CCRU home background on CCRU community relations equality and equity research

Constitutional preferences, socio-economic status and perceptions of disadvantage
among Catholics and Protestants in Northern Ireland

A Report prepared for the Central Community Relations Unit


Richard Breen
Centre for Social Research
Queen's University,
Belfast BT7 1NN

26 June 1995


Chapter 1 Introduction
Chapter 2 Constitutional preferences, party political support and national identity
Appendix to Chapter 2
Chapter 3 Variations in constitutional preferences
Chapter 4 Summary and conclusion

Constitutional preferences, socio-economic status and perceptions of
disadvantage among Catholics and Protestants in Northern Ireland

Chapter 1



The constitutional position of Northern Ireland has, since the partition of Ireland under the terms of the 1921 Anglo-Irish Treaty, been a central source of political and social division within the province. Individual preferences about the constitutional position of Northern Ireland not only differ sharply as between the Unionist and Nationalist poles. but these preferences themselves have now come to take on an unusually central role in determining what this position shall be. The UK government's position is that the constitutional position of Northern Ireland is a matter to be decided by the people of Northern Ireland. As a consequence the importance of understanding what constitutional preferences people hold and why they hold them has received a new impetus. In this report we seek to investigate this issue.

While there exist the two polar positions - Nationalist and Unionist - there may, nevertheless, be a range of more subtle gradations of both Unionist and Nationalist opinion. These gradations might concern such things as the willingness to compromise, the time frame for the preferred constitutional arrangements to be put into place. and so forth. Thus the first question that we address concerns the measurement of variations in constitutional preferences, and Chapter 2 of this report explains how we went about constructing a scale to measure individual constitutional preferences in a way that would allow us to detect a greater range of preferences beyond the simple Unionist vs. Nationalist dichotomy.

The second question we address concerns the degree of variation in constitutional preferences that exists within the Northern Ireland population and, more interestingly, the variation that exists within each of the major community groups - whom we call Catholics and Protestants (though the latter group includes all those who claim to belong to a religious denomination other than Catholic). While we expect a high correlation between community group and constitutional preferences. we nevertheless expect to see some internal variation, particularly among Catholics. where support for a united Ireland is known to be highly variable.

The third and final question we deal with concerns the explanation of the internal variation in constitutional preferences. We examine whether or not variation in, for example, constitutional preferences among Catholics and among Protestants is related to factors such as their level of education, household income, gender, social class and age.

The issue underlying these second and third questions is whether or not support for Nationalism among Catholics and Unionism among Protestants is due primarily to ideological factors (deep seated feelings about national identity which are relatively impervious to any alteration that might be brought about through changing social and economic conditions) or to economic and social position. So, in the case of Catholics, for example, the latter view would suggest that better-off members of the Catholic community would be less likely to hold Nationalist views.

However, in previous research (Breen, 1993) it was found that in so far as attitudes among Catholics towards community relations issues were concerned, these were only very weakly structured by socio-economic position. If anything they are more clearly structured by such factors among the Protestant population. This raises the possibility that, among Catholics, constitutional preferences may be less responsive to the individual's own socio-demographic position than to the kind of ideological factors to which we referred earlier.

Data sources

The analyses in this report are based on data from the Northern Ireland Social Attitudes Survey's (NISA) community relations module for the years 1989, 1991 and 1993. A total of 2614 responses were included: 866 from 1989, 906 from 1991 and 842 from 1993.


We use a range of variables in our analyses. For the most part the dependent variables concern individuals' constitutional preferences, and we discuss these in detail in Chapter 2. Here we describe the socio-demographic variables that we use as explanatory variables in our analyses.

Five such socio-demographic variables were included in the analyses in addition to religion. These were gender. household income, class, highest educational qualification and age.

Gender: 1188 respondents were men. and 1426 were female.

Household income: this is the gross annual income of the household measured in £s. Respondents were asked into which range their own household income fell. The measure used in this report assigns each respondent to the mid-point of the income range. The mean income was £13340.05, with a standard deviation of £9537.48 and a range of from £1000 to £42500. 347 responses were missing.

Class: The NISA survey contains several social class measures, for example, socio-economic group, the Registrar General's classification and the Goldthorpe classes. In this report the Registrar General's classification in used. The frequency distribution is shown in Table 1.1.

Highest educational qualification: the frequencies for the highest educational qualification of respondents are shown in Table 1.2

Table 1. 1: Frequency distribution of social class (Registrar General's categories)

R-G Classes:
I Professional etc.
II Intermediate
IIIa Skilled non-manual
IIIb Skilled manual
IV Partly skilled
V Unskilled

279 respondents reported no social class.

Table 1.2 Frequency distribution of highest educational qualifications

Higher education below degree
'A'level or equivalent
'O'level or equivalent
CSE or equivalent

15 respondents did not respond to this item

Age: the mean age of the respondents was 44.69 years, with a standard deviation of 17.45 years. Five responses were missing.

Outline of the Report

In Chapter 2 we discuss the construction of our constitutional preference scale. The NISA surveys contain a number of items relevant to this issue. Some ask respondents' constitutional preferences directly, while others approach the question in a more oblique manner. An example of the former is the question concerning preferred long term policy for Northern Ireland (with the possible answers including 'Remain part of the United Kingdom' and 'Reunify with the rest of Ireland'). An example of the latter is the question that asks respondents whether they see themselves as British, Irish. Northern Irish, Ulster, and so on.

The difficulty with using the questions in the NISA survey that ask directly about constitutional preferences is that they tend to allow only a very limited range of responses (as the above example illustrates). This does not permit us to capture the full variation that might exist in constitutional preferences. Ideally, of course, one would like questions that ask not only what constitutional arrangements the respondent most prefers but also those that he or she might be willing to accept as a second or even third choice.

In the absence of such items we seek to use information from questions which, while less direct, nevertheless have a bearing on constitutional preferences. We supplement the direct constitutional questions with these data in order to form a constitutional preference scale. This scale then allows us to distinguish finer gradations among people's constitutional preferences. We can, for example. distinguish strong Nationalists from more 'moderate' Nationalists, and similarly among Unionists.

Although the methodology of scale construction is very well known and widely used in the social sciences, the exact form of the relevant items in the NISA data set presents some nonstandard complications. Accordingly we devote some considerable time to explaining how we went about forming this scale. The technical details can be found in the appendix to Chapter 2.

In Chapter 3 we analyze constitutional preferences. So. we examine how the scores on the scale differ as between Catholics and Protestants and according to the socio-demographic factors described earlier. We then focus on the distributions of scores among Catholics and among Protestants and we investigate the degree to which such internal variation can be explained by our socio-demographic variables. We then investigate the extent to which constitutional preferences among Catholics are associated with beliefs about the position of the Catholic community vis-a-vis the Protestant community in Northern Ireland. Specifically, we investigate whether beliefs about how the security forces treat Catholics relative to Protestants; beliefs about the extent of discrimination in Northern Ireland; and attitudes towards Fair Employment legislation can be said to be related to constitutional preferences. In other words. do these preferences and attitudes form a coherent ideological position that is itself unrelated to the individual's material position (as captured in the socio-demographic measures)?

Chapter 4 summarises our results and contains our conclusions.

Chapter 4



This report has focused on constitutional preferences in Northern Ireland: that is. on people's desires about the future constitutional position of Northern Ireland. We have sought to describe the range and distribution of such preferences and to investigate the extent to which they are related to socio-demographic position and to perceptions of the relative positions of the Catholic and Protestant communities in Northern Ireland. The data we used came from the NISA Surveys 1989, 1991 and 1993.

Constitutional Preferences

In response to the simple question about long term policy for Northern Ireland, 75 per cent of those who answered wanted Northern Ireland to remain part of the UK.

In order to investigate finer gradations of preference on this issue, we formed a constitutional preference scale. as described in Chapter 2. This ranged from a low of zero (representing a respondent who favoured the union, described him or herself as "British" or "Ulster", supported the DUP, and chose the constitutional identity "Unionist") to a high of three (a respondent who favoured a united Ireland, described him or herself as "Irish", supported Sinn Fein and chose the constitutional identity "Nationalist"). Thus low scores represent extreme Unionism, high scores extreme Nationalism.

A large proportion of respondents (909 out of 2614) did not have a valid score on this scale because they had not responded to one or more of the four items making up the scale. Of those who did respond, 65 per cent scored less than one while 20 per cent scored more than two.

Community background (labelled "Catholic" and "Protestant") is the single best predictor of an individual's constitutional preferences. 96 per cent of Protestants favour the retention of the Union as against 39 per cent of Catholics. On the constitutional preference scale the mean Protestant score is 0.35 while that for Catholics is 2.15. Religious group explains over 70 per cent of the variation in constitutional preferences. However, these preferences display much greater variation among Catholics than among Protestants. We can see the reason for this if we turn to the histograms, Figures 3.1 and 3.2. Whereas virtually all Protestant scores fall in the range zero to one, the Catholic distribution is bimodal. The greater number of Catholics have scores exceeding two but a substantial minority have scores below two.

Nationalists, Unionists and "Unionist" Catholics

This led us to distinguish three groups: Protestants, who tend overwhelmingly to be Unionist; Catholics who score high on the constitutional preference scale (and are therefore Nationalist) and low scoring, or "Unionist" Catholics. The average characteristics of the three groups are shown diagrammatically in Figure 4.1.

Figure 4.1 Average characteristics of members of three constitutional groups

Protestant Unionists
Catholic Unionists
Catholic Nationalists
Favour union with Great BritainFavour union with Great Britain Favour Irish Unity
Consider themselves "British" or "Ulster" Adopt a variety of personal identity labels Consider themselves "Irish"
Consider themselves "Unionist" Consider themselves neither "Unionist" nor "Nationalist" Consider themselves "Nationalist"
Support Unionist political partiesSupport SDLP and, to lesser extent, Alliance Support SDLP and, to lesser extent, Sinn Féin
High level of trust in Stormont government. Very high distrust of Irish government More willing to trust Irish government than are Protestant Unionist Trust in Irish government. High distrust of British and Stormont governments.

While Protestant Unionists and Catholic Nationalists each represent an easily recognised constellation of constitutional and political preferences and constitutional and personal identities, those Catholics who support the Union are less easily pigeon holed and less easily matched to sterotypical depictions. This arises, we suggested, because, although they support the union with Great Britain, they are reluctant to adopt the labels of Unionism because of its identification with Unionist political hegemony within Northern Ireland.

Explaining variation in constitutional preferences

While religious group affiliation explains a great deal of the variation in constitutional preferences in the Northern Ireland community as a whole, what factors explain variation in such preferences among Catholics and among Protestants?

We first examined a number of socio-demographic factors - gender, social class. age group, education and household income. Our results present what at first may seem to be a paradox. Although there is less variation in constitutional preferences among Protestants (as Figure 3.2 displays), it is nevertheless the case that among them the relationships between such preferences and socio-demographic factors are stronger than among Catholics. Constitutional preferences vary significantly according with social class, level of education and household income among Protestants. The wealthier, better educated and higher social class Protestants tend to be less strongly Unionist. By contrast among Catholics. there is a significant gender effect, with men tending to hold more strongly Nationalist views than women. However, although there is some evidence of a class gradient in preferences (the lower the social class the stronger the Nationalism), this is not statistically significant, nor are any of the other socio-demographic factors related to constitutional preferences.

It is important to be clear about what this result is telling us about Protestant constitutional preferences. As Figure 3.2 shows, there are very few Protestants who deviate from the characteristic Unionist position. However those that do, and those who deviate the furthest, are middle class, well educated and tend to earn more. Notwithstanding, the majority of Protestants of all classes, educational levels and incomes adhere to the characteristic Unionist position.

In the final part of our analysis we examined the relationship between constitutional preferences and beliefs about the position of Catholics relative to that of Protestants in Northern Ireland. Within both groups we found much stronger relationships than was the case with the socio-demographic factors. Particularly among Catholics there were strong correlations between constitutional preferences and beliefs about whether the security forces favour Protestants: beliefs about discrimination against Catholics in the labour market; and beliefs about whether or not there exists prejudice against Catholics in Northern Ireland. The magnitude of these correlations does suggest the existence of reasonably consistent ideologies or belief-sets. Catholics who are more Nationalist are also more likely to believe that Catholics are treated unfairly by the security forces, that they have a poorer chance of getting a job and that there is prejudice against them. Among Protestants the correlations are somewhat weaker, implying that perceptions of Catholic disadvantage are somewhat more distinct from constitutional preferences than is the case among Catholics.

It is clear that simply being middle class, as opposed to working class, or highly educated, is not a sufficient condition for the reduction of Nationalist sentiment among Catholics. On the other hand, our data do not allow to say that an improvement in the material position of the whole Catholic community (and the perception of this by Catholics) will lead to any diminution in Nationalism. Although such perceptions are clearly related to constitutional preferences within the Catholic community cross-sectional data such as we have cannot help us to separate cause from effect. It may be that constitutional preferences do indeed respond to perceptions of the level of Catholic disadvantage, but it may also be the case that such perceptions are themselves shaped by the constitutional preferences that people hold.


The constitutional position of Northern Ireland has been a central issue of contention for at least 70 years. It has now come to rest in the hands of the people of Northern Ireland. To some degree our results confirm the centrality of constitutional issues in Northern Ireland. For example, political party support is very extensively determined by constitutional preferences, as we saw in Chapter 2. In these concluding remarks, however, we want to investigate what our findings might imply for the future constitutional position of Northern Ireland.

At present it appears that around 75 per cent of adults in Northern Ireland favour a long term policy of retaining the union with Britain. Among Protestants less than five per cent favour a united Ireland in the long term, while among Catholics. around 40 per cent say they want to see the union preserved.

Forecasting is a notoriously unreliable business, but this has not deterred much speculation surrounding the issue of a future Catholic majority in Northern Ireland. Extrapolating from the results of the 1991 census it has been suggested that Catholics may form the majority in Northern Ireland some time in the next century (for a sceptical view of such forecasting see Breen 1994). However, even putting aside the uncertainties attached to purely demographic forecasts, the constitutional and political implications of a Catholic majority are quite unknown. Given the current pattern of Catholic constitutional preferences a Catholic majority is far from being the same thing as a majority in favour of Irish unity. Indeed, assuming 60 per cent of Catholics supported Irish unity (and that no Protestants did) a majority in favour of unification would require that the adult Catholic population be more than five times larger than the Protestant. But, of course. it is unlikely that the current pattern of constitutional preferences among Catholics (and possibly among Protestants) will be maintained. The evidence we have suggests some shift among Catholics (and no change among Protestants) between 1989 and 1993 towards greater support for the union (as Table 3.4 shows), though this trend is not statistically significant. More generally, changing, but unpredictable, economic or political circumstances may lead to further change in constitutional preferences, but the direction of such changes cannot be forecast.

In contrast to Catholics, the Protestant community is strikingly homogenous in its constitutional preferences. This suggests that internal diversity among Protestants should not be over-emphasised. Although there is some variation among Protestants in their personal identity as between those who see themselves as "British" and those who consider themselves "Ulster" (a distinction discussed in detail by Todd, 1987), this seems to be of little significance for constitutional preferences. As Table 2.4 shows, among those who consider themselves "Ulster" there exists exactly the same distribution across the constitutional preference categories as among those who consider themselves "British". Similarly, the often heard view that a substantial body of middle class Protestants are more likely to accept a united Ireland. or some compromise that weakens the union, receives little support from our analysis. It is certainly the case that those Protestants who score higher on the constitutional preference scale (and are therefore less staunchly Unionist) are predominantly middle class, well educated and have higher incomes, but they are very few in number. Around 90 per cent of Protestants score one or less on the constitutional preference scale.

Such low scores reflect the consistency of Protestants in their responses to the four items that comprise the constitutional preference scale. This contrasts particularly with those Catholics who support the union but whose personal identity, constitutional identity and party support are sometimes in conflict with their Unionist preferences and are often somewhat ambiguous. Even among Nationalist (high scoring) Catholics there are still greater inconsistencies than are found among Protestants. Perhaps the most striking difference is to be observed in political party support. Whereas the two Unionist parties are supported exclusively by supporters of the union, of the two Nationalist parties by far the larger (the SDLP) receives a substantial share of its support from Catholics whose preferred long term constitutional preference is for the retaining of the union with Great Britain.

In a recent paper. Ruane and Todd (1992) have warned against over-emphasis on the potential for middle-ground politics in Northern Ireland. Our results reinforce this, at least in respect of the Protestant community where relatively little variation in constitutional position can be seen. Furthermore, the very high levels of distrust of the Irish government shown by Protestants (which is, as Table 3.8 reveals, if anything slightly higher than the level of distrust of the British government shown by Nationalist Catholics) would seem to suggest little willingness to accept arrangements such as cross-border bodies. Among Catholics there is greater variation and here, of course, we would point particularly to that group we have called low-scoring, or Unionist, Catholics. On the one hand this might be seen as indicative of greater potential for some kind of middle-ground politics, but (and as the corrollary to this) it can also be seen as an indicator of a possible volatility in constitutional preferences that makes predictions particularly unreliable.


1. The inclusion of this perhaps somewhat puzzling response category is presumably due to the specific question having also appeared in the British Social Attitudes survey.

2. The other advantage of using a scale is that, provided that the items in the scale are measuring the same underlying issue, the scale will have greater reliability than the single item.


Breen, Richard, 1993, Attitudes towards community relations among Protestants and Catholics in Northern Ireland, Report to PPRU.

Breen, Richard, 1994 "Over the Horizon", Fortnight, 325: 20-21.

Ruane, Joseph and Jennifer Todd, 1992, "Diversity, Division and the Middle Ground in Northern Ireland", Irish Political Studies, 77: 73-98.

Todd, Jennifer, 1987, "Two traditions in unionist political culture", Irish Political Studies, 2: 1-26.

© CCRU 1998-1999
site developed by: Martin Melaugh
page last modified:
Back to the top of this page