Conceptions and tendencies of age discrimination and attitudes towards older people in selected regions in Finland and Sweden

This study aims to access and explore tendencies in the conceptualization of age discrimination and the perceived attitudes towards older people in regions of Finland and Sweden. The analysis draws on GERDA survey data (GErontological Regional DAtabase), a repeated cross-sectional study in which data was collected in 2005 and 2010. The results indicate that the conceptions of age discrimination are changing in a positive direction, which is contrary to results shown in the Eurobarometer. On the basis of balance coefficients we show that conceived attitudes towards older people are changing as well, except for individuals in some sub-groups. We discuss the role of political rhetoric in relation to ageing awareness, the (non)individualization of society and the negotiation of age relations as tentative interpretations that strongly challenge the observed empirical tendencies.


Introduction
Few things permeate social life in the same way as the category of age. Evidence, sometimes dismissed as 'anecdotal', shows that age matters in many diverse ways (Bytheway 2011). However, we might not always acknowledge that it does, and we might not even possess the words that allow us to articulate these everyday experiences. Much less do we have 'real' insight into how these experiences evolve over time. However, a study by Tornstam (2007) has demonstrated evidence of slightly improved levels of knowledge about the ageing process among Swedes between the years 1982 and 2005.
Some of the ways in which we identify age as a constituent of experiences in everyday life are through concepts like ageism (Cherry & Palmore 2000;Coupland & Coupland 1993) and age discrimination. In this article we focus on manifestations of age discrimination and attitudes towards older people. Age discrimination is something with which many individuals are familiar (Eurobarometer, EBS 2009), but individuals rarely use the label ageism when describing or talking about such experiences. In Finland and Sweden, and the regional contexts in which we empirically examine these age-related issues, the phenomena are well known in everyday life but not by the term ageism (cf. Snellman, manuscript). However, in the article we orientate the analysis in relation to the ideological field of ageism by scrutinizing two of its central aspects, attitudes and self-reported discrimination. For the present article we define discrimination as 'behaviour that creates, maintains, or reinforces advantage for some groups and their members over other groups and their members' (Dovidio, Hewstone, Glick & Esses 2010, 10). We understand attitude (prejudice) as a thought on the individual level '(whether subjectively positive or negative) towards groups and their members that creates or maintains hierarchical status relations between groups' (Dovidio, Hewstone, Glick & Esses 2010, 7). The concepts of attitude (prejudice) and discrimination (alongside stereotypes) are often used as components in definitions of ageism. Our study is delimited to the emotional dimension of attitudes (i.e., neither behavioural nor cognitive dimensions) (Kite & Smith Wagner 2002), which we refer to as 'conceptualizations'.
We have aimed in this study to access and explore tendencies in the conceptualization of age discrimination and the perceived attitudes towards older people in selected regions of Finland and Sweden by seeking to answer four main research questions. First, how have individual conceptions of being discriminated against on the grounds of age changed between 2005 and 2010 in a cross-country regional context? Second, how have the conceptions of attitudes towards older people changed between 2005 and 2010 in different domains of society? Third, how common are these tendencies across different subgroups? And, finally, given the condition that tendencies are identifiable, how can these be tentatively understood?
Existing empirical evidence reveals a degree of stability in the prevalence of self-reported age discrimination, as well as a dramatic increase in terms of how widespread people consider age discrimination to be (EBS 2009). Different currents of modernization theory (e.g., Bauman 2000;Beck, Giddens & Lash 1994;Featherstone & Hepworth 1991;Beck 1992;Cowgill 1972) suggest that the social construction of (old) age may indeed become more negative over time as a result of individualization and an increasing idealization of youth. We arrived at our hypothesis on the basis of these theoretical assumptions and tested whether conceptions of age discrimination and negative attitudes towards older people occur more frequently as time goes by. In the analysis we have drawn on data from the GERDA (GErontological Regional Database) multidisciplinary research project.

Ways of conceptualizing age
Researchers have attempted to conceptualize age in different ways. In this section we outline previous research that has influenced our interpretations of our empirical results. We begin by outlining the discussion about ageism, which serves as an ideological background (cf. Krieger 1999) to the more specific studies of age discrimination. Other contributions function more explicitly as analytical tools, such as those of age relations (Calasanti 2003), age-coding (Krekula 2009), age positions (Lundgren 2010, Lundgren & Ljuslinder 2011, and the analyses of age-talk in everyday and institutional settings (Nikander 2009).
Ageism is a complex phenomenon (Kite & Smith Wagner 2002;Tornstam 2006) and has been defined in a large number of ways (Iversen, Larsen & Solem 2009;Snellman, manuscript). The meanings attributed to ageism ranges from narrow concepts of prejudice, stereotypes, and discrimination (Butler 1969;1975) to very broad concepts of everyday ageism (Bytheway, Ward, Holland & Peace 2007;Ward & Bytheway 2008;Bytheway 2011;Snellman 2009;Snellman, Johansson & Kalman 2012). Within this everyday discourse individuals articulate their experiences with the help of certain signifiers, for instance, old(er) or young(er) (Snellman, Johansson & Kalman 2012). Ageism is sometimes discussed in parallel with power relations based on age. Calasanti (2003) argues that there is a need to theorize age relations and old age systematically as a political location in its own right, predominantly on the basis of the experiences of older people. According to Calasanti (2003, 215), 'examining age relations … will allow us to explicate the structures that deny power to so many of the old for reasons having less to do with the aging of bodies and more to do with our construction of old age as sickness, dependence, lack of productivity, unattractiveness, and decline'.
One such theoretical concept that elucidates age relations is that of age-coding (Krekula 2009, 7), which means 'practices of distinction that are based on and preserve representations of actions, phenomena, and characteristics as associated with and applicable to demarcated ages'. Krekula illustrates how age codes can be used as i) age norms to negotiate identities ii) a means of legitimizing, negotiating, and regulating resources, iii) a resource in interaction, and iv) for the creation of age-based norms and deviation (i.e., situations where people are assigned a position of 'the other'). This reasoning clarifies that codings of age are crucial in order to understand how age relations, for instance, age discrimination, are created and effectively enforced. Krekula (2009, 15) argues that the concept 'can be understood as a logic of distinction that can be used in negotiating resources and the contingent actions that different categories allow for'.
Both Lundgren (2010) and Lundgren and Ljuslinder (2011) discuss the practice of age positioning, that is, one of the ways of negotiating and making sense of social worlds. Lundgren and Ljuslinder (2011) suggest that individuals in different contexts use age in ways that reproduce hegemonic images of older people simultaneously as they attempt to show other, more nuanced and norm-breaking images of the target group. Age positions, for instance, those connected to decline versus activity, are among other ways created by means of referring to 'population ageing'.
Another way of conceptualizing age is to draw on Nikander (2009, 864), who argues that 'chronological age and lifespan categories and other interactional formulations of age surface and are made relevant for and by us, implicitly and explicitly, and we position each other or describe and account for our own and others' actions in various everyday settings'. Some of these 'interactional formulations of age' are explored in this article, since they help us understand the interactional processes that stem from using 'age'-related survey questions, that is, how respondents position themselves (Lundgren 2012;Lundgren & Ljuslinder 2011) when confronted with questions of age-related discrimination and attitudes.

Empirical evidence of age discrimination within the EU
In western societies, the awareness of discriminatory or exclusionary behaviour has increased dramatically in recent decades. In the minds of citizens of the European Union, there is a widespread perception that discrimination based on age occurs frequently: according to the EBS (2009), 58 per cent of the European respondents -as compared with 55 per cent of the Swedish and 63 per cent of the Finnish respondents -considered discrimination based on age to be widespread. In some countries the perception of extensive discrimination seemed related to peoples' evaluation of how much effort was being made to combat discrimination in general (EBS 2009, 31). Curiously, the highest proportion among Europeans who felt that enough effort in counteracting discrimination was being made was found in Finland -68 per cent -whereas only 31 per cent of the Swedish respondents felt the same (EBS 2009, 31).
Within the European Union, the perception of the existence of discrimination based on age had undergone an increase between 2008 and 2009. When asked to make a comparison with the situation five years ago, 58 per cent of the EBS respondents believed in 2009 that discrimination based on age was widespread, as compared with 42 per cent in the previous year. This was the largest increase in the perception of the prevalence of any form of discrimination included in the study (EBS 2009, 11, 73). Simultaneously, the perception of age as a basis for discrimination against oneself, that is, self-reported discrimination, had not changed.
In part, the shift in opinion concerning age-based discrimination was attributed to changes in the economic climate in recent years, as older employees and job seekers may feel that employment opportunities are less secure today than they were prior to the economic recession. This assumption may be supported by the fact that older people were more likely than young people to report that discrimination based on age was widespread: 60 per cent of all respondents aged 40-55 and 55+ as compared with 49 per cent of respondents aged 15-24 (EBS 2009, 73, 118).

Ways of understanding social change
There is today little consensus regarding the ways in which social constructions of age come to life or how they prevail, and how and why conceptions of age generally -and age discrimination specifically -change over time.
One early attempt to understand the changing social status of older people was presented by sociologists (e.g., Parsons 1951;1964;Cowgill 1972), suggesting that older people became less appreciated in the industrial epoch due to rising standards of rationalization and effectiveness. Whereas previously older people were respected, admired, and believed to possess wisdom, their social status has declined over time as societies have become modernized, and preindustrial modes of production were replaced by highly industrial and technological structures. Cowgill (1972) has identified four central mechanisms in this process: advances in health, technological developments, urbanization, and education. As medical practices advanced and public health was improved during the twentieth century, longevity increased, which in turn increased the level of competition between age groups within labour markets. Allegedly this competition has led to a crowding out of older workers by younger ones. Similarly, advances in economic and industrial technology brought along new qualifications that favour younger workers while demoting older workers into less prestigious jobs or into retirement. As older people become excluded from the labour market, they not only lose social status and income, but they also become dependent on the younger generations.
Although some scholars (e.g., Palmore & Manton, 1974) have downplayed the relevance of this theory, it has received some support from postmodern and post-materialist sociology (e.g., Sennett 2006;Bauman 2000;Beck, Giddens & Lash 1994;Beck & Beck-Gernsheim 2002;Beck 1992;Giddens 1991;Inglehart 1990;1977). One common denominator of these contributions seems to be that cultural changes bring about different, and most likely more negative, social and political constructions of old age in a world that is becoming less socially embedded, more individualized, and more insecure. For instance, Featherstone and Hepworth (1991) have pointed out that post-modernity with its accentuation of individualization and reflexivity poses possible threats to the social image of older people because society as a whole is tending to become increasingly characterized by an admiration of youth (cf. Morganroth Gulette 2011). Similarly, gerontologists such as Estes (1979), Townsend (1981) or Walker (1981) have argued that older people have become structurally more dependent and thus less respected. Another source for ageism may arise from the process of globalization. Although the empirical evidence for a connection between age discrimination and globalization remains scant, it is likely that it has affected the images of older people negatively since they are increasingly depicted as a threat to the welfare state (cf. Lynch 2010). Whether or not these predictions actually hold true, they serve as a fruitful theoretical point of departure for this article. Not only can strengthened ageist structures lead to a cultural devaluation of old age and weaken public support for, among other things, public care services for older people, but they are also likely to have an influence on older people's selfimages and their feelings of inclusiveness. For example, in a world of growing competitiveness, economic inequality, and individualization, we may reasonably expect older people to adopt more negative selfimages and stronger sentiments of exclusion (cf. Kite & Smith Wagner 2002).

Regional data and participants
This article used survey data from the GERDA (GErontological Regional DAtabase) population study, a cross-sectional data collection carried out in two waves. The first wave of data collection was carried out in 2005 among 65-and 75-year-old people in the Ostrobothnia region in Finland and the Västerbotten region in Sweden (n=3370). The second wave of data collection was carried out in 2010, and it involved identical sampling procedures. In 2010 the age cohorts of 65-, 70-, 75-, and 80-year-old people were surveyed (n=6838). The response rates were 69 per cent in 2005 and 64 per cent in 2010 (Herberts n.d.;. In this article we delimit the analyses to individuals who were 65 and 75 years of age in 2005 and 65 and 75 years of age in 2010.
The participants were asked to answer identical survey questions on self-reported age discrimination and attitudes in the year 2005 as well as in 2010, alongside a wide range of other questions regarding, for instance, activities, values, and health. The general objective of the two waves of GERDA data collection was to attain knowledge about older people's life situation, health, and wellbeing.
The sample included all individuals living in rural municipalities, while every second individual living in the town of Vasa (Finland), and every third individual living in Skellefteå and Umeå (Sweden), were randomly sampled. In order to avoid an overrepresentation of the rural population, design weights were used (Herberts 2011). Skellefteå and Umeå were assigned a weight factor of three and Vasa a weight factor of two.
The GERDA questionnaire was externally validated among older people (n=24) in both Finland and Sweden by means of qualitative pilot-testing (Fagerström et al. 2011).

Analysed items and methods of analyses
The first question in this study was as follows: 'during the last year, have you sometimes been badly treated or discriminated against merely on the grounds of your age?' The response alternatives to this question were 'yes', 'no', and 'don't know'. This self-reported measure of age discrimination is identical to the spirit of the one used in the EBS and it has been suggested that self-reported age discrimination is a valid indicator of unfair treatment (Gee, Pavalko & Long 2007). The internal response rate on this question was high, with only 1.9 per cent (139 individuals) not responding to this question.
The other main question was: 'what point of view (Swedish 'inställning', Finnish 'asenne') do you think generally exists towards older people in our society?' The four response alternatives were 'positive', 'neutral', 'negative', and 'don't know', and this question related to eight specific domains in society: in newspapers, on TV, in advertising, in politics, in the labour market, in health/medical care, in shops/banks, and at cultural events. Owing to a relatively high nonresponse rate on this question an inclusion limit of 50 per cent was set for these eight items, which resulted in the exclusion of 149 cases from the data set.
In the analysis we examined whether reports of age discrimination and attitudes towards older people varied across the subgroups. The variables of year of data collection (2005 or 2010), age (65 or 75), gender (female or male), employment status (working or no longer working), resident country (Finland or Sweden), health (good or poor), and marital status were included in the analysis.
The analysis focused on describing the extent to which participants had experienced age-discrimination and whether participants perceived positive, neutral, and negative attitudes towards older people as a group. We carried out chi-square tests in order to analyse the magnitude of change in self-reported age discrimination and to see whether such tendencies were present across different subgroups. Balance coefficients for the whole data-set and for subgroups were calculated in order to show tendencies in the attitudinal measures. Other methods of analysing data (e.g., a multivariate analysis) would have required a deletion of many cases (owing to the response alternative 'don´t know') from the data-set in order to construct ordinal-scale data, thus reducing the descriptive value of the analysis. In short, the use of balance coefficients was seen as a more appropriate way of analysing the data. The balance coefficient is calculated as the difference between those who report a positive selfexperienced attitude and those who report negative attitudes in different domains. The coefficient varies between +100 and -100 (imaginary extremes that represent a totally positive or negative society) and represents the intensity and direction (positivity or negativity) of self-experienced attitudes towards older people. The greater the share of neutral or indecisive respondents, the closer to zero the coefficient will be.

Tendencies in age discrimination
As stated earlier, we anticipated that the extent to which individuals report age discrimination would be stable over time. However, as shown in figure 1, the analysis illustrates that 4.2 per cent of the respondents in 2005 and 2.6 per cent in 2010 reported age discrimination. Consequently, between the two points in time, and in contrast to what was expected, a minor but statistically significant decline in self-reported age discrimination had occurred (chi-square test p≤0.001). A higher share of respondents in 2010 reported that they did not have any experiences of being discriminated against  An examination across sub-groups exposed the following results. The declining tendency in age discrimination was confirmed among the Swedish (p=0.007) as well as the Finnish (p=0.001) respondents, and the tendency was also present when we examined the subgroups 75 (p=0.012) and 65 years of age (p=0.002). The tendency was statistically significant for both women (p=0.002) and men (p=0.006). However, when we examined the tendency among those still working in comparison to those retired, we found that the tendency was not significant for the former group (p=0.368), but significant for the latter (p≤0.001). Somewhat surprisingly, the tendency was significant for respondents reporting poor health (p≤0.05), but not for those reporting good health (p=0.109). Regarding martial status, the tendency was significant for those who were married (p≤0.001) and for widows/widowers (p≤0.05), but not for those living with a partner (p=0.496), were divorced (p=0.587) or unmarried (p=0.285). Thus, the declining tendency of conceived age discrimination was observed in some of the examined subgroups, but not among those who were still working, were in good health, were living with a partner, were divorced or were unmarried.
When we compared these regional results with previous findings at a European level (e.g., the EBS), we found a certain discrepancy that might suggest that conceptions of age discrimination are highly context-sensitive. When specific regions and specific subgroups of individuals are studied we can observe subtle patterns. Compared with the Eurobarometer, which was based on a random sample from the whole EU population and on all age cohorts, the GERDA-data used in our study was only based on the age cohorts 65 and 75 and targeted almost all of these respondents in the studied regions. In light of this, it is perhaps not such a big surprise that the results came out differently. These incongruent empirical results are noteworthy because they illustrate different sides of a context-sensitive phenomenon. It is not necessarily the case that conceived age discrimination is stable in all contexts. The extent to which older individuals conceive age discrimination might therefore be changing in the regions studied in Finland and Sweden. In the discussion we shall turn to tentative interpretations of what might be the reasons for the observed declining tendency.

Tendencies in perceived attitudes towards older people
When we examined how older adults in 2005 and 2010 conceived attitudes towards older people within different domains of society, we noticed changes. We observed a tendency that was similar to one found with self-reported age discrimination. Each of the eight examined domains (TV, advertising, politics, labour market, health and public medical services, shops/banks, and cultural events) exhibited a similar response pattern (appendix 1). The observed tendency demonstrates a change towards the more frequent assignment of a positive response alternative, a neutral response alternative, and, less frequently, the ascription of the negative response alternative. Additionally, the share of uncertain respondents, those who responded 'don't know' decreased between 2005 and 2010.
On the basis of these frequencies, we calculated balance coefficients in order to assess the overall attitudinal tendencies. Table 1 shows the balance coefficients for the different societal domains. The aggregated data shows an average decline in negative attitudes from -14.2 in 2005 to -10.8 in 2010. The largest difference between the two years occurred within the labour-market domain, in which the coefficient rose from -59.4 in 2005 to -51-1 in 2010. Note: the balance coefficient is calculated as the difference between those who report a positive self-experienced attitude and those who report negative attitudes. The coefficient varies between +100 and -100 and represents the intensity and direction (positivity or negativity) of self-experienced attitudes towards older people. The greater the share of neutral or indecisive respondents, the closer to 0 the coefficient will be.
A closer examination revealed that the aforementioned tendency remained quite robust across the sub-groups. The only exceptions were the Finnish and divorced respondents, which did not exhibit this tendency. The average balance coefficient for respondents in Finland was -8.8 in 2005 and -9.5 in 2010. Similarly, the average balance coefficient for divorced respondents was -15.7 in 2005 and -16.4 in 2010. The analysis revealed that the changing tendency was present among women (-15.4 in 2005, -11.5 in 2010) as well as men (-13.0 in 2005, -10.2 in 2010). The average balance coefficient indicates changing tendencies among people who were 65 as well as 75 years of age. For individuals who were 65 years of age the shift was from -17,1 to -13,7, and for the 75-year-old age group from -10.0 to -4.85. The change for Swedish respondents was from -17.3 to -11.6. Among respondents who reported that they still worked the balance coefficient changed from -11.7 in 2005 to -10.0 in 2010. Among those who no longer worked the shift was also in the same direction, from -13.7 to -8.1. For respondents who reported that they have a good health the coefficient changed from -13.4 to -10.2, and among those with poor health from -14.5 to -11.5.

Discussion
The observed tendencies can be understood in different ways, and our continued analysis yields alternative interpretations. Our analysis evolves around four broad tentative interpretations: (i) political rhetoric and ageing awareness, (ii) the individualization of society, (iii) the negotiation of age positions, and (iv) design effects. Owing to the complexity of the age discrimination and ageist processes under investigation, we do not think that it is possible to pin down any single explanation for the observed tendencies. Our interpretations derive from the dialectical relationship between the individual and the societal structure, in which individual acts are influenced -but not necessarily determined -by structural conditions (cf. Hendricks 2003). Our discussion is occupied with the conceivable chains of events that influence the respondents' conceptions.
First, one tentative interpretation is that the observed tendency is a reflection of the political articulation of improvements for older people in the local societies that have been examined. It is reasonable that self-reported age discrimination and negative attitudes towards older people actually are declining as a consequence of shifts in the ways politicians, researchers, and other stakeholders talk about older people as a group (Lundgren & Ljuslinder 2011). Judging from the average balance coefficient, we may regard the situation for older people on the labour market, for instance, still to be troublesome. But the potential contribution that older people offer to society as a whole, say, by the postponement of the retirement age and the prolongation of working lives, is today encouraged to a considerably higher extent than just some decade ago. Currently, it is even expected that we shall participate in working life for a longer period so that society can manage the financial effects of an ageing population. Moreover, mass media often construct older people as an under-exploited labour supply.
Two specific examples that are intertwined can be useful here: political rhetoric in election campaigns and the increased awareness of the ageing process. Policy-makers on a European, national, and local level often make political statements that might have affected how individuals conceive and report age discrimination and attitudes in different domains. For instance, in both countries in our study there were electoral campaigns taking place when the second wave of the GERDA study was launched. The political parties in Sweden, for instance, were vying against each other to convince the public that their own political programme would benefit retired people the most. The observed tendencies can perhaps be seen as reflection of this. By turning their attention to older people, politicians may have created a more positive atmosphere that made older people feel more appreciated than before. Startlingly, in 2011 -approximately one year after the elections -the Swedish government cancelled the tax reductions that had been promised to retired people during the election campaigns. It is not hard to imagine that the data on age discrimination and on attitudes towards older people as a group might have been very different if the respondents had known that the promises of improved economic conditions would not be carried through.
To what extent can the observed tendencies be related to changes in the meaning of ageing itself? Where does it lead us if we argue that 'all [tendencies] reflect conceptual changes in what "to age" means' (Bytheway 2011, 77)?
The awareness of what it means to age is also shaped by political rhetoric, that is, by facts and new information that are made evident in public debates by politicians based on scientific research (cf. Tornstam 2007). Palmore states that 'those with more knowledge [about aging] tend to have less negative and more neutral attitudes ' (1988, 43). This interpretation seems intriguing, since this is exactly what the empirical data reveals (cf. appendix 1). According to Palmore's line of reasoning it makes sense to ask whether the respondents had a deeper awareness (or even perhaps more knowledge) about the ageing process in general in 2010 than they had in 2005. This would seem plausible considering the fact that a current trend in society is increased knowledge about ageing (Palmore 1999;cf. Tornstam 2007). Increased research, more debates in the media, informational campaigns, attentiveness to the population ageing, and the political capital inherent in age-related issues (cf. Tornstam 2007) might have increased the knowledge about the everyday issues, difficulties, and opportunities that the ageing process actually encompasses. An improved level of knowledge about ageing might in turn have affected the respondents' answers to the questions of age discrimination and attitude. On the other hand, Tornstam (2007) points out that changes in levels of knowledge in relation to flows of information in society are not simple. For instance, he shows that levels of knowledge about the lives of older people (e.g., housing, hearing, living in institutions) do not change over time even though debates in society might be expected to influence the levels of knowledge. This observation does, however, not mean that improved levels of knowledge, generally speaking, cannot have a moderating effect on perceived age discrimination and attitudes towards older people.
Secondly, another interpretation is that the observed tendencies are related to the individualization of society (e.g., Sennett 2006), and that the respondents were less influenced by collective values in 2010 compared with the year 2005. The respondents might have taken on board more individualistic values during this short period. Previously, the respondents were perhaps more inclined to report negative conceptions of age discrimination as a way of expressing their commonality with a group that they believed to be oppressed, even though they had not been discriminated against themselves. That is, in 2010 (or perhaps as time passes) fewer people identified themselves with older people as a group or with a group being discriminated against. Individuals are well aware of the severity of age discrimination in society, as suggested in the EBS, but between the two points in time people might have come to view themselves as being more independent of that large-scale societal issue.
With regard to the contexts between the two countries, it is also plausible that the tendencies are related to cultural and historical differences. For instance, the fact that we detected a shift towards more negative age attitudes among the Finnish respondents (mirrored also in the Eurobarometer) can perhaps be a reflection of cultural differences and historical events, such as the fact that the Finnish respondents experienced the Second World War while most Swedish in the same cohorts did not experience the war in a similar kind of way. Individualistic values might therefore not be rooted, or at least not influence conceptions of age-related issues, in the same way in Finnish society.
If it is true that processes of individualization, which is one of the consequences of modernization, has influenced the inclination to opt for negative self-reported age discrimination and negative attitudes towards older people, then this would mean that despite the desirable tendencies shown in the empirical data, we might still be heading towards stable, stronger, and possibly more negative ageist structures (e.g., Sennett 2006). This opposing tendency would simply not be exposed in individual conceptualizations. Unfortunately, it cannot be ruled out that we are dealing with illusionary tendencies. Tornstam (2007) argues that observed changes of this kind might reflect persistent attitudes that merely have changed shape and taken on new forms. Tornstam has phrased this point nicely: 'the persistent pattern of stereotyping old people, in its old or new forms -"positive" or "negative" -is in reality the ageist society's way of controlling and exercising power over old people ' (2007, 56). Accordingly, precautions have to be taken so that the results of our study are not used to nurture new undesirable attitudes. We cannot, for instance, conclude that the exposed tendencies are signs of improvement.
Thirdly, the exposed tendencies may be connected to how people make sense of their social worlds by means of taking age-related standpoints. As we showed earlier, previous research has argued that age can be conceptualized in many different ways, and one of these ways is the negotiation of age and the attempts to make sense of our social world by means of these negotiations (Krekula 2009). Age, in any of its everyday forms (Snellman, Johansson & Kalman 2012), is an articulatory device (cf. Nikander 2009) that individuals use as a resource to position themselves within the social world (Lundgren 2010) by coding experience (Krekula 2009). The responses to the discriminatory and attitudinal questions can be understood as a part of a broader societal negotiation of age relations that serves to pinpoint one's political location (Calasanti 2003). Only fragments of this totality of negotiations, or individual attempts to position oneself and thereby choose a political location with the help of 'age', might actually reflect, or qualify as, shifts in age discrimination or changing attitudes towards older people. In consequence, this means that age-positioning, which in this context is the negotiation and sense-making of social worlds by means of general responses to questions concerning age discrimination or attitudes towards older people, is not used in the same way or to the same extent as previously. That is, the tendencies can be understood as a shift in how age relations are negotiated (Calasanti 2003). The respondents may not have felt the same need in 2010 to use the resource of negotiating age discrimination in order to position themselves or to choose a political location. Consequently, we are perhaps observing latent changes in 'conceptions about ages' (Krekula 2009, 15), and changes in how the respondents make sense of their age-related life-worlds when they are given a chance in a delimited context.
Finally, design effects could also have influenced the tendencies in self-reported age discrimination and attitudes. Even though the analysed questions used in 2005 and 2010 were identical, other questions differed as a result of the introduction of additional questions in 2010. One question allowed respondents to state whether they had had experiences of being treated as old, while there were also some new attitudinal measures included in the 2010 survey. The tentative interpretations discussed above might singly explain the observed tendencies. Importantly, it is also possible to understand the findings in light of a mixture of all the particular interpretations (cf. Tornstam 2007).

Conclusions
In this article we have exposed the tendencies of decreasing selfreported age discrimination and negative attitudes towards older people as a group. These tendencies did not corroborate our expectations and hypothesis, which were derived from earlier European empirical survey data as well as from theoretical contributions. The results are worthy of note insofar as they indicate a declining tendency in self-reported age discrimination and unexpected shifts in attitudes towards older people as well. We have suggested that the tendencies can be understood as time-, place-, and contextsensitive phenomena. Our discussion has revolved around the role of political rhetoric connected to an enhanced awareness of ageing, the individualization of society, individuals' negotiation, positioning, and choice of political location based on age, the potential consequences of design effects, and the possibility of a complex mixture of the aforementioned as a way of explaining the results. These theoretically guided interpretations strongly challenge our empirical results, just as the empirical results challenge previously exposed results in the Eurobarometer.
An important limitation to this study is that we do not know what kind of situation the respondents had in mind when they answered the questions. For instance, we do not know how they understood the concept of discrimination. Bytheway et al. (2007) have shown that people might relate their experiences of discrimination to many different situations. Thus, the magnitude of age discrimination or attitudes cannot be understood in an absolute sense. However, it makes sense to carry out relative comparisons between different points in time and across sub-groups since we are dealing with identical survey items (cf. Tornstam 2007).
On the basis of these tentative interpretations, future research should try to uncover the rationale that makes people identify the presence of age discrimination or certain kinds of attitudes towards older people. What are the contextual and structural factors that are involved in -or maybe even determine -people's conceptualizations of age discrimination and attitudes? By continuously monitoring the tendencies in conceptions of self-reported age discrimination and discriminatory attitudes towards older people, we can learn more about the ageist power-relations that influence individual lives.
An implication relating to policy and practice arising from this study is that we have to exercise caution in how exposed age-related patterns (e.g., stability or shifts in self-reported age discrimination) might be used: how do we allow these to have an impact on the decisions that we make? There is a serious risk that anticipatory anti-ageist discriminatory work could decrease in political decision-making and in practices that care for people if the results of our study are interpreted as improvements. This would lead to an unwarranted view that what has been accomplished is satisfactory. We need continuously to ask ourselves what these tendencies really mean and in what ways they are related to and influenced by other phenomena in society.
Västerbotten; the University of Vaasa; the Åbo Akademi University; the Umeå Univerity and the Novia University of Applied Sciences.