European Journal of Political Culture

ISSN 2784 - 0271 ISSN – L 2784 – 0271

Volume 1, Issue No.1 (March 2021), pp.28-35

Published: 30 March 2021

Some Theoretical Reflections on

Modernisation, Value Change, Cultural Change and Democracy in a Multilevel Perspective

Hermann Dülmer

University of Cologne

Germany

hduelmer@uni-koeln.de

Abstract

The aim of this contribution is to reflect on the relationship between the theoretical model for causal multilevel explanations of Colman and multilevel modelling in empirical social research. The example that has been chosen is Inglehart’s theory of modernisation, but Welzel’s theory of human empowerment is also covered briefly. The usual graphical representation of conventional multilevel models will first be elaborated using a proposal made by Muthén and Muthén. Since Inglehart’s value dimension of survival versus self-expression values is measured via factor analysis, conventional multilevel analysis has to be elaborated to a multilevel structural equation model. Within this broader framework it also becomes possible to include a macro-level variable as dependent variable in the multilevel model. The reflections on the relationship between human development and democracy might hopefully stimulate further reflections on multilevel models and encourage researchers to use advanced multilevel techniques for their own research in order to gain deeper insight into the relationship between modernisation and democracy.

Keywords: multilevel modelling, modernisation, personal and cultural values, democracy, Coleman

A prevailing aim of quantitative social science research is to explain individual behaviour or individual values and attitudes which are assumed to have an impact on individual behaviour. Factors which are used for such explanations are frequently not only personal characteristics of the individual but also characteristics of the broader social context where the individual grew up and where he or she is living. The principal task of social science, however, according to Coleman (1990: 2), is the explanation of the behaviour of social systems which are located at the macro level and not the explanation of micro-level behaviour of single individuals. Within the framework of Coleman’s (1990: 5, 11) variant of methodological individualism, system behaviour derives from actions of actors who are elements of the system. This implies that a system level proposition located at the macro level breaks into three explanations that mediate the relationship between two macro-level variables:

“one with an independent variable characterizing the society and a dependent variable characterizing the individual; a second with both independent and dependent variables characterizing the individual; and a third with the independent variable characterizing the individual and the dependent variable characterizing the society.”

(Coleman, 1990: 8)

By including this mediation in the theoretical model, the macro-level relationship should be explained and for this reason disappear as a genuine macro-level relationship (cf. also Snijders & Bosker, 2012: 12). Figure 1 displays the structure of the so-called bathtub model of causal multilevel explanations.

Coleman’s bathtub model, however, differs in several aspects from conventional multilevel analysis (Pötschke, 2020: 500). Multilevel models typically neither include the macro-micro relationship represented in Figure 1 by the arrow from Z to X, nor the micro-macro relationship from Y to W. Furthermore, the macro-level relationship between Z and W is usually not included in a multilevel model but is estimated instead via aggregate-level analysis.

A typical graphical representation of a conventional multilevel model as it can be found in many introductions to multilevel analysis (Snijders & Bosker, 2012: 10-12, Pötschke, 2020: 500) is displayed in Figure 2. The arrow from X to Y depicts a micro-level main effect from X on Y, the arrow from Z to Y depicts a main effect from Z to Y, and the arrow from Z on the arrow from X to Y depicts a cross-level interaction effect (moderation effect) from Z on the relationship between X and Y. Hence, differences in the relationship between X and Y are explained at least partially by Z.

A more elaborated representation of causal relationships in multilevel analysis has been proposed by Muthén and Muthén (1998-2017: 276). One advantage of their visualisation is that it distinguishes more clearly between the location of micro-level (within) components and macro-level (between) components of multilevel models. In accordance with structural equation modelling, observed variables are indicated by a rectangle and latent (unobserved) variables by an ellipse. The dependent Y-variable in multilevel analysis is a composite variable consisting of a within and a between variance. Hence, the observed part of Y is located at the micro level, whereas the estimated latent between variance (i.e., the variance of the means of the single macro units) is placed correctly at the macro level. The arrow from Z to Y (main effect from Z on Y) in Figure 3 shows unambiguously that a between variable can contribute only to explain the between level variance of Y. Whether the intercept and/or the slope of X is estimated with a between level variance component (random component u) is indicated in the micro-level part of the model by a filled circle. The filled circle at the end of the arrow from X to Y represents a random intercept that is referred to as the latent variable Y in the macro-level part of the model; the filled circle on the arrow from X to Y represents a random slope that is referred to as X_u in the macro-level part of the model. The latent variable X_u represents the macro-level variability of the slopes around their grand mean. This variability can be reduced (i.e., at least partially explained) by Z. Therefore, the arrow from Z to X_u indicates a cross-level interaction effect from Z on the relationship between X and Y. The arrow on the observed within variable Y represents the residual variance of the intercept of Y; the arrows on the two latent variables Y and X_u represent the respective random component of the intercept and the slope. The covariance between the random components is indicated by an arrow with two heads.

Prominent examples of multilevel theories within the field of political culture research are Inglehart’s (1997; Inglehart & Welzel, 2005) theory of modernisation and Welzel’s (2013) theory of human empowerment. Inglehart’s (2008: 131) theory of value change starts at the macro level with the observation that the “remarkable economic growth that occurred during the era following World War II, together with the rise of the welfare state, brought fundamentally new conditions in advanced industrial societies”. This historically unprecedented change in the macro-level conditions (Z) of advanced industrial societies caused a gradual shift at the micro level from materialistic to postmaterialistic value priorities (Y) in younger cohorts (X) born after World War II. This theoretical assumption is based on a scarcity hypothesis which states that an “individual’s priorities reflect the socioeconomic environment: One places the greatest subjective value on those things that are in relatively short supply” and a socialisation hypothesis which states that “one’s basic values reflect conditions that prevailed during one’s preadult years” (Inglehart 1990: 68, cf. also 1997: 33 and 132).

Whereas the scarcity hypothesis implies that throughout their lives people retain a certain capability to adapt their value priorities to short-time changes in their environment (period effects), the socialisation hypothesis assumes that the basic value priorities of an individual are internalised during their formative years and remain quite stable for their whole life (cohort effects). Therefore, value change is expected to be characterised by short-time period effects that are superimposed on long-term cohort effects reflecting the conditions that prevailed while a cohort was growing up (cf. Inglehart, 1990: 79-82, 1997: 34). The direction of value change is deduced by loosely referring to Maslow’s (1954; cf. Inglehart, 1977: 22-23) hierarchy of needs. Although he does not follow Maslow’s needs hierarchy in detail, Inglehart (1997: 33; 1990: 134) adopts a basic distinction between basic materialist needs for economic and physical security, and higher order postmaterialist needs such as esteem and self-expression. The latter value orientations are not assumed to take priority until the former are taken for granted.

Basically, multilevel theories can be tested empirically for many countries via cross-sectional data from one point in time, or for one country via repeated cross-sectional or via panel data. The macro level consists in the first case of countries, in the second case of time and in the last case of respondents. For panel data time is embedded in the personal context of respondents and for this reason is modelled at the micro level. For a strict test of causal relationships panel data are ideal.[1] Since panel data are rarely available for a bigger number of countries, researchers interested in cross-cultural comparisons usually use cross-sectional data for conducting multilevel analysis. In this case one has to rely on the assumption that cross-level differences between macro units (countries) follow the same causal mechanisms as the cross-time differences within a macro unit (temporal sequence as a necessary condition for causality). For testing Inglehart’s theory via cross-sectional data, the macro-level variance of the latent variable Y (cf. Figure 3) consists in differences in the average of materialist-postmaterialist value orientations across countries. In multilevel analysis, an attempt is usually made to reduce this variability by regressing (the latent macro-level variable) Y on Z. The observed variable Z might be the Gross National Product per capita (GNP pc) as an indicator for a society’s socioeconomic development (cf., for instance, Inglehart, 1997: 150-51) or the Human Development Index (HDI) as a broader indicator for modernisation (cf., for instance, Dülmer et al., 2015).[2]

A cross-level interaction effect can be expected if the countries in the sample data are rather heterogeneous regarding experienced change in their levels of security as measured by Z (GNP pc or the HDI). Inter-cohort value change toward postmaterialist value priorities should, according to Inglehart (1997: 46-47, 143), be bigger in countries where the socioeconomic conditions have strongly improved over time than in countries with lower or even no socioeconomic development. If the most developed countries (Z) have also experienced the strongest socioeconomic growth, then the biggest inter-cohort differences regarding their value orientations can be expected in highly developed post-industrial countries. Hence, the impact of cohort membership (X) on individual value orientations (Y) should depend on Z. Trying to model Coleman’s relationship between Z and X (more precisely, the macro-level variance of X), however, does not make sense, since value priorities of cohorts but not cohort membership itself depend on macro-level conditions. Cohorts in our case represent all conditions that had an impact on the individual socialisation process.

In Inglehart’s (1997; Inglehart & Welzel, 2005: 52) later work the polarisation between materialist and postmaterialist values became a central component of a broader latent dimension of value shift from survival to self-expression values. The latent dimension is measured via factor analysis. Since socioeconomic development is linked with changing values at both the country level and the individual level (Inglehart & Welzel, 2005: 61), these values and their determinants can be analysed at both levels simultaneously via multilevel structural equation modelling (cf. Dülmer et al., 2015). Such models, however, are only identified when a large number of countries can be included. At the micro level they are also faced with the problem of measurement invariance (cf., for instance, Meitinger et al., 2020), a problem that has most recently been critically addressed by Welzel et al. (2021).

In order to cover Coleman’s bathtub model more completely, one can easily extend the multilevel structural equation model by including an indicator for W (observed dependent between-level variable). Examples for such indicators are different aspects of democracy. The link between socioeconomic development, self-expression values and the emergence, survival and strength of democracy is explained by Inglehart and Welzel (2005: 151-157, 171): socioeconomic development tends to make people more independent and nurtures a sense of existential security and autonomy; a growing sense of existential autonomy leads people to give priority to self-expression values that emphasize emancipation and autonomy over authority; rising self-expression values lead people to demand institutions that allow them to act according to their own choices and motivate them, accordingly, to seek the civil and political rights that define liberal democracy. Although this argumentation includes micro-level propositions, existing macro-level variance can be modelled in multilevel analysis only by indicators that cover macro-level differences.

Whereas Inglehart’s (1997: 49, 51; cf. also Welzel, 2013: 61) selection of indicators for the survival versus self-expression dimension is based on exploratory factor analysis, Welzel (2013: 59, 66-67) derived the indicators for his more narrowly defined emancipative values from theoretical considerations. Although the theoretical explanations of Inglehart and Welzel differ, both approaches can be used for explaining the link between human development and democracy. This provides the opportunity to conduct an empirical theory comparison.

The aim of this article was to reflect on the relationship between human development and democracy. This might hopefully not only stimulate deeper reflections on multilevel models but might also encourage researchers to use more advanced multilevel techniques for their own research. In this way this article might contribute to stimulate further scientific progress that fosters deeper insight in the area of political culture.

References

Coleman, James S. (1990). Foundations of Social Theory. Cambridge, The Belknap Press of Harvard University.

Dülmer, Hermann, Ronald Inglehart, and Christian Welzel (2015). Testing the Revised Theory of Modernization: Measurement and Explanatory Aspects. World Values Research 8 (2), 68-100.

Inglehart, Ronald (2008). Changing Values among Western Publics from 1970 to 2006. West European Politics 31 (1-2), 130-146.

Inglehart, Ronald (1997). Modernization and Postmodernization. Cultural, Economic, and Political Change in 43 Societies. Princeton: Princeton University Press.

Inglehart, Ronald (1990). Culture Shift in Advanced Industrial Society. Princeton: Princeton University Press.

Inglehart, Ronald (1977). The Silent Revolution. Changing Values and Political Styles among Western Publics. Princeton: Princeton University Press.

Inglehart, Ronald and Christian Welzel (2005). Modernization, Cultural Change, and Democracy. The Human Development Sequence. Cambridge, Cambridge University Press.

Klein, Markus and Manuela Pötschke (2004). Die intra-individuelle Stabilität gesellschaftlicher Wertorientierungen. Eine Mehrebenenanalyse auf der Grundlage des sozio-ökonomischen Panels (SOEP). Kölner Zeitschrift für Soziologie und Sozialpsychologie 56 (3), 432-456.

Maslow, Abraham K. (1954). Motivation and Personality. New York, Harper and Row.

Meitinger, Katharina, Eldad Davidov, Peter Schmidt, and Michael Braun (2020). Measurement Invariance: Testing for It and Explaining Why It is Absent. Survey Research Methods 14 (4), 345-349.

Muthén, Linda K. and Bengt O. Muthén (1998-2017). MPlus. Statistical Analysis with Latent Variables. User’s Guide, 8^th edition. Los Angeles, Muthén & Muthén.

Pötschke, Manuela (2020). Mehrebenenmodelle. In: Claudius Wagemann, Achim Goerres, and Markus B. Siewert (eds.): Handbuch Methoden der Politikwissenschaft. Wiesbaden, Springer.

Snijders, Tom A. B. and Roel J. Bosker (2012). Multilevel Analysis. An Introduction to Basic and Advance Multilevel Modeling. 2^nd edition. Los Angeles, Sage.

Welzel, Christian (2013). Freedom Rising. Human Empowerment and the Quest for Emancipation. Cambridge: Cambridge University Press.

Welzel, Christian, Lennart Brunkert, Stefan Kruse, and Ronald F. Inglehart (2021, forthcoming). Non-Invariance? An Overstated Problem with Misconceived Causes. Sociological Methods & Research.

__________

[1] Klein and Pötschke (2004), for instance, used the West German data of the German Socioeconomic Panel for analysing the impact of age, cohort, and period effects on materialist and postmaterialist value orientations (measured on a 4-point scale ranging from pure materialists to materialist and postmaterialist mixed types to pure postmaterialists). In their panel analysis they found strong evidence for cohort effects. However, they also found evidence for non-linear life-cycle effects (cf. Klein & Pötschke, 2004: 451).

[2] In multilevel analysis with repeated cross-sectional data from one country, the variability of the latent variable Y (value priorities) might be caused by period effects that go back to different inflation rates Z (Inglehart’s materialist-postmaterialist index includes the item “fighting rising prices” which is sensitive to the inflation rate, cf. Inglehart, 1997: 133-137). Although the effect is modeled at the macro-level, it is mediated via the bathtub.

Corresponding Author:

Dr. Hermann Duelmer, Private Lecturer (Privatdozent), Institute of Sociology and Social Psychology, University of Cologne, Germany. Contact Address: hduelmer@uni-koeln.de

Copyright @ 2021, Hermann Dülmer

European Journal of Political Culture Vol.1(1):28-35.

Open Access

This article is distributed under the terms of the Creative Commons Attribution-Noncommercial-NoDerivatives 4.0 International License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.