Do we have to combine Values in the Schwartz' Human Values Scale? A Comment on the Davidov Studies

  • Desiree Knoppen
  • Willem Saris
Keywords: Schwartz Human Values scale, ESS, discriminant validity, factor model, misspecifications


This paper addresses the question if it is meaningful to group basic values from the values model of Schwartz, which is a vital component of the European Social Survey (ESS), and presents an alternative approach. The suggestion to group values is raised in several studies led by Eldad Davidov, which more precisely conclude that only four of the original ten values can be studied as such (Hedonism, Stimulation, Self-direction, and Security), whereas the remaining six values have to be grouped in three pairs in order to solve the problem of a lack of discriminant validity. This paper indicates that the grouping was necessary because of the choice strategy of items for the ESS. The items chosen for the different values are chosen in such a way that the correlation within the value is relatively low and sometimes the correlations with items of other values is higher. We show this in three steps: the first one based on a German-study, looking for the correlation between the values in the original Portrait Values Questionnaire (PVQ). Secondly, the ESS selection from this wider set of items is used to show that this choice leads to much higher correlations between the values. Thirdly, an analysis with an alternative choice from the same PVQ set of items is done to show that this high correlation is not necessary. In this way we show that the high correlation between the values in the ESS is due to the selection of the items in the ESS and is not intrinsic to the values studied.