Does Correction for Measurement Error Have an Effect on the Structure of Basic Human Values?

Abstract

It is well known fact that survey data contains always measurement errors, which can in different ways bias the results of the analyses (Alwin, 2007). In quantitative research this problem has been usually ignored due lack of quality estimates and complexity of correction procedure. Although this argument is not valid anymore, as there exists a new approach to determine quality of specific question, based on a prediction of the quality using the program SQP2, which in turn allows correcting for measurement error by following a relatively simple procedure (Saris & Revilla, 2015). During the recent years, there have been an increasing number of studies researching the structure of basic human values (Schwartz, 1992), which has resulted with improvements in the theory and led to a new 19-factor value model (Schwartz et al., 2012; Saris, Knoppen, & Schwartz, 2013). Although none of recent studies have considered the measurement error, which can potentially also have an effect on the structure. Therefor in this paper we will try to overcome this shortcoming and will be analysing the effect of the correction for measurement error on the structure of basic human values. We use same representative data from Estonia that was used in study of Lilleoja and Saris (2014), which enables to compare value structures before and after the correction for measurement error. Due to the existence of large ethnic minority in Estonian society, the sample allowed additionally testing the equivalence of the measurement in the two subpopulations after correcting for measurement error. This study shows that the correction for measurement error provides additional support for the validity of new value structure and it also shades some new light on the cross-cultural equivalence.

Keywords

• measurement error
• Schwartz human values scale
• equivalence
• PVQ-40