Keywords: Political behavior, participation and culture

Opinion leaders may not be equally distributed throughout society, but primarily belong to the higher classes of society! Implying an elite model of opinion leadership, contradicting Lazarsfeld et al.’s (1944) study ‘The People’s Choice’, and underresearched ever since introduced by Katz and Lazarsfeld’s (1955) work on political opinion leadership in the USA, this idea has been subject to longstanding controversy. Many studies have touched upon social status and yielded inconclusive results (Weimann, 1994). As public opinion research moves towards more generic and empirically substantiated approaches, we lack a rigorous assessment of the elite model’s crossnational validity across Europe, its generalizability across topic areas other than politics, and the mechanisms causing the stratification of opinion leadership. Our research undertakes the first theoretical and empirical effort cumulatively devoted to these gaps. We complement seven decades of opinion leader research to advance our understanding of whether and why opinion leaders may concentrate in high status levels.
Our empirical case builds on three complementary survey studies. Study 1 establishes the elite model’s crossnational applicability by submitting general population data from the European Union’s 28 membership states (N = 27,801) to multilevel analysis (Hox, 2010). Study 2 investigates the elite model’s validity across 25 topic areas. We analyze single source respondent-level data from Germany (N = 23,854) using structural equation modeling (Hancock & Mueller, 2006). In continuation of Studies 1 and 2, Study 3 exemplifies for which reasons opinion leaders may primarily belong to higher strata. We submit general population data on economic and financial opinion leadership in Germany (N = 10,100) to mediation analyses (Hayes, 2009). All measures are in line with prior studies, but using large-scale representative data sets and advanced statistics yields major gains in precision and power. Using multiple data sets casts additional light into the elite model’s robustness across samples and time (Loken & Gelman, 2017).
We find that the elite model validly contributes towards a generic understanding of opinion leadership. Traditionally observed stratification of American political opinion leaders readily generalizes across Europe. The same applies to topic areas involving similar civic interests as politics (e.g. economics, culture, science), especially when compared to consumer and household matters. This finding expands McCombs and Poindexter’s (1983) conceptualization of citizens getting involved into problems facing society rather than merely looking into the narrower settings of voting campaigns and their everyday lives. Stratified opinion leadership is decisively driven by the factors commonly named, i.e. elite’s print and online media use, social networks, and educational levels. However, fully explaining stratified opinion leadership requires updating prior explanations against elite’s subjective knowledge and self-confidence (Trepte & Scherer, 2010). Moreover, findings yield stability in the stratification of opinion leadership, both temporally and across samples. Findings are discussed against 1) mixed results and limitations of prior studies, 2) the major changes of the media landscape since Katz and Lazarsfeld (1955), and 3) the increasing advent of anti-elite attitudes in many countries (Aalberg et al., 2016).