Examining National Sentiments in Britain. Roger Scully Wales Governance Centre Cardiff University

Transcription

1 Examining National Sentiments in Britain Roger Scully Wales Governance Centre Cardiff University Paper presented to the Annual Conference of the Elections, Public Opinion and Parties specialist group, Oxford, September 2012 Abstract: Attitudes towards the nation have become increasingly central to UK politics in recent times. While some aspects of this, such as national identities and constitutional preferences, have been studied in detail, little is known about the sentiments that may underpin such identities and preferences. This paper aims to address these deficiencies in current knowledge. Drawing on a hitherto neglected series of items in parallel surveys conducted in 2003 in England, Scotland and Wales, it demonstrates that these items can be combined to form reliable, multi-item scales of sentiments towards Britain and towards England, Scotland and Wales. The paper then examines the relationship between sentiments towards Britain and those towards the sub-state nations of Britain; explores the factors that predict national sentiments; considers the relationship of national sentiments to party support; and assesses the further implications of the findings. Key Words: National sentiment; Britain; England; Scotland; Wales.

2 Nationalism has been arguably the most important phenomenon in global politics over recent centuries (Elias 1990: 146). Yet outwith the singular exception of Northern Ireland, it has typically been marginal to the study of UK politics. Questions about competing senses of nationhood in the UK, and their implications, barely arose until quite recently; a banal unionism, shared by the vast majority of people, could safely be assumed (Kidd 2008). The referendum on Scottish independence due in 2014 is merely the most obvious illustration that this assumption is now invalid. In response, significant attention has come to be given to the politics of the nation in the UK. There have been major recent studies of the history, structure and ideology of the nationalist parties in Scotland and Wales (Lynch 2002; McAllister 2001; Mitchell et al 2011; Wyn Jones 2007, 2013). Recent years have also seen important analyses of the changing character of, and growing challenges to, British unionism (McLean and McMillan 2005; Kidd 2008). The establishment and functioning of national devolution in Scotland, Wales and Northern Ireland has been explored in detail (Rawlings 2004; Trench 2004). And many aspects of public attitudes have been given attention. Studies have used survey data to examine the varying senses of national identity across the different parts of the UK (e.g. Bond and Rosie 2010). Opinions on matters such as the forms of government that should operate over the nations of Britain have also been probed, and analysts have examined the extent to which these opinions are shaped by factors such as national identity (Ormston and Curtice 2010; Pattie et al 2004). But even within the field of public attitudes research, important aspects of the politics of nationhood in the UK have received scant attention. Notably, while important work has told us much about the national identities that people in the different nations of the UK affirm,

3 and their preferred forms of government, almost no systematic work has investigated the deeper sentiments about the nation that underpin people s sense of nationhood. Little work has explored British national sentiment (cf. Heath et al 1999); no major published studies have investigated sentiments towards the constituent nations of Britain, nor examined the relationship between such sentiments and those concerning Britain itself. While this remains true, our understanding of the politics of nationhood in contemporary Britain must necessarily remain incomplete. This paper addresses these deficiencies in current knowledge. Drawing from a (hitherto neglected) series of items run in parallel surveys conducted in 2003 in England, Scotland and Wales, it demonstrates that these items can be combined to form reliable, multi-item scales of sentiments towards Britain and towards England, Scotland and Wales. After this, the paper goes on to examine the relationship between sentiments towards Britain and sentiments towards the sub-state nations. We then explore the factors that predict national sentiments, and conduct an initial exploration of the relationship between national sentiments and party support. Finally, the conclusion assesses the further implications of our findings. The Study of National Sentiments in Britain Studying national sentiments effectively requires a clear idea of what the term means: something that is less straightforward than might first be imagined, for the study of nationalism is a field characterised by considerable theoretical and conceptual contestation. This paper follows the understanding of Breuilly (1996), who distinguishes between

4 nationalist doctrines, nationalist political movements, and national sentiments. The latter concerns attitudes towards and about a nation, and will be the focus for this paper. 1 In the study of British politics, national sentiments remain largely unexplored territory. There have been many popular discussions of attitudes to, and understandings of, Britishness, and similar attention to the constituent nations of Britain. 2 Important work has also examined, in individual cases and among elite groups, how attitudes to the nation underpin nationalist movements. 3 But very little work has systematically explored public attitudes to Britain or to the constituent nations of Britain. And no published work at all has examined what must surely be a crucial issue the relationship between these two forms of national sentiments. To what extent, and in what ways, do attitudes about Britain relate to those about England, Scotland and Wales? The landmark indeed, virtually the only published study of British national sentiment (BNS) is that of Heath et al (1999). In line with the definition offered above, this study distinguished between national identity, which can be thought of as a categorical selfconcept, and nationalist sentiment, which can be thought of as a dimension (1999: 156). The study sought to develop a multi-item scale to measure this attitudinal dimension; to chart whether the supporters of the different parties differ in their acceptance of official British 1 Breuilly observes, on the distinction between doctrines, movements and sentiments, that we know that nationalist doctrines and nationalist politics frequently arise in societies and regions where much of the population lacks any strong or distinct sense of national identity. We also can point to cases where there are widely shared national sentiments but where these have not been associated either with the elaboration of nationalist doctrines or the emergence of significant nationalist movements (1996: ). 2 Prominent examples include, on Britishness, Lowe and McArthur (2010) and (somewhat more scholarly in nature) Gamble and Wright (2009); Morris (2000) on Wales and Welshness; and Paxman (2007) on Englishness. 3 Wyn Jones (2007, 2013) provides a detailed qualitative study of the ideology, including understandings of Wales and Welshness, of Plaid Cymru leaders during the twentieth century. Mitchell et al s (2011) study of the Scottish National Party combines the qualitative study of party elites with quantitative measurement and analysis of the attitudes of the party membership.

5 nationalism (1999: 157); and then to explore the relationship between BNS and more wellestablished measures of economic left-right attitudes and social libertarian-authoritarianism. Heath et al s BNS scale was constructed around a series of items to which respondents were given five possible responses, ranging from strongly agree to strongly disagree. These items, used in several waves of the British Election Panel Survey (BEPS), were the following: - Britain has a lot to learn from other countries in running its affairs - I would rather be a citizen of Britain than of any other country in the world - There are some things about Britain today that make me ashamed to be British - People in Britain are too ready to criticise their country. 4 The BNS scale aggregated responses to all items (with appropriate re-coding so more pro- Britain answers scored high in each instance). BEPS respondents scores on the BNS scale were shown to be related to responses on the Moreno national identity scale in Scotland (with those scoring higher on British sentiment more likely to affirm a more British national identity), and to be associated with support for different political parties (with Conservatives and BNP supporters scoring higher on British sentiments than did supporters of other parties). Heath et al also made some effort to explore the social bases of such sentiments (with British sentiment found to be stronger amongst older and less well-educated respondents, as well as adherents to the established churches in England and Scotland). However, little relationship 4 The core BNS scale items are supplemented in some analysis by two further items included in the 1997 BEPS wave: - The government should do everything it can to keep all parts of Britain together in a single state - Britain should cooperate with other countries even if it means giving up some independence. The second item in the main BNS scale is identical to one included in two International Social Survey Programme multinational studies, conducted in 1995 and 2003, which had their own scales of what some authors term national sentiment (Hjerm and Schnabel 2010); however, the label of chauvinism used for this scale by other scholars (e.g. Coenders and Scheepers 2003) appears more appropriate given its rather narrow focus.

6 was found between the BNS scale and either economic left-right or social libertarianauthoritarian attitudes. Although Heath et al s study has been widely cited, there have been disappointingly few subsequent attempts to apply their BNS scale, to extend or refine it, or to develop alternative measurements of national sentiments in Britain. And while the original study remains a reference point in the literature, it is not beyond criticism. At least three limitations in the study can be identified. First, the BNS scale was simply not empirically very successful. Heath and colleagues themselves acknowledged the Alpha reliability coefficient (.35) for the four-item scale in their original study to be meagre. 5 Moreover, when the items used in the scale were factor analysed (along with items from the standard economic left-right and social libertarianauthoritarian scales), two of them each loaded heavily on two separate factors. If British national sentiment is indeed a single dimension, it is not clear that the Heath et al scale taps it very effectively. Nor is the scale very obviously successful as a tool for understanding other attitudes. Given these empirical weaknesses, it is perhaps unsurprising that the scale has been deployed in few subsequent studies. Second, the construct validity of the scale can be questioned. 6 Does the BNS scale effectively measure British national sentiment? This is ultimately an interpretative judgement about the scale items and the underlying concept that they are intended to measure. 7 The four items in 5 The Alpha coefficient for Heath et al s British sentiment scale improves (from.35 to.51) when the two additional items available in the 1997 BEPS wave are included. 6 A specific form of construct validity content validity is relevant here. Content validity is concerned with whether a measure taps all relevant facets of the underlying construct (Carmines and Zeller 1991). 7 Methods have been developed to assess objectively the construct validity of measures most obviously, that one should expect high correlations between the chosen measure and variables that are known to be closely

7 the core BNS scale all relate to pride in contemporary Britain and its standing in the world, with the second of the four also tapping to some extent into xenophobia. 8 This offers a rather narrow and partial measurement of sentiments about Britain as a nation. To give but one example, the BNS scale items include nothing about attitudes towards popular symbols of Britishness. It seems likely that a more broad-ranging scale, tapping sentiments about Britain from a variety of directions, might be constructed: an initial attempt at this is made in the following section. Finally, we must emphasise that examining British national sentiment is not the same thing as examining national sentiments in Britain. The UK is a state with differing senses of nationhood and national identity: the serious examination of national sentiments in Britain must seek to understand sentiments about these different senses of nationality. Such a line of critique is arguably quite unfair to Heath et al, as it involves criticising their study for not doing something it never sought to do. But the basic point surely remains intact: the serious study of national sentiments in Britain requires extending the scope of analysis beyond the parameters that Heath et al observe. What is needed, in short, is the construction of reliable and valid measures of sentiments both about the different nations of Britain and about Britain itself. This is what the following section of the paper attempts to provide. Measuring National Sentiments in Britain related to the underlying construct. But such a strategy is problematic, if not impossible, where no such variables exist, as is the case for measuring national sentiments in Britain. 8 It should be acknowledged that the two supplementary items from the 1997 BEPS wave extend the scope of the BNS scale somewhat.

8 Developing reliable and valid measures of national sentiments in Britain is made particularly difficult by three major limitations in available data. Few major social surveys include sufficient numbers of respondents from all three main British nations to permit analysis of attitudes in each. Few of those that do include sufficient respondents from all three nations have asked questions relating to national sentiments. And the small number of studies clearing both these first two hurdles have generally not asked about both Britain and the constituent nations of Britain. However, a parallel series of surveys conducted in 2003 in England, Scotland and Wales did include several sets of questions, answers to which can be combined to form measures of national sentiments in Britain. 9 This is not to claim that the scales thus derived constitute the definitive measurement of such sentiments. But they arguably maximise what can be learned from the limited existing data. The questions deployed here address national sentiments from several, rather different, directions. The first we examine is concerned with attachment to the nation. Respondents were asked to rate, on a four-point scale running from Very to Not at all, how closely they felt attached to your local area, to Britain, and to Europe. Respondents from England, Scotland and Wales were also asked about their respective level of attachment to England, Scotland and Wales; English respondents were in addition asked about your region. Results for England, Scotland and Wales in 2003 are presented in Table 1. Unsurprisingly, Europe attracts the lowest sense of attachment in all three nations. In other respects, however, the nations differ. For England, we observe fairly similar levels of attachment to the locality, region, England and Britain: there is no clear primary referent of attachment. In Scotland, attachment to Scotland is greatest, with the locality coming second and Britain a rather 9 The three surveys we draw upon are the 2003 British Social Attitudes Survey (from which we draw our sample for England), the 2003 Scottish Social Attitudes Survey, and the 2003 Wales Life and Times study. Sampling for all three surveys was conducted face-to-face by the National Centre for Social Research.

9 distant third. Welsh respondents also accord lower levels of attachment, on average, to Britain than their locality or Wales. But distinctive to Wales is the relatively strong attachment to localities, which surpasses even levels of attachment to Wales itself. Popular contemporary rhetoric that describes Wales as a community of communities seems to have some basis in reality. 10 TABLE 1 ABOUT HERE A second set of questions asked directly about national pride: how proud respondents felt to be British and to be English, Scottish or Welsh. Results from these questions are shown in Table 2. As can be seen, in England, levels of pride in Englishness and Britishness are very similar. In both Scotland and Wales, by contrast, reported pride in the sub-state nationality is notably greater than in Britishness, with the gap particularly large in Scotland. The one respect in which Wales stands out is in having the highest number of respondents who reject the label of the sub-state nationality entirely. 11 TABLE 2 ABOUT HERE A third set of questions explored attitudes towards those ubiquitous symbols of nationhood, national flags. The results, reported in Table 3, demonstrate little hostility to any flags. Even those people mainly in Scotland and Wales who reject the label British are more likely 10 The characterisation of Wales as a community of communities dates back at least to ideas propounded by the writer and political activist Saunders Lewis (see Jones and Thomas 1991 for an introduction to Lewis thinking). Interestingly, while Lewis remains a deeply controversial figure, this element of his thinking has become a widely-shared feature of contemporary political discourse in Wales. 11 Many of those who rejection any sense of Welshness come from the significant proportion of the Welsh population (approximately twenty per cent, according to the 2001 census, compared to some eight per cent in Scotland) who were born in England. Of respondents in our 2003 Welsh survey who reported having been born in Wales, only 1.5 per cent indicated in response to this national pride question that they not Welsh. Of those who reported having been born outside Wales (the vast majority of whom were born in England) some 55.0 per cent chose the not Welsh option.

10 to feel indifference to the Union Jack than hostility. We can also see from the table that pride in the Welsh flag was very high even higher than levels of Scottish pride in the Saltire. TABLE 3 ABOUT HERE A final question used in the 2003 parallel surveys to some extent required survey respondents to prioritise one sense of nationhood, by asking them what nationality they would choose to have on their passport were they able so to choose. As is shown in Table 4, most English respondents opted for British. But a substantial majority in Scotland opted for Scottish, and a smaller majority in Wales chose Welsh, although the proportion choosing British was notably higher in Wales than Scotland. TABLE 4 ABOUT HERE Though the specific results for each of these questions are of some interest, of more concern for current purposes is to explore the extent to which answers to these questions may be interrelated. Do these four very different questions probe essentially separate attitudes, or do they tap into a more general phenomena that we might plausibly label national sentiments? To explore this issue we draw together elements of the above questions into two distinct sets of variables concerned, respectively, with sentiments about Britain and about England, Scotland or Wales. Our putative British national sentiment variables are: - Answers to how closely attached do you feel to Britain? - Reported level of pride in being British

11 - Reported pride in the Union Jack flag; and - Whether or not respondents indicated that they would choose to be described as British on their passport. 12 The English/Scottish/Welsh national sentiment items are: - Answers to how closely attached do you feel to England/Scotland/Wales? - Reported level of pride in being English/Scottish/Welsh - Reported pride in the Cross of St. George/Saltire/Red Dragon flags; and - Whether or not respondents indicated that they would choose to be described as English/ Scottish/ Welsh on their passport. 13 Correlations (not shown) of the individual variables for each of the three nations indicate the patterns of association between them to be fairly similar in Scotland and Wales, with England differing a little. In both Scotland and Wales, all the British items are strongly intercorrelated; so too are all the Scottish/Welsh items. The inter-relationship between the British and Scottish/Welsh items is less consistent, although in several instances there are significant negative correlations. In England, while all the British items are positively related, and so too are most of the English items, there are also many positive correlations between British and English items (although the English passport item correlated negatively with both attachment to Britain and pride in Britain). 12 For the British sentiment scale, items were coded as follows. Closeness of Attachment to Britain was coded 2 for Very, 1 for Fairly, -1 for Not Very and -2 for Not at all. Pride in being British was coded 2 for Very, 1 for Somewhat, -1 for Not Very, and -2 for Not at all. Pride in the Union Jack Flag was coded 2 for Very Proud, 1 for A Bit Proud, 0 for Indifferent, -1 for A Bit Hostile, and -2 for Very Hostile. The British Passport item was coded 1 for those indicating they would choose to be described as British on their passport, 0 otherwise. 13 For the English/Scottish/Welsh sentiment scale, items were coded as follows. Closeness of Attachment to England/Scotland/Wales was coded 2 for Very, 1 for Fairly, -1 for Not Very and -2 for Not at all. Pride in being English/Scottish/Welsh was coded 2 for Very, 1 for Somewhat, -1 for Not Very, and -2 for Not at all. Pride in the St. George/Saltire/Red Dragon Flags was coded 2 for Very Proud, 1 for A Bit Proud, 0 for Indifferent, -1 for A Bit Hostile, and -2 for Very Hostile. The Passport item was coded 1 for those indicating they would choose to be described as English/Scottish/Welsh on their passport, 0 otherwise.

12 These patterns are more formally confirmed by a factor analysis for each set of variables. 14 For both Scotland and Wales we observe neat, two-factor solutions, with all the British items loading heavily on one factor and all the Scottish/Welsh items loading heavily on the other. 15 In England, the results are rather more complex. A three-factor solution emerges from the data: the two national pride questions and those relating to flags load heavily on the first factor; the second factor is loaded on by both passport items; while the third factor is strongly associated with the attachment items. This indicates a more complex pattern of national sentiments in England than in the two other British nations, but reinforces the point that sentiments towards England and Britain are positively correlated. When we aggregate responses from the individual items into scales of national sentiments, we again find very similar results in Scotland and Wales. 16 In Wales, the Welsh sentiments scale produces a reliability coefficient of.73, and the British scale one of.68; in Scotland, the Scottish sentiment scale has an Alpha of.71 and the British scale one of.69. These are highly satisfactory results, and suggest that in both countries the two scales are reliable measures of coherent phenomena. 17 In England the findings are slightly less strong: both the English and 14 The factor analysis method used was Principal Components, with varimax rotation. 15 In Scotland, the first factor, with an eigenvalue of 2.65, was heavily loaded on by all four Scottish sentiment items; the second factor, with an eigenvalue of 2.23, was heavily loaded on by all four British sentiment items. These two factors combined explained a total of percent of item variance; no other factors approached an eigenvalue of 1. In Wales, the first factor, with an eigenvalue of 2.97, was heavily loaded on by all four Welsh sentiment items; the second factor, with an eigenvalue of 2.00, was heavily loaded on by all four British sentiment items. These two factors combined explained a total of percent of item variance; no other factors approached an eigenvalue of 1. In England, the first factor, with an eigenvalue of 2.95, was heavily loaded on by both national prid variables and both national flag variables; the second factor, with an eigenvalue of 1.91, was heavily loaded on by the two passport items; the third factor, with an eigenvalue of 1.21, was heavily loaded on by the two national attachment variables. The three factors combined explained a total of percent of item variance; no other factors approached an eigenvalue of Both scales are constructed by adding the values of the individual variables, and thus run from a maximum of 7 to a minimum of -6. A Cronbach s Alpha coefficient of.70 or above is generally regarded as indicating a reliable multi-item scale (see Pennings et al (1999: 97)). 17 To the author s knowledge, no subsequent survey in England or Scotland has repeated all the items included in the 2003 parallel studies. However, the pre-referendum wave of the 2011 Welsh Referendum Study did repeat all the items. This obtained very similar findings to the 2003 survey in Wales: factor analysis of the items

13 British sentiment scales produce Alpha coefficients of.60. Nonetheless, this is still far superior to the findings obtained by Heath et al (1999) for their BNS scale. A final question to address in this section is the levels of national sentiments that our scale indicates exist in the three British nations. Average scores on the two national sentiment scales for England, Scotland and Wales are displayed in Table 5. The findings are intuitive, reinforcing confidence in the validity of the measures. On the British sentiment scale, we find England to have the most positive average levels of BNS and Scotland the least positive, with Wales in the median position. The picture is the reverse for sub-state national sentiment: levels of such sentiment are greatest in Scotland and lowest in England, with Wales again falling between the two. TABLE 5 ABOUT HERE To summarise, while there is a paucity of good data on national sentiment in Britain, in this section we have been able to show that a number of survey items included in parallel surveys in 2003 can be exploited to develop scales that tap into national sentiments in England, Scotland and Wales. These scales appear to be highly reliable in Wales and Scotland, although slightly less so in England. The following section of the paper will now explore what relationship, if any exists between national sentiments towards Britain and those concerning England, Scotland and Wales. produced a very similar two factor solution, while reliability analysis produced an Alpha coefficient of.82 for the Welsh sentiment scale, and.71 for the British sentiment scale. This evidence provides valuable reassurance that the 2003 findings are robust across time (and also sampling method, given that the 2003 surveys were implemented via face-to-face sampling, and that in 2011 via the internet).

14 The Inter-relationship of National Sentiments It is not theoretically or intuitively obvious how sentiments about Britain should be related to those about the sub-state nations of Britain. When national identities are examined, Britishness is often as in the classic Moreno national identity question placed somewhat in opposition to an identity with one of the individual nations of Britain. 18 In a similar manner, it may be that those who have the strongest national sentiments about England, Scotland and Wales are also those most eager to distinguish these nations from Britain. But it is quite conceivable that some people may have strong positive or negative sentiments about both Britain and one of its sub-state nations; or, indeed, that many people may have no strong sentiments about either. Elements of history and contemporary politics, however, suggest that the most plausible expectation may be one of differing patterns between the three British nations: In England, where British identity and the British state have often been intermingled and confused with England, we should be most likely to see a positive correlation between national sentiments about Britain and those about the sub-state nation. In Scotland, where a strong Scottish nationalist movement has existed for some years, where many people affirm an exclusively Scottish (not British) national identity, and where as we have already seen levels of BNS are lowest, we should be most likely to see a negative correlation between sentiments towards Britain and those concerning the sub-state nation. 18 The Moreno national identity question was first developed by Juan Linz and Luis Moreno in the context of research on Scotland and on Spain s autonomous communities, primarily the Basques and Catalans. (For discussion of the background to the development of the question, see Moreno 2006). The question asks respondents to choose a single national identity from a range of options along a dimension whose extreme points are (in the case of Scotland) Scottish not British and British not Scottish.

15 Wales, we hypothesise, will be somewhere between England and Scotland in the relationship between it and Welsh sentiment, just as it was in levels of both British and sub-state national sentiments. This reflects the rather weaker status of Welsh nationalism and of exclusively Welsh national identity when compared to their equivalents in Scotland. To explore the inter-relationship between the two forms of national sentiments, we compare the mean levels of BNS among those with negative, neutral to mildly positive, and strongly positive levels of sentiments about England/Scotland/Wales. 19 Results are shown in Table 6. As can be seen, our expectations regarding England are wholly fulfilled. The small number of respondents reporting negative sentiments towards England also reported very low average levels of positive sentiments towards Britain. Those with neutral or mildly positive sentiments about England also felt somewhat more warmly about Britain, while those with the strongest positive sentiments towards England also tended to have the most positive attitudes to Britain. The strongly positive (Pearson s r =.54; p <.000) bivariate correlation between the two forms of national sentiments confirms the clear nature of this relationship. TABLE 6 ABOUT HERE By some contrast, however, at least some of our expectations about Scotland and Wales are confounded. In Scotland, where we see significantly lower average levels of BNS than in either England or Wales, we nonetheless also see an apparently rather weak positive relationship between sentiments towards Scotland and those towards Britain. However, the 19 Levels of BNS reported in Table 6 are mean averages as measured on the scale running from -6 to 7 (as explained in note 15). We should note that, by using measurements of national sentiments that derive from parallel items, the chances of observing any relationships between British national sentiment and English/Scottish/Welsh national sentiments should be maximised (when compared with potential measures that sought to tap these underlying constructs in rather different ways.)

16 apparent relationship shown in Table 6 is dependent on the very small numbers of respondents who reported negative sentiments towards Scotland. Overall, a bivariate correlation suggests that the two forms of national sentiment are essentially unrelated (r =.02; p =.43). Rather contrary to our expectation, what individuals feel about Scotland seems to have little relationship to their sentiments regarding Britain. Also somewhat contrary to expectations it is in Wales where we observe a mildly negative relationship between the two forms of national sentiments. The figures in Table 6 indicate that higher levels of Welsh national sentiment among respondents are associated with slightly lower levels of British national sentiment (and vice-versa). However, a bivariate correlation indicates that any such relationship between the two sets of attitudes is strikingly weak (r = -.11; p <.01). For the most part, and very similar to our findings from Scotland, what people feel about Wales seems to have rather little bearing on what they feel about Britain. Explaining National Sentiments Having established our measures of national sentiments in Britain, gauged the levels of such sentiments and explored their inter-relationships, a number of further questions arise. One set of questions, to be examined in the following section, is the relationship between national sentiments and party support. Another set of questions, which we examine now, concern the factors that might help account for national sentiments. Heath et al (1999) found a number of social background variables to be significantly related to their BNS scale. These were:

17 Age (with older age cohorts having higher levels of British sentiment); Education (with the more highly educated scoring lower in BNS); Religion (with British sentiment being significantly associated with membership of the established church in both England and Scotland; 20 Social Class (with middle-class respondents having greater BNS); and National identity (with a more strongly Scottish national identity associated with lower levels of British sentiment). Here, we explore whether these relationships hold true for either the alternative BNS measure developed here or for English, Scottish and Welsh national sentiments. Using our national sentiment measures as the dependent variables for our analysis, we report in Table 7 OLS regression estimates (with robust standard errors and model fit statistics) for a simple multivariate model in England, Scotland and Wales. The model includes measures of all those factors found to be important by Heath et al; we also include a dummy variable for the sex of respondents. (Detailed of how all variables in this analysis were specified are given in the Appendix). TABLE 7 ABOUT HERE While there are some results that differ slightly, and there is in general a somewhat better model fit in Wales and Scotland than in England, the picture overall is strikingly similar across the three nations. In all three, the strongest predictors of variation in BNS are age and religion. In comparison to the reference category of those 65 and older, those from younger 20 It may be useful to remind ourselves of the differences in the official status of churches in our three nations. The Anglican Church of England is the established state church in England only. A wholly different church, the Presbyterian Church of Scotland, enjoys established status in Scotland. In Wales, the Anglican church was disestablished in 1920 after a long popular campaign.

18 age cohorts in England, Scotland and Wales evince significantly lower levels of British sentiment. We also find that membership of the Anglican church in England and Wales, and of the Presbyterian Church of Scotland in Scotland, is associated with significantly greater levels of BNS. For BNS, patterns of association with national identity are somewhat complex, although in both Scotland and Wales but not England to anything like the same extent those rejecting a British element to their national identity score much lower on our BNS scale. For sentiments about the sub-state nation, although there are some significant correlations with some of the other variables, by far the strongest predictor is national identity. Those whose national identity is exclusively or mainly with the sub-state nation, rather than Britain, evince much greater levels of positive sentiments for England, Scotland and Wales. Our findings, rather in line with those of Heath et al (1999), do indicate a generally strong relationship between national sentiments and national identities. But national identities appear to be far from the only factor shaping sentiments towards the nation. Sentiments towards Britain appear to be related also to two other social markers church membership and age. However, with the established churches in long-term decline, and at least the possibility that the age differences found here are more generational than life-cycle in nature, the implications of our findings are not particularly positive for the maintenance of British national sentiment over the longer-term. National Sentiments and Party Support

19 In this final main section of the paper we turn attention to another issue explored by Heath et al (1999) in their study of BNS: the relationship between national sentiments and party support. Heath and colleagues found that levels of BNS differed notably across supporters of the different parties; they also found that their BNS measure made a significant contribution towards explaining party support. Here, we explore these same questions but for our twin measures of national sentiments in the three British nations. Heath et al measure party supporters via question on attitudes to parties; given that such data are not available for all three surveys we use we instead focus on a more traditional party identification. Table 8 reports the mean average scores on both national sentiment scales, for each of the three British nations, of identifiers with the main parties. The results show the differences between the parties to at times be greater than those, reviewed earlier, between the nations. Conservative identifiers score much the highest levels of British sentiment in all three nations; perhaps unsurprisingly, identifiers with the nationalist parties score lowest on this measure in Scotland and Wales. However, it should be noted that, on average, nationalist identifiers score low positive levels of British sentiment, rather than lapsing into negative territory suggesting that something close to indifference, rather than active hostility, characterises the attitudes towards Britain of identifiers with these parties. Identifiers with the nationalist parties in Scotland and Wales also, rather unsurprisingly, score highest on levels of Scottish and Welsh sentiments respectively. But identifiers with each of the three main unionist parties also tend to have positive attitudes towards Scotland and Wales. In England, reinforcing the positive correlation observed earlier between British and English sentiment, Conservative identifiers are again those who score highest on average on the latter. TABLE 8 ABOUT HERE

20 The contribution of BNS to party support was measured by Heath et al through a multivariate analysis of recorded voting behaviour in European and general elections. Such a measure of party support at the ballot box is less easy to construct using the data available to us. Measures of voting in the 2003 devolved elections are available for Scottish and Welsh respondents, but no such equivalent measure exists for those from England. We therefore once again use party identification as our dependent variable, specifying multinomial logistic regression models where we compare those identifying with any of the major parties (three in England, four in both Scotland and Wales) with the reference category of those identifying either with other parties or with none. Table 9 presents outline results from a series of models for each of the three nations. TABLE 9 ABOUT HERE Model 1 in each nation is a simple social demographics model that includes the same measures of age, sex, social class and educational attainment as used in the analysis reported in Table 7. This model displays limited explanatory power in all three nations. Model 2 simply adds a measure of national identity to the variables in Model 1. This addition makes little improvement to the model fit in England, but produces a rather better fit in both Scotland and Wales, where more strongly British identity is associated with identification with the Conservatives, and greater levels of Scottish/Welsh identity with support for the nationalist parties (and also, in Wales, the Labour party).

21 The sole explanatory variable in Model 3 is respondents placement on a standard socioeconomic left-right scale constructed from answers to a series of attitudinal questions. 21 As we see, this attitudinal scale has a very limited association with party identity in any of the three British nations. Model 4 is similarly parsimonious in including only our two national sentiment scales. As the outline results indicate, this model obtains at least as good a fit as that for socio-economic attitudes in England, and a substantially better one in both Scotland and Wales. Finally, we specify an aggregate model that includes all the variables from models 1-4. This aggregate model obtains the best fit in all three nations, indicating that the different categories of variables all have some contribution to make towards explaining party identification. But, most importantly for current purposes, the detailed results for this final model (not shown) indicate that even when such a range of other factors are controlled for, there remains a robust association between national sentiments and party identification in all three nations. In England, there is a strong a significant positive relationship between BNS and Conservative identification, and a similar (if slightly weaker) relationship with Labour party identification. In Scotland, BNS is linked strongly to Conservative identification, while Scottish sentiment is associated closely with identification for the SNP. And in Wales, we again see a strong link between BNS and Conservative identity; Welsh sentiments are significantly related to identification both with Plaid Cymru and with Labour. The analysis here does not claim to offer a full and final model of the factors underpinning party identification. But it does indicate that the relationship between national sentiments and party identification, at least, is sufficiently robust to persist in all three British nations within a model that accounts for a number of other plausible independent variables. The findings also raise the possibility that if national sentiments can help us account for party identity, then 21 For England and Wales, we use an identical socio-economic left-right scale, as developed and reported by Heath et al (1994). Only two of the five questions used in this scale, however, were asked in the 2003 SSA survey. We therefore use three alternative ones to construct a similar scale. (Details available from author).

22 they may also have a role to play in explaining other measures of party support, including voting behaviour. At the very least, our findings indicate that this is an issue that warrants further investigation. Conclusions This paper has, arguably, bitten off rather more than it has been able to chew. We certainly cannot claim that it has offered the final word in the measurement of national sentiments in Britain, nor in assessments of the relationships between such measures and other variables. The contribution of this paper has been far more provisional in nature; however, it may still be significant. We have established that, while there is a marked paucity of relevant available data measuring matters pertaining to national sentiments, items from a series of parallel surveys conducted in 2003 can be combined to form reliable and valid measures of national sentiments towards both Britain and the sub-state nation in the three main British nations. It has also been shown that more recent evidence from Wales strongly supports the continuing reliability of these national sentiment scales. It has further been shown that, at least when measured in 2003, levels of British national sentiment were lower in Scotland than Wales, and lower in Wales than England. Sub-state national sentiments, however, were the opposite strongest in Scotland and weakest in England. Nonetheless, the inter-relationships between the two forms of national sentiment are more interesting and complex than many would perhaps expect. Sentiments towards Britain and towards England in England are strongly positively correlated, while the equivalent measures in Scotland and Wales are essentially

23 unrelated. There is little sign that national sentiments in either of the Celtic nations have become negatively related: indeed, as we have seen, even among identifiers with the Scottish and Welsh nationalist parties Britain generally evokes indifference rather than active hostility. We have also seen that national sentiments appear to be correlated with age group and with national identity, while sentiments themselves appear to have some potential as a variable for the explanation of party support and maybe voting. In general, we believe that the results presented in this paper should be interpreted cautiously. But we do suggest three implications that might be drawn from them. First, the findings here indicate strongly that national sentiments in the UK are both complex and potentially important. Therefore, there is a need for political science to study them more seriously than it has done hitherto. Second, and following directly from the first point, there is a clear need for greater efforts to gather and analyse data relating to national sentiments. Some recent work has suggested that an English identity more distinct from a British one has, over recent years, begun to emerge much more clearly (Lodge et al 2012). Further data is needed to discover whether the relationship between sentiments towards England and towards Britain may also have changed in any significant manner in the years since A final point concerns the practical implications of our findings. Nationhood has become far more contested in Britain in recent times. The 2014 independence referendum in Scotland will make it more contested still. The stark choice imposed by a referendum may require people in Scotland to decide whether they prioritise their Scottish or British identity. But our findings indicate that, even in Scotland and even among those with the strongest sentiments towards Scotland, there is little sign that such attitudes are either motivated by, or prompt,

24 hostility towards Britain. In that sense, thus far Britain appears to have avoided some of the most negative national sentiments that can be generated by nationalist movements.

25 Tables Table 1: How Closely Attached do you feel to? (%) A. Wales Local Area Wales Britain Europe Very Fairly Not Very Not at all Don t Know <1 <1 <1 1 Number of respondents = 988 B. Scotland Local Area Scotland Britain Europe Very Fairly Not Very Not at all Don t Know <1 0 <1 <1 Number of respondents = 1508 C. England Local Area Region England Britain Europe Very Fairly Not Very Not at all Don t Know <1 1 <1 <1 1 Number of respondents = 3742 Sources: Wales: Welsh Life and Times Survey; Scotland: Scottish Social Attitudes 2003; England: British Social Attitudes 2003

26 Table 2: National Pride (%) A. How Proud are you of being British? Wales Scotland England Very Proud Fairly/Quite Proud Not very Proud Not at all Proud Not British / Don t Know Number of respondents B. How Proud are you of being Welsh/Scottish/English? Wales Scotland England Very Proud Fairly/Quite Proud Not very Proud Not at all Proud 2 <1 2 Not Welsh/Scot/English / Don t Know Number of respondents Sources: as Table 1 Table 3: Pride in National Flags (%) A. How do you feel about seeing the Union Jack? Wales Scotland England Very Proud A Bit Proud Indifferent* A bit hostile Very hostile Number of respondents B. How do you feel about seeing the Red Dragon/Saltire/Cross of St George? Wales Scotland England Very Proud A Bit Proud Indifferent* A bit hostile <1 <1 2 Very hostile <1 <1 1 Number of respondents Sources: as Table 1 *Combination of Not feel much either way, it depends and don t know responses Table 4: If allowed to choose Nationality on Passport (%) Wales Scotland England British English Welsh/Scottish* European Other Number of respondents Sources: as Table 1 (*Welsh in Wales, Scottish in Scotland, either in England)

27 Table 5: Mean Average National Sentiment Levels, England, Scotland and Wales British Sentiment English/Scottish/Welsh Sentiment England Scotland Wales Table 6: Mean Levels of British National Sentiment by levels of English/Scottish/Welsh national sentiment Levels of English/Scottish/Welsh Sentiment England Scotland Wales

28 Table 7: OLS Regression Estimates (Robust Standard Errors) for Social Bases of National Sentiments England Scotland Wales British sentiment scale English sentiment scale British sentiment scale Scottish sentiment scale British sentiment scale Welsh sentiment scale Age (Reference category = 65+) (.29)*** (.26)*** (.25)*** -.83 (.23)*** -.32 (.24) -.18 (.27) -.63 (.24)** -.51 (.23)* -.35 (.22) -.23 (.22) (.27)*** (.21)*** (.19)*** (.20)*** -.78 (.19)*** -.32 (.20) -.16 (.15) -.54 (.14)*** -.34 (.14)* -.09 (.14) (.34)*** (.26)*** -.78 (.25)** -.90 (.25)*** -.08 (.25) (.28)*** -.42 (.22) -.45 (.21)* -.25 (.21).09 (.44) Female -.23 (.15).40 (.14)**.35 (.12)** -.03 (.09) -.09 (.16).04 (.14) Highest qualification (Reference category = no qualifications) Degree Other qualifications Religion (Reference category = None) Anglican Catholic Other Christian Non-Christian Church of Scotland Social Class (Reference category = working class) Salariat Petty Bourgeoisie Routine non-manual Foremen/Technicians National Identity (Reference category = British not English/Scottish/Welsh) More British than English/Scottish/Welsh Equally British and English/Scottish/Welsh More English/Scottish/Welsh than British English/Scottish/Welsh not British -.60 (.24)* -.21 (.18) 1.13 (.17)***.15 (.27).66 (.22)** -.16 (.30) (.18).32 (.31).18 (.20).18 (.26).98 (.25)***.60 (.21)**.23 (.23) -.56 (.25)* -.80 (.23)*** -.28 (.17).83 (.16)***.40 (.25).68 (.21)** -.21 (.28) (.17) -.14 (.29).07 (.19).24 (.24) 1.10 (.23)*** 1.61 (.20)*** 2.60 (.21)*** 2.50 (.23)*** -.46 (.21)*.09 (.15) (.19)* 1.00 (.20)*** -.38 (.57) 1.03 (.14)*** -.04 (.16).38 (.24) -.25 (.18) -.22 (.21) 1.03 (.35)**.70 (.24)** -.74 (.23)** (.23)*** -.19 (.16).04 (.11) (.14).19 (.15) -.30 (.41).63 (.10)***.08 (.12).36 (.18)* -.04 (.13) -.01 (.16) 1.06 (.25)*** 2.84 (.17)*** 3.95 (.17)*** 4.34 (.17)*** -.56 (.28)* -.33 *.19) 1.04 (.20)***.15 (.30).67 (.22)** -.50 (.50) -.04 (.32) -.41 (.21)* -.17 (.29).01 (.22) 1.04 (.32)**.56 (.25)* (.25)*** (.27)*** -.04 (.23).01 (.15).34 (.16)*.17 (.25).54 (.18)** -.21 (.41) (.27).22 (.17).35 (.24) -.11 (.18) 1.70 (.27)*** 4.77 (.20)*** 6.06 (.21)*** 6.46 (.22)*** Adjusted R Weighted number of respondents * p <.05; ** p <.01; *** p <.001