Tuesday 16th July
Wednesday 17th July
Thursday 18th July
Friday 19th July
Download the conference book
Combining cross-nation and longitudinal perspectives in substantial social research |
|
| Convenor | Professor Jaak Billiet (KU Leuven) |
This session relates to the submissions that are encourages for ESRA 2013: methods for cross-national data analysis; longitudinal surveys; substantive topics.
It is inspired by past workshops and seminars within the context of QMSS_2 in the 2008-2012 period, in particular the program of research group 443 on Cross-nation comparisons.
This session will be focused on conceptual and methodological problem of overtime analysis in comparative research in the context of substantive studies. While there is a real interest in applying an overtime perspective among cross-national survey researchers, the combination of cross-national comparison with longitudinal analysis is an emerging rather than an established field. Against this background, the prime aim of the proposed session would be to discuss issues and tools in overtime analysis related to substantive studies (research examples) wherein best practice is shown.
In view of the scarcity of cross-nationally comparable panel data in social science research, we define overtime analysis in the broadest possible way. Specifically, in addition to the analysis of cross-national panel data, which is still rather rare, we would include different kinds of data and analysis under "overtime analysis": the analysis of repeated cross-sections to answer theoretical questions about trends or change over time in a comparative way, e.g. European Social Survey, European Election Study, etc.; macro-level time series data; combined multi-time and multi-level analysis; repeated measures with multi-level data...
Apart from technical and statistical issues in the context of substantive studies, theoretical and conceptual papers are also welcomed in which design aspects and the theoretical validity of variables at different levels are discussed.
Literature:
Billiet, J. (in press). Quantitative Methods with survey data in comparative research. In: Patricia Kenneth (Ed.). A Handbook of Comparative Social Policy. (3d edition, in press)
The median-voter hypothesis claims a positive relationship between demand for redistribution and economic inequality. Many economists take this relationship as given. Empirical evidence for this relationship, however, has been rather ambiguous in previous research. The major shortcoming of previous research is that it is either purely cross-sectional or longitudinal but based on aggregated data. This paper uses an innovative method to overcome this shortage. A multilevel-hybrid-model for pooled cross-section data is used to estimate cross-sectional (between-country) and longitudinal (within-country) effects of economic inequality, while simultaneously controlling for individual-level compositional effects. The model, furthermore, allows estimating cross-level interactions. The analysis is based on data from the European Social Survey (2002-2010). The results suggest a strong and robust within-country effect of inequality on demand for redistribution, supporting the assumption of the median-voter hypothesis. This relationship is not found between countries. From an econometric point of view, this points to an unobserved variable at the country-level. The paper considers welfare regimes as a possible factor capturing these unobserved country-differences. However, none of the existing welfare regime typologies performs well, neither in terms of capturing unobserved heterogeneity, nor in general terms of explanatory power. All in all, the paper founds solid support for the proposition that demand for redistribution is positively related to economic inequality, but it casts general doubt on the values of cross-sectional analysis and the welfare regime approach.
Focusing on macro-level processes, this article combines Decennial Census and Current Population Survey data to simultaneously investigate intra-group and inter-group patterns in immigrant intermarriage using structural and cultural explanations. Covering a 130 year period, the results of our multilevel analysis for 140 national-origin groups indicate that structural characteristics are important when investigating the question why some origin groups become more "open" over time while others remain relatively "closed". Accordingly, the likelihood of intimate relationships bridging group boundaries is more likely to increase over time when the relative size of an immigrant group decreases, sex ratios grow more imbalanced, the origin group grows more diverse and social structural consolidation decreases. Cultural explanations also play a role suggesting that an origin group's exogamous behavior in the past exerts long-term effects and exogamous practices increase over time when the prevalence of early marriage customs declines. From a cross-sectional perspective, the findings support prior research suggesting that the likelihood of exogamy increase with declining group size and with more imbalanced sex ratios.
In many European countries the debate over secularization and religious change is far from having reached unanimous conclusions. In Italy, for example, some scholars underline an unbroken trend toward a decrease of religiosity, while other scholars highlight the signs of a religious revival especially in younger generations.
The aim of the paper is to define the religious change in church attendance in European countries for a period as long as possible. The analytical strategy is to consider more surveys together in order to widen the observation window and to get more reliable estimates. Many surveys are suitable for the analysis so it is necessary to apply some selection criteria. Two criteria are going to be applied: surveys should be repeated - since we are interested in aggregate change - and they should come from non-commercial institutes given the open access to the data. Applying these criteria and taking into account the comparative constraint, we will analysis religious change in up to 15 EU countries using the following five studies: Eurobarometer, European social survey (Ess), European values study (Evs), World values survey (Wvs), International Social Survey Programme (Issp).
The joint analysis of many surveys gives also the opportunity to compare different data and evaluate their reliability. The question of convergence of results will be addressed, in levels of church attendance and their development over time. To evaluate systematically time change and survey effects a multilevel analysis is developed, where individuals are nested in surveys administered over different time points.
While anti-immigration attitudes in Europe have been investigated quite extensively, research that focuses on the evolution of attitudes towards immigration remains scarce (Semyonov, Raijman, and Gorodzeisky, 2006; Meuleman, Davidov & Billiet, 2009). Existing studies show that, since the mid-90s, evolutions in anti-immigration sentiment are not monotonous and vary widely across European countries. Furthermore, observed trends can be interpreted in terms of realistic group conflict theory: Unemployment rates and immigration flows partly explain the cross-country variation in attitude trends (Meuleman, Davidov & Billiet, 2009).
In this paper, we use the current economic crisis as a natural quasi-experiment to revisit realistic group conflict theory. We compare short-term trends (2006 - 2008 - 2010) in anti-immigration attitudes across 22 countries, using European Social Survey data. To analyse this cross-national repeated cross-section data, we develop a multi-level model in which countries are considered as the highest level of analysis. Varying trends are modelled by including time as a fixed effect with random slopes.
The analyses indicate a general decrease in anti-immigration attitudes between 2006 and 2008 as well as a sharp increase between 2008 and 2010, although trends vary considerably across countries. Interestingly enough, the mechanisms driving attitude changes are not stable across time. While labour markets changes provide a solid explanation for the evolution between 2006 and 2008, this no longer seems to be the case in times of deep economic crisis (changes between 2008 and 2010). Possible explanations for this remarkable finding are discussed.
The trust among people is considered as a key factor related with social capital in a country. The extent which people trust on others has been shown in some studies is relevant (Levi, 1998:174; Hardin, 1992; Sztompka, 1995, Zmerli et al., 2007).
Social (generalized) trust is measured among Spanish and Norwegian citizens during five time periods from 2002 to 2010.
Social trust is a complex construct and will be treated as a latent variable using Confirmatory Factor Analysis (CFA). European Social Survey (ESS) data will be used to test whether the typical trust indicators hold across all periods and for both countries.
In this paper we will analyze to what extent trust can be comparable as a construct in different countries with significantly different levels of social trust. In order to study its comparison, tests for measurement invariance of CFA Social Trust will be used. Latent means will also be studied across time and countries.
Also, the paper aims to analyze and compare the effects of key variables for social trust in different contexts, with different levels of social trust. The main working hypothesis in this paper will be that there is a trend or constant over time for the predictive factors of social trust. We will use Structural Equation Modeling methodology in order to analyze and compare the effect of key correlates for trust, such as satisfaction with democracy, satisfaction with economy, satisfaction with life, age, gender, education.
Measurement invariance for complex concepts (constructs), such as social trust, is required when cross-national comparisons are of interest in order to assure a proper equivalence across countries. It is also possible to use measurement invariance for a specific country across time, if invariance holds constructs have the same meaning across time and adequate trend analysis of means for each time period can be reported.
Since data, European Social Survey (ESS) for instance, are available for several waves, comparison for both countries and across time can be a good combination in order to study the equivalence of the same concept across time and countries together in a single model.
This combination can be carried out using multiple-group confirmatory factor analysis (MGCFA) in order to test measurement invariance. Configural (structure), metric (relationships) and scalar (latent means) invariance will be tested on the latent variable social trust. In the case invariance holds, comparisons across groups can be correctly interpreted.
This situation requires large number of groups; since groups used are some ESS participant countries and time (5 ESS waves from 2002 to 2010, in this case).
Finally, consequences of inaccurate measures for comparisons will be reported. Latent means (which can be compared across groups when scalar invariance holds) are compared with traditional sum score means. Advantages of using measurement invariance will be emphasized.