Welfare Attitudes Outline • Review of Welfare Policies • Discussion of Svallfors’ methodology and the relationship between attitudes and policies • My criticism of Svallfors’ Swedish article for the Swedish case and showing my alternative results • Discussion of the CR Methodological purpose • Review of Cronbach alfa and factor analysis • Review of regression • Brief introduction of Structural Equation Modelling • Showing that we can get different results using different methods and different definitions Some Basics on Welfare Policies • Liberal= means testing, selective, low levels of support, leads to stigmatization • Social democratic=universal, social citizenship, have right to benefits because of being a citizen • Conservative= conserving already existing hierarchies, favoring some groups over others, traditional family roles Dynamics • Which countries belong to each group? • Means-tested policies lead to stigmatization, so are less popular even within social democratic countries • Universal policies benefit the middle class so have great support even within liberal countries (national health service in the UK, social security in the USA) • It is more difficult to make cutbacks (i.e. Retrenchment) in conservative and social democratic countries than in liberal ones Individual level dynamics • With data bases can examine which groups of people are more likely to support welfare policies and which are more likely to oppose them • We can also create scales to measure support for welfare policies • Svallfors from Umeå in Sweden is one of the most important social scientists writing about this, so I naturally wrote an article criticizing him Svalfors measurement of welfare support Qustions about the scale • Why did Svallfors think it was better to use a scale than only one question? • Why did he create the same scale for all countries? • Why not just choose questions that one thinks measure welfare support rather than use statistical methods like Cronbach alfa? Interpretation • Three questions scaled well, so were included • Cronbach alfa used • .7 is a good score for Cronbach alfa, most countries slightly below, Austria not reliable • Norwegians most supportive (highest means) • Scale 1-6, where 6 is highest • Americans least supportive (lowest means) • Conservative countries as supportive as social democratic • Radical-liberal more positive than truly liberal • No relation to policies and attitudes (he discusses this more in other articles) • Thus, institutions more important than attitudes. Independent Variables • Class • Gender • Unemployed, Retired, employed, not in labor force (= Receiver of benefits ) • Working in the public sector (other articles) • Receiver of benefits (other articles) • Income (other articles) Interpreting the multiple regression • Gender significant for each country • What do the stars (*) mean? • How do you interpret this? • Class: some groups are significant, some are not • Problem in classification of class – cannot be single variable with this method • Why is class negative? • Which variables are not important? Interpreting R^2 • Not usual to write in % but rather .124 • What does it mean? • Is the explanatory value of these models strong or weak? Svallfors article in Swedish • one-dimensional scale • More independent variables • Mostly questions on responsibility • Concludes that differences in attitudes not related to policies (except for ”truly liberal USA) • Institutional differences more important Questions on government spending Interpreting government spending • Respondents from GB most positive • During Thatcher era they felt that spending should be increased • Lower starting point than in Sweden • Still does not say what kind of policies people would like (universal, selected, means-tested, etc.) • Outcome depends on starting point • Sweden in 1^st or 2^nd place except for eduction • In the USA long tradition of publicly financed public education. The first country with free public schools. Questions on government responsibility Intepreting government responsbility • Again the USA is the outlayer • No big difference among the other countries • Problems in interpreting ”responsibility” • The government could be ”responsible” for healthcare by providing it (national health service) or by regulating it (liberal alternative) or by providing a publicly supporting health insurance (conservative alternative) • Even in the USA high support for aid to the elderly (social security is a universal program) but low support to providing jobs for unemployed (selective program) Income equality Svallfors does not look at questions on equality, but provides another dimension • Sweden stands out more • One exception is cuts in government spending, in which GB is slightly higher because of Thatcher • The other exception is wage control, because in Sweden the unions are strong and do not want governmental intervention in the wage sector • But neither of these two questions scale well, so not included in the scale of supporting equality The importance of the equality dimension • People in other European countries also want generous social policies • But Swedes see equality in itself as a positive goal • Swedes are also willing to pay taxes for financing generous social policies • This cultural factor makes it more difficult to cutback on social programs in Sweden and makes it easier to gain support for financing expansions of social programs Svallfors model using SEM program Improvements on Svallfors with SEM • I create a two-dimensional model for measuring welfare support (using CONFIRMATORY factor analysis) • With CFA we begin with theory, with Exploratory we begin with data • Factor 1=big public sector • Factor 2=equality • By doing this the explained variance (R^2) increases from 14-16% to 24% (in the English article R^2 was only 3-12%) • The model is more parsimonious as gender is no longer significant and receiver can be eliminated as the model makes a closer fit without it • We see that women are more supportive of welfare policies because they are more likely to be employed in the public sector, not because they are inherently more leftwing or more likely to receive benefits • These are models 1 and 2 below The SEM Regression Table Adding Marx • Instead of the 7-scale Goldthorpe measurement, I used a 3-scale Marxian • Working Class, Professional, Capitalist • I included the class of each respondent’s partner • This almost doubles the explained variance • See model 3 Adding the connection between attitudes and voting Intepreting the addition of voting • Models 4 and 5 • We see that welfare attitudes are highly correlated with voting • Those who are support generous welfare policies are more likely to vote for leftist parties • With SEM we can see the connection between structural factors, attitudes and voting Comparing Sweden to the CR Intepretation • Class important in Sweden • In the CR classes were in flux, so people did not know their class interests • But income was very important • Social policies more important for Swedish voters than Czech • In CR people could support generous welfare policies and still vote for Klaus (charismatic) • Anti-communist feelings more important than social policies • Thus, many authors were wrong about Czech politics being class-based and revolving around economic issues Summary • Using factor analysis we see that welfare has an equality dimension • Suddenly culture matters: Swedes have a culture supporting equality and are willing to pay higher taxes to support welfare • People in other countries like benefits, but do not like paying for them • The Marxian definiation of class better explains attitudes • With SEM we can examin the relationship between attitudes and behavior • SEM also showed that working in the public sector explains women´s greater support for welfare policies • Comparing Sweden to the CR shows that class was not important as previous authors stated and parties do not necessarily compete based on economic issues