Sports and Nationalism in a Globalized World

The topic of this article is sport nationalism. Using survey data from 25 countries, the study shows generally high levels of sport nationalism. The results indicate, however, that Western Europe has a lower level of sport nationalism while some less developed countries and Eastern Europe have higher levels. Furthermore, this research looks at how sport nationalism depends on individual and national characteristics. At the individual level, age, religion, income, sport participation, and sport attendance show a positive correlation with sport nationalism, whereas education has a negative correlation. The effect of age varies between countries. More democratic, prosperous, and globalized countries are the least sport nationalistic. Looking at how individual and national factors interact, the individual effect of education depends on the national level of cultural globalization (i.e., a greater negative effect in more globalized countries), and the individual effect of income depends on gross domestic product.

. For the purpose of this study, football clearly illustrates both the common global focus on sports and how this concern with sports provides excellent material for promoting and demarcating national idiosyncrasies.
Besides sports offering a golden opportunity to see how individual, national, and global factors together affect national identities (Skey 2013), studies of sports and nationalism are important because sports are ubiquitous in late modern societies and very often involve overtly nationalist dimensions; these are apparent during international sports events such as the Olympics and World Cup, but are also found in domestic sports. Furthermore, since sports seem so important for many people, and so many resources are devoted to them, it is timely to question how they might have political and democratic importance. Sports being of consequence for national identity is one possible way to approach this question. Even though a huge literature on nationalism exists, 3 a common critique has been the "fail [ure] to see the everyday nationalism that organizes people's sense of belonging" (Billig 1995;Calhoun 2007: 27;Edensor 2006;Skey 2013). Although studies on sports and nationalism are plentiful, 4 studying nationalism through sports is a suitable answer to this critique because it shows exactly how common people assign meaning to their respective countries in light of an everyday phenomenon.
Therefore, the purpose of this study is first to examine the extent to which people in different countries actually experience some kind of national pride related to sports, and next to ask whether national pride related to sports reflects individual differences (e.g., income) and/or national variations (e.g., gross domestic product [GDP]). Second, the study also investigates whether the effects of individual characteristics on nationalism reflect national differences (e.g., the variation in the effect of income on nationalism between countries). The third aim is to interpret these findings in light of the larger discourses on sports and nationalism by asking whether sport nationalism adheres to traditional understandings of nationalism or there are indications of other patterns of nationalism related to sports.
The study is based on an International Social Survey Programme (ISSP) data set (2007 Leisure and Sport) covering 25 countries.

NATIONALISM AND SPORTS: THEORETICAL PERSPECTIVES
At the core of this study is the phenomenon of sport nationalism, operationalized through a survey question asking respondents how proud they are when athletes from their respective countries succeed. The first aim of this section is to situate this rather narrow form of sport nationalism more clearly within the general discourse on nationalism and to identify a conceptual framework. Thereafter, the purpose is to suggest, hypothetically, how various individual and national characteristics contribute both alone and together to explain the differences in this type of sport nationalism.
What is Sport Nationalism?
Obviously, whether people are proud of their national athletes represents only one part of the larger picture of nationalism, and to interpret our empirical results properly requires a theoretical clarification: what is sport nationalism and how does it come about? In historical terms, the question we need to answer is how "the gap between private and public worlds was [also] bridged by sport" (Hobsbawm 1992: 142). Sociologically, the challenge is to establish how narratives on nationality are produced in an endless and continuous stream of reminders ("flaggings") of who we are as inhabitants of our countries (Billig 1995). Accordingly, what we need to understand is how sports matter for nationalism in some type of socialpsychological perspective showing how individual and collective factors act together and generate stories and identities telling us who we are as individuals as part of larger national collectives.
The concept most commonly used to grasp how various ideological elements link individual and collective experiences to national characteristics is national identity. Developing a national identity is a means of self-categorization, where we categorize and learn stereotypic norms that we then assigned to "us" and "them." A helpful conceptual framework for better understanding the role of sports in such processes is found in Tilly (2005: 131-52), whose concern is how identities that depend on boundaries between "us" and "them" are produced. Tilly distinguishes between two stages in such processes: the social mechanisms precipitating boundaries and those constituting boundaries. Although several of the precipitative mechanisms-encounters, imposition, borrowing, conversation, and incentive shiftsoccasionally help produce sport-based national narratives (Barthes 1972;, one of the constitutive mechanisms-activation-seems particularly relevant in the case of sports and nationalism. As an answer to Billig's (1995) question of how we "do not forget" that we are part of a country, activation of what are mostly already existing stories of who we are is a main role played by sports in the continuing, successful, everyday production of national identities. We (Brazilians, Germans, etc.) are a football country, we play football in our special way (juego bonito, machine-like, etc.), and we have our team and players.
Returning to the general discourse on nationalism, we can see how sports, as they take on ritual forms, might activate national sentiments and feelings. In this way, sports become central symbols in modern societies' nationalism, by producing and, perhaps most important, activating the stories about who we are as members of countries. National pride resulting from sports is one way to conceptualize the result of such processes (Evans and Kelley 2002;Kavetsos 2012;Kavetsos and Szymanski 2010;Smith and Kim 2006). The point is that being proud of national athletes' successes-we are good at football, and play football in a very attractive way (which is a consolation when losing)-represents an experience contributing to, and activating feelings of, sameness among the people of specific countries, which is partly what banal nationalism (Billig 1995) is all about. The question to consider in the next section is how such processes might occur; how to expect and explain differences in sport-related national identities.

Explanations of Sport Nationalism: Hypotheses
In general, it seems reasonable to presume that sport nationalism has two sources. On the one hand, sport nationalism might align itself with nationalism: individuals with strong national sentiments are also those proud when national athletes succeed. On the other hand, sport nationalism might also depend on affection for sports: those interested in sports may feel SPORTS AND NATIONALISM IN A GLOBALIZED WORLD "nationally proud" of their athletes despite not otherwise expressing strong national feelings. Based on these two sources, both individual and national factors are presented below, and then how these individual and national factors might combine in random effects and interactions is discussed. The point is to indicate how these factors might play a role in the processes of generating national pride from sports.

Individual Characteristics
As Calhoun (2007) and Smith (2009) pointed out, national identities are often based in deep cultural layers, and among these, "the links between religion and national consciousness can be very close" (Hobsbawm 1992: 67). Religion could be expected to further nationalism, partly because it is, like sports, often linked to rituals, places, and people with a national flavor. Furthermore, in the context of comparative survey research, culture is often linked to cognitive skills (Bollen and Medrano 1998;Inglehart 1990); the idea is that more educated people have a better understanding of otherness, more easily imagine themselves as a part of larger social groups, and thereby end up being more cosmopolitan (Beck 2009). In our data, this should imply that people with higher education are less prone to becoming proud because of sports than those with less education. In addition to culture, one should expect material resources to matter in terms of nationalism. The crux of such arguments is that people with secure finances feel less threatened by "others" than those with insecure finances. In our context, this implies that people with high incomes should be less nationalistic than those who earn lower salaries.
Results from previous studies indicate that the effects of gender depend on the form of nationalism under study (Kunovich 2009). As such, it is not obvious how gender might affect sport nationalism, but if it does, it could be reasonable to assume that because men are often more dedicated to sports than women, they have a stronger tendency for sport nationalism. Next, previous studies seem to show that older people in general are more nationalistic than younger people. This effect is probably (partly) due to younger generations' more cosmopolitan attitudes (in part reflecting education), but could also result from experiences specific to different generations and cohorts in various countries, related both to sports events and nonsports events associated with wars and politics. When it comes to the direct effects of sports on nationalism, two rather straightforward assumptions can be added: participation in sports and physical activity and attendance at sports events will both probably further sport nationalism. If it is correct, as Tännsjö and Tamburrini (2000) suggest, that sports foster nationalism, this effect could be rather strong.

National Characteristics
Several national characteristics might influence sport nationalism, some of which potentially reflect unique national experiences that are difficult to cover fully in a large comparative study like this. Hence, to enable the comparison of a set of important and similar national factors relevant for these processes across countries, the focus has been placed upon characteristics present, and presumably important, in all countries. These are structural factors related to individual incomes and national prosperity, political factors involving democratic institutions, and cultural factors reflected in individual education and the extent to which countries are culturally globalized. In addition, one variable intended to capture a country's interest in sports has been included.
For economic development, GDP has been applied. The social mechanism here is in line with Hechter's (2002) perspectives that economic competition promotes nationalistic antagonism; therefore, richer countries are less (sport) nationalistic. For politics, a measure of democratic governance reflecting the institutional strength of democracy has been included, comprising dimensions of free and fair elections, political stability, government effectiveness, regulatory quality, rule of law, and control of corruption (Kunovich 2009: 584). The assumption is that stronger democracies are more confident in their institutions, which in turn will make their inhabitants less prone to developing stronger nationalistic attitudes, also based on sports (Kunovich 2009). Cultural globalization measures the spread of ideas, information, and people, and is assumed to have an effect parallel to education at the individual level; that is, more globalized countries have more cosmopolitan citizens, which in turn means they are less nationalistic. The final variable indicates the extent to which countries tend to prioritize and succeed in sports, and even though a recent study questions this association , this study intends to test the hypothesis that a country's focus on elite sports strengthens the prevalence of sport nationalism in that country.
One central theme in the literature on nationalism is the distinction between a civic/ democratic and primordial/ethnic nationalism, often linked to an east-west divide (Björklund 2006;Jayet 2012;Smith 2009). The east-west context could also impact sport nationalism, although it is difficult to get at this divide except in the case of the countries in the study having (more or less) geographically clear eastÀ west positions. Furthermore, there are obvious differences-urban/rural, industrial/postindustrial, particular sports traditions, national cultures of various types-linked both to nationalism and sports that are not captured by the variables included in this study.

Random Effects and Interactions
To grasp the interdependency between the effects of individual and national characteristics, a multilevel approach was used. For some of the variables, the expectations of the effects are unclear or there are reasons to expect that these effects will vary between countries. This is true for age, the effect of which could depend on specific national events, sports-related or otherwise, so it has been included as a random effect; that is, the effect of age on sport nationalism is allowed to vary freely by country. Gender roles vary drastically, and women in countries with more liberal gender regimes might be more interested in sports and thereby more easily made proud by the achievements of athletes. Nevertheless, it is difficult to have clear expectations about the effect of gender, so the variable is, accordingly, included as a random effect in the models.
The effects of both material (income) and cultural (education) resources are dependent upon establishing a type of boundary between "us" and "them." These boundaries form the basis for comparisons with others, and the result is supposed to give rise to different levels of nationalism. This makes it reasonable to assume that the effects of these two variables at the individual level might depend upon parallel characteristics at the national level; the way cultural and SPORTS AND NATIONALISM IN A GLOBALIZED WORLD material boundaries develop at the individual level depends on the cultural and material resources at the collective level. For countries with poor economic resources and a less globalized culture, it is assumed that the effect of income and education will be more weakly negative than in more prosperous and globalized countries. Hence, two sets of variables have been included as interactions in the model.

DATA AND METHODS
The data applied in this article are from the International Social Survey Programme 2007 study on Leisure and Sport in 25 countries. Data were collected from 2006 to 2008. The study covers themes such as participation in, the meaning of, and satisfaction with leisure activities; time devoted to such activities and these activities' connections to other spheres of life; social and political participation; and social determinants and consequences of leisure. The sampling procedures varied between countries in terms of the units addressed (addresses, households, or individuals), age cutoffs, and number of stages. The surveys were administered as face-to-face interviews or in a self-completion format. The response rates varied from about 20 percent to above 90 percent. In short, the type and quality of the data vary. 5 An important question to ask is how the timing of various sport events might have influenced the results of the ISSP surveys. On the one hand, this topic is too large to cover properly in this study, because it is based on data collected over a period of three years (2006À 8) for 25 countries, each with a specific sports culture. This makes it very difficult to speculate on the ways events eventually influence sport nationalism. On the other hand, to provide background, it makes sense to look briefly at a few of the most spectacular events leading up to the surveys. The first major event was the 2006 Winter Olympics in Turin, Italy. The winners, as estimated in Olympic points, were Germany, Canada, the United States, Austria, and Russia. Second, Germany organized the FIFA World Cup in the summer of 2006. Italy beat France in the final with Germany and Portugal following. The most significant sports event in 2007 was probably the Athletics World Championships in Japan, where the United States, Kenya, Russia, Ethiopia, and Germany were the most successful based on the medal table. 6 In short, 2007 appears as a normal in-between year for international sport events following the FIFA World Cup and the Winter Olympics in 2006 and preceding the Summer Olympics of 2008 in Beijing. These large events had a Western dominancethe United States and Germany most consistently-but Russia was also successful at the Olympics and the Athletics World Championship.
The purpose of this study is to determine how individual and national characteristics alone and together influence sports and national identities. For substantive reasons (the level of nationalism and the effects of individual variables are expected to differ between countries) and statistical reasons (correlated residuals), the best way to analyze such data is by using multilevel models (MLMs) (Gelman and Hill 2007;Raudenbush and Bryk 2002;Snijders and Bosker 1999). 7 One the one hand, MLMs are rather straightforward extensions of ordinary regression analysis (Bickel 2007) where fixed regression coefficients are interpreted as ordinary regression coefficients. At the same time, the point is also to let intercepts and independent variables depend on the group level (countries) through random effects and interactions, thereby illustrating the variation in effects depending on the national level.
The dependent variable was measured through a question asking "How proud are you when [Country] does well at an international sports or games competition?" The answers range from "I am very proud" through "somewhat proud" and "not very proud," to "not proud at all" (see Table 1). There is also a "can't choose" option (removed from the data). Since a relatively high proportion of respondents stated they are "very proud," the distribution of answers is rather biased, so logistic MLMs were chosen comparing only those being "very proud" when their country does well at an international sports or games competition-representing the true sport nationalist-with all the less proud responses together. Even though this article's somewhat narrow definition of nationalism as being proud of national athletes is a rather restricted view, it has an advantage methodologically as the question of "How proud …" is specific in a way that makes the common problem of diverse understandings across countries less problematic than with more general questions (Heath, Martin, and Spreckelsen 2009).
Religion was measured based on attendance at religious services, while to understand the relative effects of education and income within and across countries, each country was divided into quintiles reflecting years of education and family income. Because of low response rates to the question of income (25 percent missing), values for this variable were assigned through regression predictions (Gelman and Hill 2007: 535). Sport participation reflects how often respondents "take part in physical activities such as sports, going to the gym, going for a walk," whereas sport attendance reveals how often people "attend sporting events as a spectator." Economic development is measured using GDP (2007). Cultural globalization is taken from the KOF index of globalization (Dreher 2007), and comprises the spread of ideas, information, images, and people (http://globalization.kof.ethz.ch). Democracy, expressing political openness and stability, is measured through the extent to which countries have free and fair elections, political stability, bureaucratic and regularity quality, and rule of law (http://www.worldbank. org/wbi/governance/data). There are several arenas for competition in international sports, and an index of the results from three of the most important of them was used: success in  8 To help interpret the results, variables were means-centered and the variables at the national level are also standardized. Statistical and graphical analyses were conducted using the program R (R Core Team 2014).
The results are presented in three steps. First, the bi-and univariate information on the dependent and independent variables is given. Second, the basic information on the MLMs (fixed effects, random effects, and model fits) is given. Third, because MLMs are often difficult to interpret (Ferron et al. 2008), especially when logistic regressions and interactions are involved, figures are presented showing not only the random effects-how the effects of age depend on the country-but also the interactions-how individual effects such as income depend on national characteristics such as GDP. A second reason for providing figures to support the interpretation is the fact that large samples such as those used in this study can easily cause small effects to be significant, so it is crucial to be aware of the substance of these effects besides their significance (Gill 1999). It is also timely to remember the obvious fact that the effects found in these analyses, even when interpreted in light of a suggested social mechanism, do not necessarily represent causal relationships (Berk 2004). The presentation of the models reflects the problem of multicollinearity among three of the independent variables, and this is commented upon in the presentation of the findings. There are no serious outliers (Nieuwenhuis, Pelzer, and Te Grotenhuis 2009) in the data. Figure 1 answers the question of the overall level of sport nationalism, and there are obviously both similarities and differences across countries. The scale used for the sport nationalism variable varies between 1 and 4, and given that all countries have means between 2.9 and 3.8, one could say that the similarities are more striking than the differences: all countries tend toward a relatively high average value on this variable. Sport nationalism is, generally, a widespread phenomenon; many people feel very proud when their national athletes do well. 9 Nevertheless, differences in levels of sport nationalism between countries are also considerable. A first finding is that West European countries are prominent among the countries with low levels of sport nationalism, yet in other "western" countries such as Australia, New Zealand, and the United States, the level is higher than the world average. Second, the East European countries are mostly above average, though there are substantive differences between them. For example, the Czechs are below average while Poland is among the most sportnationalist countries. Three less-developed countries-the Dominican Republic, South Africa, and the Philippines-are the most sport-nationalist countries, whereas three Latin American countries-Chile, Mexico, and Uruguay-are close to the average. Of the Asian countries, South Korea is rather sport-nationalist, while Japan and Israel are below average. One part of the results from the MLMs consists of reporting the fixed effects, that is, the effects common to all cases. A second part of the results, the random effects, tells us how selected effects-here the intercept and age-actually also vary across countries. A third part reports the fit of the model (see Table 2).

RESULTS
In this study, the three level-2 variables of GDP, democracy, and cultural globalization are indeed correlated, 10 and this results in disturbances (see Shieh and Fouladi 2003) in the final model in which these variables are applied together. To show how these disturbances affect the main model (5), a set of models (1-4) has been included showing how the model operates with and without each of these variables, and their corresponding interactions. What these MLMs illustrate is that the individual fixed effects are only marginally affected by these level-2 variables, and that the random effect based on age is almost identical across the models. The value of the intercept is more or less constant across Models 2-5, but the random effect varies, and it seems that GDP is the level-2 variable reducing this effect the most (Model 2). When all of the level-2 variables are included in the model, only GDP appears significant, but when each of the level-2 variables are applied alone, they are all clearly significant. 11 In Model 5, however, the effects of cultural globalization and democracy do not come to the fore, but based on the findings in Models 3 and 4, it can be assumed that a high correlation between these variables disturbs the calculation of their common effects. The subsequent interpretations are, with the above precautions, based on Model 5.
With a dichotomous dependent variable, logistic regression is applied, and given that all of the independent variables are mean-centered, the intercept represents the logit of an average FIGURE 1 Sport nationalism in various countries: "How proud are you when [Country] does well at an international sports or games competition?" Answers from 1 to 4. National means (points) with standard deviances (lines). Dotted vertical line: Overall mean.  Signif. codes: ***p < 0.001, **p < 0.01, *p < 0.05.

SPORTS AND NATIONALISM IN A GLOBALIZED WORLD
person in the sample. Initially, not taking the random effects into account, this implies that the average probability of being a sport nationalist is (from the full model) 0.55, but the standard deviation of the random effects indicate that this value differs considerably between countries, so the probability of being a sport nationalist varies from 0.35 to 0.74, �2 standard deviations from the fixed intercept. Furthermore, gender seems to be the only individual characteristic included in the study without any significant effects on national pride in sports. This study was open to a variation in effects based on gender given the large differences in gender equality among the countries partaking in this study; yet the gender effect seems of minor importance, including when acting as a random effect variable (not included in final models). The fixed effect of age is positive, indicating that older people are more sport nationalistic than younger people, and in general, the probability of being a sport nationalist increases by about 0.03 for each age quintile. Yet, interestingly, reflecting the variation in the social meaning of age across countries, which is also reflected in sports experiences, age has significant random effects. In most countries (95 percent) the effects based on age vary between À 0.15 and 0.34; that is, for some of the countries, age has a negative correlation even though the fixed effect is positive. Whereas the average person has a 0.55 probability of being a sport nationalist, the probability for a person with an age effect at one standard deviation from the fixed effect changes by about 0.06, and at two standard deviations it shifts by 0.11. Figures 2aÀ 2d illustrate the random effects of age, and demonstrate that even though the effect in most countries is clearly positive, in some countries such as Sweden, the Czech Republic, South Africa, and the Dominican Republic, it seems to be negative.
As expected, education has a negative correlation to sport nationalism, and the probability of being a sport nationalist-all other factors being equal-decreases by 0.02 for each upward quintile of education. One concern was whether the effect of education would be the same across all countries, considering whether cultural globalization might cause education to have a stronger or weaker effect in different countries. The results seem to confirm such an assumption ( Figure 3); the negative effect of education is stronger in more culturally globalized nations. In Switzerland, a culturally open country, the effect of a 1-quintile downward shift in education gives a 0.05 reduction in the probability of sport nationalism; in Japan, an average country in terms of cultural globalization, the effect of education is reduced to À 0.02, whereas the effect of education is close to zero (0.004) in the Philippines. These variations in educational effects could indicate that the cultural effect at the national level strengthens the effect at the individual level, so that the knowledge and international/cosmopolitan orientation provided by education are stronger in countries integrated into the global culture than in less globally integrated countries.
Similar to education, material resources (income) was assumed to have an overall negative correlation with sport nationalism, but again the question arises as to whether this effect might depend on the economic level of the country (GDP). The results (Figure 4) show that this assumption is valid. In the more prosperous countries such as Norway, the effect of income is negative, while in France, the income effect is close to zero, and in the Philippines, among the poorer countries in this survey, the effect of income is positive. These results arguably support the relevance of the suggested social mechanism generating these correlations, namely, that prosperity makes people less prone to fear or envy others, and that this effect is strengthened by the same effect at the national level. However, this finding also makes it clear that not

SPORTS AND NATIONALISM IN A GLOBALIZED WORLD
only does this effect depend on the interaction between levels, it is also generally (fixed effect) opposite to the negative effect that was assumed.
Religious devotion has a positive effect on sport nationalism, in line with the assumptions. The two variables included in this sports-focused study that were not included in previous studies of nationalism show that being involved in sports, either as a participant or an attendant, increases the possibility of having sport nationalist feelings. A physically active person (upper quintile) has a 0.05 higher probability of being a sport nationalist than an inactive person (lower quintile), and a 0.03 higher probability than a person with average levels of physical activity. The probability of a person who attends sporting events being a sport nationalist is 0.13 higher than that of a person who does not watch live sports. The effect of national-level characteristics should be viewed with some caution. The variable indicating how important sports are for a country was not significant and was dropped from the analyses, but the remaining three variables all seem to matter and all have negative effects: high GDP leads to lower levels of sport nationalism, which also seems to be the case for cultural globalization and democracy. As already illustrated, GDP and cultural globalization also matter for the individual effects of, respectively, income and education.

DISCUSSION AND CONCLUSIONS
The topic of this article is the prevalence of sport nationalism, how it varies between countries, and how such differences might depend on factors at individual and national levels.
Given the pervasiveness of sports in late modern societies and the many evident links to nationalism, it comes as no surprise that sport-related national identities-being proud when national athletes succeed-are relatively strong and widespread. Even though, at first glance, a high level of sport nationalism makes countries appear more similar than different, there are also strong and systematic differences between countries' sport nationalism. The countries/ regions with low levels of sport nationalism are all West European (Switzerland, Finland, Norway, Flanders, and France), whereas most of Eastern Europe (Poland, Croatia, Russia, Latvia, and Slovenia) have higher than average values. This could indicate that nationalism, as it is found in relation to sports, reflects some type of familiar east-west difference in nationalism. There are, however, two precautions to this conclusion. First, the Czech Republic, despite being an East European country, is low on sport nationalism, and several Western countries-Australia, New Zealand, and the United States-are above the world average of sport nationalism. Second, the east-west nationalism distinction mostly comes with an idea of substantive differences (ethnic versus civil) in nationalism, and even though sports could easily be interpreted as a factor related to ethnic nationalism, our data do not contain information on the content of sport nationalism that would allow us to draw conclusions on such issues.
Furthermore, in many cases it seems that differences in sport nationalism are related to economic and cultural resources, and as has been determined, variations at the national level suggest national differences other than east-west. In general, countries with low GDPs and low levels of democracy and cultural globalization are clearly more sport nationalistic than other countries. This indicates that nationalism (as related to sports) might play different roles in various countries; it points beyond the classic east-west distinction and reveals the need for empirical and theoretical refinements.
Next, the aim was a more complex understanding of how sport nationalism is a result of material and cultural factors. For cultural matters, the hypotheses were confirmed: highly educated people everywhere tend to be less sport nationalistic than less educated people. The assumption relation to the interactions between individual and national levels was also supported: education has a stronger negative effect on sport nationalism in culturally globalized countries. For income, the basic assumption of a negative effect was not supported, and high income seems, in general, to further sport nationalism. However, the hypothesis of an interaction between the effect of income on the individual level and a country's prosperity (GDP) was supported: the effect of income is negative in more prosperous countries, but positive in poorer countries. The initial general hypothesis was thereby rejected; however, due to the interaction effects, it seems valid in a few wealthy countries. Further differences at the individual level were found, mostly in line with the hypotheses, to show that older people in general are more sport nationalist than younger people, but also that there are exceptions to this tendency such that age has a very weak and even opposite effect in some countries. Gender is less important in this picture, whereas attending religious services has a positive correlation to sport nationalism. The assumption that sport nationalism is not only about nationalism but also the more ordinary everyday phenomenon of sports is confirmed, as both sport activity and sport attendance further sport nationalism. On the national level, three of the explanatory factors involved had the expected effects. Prosperous countries, culturally globalized countries, and countries with strong democracies are all (relatively) low in sport nationalism. A country's interest in elite sports, as measured in this study, seems not to matter for sport nationalism.
These findings have several implications for the general sociology of nationalism. First, we can see that sport nationalism is present everywhere, yet in different guises. This points toward the relevance of the example of football as fuel for nationalism in a global age; a common narrative on sports, of which football is only the most visible, is given particular national features. In theoretical terms, sports activate ideas and memories of who we are as citizens of a country. Second, the results indicate that an east-west distinction is operative and useful; however, the results also blur this distinction and point toward other geographical distinctions (north-south, continent-based, rich-poor, and more or less globalized) that make a difference in sport nationalism. As Billig (1995) claimed and previous research has shown Smith 2001a, 2001b;Shulman 2002), these results expose how nationalism is embedded in the mundane and concrete everyday lives of individuals, and thereby supports the need for further studies of this kind to fully understand what nationalism is, how it is created, and what effects it might have. The complex interactional patterns of individual and national factors should also carry importance for some aspects of a sociology of globalization, showing how sports, being mostly very local, have national and even supranational bearings, and therefore are good examples of "glocalization" (Giulianotti and Robertson 2007Robertson , 2009Robertson 1995).
For sociologists of sports, the question of whether and how sports might play a role in different national cultures becomes relevant and timely with the ubiquity of sport nationalism. This implies at least two challenges. In the sociology of sports, nationalism is often understood in a historical and case-oriented context that sometimes fails to notice the individual characteristics found to matter in this study, and the sociology of sports could generally be more open to the interplay between individual and collective factors. Moreover, the importance of sports with respect to a political issue such as nationalism also indicates that the "banal phenomenon" of sports could be studied more systematically with respect to other "grand theoretical" issues such as democracy, integration, and justice. A better understanding of the role of sport nationalism in society could also shed light on two of the recurrent questions of sport sociology, namely, why some sports are popular in some countries and not in others (Foer 2004;Mandelbaum 2004;O'Donnell 1994;Szymanski and Zimbalist 2005), and how interest in sports might travel between countries (Eriksen 2007;Kaufman and Patterson 2005).
There are still many questions related to nationalism in general, and sports and national identities in particular, and several challenges for future research in the field stand out. Inspired by Tilly's (2005) work on identities, the theoretical thrust of the article stated that sports first and foremost could be presumed to work as a kind of activator (rather than a precipitator) of national identity, that is, as a reminder and trigger of national memories and stereotypes commonly not originating from sports. However, there is a need for theoretical development, and a processual and dynamic approach along the lines inspired by Tilly's conceptual framework might prove particularly fruitful. Such an approach, first of all, might raise issues such as how sport is relevant to the precipitative and constitutive social mechanisms, how different everyday experiences collectively matter for national pride and identities, how sport nationalism depends on more collectivistic and individualistic (tight and loose) national cultures (Gelfand et al. 2011), and how such banal nationalisms influence the larger political picture-civic or SPORTS AND NATIONALISM IN A GLOBALIZED WORLD ethnic-addressed in the general discourse on nationalism. Second, since many countries have specific sports closely linked to their national identities, there is obviously a question of how various types of sports might have different roles to play in the kind of national feelings they arouse. In asking about the effects of various sports, one could also look at different sport arrangements by asking how different sport events matter for nationalism in different countries, or if countries dedicated to specific sports-for example, football, cricket, ice hockey, baseball-develop some type of common and/or regional identities based on these sports. Third, the question of how success in specific sport events eventually influences national identity should be studied more elaborately. To answer this question experimentally, one could conduct surveys before, during, and after big events. Even though this study included a brief discussion of some relevant sport events and tested a measure of success at mega sport events (without significant effects, so excluded from the models), it is still reasonable to assume that success (or failure) at such events might influence nationalism, at least for a brief period. Fourth, there is the question of how the fact that supporters often prefer their local or regional teams to national teams matters for nationalism. Fifth, even though there are studies comparing the different contents of nationalism (Evans and Kelley 2002;Hjerm 1998b), it could be interesting to view sports in conjunction with other banal examples of nationalism. Finally, the data used in this study made it difficult to go into detail on questions of how sport nationalism relates to the more traditionally ideological aspects of nationalism, but interesting questions exist related to whether sport nationalism is primarily furthering ethnic nationalism or civil nationalism. It would be useful to know what types of ideological elements are actually activated by sports in various settings, times, and places, and which of these are specific to sports. There are also several factors at the individual level not available in this data set, such as political orientation and ethnic background, that could be of interest for future studies.

NOTES
6. For those interested in delving into details for how results from sport events might matter for the results of this study, there are Wikipedias on the sport results from these years, see https://en.wikipedia.org/wiki/2006_in_sports and https://en.wikipedia.org/wiki/2007_in_sports. 7. The correlation between individual and national characteristics is confirmed by an intraclass correlation coefficient of 0.12.
8. For the Olympics, countries get points based on numbers of medals, and the measure used here gives measure relative to the highest score in the two Olympics. For soccer, the measure is based on the FIFA ranking in June 2007, and again, a country's measures are relative to the best country(Italy).
9. These results are based on analyses of the "original variable" before it is dichotomized for logistic regression analyses.