Attitudes to the Welfare State: A Systematic Review Approach to the Example of Ethnically Diverse Welfare States

by Trude Sundberg
University of Kent

Sociological Research Online, 19 (1) 28
<http://www.socresonline.org.uk/19/1/28.html>
10.5153/sro.3337

Received: 12 Nov 2013     Accepted: 25 Feb 2014    Published: 28 Feb 2014

Abstract

There has been a growing interest in and expansion of research on welfare attitudes and the legitimacy of welfare states as they face both internal and external pressures at social, political and economic levels. The number of cross-national surveys is steadily increasing and many studies examine developments in social provision and public attitudes towards welfare. However, there is a lack of a clear and comprehensive overview of knowledge about tendencies in support for the welfare state in light of these pressures. Moreover, there is a lack of understanding of what impacts variations in attitudes and the relationship between attitudes and other constructs such as perceptions, values and stereotypes, which all form part of support for the welfare state. The article reports on findings from a project using tools from the systematic review tradition in an innovative way to achieve a comprehensive and systematic overview of current knowledge. The article has three main contributions; firstly, it adds to our understanding of the relationship between attitudes, perceptions, beliefs, stereotypes and values, secondly, it adds to attitude theory and thus enhances our understanding of welfare attitudes and lastly it illustrates how it achieves the first two through applying adapted tools from the systematic review tradition. Systematic Review as a method originated in natural science and medicine, and the paper examines how it can be successfully transferred to issues in social science. The article argues in favour of emphasising the importance of differentiating and investigating what is known and not known by examining the relationship between immigration and support for the welfare state. Thus, an adapted systematic review is found to enhance our understanding of trends in welfare support by use of theoretically framed research synthesis.

Keywords: Attitudes, Values, Perceptions, Stereotypes, Welfare State, Immigration, Ethnic Heterogeneity, Support

Introduction

1.1 Societies across the world are facing an array of changes ranging from an ageing population to economic crises and changing social structures. For the welfare state this raises questions such as: how social needs may change, what pressures are created for welfare states and how welfare states, citizens and politicians respond to these changes and pressures. Studying potential changes in citizens' support for institutions such as the welfare state is important for at least two reasons. On the one hand, politicians need to decide how to respond to the challenges, and their decisions may be influenced by attitudes and values found amongst citizens. On the other hand, reforms and changes in institutions, which may be seen as institutionalised norms, may themselves impact on citizens' values and attitudes.

1.2 This article reports on findings from a wider project analysing support for the welfare state by use of an adapted systematic review approach. A systematic review can be defined, in a traditional approach to the method, as:

"A systematic review has clearly identi?ed questions, searches for relevant research following a procedure speci?ed in advance, has criteria for which studies to include, has criteria for the information to be considered from these studies, appraises studies using clear criteria for what is good and what is less good research and synthesises the results in a transparent fashion, sometimes using statistical pooling" (Bradshaw et al. 2000: 6).

1.3 The empirical data in the article are studies analysing trends in support for the welfare state when related to the impact of higher levels of ethnic diversity due to increasing immigration. In this area, scholars, mainly from the USA, have found that higher ethnic diversity lead to lower support for the welfare state. Thus, scholars have argued that similar trends will appear as Europe becomes more ethnically diverse. Furthermore, some argue there a link between the rise of right-wing politics and immigration, which can reinforce 'us' versus 'them' resentment (Van Oorschot 2008). At the same time, immigration is argued to be a politically important issue due to its political salience and politician's emphasis on immigration as a political problem, which may negatively influence attitudes (Larsen 2008). However, this article argues against theories arguing in favour of general mechanisms and show that alternative approaches such as contact theory and theories emphasising that institutions may act as barriers to the negative impact may be better placed to explain trends in Europe.

1.4 This article illustrates trends in the relationship between increased immigration and support for the welfare state and tests whether the USA experience is unique by use of adapted systematic review tools. It also demonstrates how theoretically framed research synthesis building on realist synthesis, emphasising underlying explanations in research synthesis, can help increase our understanding of a phenomenon. Furthermore the article adds to our understanding of the relationship between attitudes, perceptions, beliefs, stereotypes and values by analysing trends separately for attitudes and values. One of the main arguments in the article as well as the larger project it is based on is that systematic review tools may be used fruitfully in the Social Sciences, and here Social Policy, however, one must appreciate the diversity and specific characteristics of Social Sciences when doing so. The result of the latter is that it is difficult to apply some principles from the Systematic Review tradition such as replication, homogeneity and hierarchies; however at the same time it is illustrated how an adapted version of the method may help increase our understanding in a specific area. Responding to these challenges the article illustrates one way of carrying out a review that contributes to knowledge. Thus it shows the utility of systematic review tools, whilst at the same time appreciating these issues, and thus adds to a growing literature that discusses ways in which reviews can be carried out successfully in the Social Sciences.

1.5 In studies on immigration and welfare support, one dominant school of thought, building on arguments put forward by American scholars, underlines the importance of the degree of ethnic heterogeneity in explaining why the population in some countries supports welfare policy (Alesina & Glaeser 2003). Their argument is that diversity has a negative effect on the generosity of welfare states as it weakens social ties and feelings of a common identity that laid the ground for encompassing welfare arrangements (Van Oorschot 2008). On the other hand, some argue that immigration may have a positive influence on support in accordance with cultural theory, whereby increased diversity widens people's horizons and makes them more accepting of other cultures (ibid). Much emphasis has been put on testing and evaluating whether US-based findings (e.g. Alesina and Glaeser 2003), which predict a negative influence of higher ethnic diversity on support for the welfare state, are replicated in other parts of the world. Thus, the focus of all the studies included in this article is on effects of either level of immigration or another measure of ethnic fragmentation and/or attitudes towards immigrants on a measure of welfare support. In other words, as Europe becomes more ethnically fragmented, authors such as Alesina and Glaeser (2003) would expect support for the Welfare State to decrease[1]. On this basis, and given the a lack of consensus on the impact of increasing immigration on support for the welfare state in Europe, combined with a high number of studies produced, the area represents a good 'testing ground' for the use of systematic review. In addition to analysing trends in support, the aim is to test whether trends differ by constructs (i.e. stereotypes, attitudes and values). Whilst support for the welfare state is a concept that encompasses all of these constructs, the literature in the area does not tend to differentiate between them, and this article illustrates why it is important to do so by use of an adapted systematic review.

1.6 In the following, there will be a focus on differentiating between the constructs (e.g. attitudes versus values) as a framework of analysis, in order to relate studies in the area to each other and help those studies that are seen as contradictory to 'talk' to each other. This is argued to be a way in which we can better synthesise the diversity of studies in the Social Sciences. The structure of the article will be; a brief discussion of approaches and understandings of the constructs in question, followed by a discussion of some common explanatory approaches in the area. This will be followed by a discussion of trends in stereotypes, and a discussion of main effects and trends in attitudes and values.

Methods

2.1 This article and project make use of tools both from systematic review methods and realist synthesis to take into account the complexity and multilevel nature of the area studied. To achieve a comprehensive review and understanding of dynamics in support for welfare and immigration, the research design takes the form of a review striving to follow the transparent and orderly characteristics of a systematic review in the data collection and early analysis stages of the project, as per the definition provided by Bradshaw (2000). This meant that a comprehensive search strategy, inclusion and exclusion criteria were set prior to starting the review in line with traditional systematic review methodology. A pilot was carried out and inclusion criteria and search strategies were amended as a result of the pilot as well as in collaboration with staff from the Evidence Network, and welfare attitude scholars from the disciplines of Economics, History, Politics, Sociology, Criminology, Anthropology, Social Policy and Psychology. The latter was done to appreciate the heterogeneity of the field and to minimise bias. Both inclusion criteria and search strategies were designed to appreciate the diversity of studies in the area. Furthermore, the Social Sciences have been argued to be particularly prone to a changing terminology in the Social Sciences (Gomersall and Grayson 2003). In this study this was controlled for by consulting a variety of academics when constructing the search strategies and inclusion criteria. Longer and more detailed inclusion criteria would have been another strategy, but it would have been problematic. For one, it would have required the knowledge only acquired by a review to be detailed enough, it would also have been very time consuming as well as difficult to use. Furthermore it would not be sufficiently flexible to appreciate the heterogeneity of the studies in the area.

2.2 In the screening and data extraction stages, due to the vast amount of hits and diversity of studies (a total of 458,791 items), alternative, searches were carried out within the hits by use of the original search strategies (see the Appendices for databases searched, inclusion criteria, search strategies and no. of included, excluded and duplicate studies). To establish the best way of doing this it was necessary to work closely with the staff at the EPPI center to determine the best strategies to carry out the review. In accordance with them it was decided to use strategic searches within the review itself to identify the most relevant studies by using the same search terms as those used for the database searches. This article focusing on results from the research synthesis of studies in the area of immigration and welfare support, illustrates alternative approaches to research synthesis in the Social Sciences and value the variety of methods and findings of the studies included. As mentioned the approach used includes elements from realist synthesis, to achieve a comprehensive review of 'what works' that fully appreciates the heterogeneity found in the Social Sciences.

2.3 The method used in the project the article reports on emphasise the importance of the principles of inclusiveness, heterogeneity, iteration and judgment if one is to carry out a successful review in the Social Sciences. This is because studies within the Social Sciences are characterised by a vast diversity of methods used and areas studied, which means that an emphasis on homogeneity could mean that one misses out on important knowledge about and understanding the issue at hand. During the process of the project an approach making use of tools from systematic review as well as some of the principles of realist synthesis was developed. The latter as it enables and values iteration, value the heterogeneity found in the field as well as emphasising the understanding of the explanatory mechanisms at play. Realist Synthesis emphasise explanations of 'what works in what circumstances?' through comparing and contrasting findings and theories, rather than using a cumulative additive approach of 'what works' that does not take into consideration context and underlying mechanisms. This is deemed important as a way of appreciating context as well as a way of creating greater understanding of what explanations are found to be confirmed when synthesising findings. Here, the synthesis is carried out through use of a theoretical framework, which will be outlined below, to which studies' findings are added, creating a roadmap with patterns outlining which explanations are found to be confirmed and which are not. Through this approach the article demonstrates how a framework differentiating effects by constructs is found to be better suited to explain tendencies in welfare state support related to changes in immigration and ethnic diversity than a traditional systematic review approach.

2.4 One of the main issues making a traditional systematic review approach difficult in the Social Sciences is the heterogeneity of studies in the area. In the research synthesis, the focus here, this makes it problematic to calculate effect sizes due to use of different methods, independent variables and outcomes. Rather than only embarking on a narrative exploration of the overview, the article will give examples of translations of data building on ideas from Onwuegbuzie (2003). He describes approaches to alternative ways of calculating and synthesising effects in systematic reviews of qualitative studies. These enable different studies to be compared and related to each other through giving value to the direction of the effect of a variable, e.g. gender's effect on redistribution, by giving a study a value of 1, 0 or -1. Although there are suggestions in the literature on how to calculate effect sizes based on standardised betas taken from multiple regression models (Aloe 2009) standardised betas are not commonly part of multilevel models (often used in attitude studies) nor always presented in the literature in this area, thus making use of these techniques almost impossible. It is argued that the approach described by Onwuegbuzie (2003) enables further synthesising that can be used to create percentages and even used in factor analysis (Onwuegbuzie 2003). Due to space and time limitations, I have not included the latter in this article. One issue with this type of directional value is that one loses the ability of comparing the strength of effects, something that needs to be critically discussed and addressed in a systematic review. A further issue in synthesising evidence in the Social Sciences, which may be even more problematic than in other disciplines is the importance of national and cultural factors in impacting social meaning.

2.5 The approach taken here is in line with other critics of systematic review as it is critical of the role of replication, judgement, hierarchy of evidence and predicated research processes. The selection of studies is guided by the recognition of the importance of heterogeneity at every step of the research process, to ensure inclusion of all relevant knowledge in the area. This inclusive approach has enabled me to identify issues such as limitations of available data and surveys, issues around operationalization and lack of theorisation, and a lack of discussion of how support is measured in the relevant literature. It is important at this point to highlight that these are but a few examples of the weaknesses and strengths of the method. The following section will first give an overview of approaches to immigration and welfare state support, and then outline the different constructs used (e.g. attitudes versus values). The latter will be used as a framework for the research synthesis that follows.

Immigration and welfare support

3.1 Immigration is one of the key social changes bearing social, political and economic issues for many countries. I will now provide an overview of some of the common approaches to analysis of the relationship between immigration and support for the welfare state, these are examples of expected effects and approaches to the analysis of immigration and support for the welfare state rather than being a full literature review as the analysis also provides an overview of the field.

3.2 The school of thought dominated by American scholars underlines the importance of the degree of ethnic homogeneity when explaining why the population in some countries supports welfare policy while in others it is more reluctant (Larsen 2008; Alesina & Glaeser 2003). Their argument is that ethnic diversity has a negative effect on the generosity of welfare as well as citizens' support for the welfare state. Building on and related to these types of negative effects, some sociological explanations often emphasise that increased immigration in Europe may potentially weaken the social ties and feelings of a common identity that laid the ground for encompassing welfare arrangements (Van Oorschot 2008). Thus, based on this theory we would expect the majority of studies to find a negative relationship between increased immigration and welfare support in Europe and other parts of the world when they experience increased ethnic diversity.

3.3 Ethnic heterogeneity, exemplified by authors such as Alesina and Glaeser (2003) and others based on the American experience, is one of several theories aiming to explain relations between immigration and support for the welfare state. Amongst other mechanisms that are seen to explain a relation between immigration and a potentially declining support for the welfare state in countries that experience increased immigration are: multiculturalism, costs of immigration and competition over welfare, welfare state chauvinism (a nationalistic belief in keeping benefits to 'our own'), differences in work ethic between immigrants and the native population; institutional change and the emergence of an ethnic 'underclass'"(Goul Andersen 2006: 1). Another school of thought emphasises institutional structures and claims that "attitudes toward welfare policy can be (partly) explained by cross-national differences in the institutional structure of the different welfare regimes" (Larsen 2008: 146). Furthermore, some argue that there is a link between the rise of right-wing politics and immigration, which can reinforce 'us' versus 'them' resentment (Van Oorschot 2008). Others argue that left-wing politics "substantially counteracts the impact of greater diversity on the European social model" (Taylor-Gooby (2005) quote from Van Oorschot 2008:5) and some argue that a comprehensive welfare state forms a barrier against diversity-based retrenchment. Based on previous studies, I expect the research synthesis to show that an increase in immigration will lead to a decrease in support for the welfare state, as Van Oorschot (2008) found that "the informal solidarity Europeans feel towards migrants is low, when compared with solidarity shown towards other vulnerable groups" (Van Oorschot 2008: 12).

3.4 However, alternative theories argue that welfare state arrangements work as a barrier and may ward off possible negative consequences. Furthermore, negative effects may be countered or not found to be the case at all if one takes into account the ethnic diversity and multiculturalism at the point of establishing a welfare state; "Once built, institutions are resistant to change. Institutions have a strong impact on perceptions, norms and values. Thus it would seem likely that reactions to immigration are different in different welfare states" (Goul Andersen 2004: 6). Like Goul Andersen (2004), I am sceptical to a mechanical use of the American experience as a predictor of European trends. This is because different parts of Europe has seen a long process of diversification and break with tradition (Goul Andersen 2004: 7), even if these have not been as racially fragmented as in the USA, and changes have been absorbed by the European society rather than resulting in stronger cleavages and decreased support for the welfare state. As a result, a competing hypothesis, contact theory, would be that immigration rates have "no influence on relative solidarity towards migrants, and even that a higher rate of foreign-born citizens goes together with higher relative solidarity" (Van Oorschot 2008: 12). What follows will analyse these trends by focusing on trends in the different constructs separately to assess whether different effects are found on support for the welfare state.

Constructs- Attitudes, values, perceptions and stereotypes

4.1 In the literature, scholars tend to focus on attitudes, values, stereotypes and perceptions when analysing support. Here, I explore to what extent trends in these may vary by construct used (i.e. values versus attitudes), and use it as a framework for research synthesis. This enables a theoretical framework in the synthesis of the studies, which appreciates the variety of the studies included and gives importance to explanations in the research synthesis stage of systematic reviews in the Social Sciences. The constructs are important in relation to welfare states on many accounts; one may argue that attitudes to welfare and changes in these may impact both welfare institutions themselves as well as social interactions amongst various groups in society. The characteristics and properties of the constructs in question here are rarely discussed at much length in the literature focussing on support for the welfare state, and this article aims to start addressing the issues related to this limitation. Attitudes, perceptions, stereotypes and values all form part of what is included in studies done in the areas of comparative support for the welfare state (Pettersen 1995). This furthers the importance of assessing how and to what extent we can observe similar trends across the constructs and how they may be related to each other.

4.2 We can define attitudes in the following way; ". . . an attitude is more generally seen as a disposition to respond in a favourable or unfavourable way to given objects" (Oskamp 1991: 9). When differentiating between attitudes, beliefs and values, Oskamp argues that beliefs are cognitive ideas about the attitude object, whereas attitudes may be seen as affective dimension towards an object and again values may be seen to represent important life- goals or societal conditions (1991: 9). Furthermore, we can expect perceptions and stereotypes to inform attitudes and values and be related to beliefs, as they all are related to normative beliefs and opinions about particular objects. However, stereotypes can be argued to be specific to an image of a group rather than a reaction; for as Molden & Dweck (2006: 199) note, "At their core, stereotypes are a constellation of traits and abilities that are ascribed to anyone who is a member of a particular social group (e.g., librarians are typically viewed as being quiet, reserved, and conservative)". As Svallfors (2006) and Oskamp (1991), this article defines attitudes as reactions to and beliefs towards an object, in this case the welfare state and its parts (Davidov et al. 2008: 584). These are different from values, which can be defined as "the belief that some end-state or mode of conduct is preferable over another end-state mode of conduct" (ibid). Values are, in line with this definition, something more fundamental and deeper seated than attitudes.

4.3 Haller (2002) differentiated attitudes and values by developing a framework focussing on levels of generality. Placing values and attitudes on different levels of generality enables us to understand and analyse these better. This emphasises the importance of a more sophisticated theorisation of attitudes, a position shared by this project. The three different levels, 'Universal values', 'Societal' values and 'Situational' value orientations (Haller 2002) operate at different levels of abstractness and change more/less easily. Firstly, "universal values are those very basic human values that are known in any literate civilization. . . We can think here of values like equality and freedom, justice and respect" (Haller 2002: 143). 'Societal values or value orientations' are more concrete values tied to specific social contexts and populations e.g. equality which value has changed "both in content (equality of opportunity, equality of treatment of men, etc.) and in coverage (equality between estates, between men and women. . .)" (ibid). The latter can be used in cross-national studies such as this, and is distinguished by measures emphasising underlying values that are tied to social context such as views on inequality, a measure used in support for the welfare state literature. The differentiation of attitudes and values also enables us to analyse changes over time, persistence of trends as well as consolidation of societal changes. These may all be important to take into account as values ('Societal' values) change slowly and are deeply ingrained in institutions and countries whilst attitudes ('Situational' values) may change more rapidly. Thus, the research synthesis in this article separates attitudes ('Situational values') from values ('Societal values') to explore and discuss differences between these. The following will discuss the results of the research synthesis, starting with a discussion of findings in regards to stereotypes, then a section synthesising studies focusing on attitudes and values as a measure of welfare support.

Impact of immigration in changing welfare support

Ethnic heterogeneity & stereotypes

5.1 The main argument tested in many of the European focused articles and books in the area of immigration and support for the welfare state is the 'ethnic heterogeneity hypothesis', which has been dominated by American scholars (Larsen 2008). Alesina and Glaeser (2003) are amongst the proponents of this theory; they find that:
"a less generous welfare state is in part due to the fact that the white majority does not want to redistribute in a way that would favor racial minorities. Racial differences between rich and poor facilitated the propagation of views such as 'all poor are lazy' precisely because racist views associate laziness with different skin colors" (Alesina & Glaeser 2003: 218).

5.2 From the above it is clear that stereotypes are an important part of the theory expecting a negative impact of increased immigration. Gilens (2000) demonstrates that this can be explained by American uniqueness where the racial divides and stereotypes fall along the same lines as poverty. As Alesina and Glaeser (2003), Gilens (2000) argues that negative attitudes to the poor in the USA are driven by the stereotype of African-Americans as lazy. Although Gilens (2000) supports the ethnic heterogeneity argument, he underlines that it is "the perception that most people currently receiving welfare are undeserving" (Gilens 2000: 174) that is the main driver for the lack of support for welfare in the USA. Thus, the basis for the ethnic heterogeneity thesis is built on stereotypes and perceptions of poor as lazy. How has this been studied then in work analysing the phenomena in Europe? The following section will show how studies on Europe have tended to focus on attitudes and values, rather than stereotypes, when analysing the topic. In one of the very few studies analysing stereotypes in Europe, Larsen (2011) finds that the negative views and stereotypes of African-Americans found in the USA are also found in the UK, Sweden and Denmark, and thus the American experience may not be seen as unique. Unfortunately the review did not include any further studies focusing on stereotypes comparatively as no studies fulfilling the inclusion criteria were identified. The following will discuss the differences in findings depending on whether studies focus on attitudes or values.

Impact of immigration in changing welfare support - attitudes and values

5.3 How then might differentiating between attitudes and values help explain differences in impact of immigration on support for the welfare state? The research synthesis is based on 13 studies identified in the larger project, and is used to illustrate differences across attitudes and values as well as to give an overview of trends in the relationship between increased immigration and welfare support. Overall in the research synthesis, I find some support for contact theory, arguing that increased contact between ethnicities occurring as a result of more diverse societies will have a positive rather than a negative effect on support. This theory appears to explain positive effects of immigration on concern for the living condition of needy groups on a macro-level. On the individual level self-interest is found to play an important role across individual level variables. Furthermore, compared to the findings from the wider project that analyses the relationship of other macro-level changes such as increased unemployment socialisation is found to be important when we analyse the impact of increased immigration. This is based on the finding that education has a positive impact across studies analysing the impact of immigration on welfare support. Thus, when it comes to mediating the relationship between immigration and support for the welfare state, we can expect that different moderating and mediating effects across the macro and individual level depending on what welfare pressure is analysed. This adds evidence in favour of a 'political man', rather than a homo economicus or homo sociologicus. This is because citizens' self-interest and socialisation are shaped by the institutional framework they live within, and there is also evidence that suggest that these factors explanatory power change as citizens respond to institutional and societal changes.

Table 1. Directional Effects of key variables on welfare support

5.4 Making use of the approach adapted from Onwuegbuzie (2003), Table 1 displays the directional effects (1 for positive, -1 for negative or 0 for no effect) found to be statistically significant within and across the 13 studies that focus on attitudes and values. As can be seen from the Table 1 all but two studies include macro-level measures of changes in national immigration levels in their models. I assume that redistribution is a value, despite the fact that some studies' measurements of redistribution focus on specific objects. Therefore it may also be considered an attitude, although the measurements normally include elements of both (values and attitudes) thus underlining the lack of clear distinction between the two in the literature. For the individual level variables, empty cells mean that the variable was not included in a study. The individual level variables in Table 1 were chosen as they were the most commonly used variables across the included studies. Other variables were also included in some studies, however the article emphasises the variables most studies have in common as the aim is to illustrate trends across studies and how one may synthesise these. As a result I have included attitudes to immigrants, age, education, and gender. When it comes to other issues around measuring different types of immigrants, i.e. difficulties of measuring illegal immigration, as well as their relative distance to majority populations I am bound by the measures used in the included study. Thus, I will only be able to discuss the included variables. Moreover, the effects reported are effects controlling for the other variables included.

5.5 Higher levels of immigration are found to have a positive and socialisation effect on concern for immigrants relative to other needy groups such as the elderly in the studies making use of this outcome measure, which can be seen as an attitude as it is related to a specific group. This supports contact theory, rather than theories based on the USA experience. However, it has weakly negative effects on values such as redistribution and inequality. Thus, the choice of construct (attitude or value) to measure support for the welfare state matters. The most important variable found to have an effect in studies that use several measures of immigration is 'shares of foreign born' (% foreign born population relative to the total population). Although sometimes attitudinal such as when measuring attitudes towards redistribution towards certain group, redistribution is closer to a value construct as it measures views of inequality in the society. Thus, we find different effects of immigration when using a measure that includes values, than when we measure attitudes, here measured as concern for immigrants relative to other needy groups. Thus, it is important to pay attention to whether or not this may change over time as it is argued that values change much less rapidly than attitudes. In other words, it is important to understand whether the slightly negative effect of 'immigration' on 'redistribution' in current studies may change in the future, as many citizens begin to have more contact with immigrants.

5.6 An overall and strong finding of this synthesis is that negative attitudes towards immigrants are negatively related to support for the welfare state. This is an important finding for European attitude research as negative out-group perceptions and stereotypes of other ethnicities are a fundamental part of the American experience (Alesina and Glaeser 2003) as seen in the discussion above. Importantly, we also see how these negative attitudes, consistent across studies, may impact reported voting behaviour, through policy-bundling effects according to Verneby and Finseraas (2010). Going back to the political saliency of immigration this is important to have in mind when and if politicians 'play the race card' in political discussions. In addition, Koning (2011) finds that welfare chauvinism does not necessarily correlate with immigration levels. These findings shed some light on whether higher levels of immigrants lead to more or less negative attitudes towards immigrants. However, he does not carry out any multivariate analysis of the relationship thus limiting the ability to generalise about the effects beyond the bivariate findings of the study. Furthermore, Senik et al. (2008), studying differences between natives and immigrants' attitudes, testing welfare magnet arguments, finds that whilst some may expect immigrants to be more positive towards the welfare state, immigrants are slightly more conservative than natives but adapt within one generation. In other words, there is no evidence of immigrants being more positive towards the welfare state than the native population. This goes against the USA case, where the black population, in line with self-interest explanations, is found to be more positive than the white to the welfare state.

5.7 The studies included have not covered the relationship between immigration and negative attitudes towards immigrants in any depth and this is an important area to expand on in future studies. Crepaz (2008) studies the impact of multicultural policies on support for the welfare state and finds a positive relationship. This is an additional element to that studied by most in the area as he includes immigration policies, which is an area that merits further investigation. However, one needs to take in account which countries have multicultural policies and consider whether these already have higher levels of support for the welfare state (e.g. Sweden).

5.8 So far, I have argued that the analysis of a range of studies enabled by systematic review tools may help researchers identify tendencies that can be further tested in future research. Whilst one could argue that these findings may be found by use of a traditional literature review, I argue that traditional literature reviews do not analyse nor identify their arguments through a systematic search of available evidence nor do they synthesise these in a systematic manner. In addition, traditional literature reviews do not tend to embark on systematic synthesis of previous evidence that lead to comprehensive analysis of directional effects found within and between study findings. Furthermore, the combination of findings enables researchers to start putting together a type of highway rule set as Pawson (2006) argues. Thus, the findings emphasise explanations and diversity in the research synthesis. This has the aim of informing and strengthening the explanatory power of the different theories'. The findings suggest that whilst higher immigration may lead to lower support for redistribution, it can also lead to higher concern for the living conditions of immigrants. In other words there is no clear evidence supporting one theory only, but different theories are found to be important depending on whether or not we measure welfare support as an attitude or a value. On the other hand, negative sentiments towards immigrants are consistently found to impact both measures of welfare support and voting suggesting that attitudes and values are related to each other and with behaviour.

5.9 In terms of the adaptive approach used it is found that synthesising the evidence by giving an overview of directional effects may help us appreciate the heterogeneity of the studies whilst at the same time synthesising the effects across them beyond what a common literature review would be able to do. However, a weakness is that it is difficult to take it much further than this without running into the same issues around loss of heterogeneity. Furthermore, the lack of findings and studies that analyse changes over time means that this aspect cannot be explored further. Looking at the publication dates of the included studies, immigration appears to be a recent focus for scholars in the area, and it will be interesting to gain further understanding of changes over time in future studies.

5.10 I now turn my attention to the effect of more typical socio-economic and demographic variables at the individual level on support for the welfare state. I will focus on three key variables that most studies include; age, gender and education. The effects reported are effects when controlling for the other included variables. These individual variables are also important in relation to the various other welfare pressures, the ageing population as well as changing family structures and labour markets. Women are consistently found to be more negative than men when studies control for either (or both) attitudes towards immigrants/immigration and levels of immigration. This can be explained by self-interest as women often use welfare services and benefits more than men and thus may conceive of immigrants as more of a threat. On the other hand, age is found to be positively related to welfare support even when controlled for immigration variables (macro and micro), and the two studies that find them to be negative, Verneby and Finseraas (2010) and Van Oorschot and Uunk (2007), do so due to their research design and operationalization. Van Oorschot and Uunk's (2007) dependent variable is concern with immigrants relative to the elderly, which may explain why the elderly may be more concerned with their own group. Verneby and Finseraas (2010) operationalize age as 1= older than 45 whilst 0= younger. The latter is done as age is claimed to have a curvilinear effect, something which should be further researched.

5.11 As argued in the section on macro-level variables it is important to take into account mediating and moderating effects. One finding that underlines this is the fact that controlling for immigration levels is related to negative effects of being female. In addition it shows that there are different effects of different types of welfare pressures. In other words, higher levels of immigration may mean that women are more negative towards the welfare state; however, the impact of an ageing population is different. It is important to be able to make these observations, paying attention to the multi-dimensionality and complexity of the area, and a review of this type enables these types of analyses. It furthermore emphasises the need to give attention to mediation and relationships between variables at macro and micro level in analyses of support for the welfare state. This again shows the use of systematic review in the synthesis as one should put greater emphasis on operationalization and comparability of the included studies, something that is not part of a traditional literature review. Evidence suggests that both socialisations as well as self-interest explanations are significant. A socialisation effect is found with education in the area of immigration as it is positively related to welfare support when controlled for immigration and/or attitudes to immigration/immigrants. On the other hand findings strengthen self-interest explanations as individuals are found to be more negative towards when controlling for immigration attitudes and levels.

Table 2. Findings & Constructs

5.12 Table 2 positions the studies by use of the theoretical framework outlined earlier in the article differentiating between attitudes and values. Thus, showing how theory may be used to drive research synthesis in a systematic review in the Social Sciences. It is interesting to observe that immigration levels seem to have a positive and socialisation effect at the level of attitudes, which according to Haller (2002) may change quicker than the more ingrained values. This may suggest that the positive effect observed at the attitude level may not have crystallised at the value level yet. Thus, based on this synthesis it may be argued that the weak negative impact of immigration levels on redistribution can become positive if the socialisation effect impacts values over time. The latter is crucial when one aims to understand whether the ethnic heterogeneity hypothesis may occur in Europe. Furthermore, by way of synthesising evidence we also find that the macro-level variables may have socialisation effects through welfare regimes mediating the effect of immigration when redistribution is the outcome. This may give evidence of the more ingrained values represented in cultures and embodied in institutions that serve as filters for rapid changes. Overall, this is an example of how systematic review approaches may help identifying mediating variables difficult to identify and deal with in a single study through its provision of an overview of a wider range of evidence.

5.13 Self-interest is found to play a role if higher immigration is combined with worsened socio-economic conditions in a country (as scarcity impacts negatively when measured by GDP and unemployment levels as in Mau and Burkhardt 2009). In other words, a combination of changes at macro level with worsening GDP and unemployment combined with immigration may mean that positive effects such as socialisation effects explained by contact theory can have less explanatory effects as scarcity and self-interest come into play. It is important for studies to further explore this in their models by commenting on mediation and relationship between variables so as to further our understanding in this area. This is particularly important given our current global situation where economic and political crises are combined with increased migration. The identification of mediating variables both at macro and micro level is crucial if one is to understand a phenomenon more fully, and particularly to inform future studies.

5.14 The different effects of macro and micro level variables also underline how macro level measures of institutions such as welfare regimes may capture values in and of itself which can be said to be a result of culture and thus be related to socialisation effects. On the other hand, we have seen the dominance of self-interest on the micro-level, most notably in age and gender. However, attitudes, here measured by concern for immigrants, are impacted positively through what could be explained by socialisation effects of higher immigration levels. Thus it is not so that socialisation necessarily is more important to explain tendencies on one level of generality of values than the other. This underlines the importance of flexible and dynamic theoretical approaches in the area of welfare support that can incorporate different dynamics and explanations that appreciate the differences between constructs. It also highlights the need to understand mediation and moderation across macro and individual level variables. An appreciation of this can be found amongst some newer theorists in the areas that emphasise that citizens are socialised into but then adapt to their institutional settings. Thus, we see a more dynamic approach that goes further than self-interested or socialised support to a theory where moral economies of reciprocity are created and where both approaches are interrelated. As a result we get a more complex picture where individuals' support for the welfare state is related not only to the regime-dependent institutional reality, norms and self-interest but also to their position as 'embedded individuals' influenced by experiences and norms in family relations and social networks.

Conclusion

6.1 In summary, this article makes both methodological and substantive contributions. Firstly, it underlines the strengths and weaknesses of systematic reviews in the synthesising process. Although meta-analysis is made difficult by the range of different methodological and theoretical approaches and by difficulties in synthesising across the heterogeneity of studies in the Social Sciences, a systematic review synthesis framed and shaped by theoretical frameworks has been found to help researchers identify what values/attitudes are impacted by what, why and in what circumstances. Secondly, it has highlighted important variables and concepts that might provide a more nuanced understanding of the ethnic heterogeneity hypothesis. This systematisation may enable a researcher to organize evidence in new manners that may shed new light on mechanisms at play in a particular area.

6.2 To illustrate the importance of differentiating between different constructs, I carried out an in-depth analysis of the relationship between immigration and ethnic diversity on support for the welfare state. There are practical, methodologically principled and synthesis related issues that arise when conducting a systematic review in the area of support for the welfare state which are also argued to be relevant for the Social Sciences in general. Thus, the article suggests fruitful ways of approaching synthesis of research in this area by use of adapted systematic review tools. Systematic reviews may serve the purpose of giving a systematic, comprehensive review of evidence in areas in need of this. Furthermore, mapping exercises of current knowledge may help create bases for further development of theory as well as helping researchers to get a more comprehensive overview of the evidence of what dynamics are at play. Moreover, it has been argued that theory driven research synthesis is important and that iteration, refutation and heterogeneity should be valued in a systematic review conducted in the Social Sciences.

6.3 In terms of findings about trends in support for the welfare state, we find that the American experience is not replicated in Europe when measuring support as attitudes and values. Although there is some support for this explanation in terms of the effect on redistribution, it is argued that one needs to take into account differences in effects on different constructs. Thus, whilst there is some evidence for a negative effect of higher levels of immigration on support for redistribution (a value), the opposite is found for attitudes, 'concern for immigrants'. This is another benefit of synthesising evidence by use of systematic review tools. It is argued that there is a need for further critical discussion of operationalization in studies in this area. This is further supported by differences in effect by measure. Theoretically, contact theory is supported at the situational level, through concern for needy groups in society, whilst self-interest explanations are supported when we synthesise effects on the individual level, when looking at gender and macro-level and the effects of GDP and unemployment rates. Education is consistently found to have a socialisation effect on citizens' support for welfare state when increasing immigration is taken into account, which is not so in relation other pressures such as increased economic scarcity. It has also been argued and shown that these findings imply the importance of, and need to further explore, the relationship between independent variables emphasising moderation, mediation and confounding factors. Overall, the evidence confirms a dynamic theoretical approach emphasising the need to take into account the interplay between institutional context and individuals' attitudes, thus placing the article alongside a school of thought arguing that individuals attitudes adapt to changing institutional and societal context. Consequently, theoretical explanations need to allow the possibility that self-interest changes as context does, e.g. what is in an individual's self-interest when it comes to increased immigration may change when we allow for the impact of the economic crises.

Appendix 1 Databases searched

Appendix 2 Inclusion criteria of project the article is based on

  1. All studies need to be related to support for the welfare state and thus measured as attitudes, or beliefs, values, perceptions and stereotypes as discussed.
  2. As discussed in the literature review support can be studied as:
    1. Support for redistribution, basic income, and equality, provide jobs to all, support for equity, need or equality.
    2. Support for spending on needy/poor, on the welfare state and on specific programmes.
    3. Support for the welfare state in general and specific programs taken to represent the welfare state (policy areas/specific benefits and services).
    1. Range and degree of welfare state legitimacy; "range regards the issue of whether government should or should not take up welfare responsibilities and in what range of policy areas. Degree concerns the issue of how much government should spend on welfare provision" (Van Oorschot, 2010).
  3. All studies should deal with attitudes related to welfare pressures.
  4. All studies need to cover the area of Europe and be about one or more European countries. Europe is defined as the EU plus Lichtenstein, Norway and Switzerland.
    1. Studies covering USA, Canada, Australia and New Zealand will only be included if they also cover a European country as defined in this project.
    2. If studies cover a European country and other areas they will only be included if these other areas are USA, Canada, Australia and New Zealand.
  5. All studies should be cross-national, thus need to include two or more countries
  6. Studies to have been published/written/presented from 1994 to end of 2010.
  7. Only studies written in English will be included because:
    1. The reviewer will not be able to read all other languages.
    2. The reviewer would only be able to include studies written in a few languages in the European regions, which would bias the study towards those countries.
    3. Studies written in English assumed to be the most influential in international literature
  8. Unpublished/'Grey' literature and published literature included
  9. Qualitative cross-national and quantitative studies will be included, however, purely descriptive studies will not be included

Appendix 3 Search strategies

Following the guidelines from the Centre for Reviews and Dissemination latest guidelines (2009) for doing systematic reviews the search terms were guided by the main objective and the PICOS (population; intervention; comparator; outcomes; study design). This guided the development of search terms i.e. Population: studies about attitudes; Intervention: studies using data from social surveys; Comparator: other countries, other attitudes, attitudes at other points in time; Outcome: findings about tendencies; Study design: all but descriptive. BOOLEAN strategies for searching were used. To achieve a comprehensive review despite the frequent changes that tend to occur within the Social Sciences the study consulted a variety of academics when constructing the search strategies. To inform the search terms the reviewer got input from researchers within the areas of Economics, History, Politics, Sociology, Criminology, Anthropology, Social Policy and Psychology at the University of Kent as well as from authors of the studies included in a pilot study to ensure that the search terms encompass the widest possible range of attitudes and to minimise bias.

Full search strategies can be downloaded here: http://trudesundberg.com/2014/01/28/search-strategies/

Appendix 4 Total inclusions and exclusions in main project

Total number of studies taken into account:31,670
Included in 1st screening:746
Included in 2nd screening:84
Total exclusions:30,387
Duplicates 537

Notes

1This is related to another finding of the main project that will not be discussed in this article, which underlines the need for critical discussions around operationalization of immigration and ethnic diversity, as these are two different concepts. [0]

References

ALESINA, A. and E., Glaeser (2003). Fighting poverty in the US and Europe. Oxford: Oxford University Press.

ALOE, A. (2009). A Partial Effect Size for the Synthesis of Multiple Regression Models. PhD Dissertation, Florida State University.

BRADSHAW, J. R., Baldwin, S. et al. (2000), Methodologies for socially useful systematic reviews in social policy, York: University of York.

CREPAZ, M. (2006). Chapter 3 'If you are my brother, I may give you a dime!' Public opinion on multiculturalism, trust, and the welfare state. In: Kymlicka Will, Banting Keith Multiculturalism and the Welfare State: Recognition and Redistribution in Contemporary Democracies. Oxford: Oxford University Press.

CREPAZ, M. (2008). Trust beyond borders: immigration, the welfare state, and identity in modern societies. Ann Arbor MI: University of Michigan Press.

CRUTCHFIELD, R. and Pettinicchio, D. (2009). 'Cultures of inequality': ethnicity, immigration, social welfare, and imprisonment. Annals of the American Academy of Political and Social Science. Vol. 623, No. 1, p. 134-147.

DANCYGIER, R and Saunders, E. (2006). A new electorate? Comparing preferences and partisanship between immigrants and natives. American Journal of Political Science. Vol. 50, No. 4, p. 962-981.

DAVIDOV, E, Meuleman, B, and J. Billiet (2008). Values and Support for Immigration: A Cross-Coutnry Comparison. European Sociological Review, Vol. 24, No. 5, p. 583-599.

FINSERAAS, H, (2008). Immigration and preferences for redistribution: an empirical analysis of European survey data. Comparative European politics, Vol. 6, No. 4, p. 407-431.

GILENS, M. (2000) Why Americans Hate Welfare: Race, Media and the Politics of Anti-poverty Policy. Chicago: University Chicago Press.

GOMERSALL, L. and Grayson, A. (2003) A difficult business: Finding the evidence for social science reviews.  Working Paper 19, London: ESRC UK Centre for Evidence Based Policy and Practice.  

GOUL ANDERSEN, J. (2004) Immigration and the Legitimacy of the Scandinavian Welfare State. Keynote paper form 13th Nordic Migration Conference, Aalborg/AMID 18-20 November 2004.

GOUL ANDERSEN, J. (2006) Immigration and the Legitimacy of the Scandinavian Welfare State: Some Preliminary Danish Findings. AMID Working Paper Series 53/2006

HALLER, M. (2002), Theory and Method in the Comparative Study of Values: Critique and Alternative to Inglehart, European Sociological Review, Vol. 18, No. 2, p. 139-158. [doi://dx.doi.org/10.1093/esr/18.2.139]

KONING, E. (2011). The Real and Perceived Economics of Immigration. Paper presented at 2011 Canadian Political Science Association conference. Waterloo, Ontario, Canada.

LARSEN, C. A. (2006). The institutional logic of welfare attitudes: how welfare regimes influence public support. Aldershot: Ashgate.

LARSEN, C. A. (2008). The Institutional Logic of Welfare Attitudes. Comparative Political Studies. Vol. 41, No. 2, p. 145- 168. [doi://dx.doi.org/10.1177/0010414006295234]

LARSEN, C. A. (2011). Ethnic Heterogeneity and Public Support for Welfare: Is the American Experience Replicated in Britain, Sweden and Denmark? Scandinavian Political Studies, Vol. 34, No. 4, p. 332-353. [doi://dx.doi.org/10.1111/j.1467-9477.2011.00276.x]

ONWUEGBUZIE, A.J. (2003). Effect Sizes in Qualitative Research: A Prolegomenon. Quality & Quantity, Vol. 37, p. 393-409. [doi://dx.doi.org/10.1023/A:1027379223537]

OSKAMP, S. (1991) Attitudes and Opinions 2nd edition. New Jersey: Prentice-Hall Inc.

MAU, S. and Burkhardt, C. (2009). Migration and welfare state solidarity in western Europe. Journal of European social policy. Vol. 19, No. 3, p. 213-229. [doi://dx.doi.org/10.1177/0958928709104737]

MOLDEN, D. C. and Dweck, C.S. (2006). Finding 'Meaning' in Psychology. American Psychologist, Vol. 61, No. 3, p. 192-203. [doi://dx.doi.org/10.1037/0003-066X.61.3.192]

PAWSON, R. (2006). Digging for Nuggets: How Bad Research Can Yield Good Evidence. International Journal of Social Research Methodology, Vol. 9, No. 2, p. 127-142. [doi://dx.doi.org/10.1080/13645570600595314]

PETTERSEN, P. A. (1995). Chapter 8. The Welfare State: The Security Dimension. In: Borre, O. and Scarbrough, E. The Scope of Government. Oxford: Oxford University Press.

SENIK, C.; Stichnoth, H. and Straeten K. (2009). Immigration and natives' attitudes towards the welfare state: evidence from the European social survey. Social Indicators Research. Vol. 91, No. 3, p. 345-370. [doi://dx.doi.org/10.1007/s11205-008-9342-4]

SVALLFORS, S. (2006). The Moral Economy of Class: Class And Attitudes in Comparative Perspective. Stanford: Stanford University Press.

TAYLOR-GOOBY, P. (2005) Is the Future American? Can left politics preserve European welfare states from erosion through growing 'racial' diversity? Journal of Social Policy, Vol. 34, No. 4, p. 661-672. [doi://dx.doi.org/10.1017/S0047279405009207]

VAN OORSCHOT , W. (2006). Making the difference in social Europe: deservingness perceptions among citizens of European welfare states. Journal of European Social Policy. Vol. 16, No. 1, p. 23-42. [doi://dx.doi.org/10.1177/0958928706059829]

VAN OORSCHOT, W. (2008). Solidarity towards immigrants in European welfare states. International journal of Social Welfare, Vol. 17, No. 1, p. 3-14. [doi://dx.doi.org/10.1111/j.1468-2397.2007.00487.x]

VAN OORSCHOT, W and W, Uunk (2007a) Welfare spending and the public's concern for immigrants -Multilevel evidence for eighteen European countries. Comparative politics Vol. 40, No. 1, p. 63-+. [doi://dx.doi.org/10.5129/001041507X12911361134433]

VAN OORSCHOT, W and W, Uunk (2007b) Multi-level determinants of the Public's Informal Solidarity towards Immigrants in European Welfare States [Internet site] Available: http://arno.uvt.nl/show.cgi?fid=94175 Accessed 18 April 2011.

© Sociological Research Online, 1996-2014

UniS: University of Surrey logo University of Stirling logo British Sociological Association logo Sage Publications logo Electronic Libraries Programme logo Epress logo