of the design of a referendum and, in so far as people do have such views, what these are. We have seen that, when prompted, significant numbers of respondents did voice views on a range of aspects of referendum design, and (especially) on the question of how a Secretary of State should determine whether a referendum is required. What is striking, however, is that there were few issues on which clear views were consistently held among respondents and few issues on which there were marked divergences across the three political identities. In fact, we observed only three issues on which nascent divergences of that kind might be emerging: the referendum franchise; the referendum threshold; and whether it would be desirable to hold a citizens’ assembly in relation to unification. The first two of these are hardly surprising: different design choices here would have predictable effects on the likely outcome of any vote. The last, meanwhile, clearly relates to the fact that Sinn Féin and pro-unification civic groups have championed the cause of a citizens’ assembly designed to prepare for unification (e.g. Sinn Féin, 2022). Such deliberative bodies can have great value in bringing people together to discuss complex issues. As we noted above, however, the divided state of opinion towards assemblies would pose challenges: bringing all parts of the community together in dialogue could well prove impossible. It might be better to use deliberative bodies to discuss specific policy issues, or for holistic discussion of multiple possible constitutional futures, so as to allow everyone to take part without feeling threatened or excluded.

Even on these three issues, however, only minorities – often very small minorities – of respondents expressed views. Meanwhile, the other oft-expressed thought among respondents was that the experience of the Brexit referendum should not be repeated and that any vote on the constitutional question should be designed to allow voters to make a more informed choice. It is noteworthy that the report of the Working Group on Unification Referendums on the Island of Ireland (2021) agreed with this aspiration, but also highlighted several considerable challenges that would need to be addressed in order to meet it.

Conclusion

We began this article by noting why understanding public attitudes to the process of a referendum in Northern Ireland matters. Many aspects of that process have not yet been defined, and changing that could prove highly contentious. Should a referendum come about, policymakers will need to understand what is seen as legitimate or illegitimate in different parts of the community.

The evidence that we have gathered suggests that the opinion environment at present is relatively permissive: entrenched or polarised views on how any referendum should be conducted have not yet emerged. As we have emphasised, the responses to our consultation cannot be taken as representative of public opinion across Northern Ireland – and we hope that this work may enable further research using representative sampling methods in the future. But the skew is likely to be towards those who are more engaged and have stronger views, which strengthens rather than weakens the significance of a finding of only weakly developed views.

That suggests that a good case exists for the British and Irish governments to cooperate now on agreeing the format of any future referendum, resolving the matter when it is not a hot political issue, rather than waiting till it has become so. Of course, the counterargument to that is that picking the matter up now would cause it rapidly to become a hot issue, and would also be taken as a signal that the governments were expecting a vote in the near future.

There is a quandary here. The rules of the game are best agreed when the game is not being played. But discussing the rules sets the game running. What the results presented here do suggest, however, is that there exists space for actors who would have no direct role in any future referendum – such as academics and civic organisations whose interest is in enabling a fair and legitimate process, not in influencing the outcome – to argue for the design of a referendum to follow certain principles. The state of public opinion is not such that such an endeavour would be fruitless.

Acknowledgements

The authors thank everyone who took the time to promote or respond to our public consultation. Reading people’s views on these subjects was exceptionally rewarding and enlightening. The authors also thank the team at the Constitution Unit and the members of the Working Group on Unification Referendums on the Island of Ireland. In particular, we are deeply grateful to the research volunteers who assisted our analysis: Joe Cardwell, Robert Liao, Allison O’Malley Graham, and Oli Maddison. The authors would also like to thank Dr Jamie Pow at Queen’s University Belfast for his thoughtful comments on a draft of this paper, and for his assistance with disseminating the survey. Finally, authors are grateful to two anonymous reviewers at Irish Political Studies, who provided insightful and encouraging comments.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the British Academy under Grant number IC4/100198, and by the Joseph Rowntree Charitable Trust under Grant number 10774.

Notes on contributors

Alan Renwick

Alan Renwick is Professor of Democratic Politics and Deputy Director of the Constitution Unit at University College London. He was the Chair of the Working Group on Unification Referendums on the Island of Ireland.

Nadia Dobrianska

Nadia Dobrianska is a PhD student at University College London and a former research volunteer at the UCL Constitution Unit.

Conor J. Kelly

Conor J. Kelly is a PhD student at Birkbeck College, University of London and a Research Assistant at the UCL Constitution Unit. He was the Working Group's Research Assistant and Project Manager.

Charlotte Kincaid

Charlotte Kincaid was the Constitution Unit's Impact Research Fellow, leading the Working Group’s communications and engagement work.

Notes

1 The members of the Working Group were: Alan Renwick (Chair), Oran Doyle, John Garry, Paul Gillespie, Cathy Gormley-Heenan, Katy Hayward, Robert Hazell, David Kenny, Christopher McCrudden, Brendan O’Leary, Etain Tannam, and Alan Whysall.