Zsuzsanna Suranyi

Department of Psychology

Karoli Gaspar University

Application of model based clustering in revealing and comparing national characters

  Zuckerman, Kuhlman, Joireman, Teta, and Kraft (1993) have proposed a five-factor model of personality with a psychobiological basis. Their questionnaire has recently been developed into a hierarchical structure of factor-facet version, Zuckerman-Kuhlman-Aluja Personality Questionnaire (ZKA-PQ, Aluja, 2010). This instrument contains five factors with four facets per factor: a) Aggressiveness (Physical Aggression, Verbal Aggression, Anger, and Hostility); b) Activity (Work Compulsion, General Activity, Restlessness, and Work Energy); c) Extroversion (Positive Emotions, Social Warmth, Exhibitionism, and Sociability); d) Neuroticism (Anxiety, Depression, Dependency, and Low Self-Esteem); and e) Sensation Seeking (Thrill and Adventure Seeking, Experience Seeking, Disinhibition, and Boredom Susceptibility/Impulsiveness, Aluja et al., 2010). The facets within one factor correlate with each other at a high level. The authors therefore say that factors should be considered as basic dimensions of personality. It is believed that if structures of personality traits show cross-cultural stability then they can be regarded as human universals.
  However, the invariance of the factorial structure does not mean that the typical patterns of values on the facets/ factors are also the same across cultures. Many cross-cultural studies have focused on comparing mean personality profiles as typical national characters. These descriptions of national characters provide a stereotypical view on nations; however, mean personality profiles do not necessarily apply to all individuals. Unfortunately there is no evidence that mean profiles are in congruence with national stereotypes. A common misinterpretation is identifying profile means as types and attributing the characteristics of these centers to all members of the nation. There might be more than one typical personality pattern within a nation and maybe the mean profile is not a typical one (just the mean of different personality profiles).
  This lecture aims to present the application of person-oriented methods (cluster analytic approach) in cross-cultural personality research. We can identify types of profiles with cluster analyses; however, classical algorithmic methods (such as K-means and hierarchical clustering) have been criticized for many methodological problems, such as finding an optimal cluster number and clustering of atypical individuals. Therefore I will discuss applications of a more appropriate method – model-based clustering – for this class of analyses. This approach formally models the data as coming from several subpopulations, via a finite mixture model (Fraley & Raftery, 2007).
  I have applied k-means clustering and model-based clustering in cross cultural studies with ZKA-PQ questionnaires. Model based clustering has proved to be a more adequate clustering method for revealing national types. Possible cultural contexts, clinical applications, methodological problems and the importance of interpretation of national types will be discussed.

Back to Colloquium Series