Psychometric properties of Eysenck personality questionnaire-revised short scale (EPQ-R/S) in three large Czech cohorts: people with epilepsy, people with neurological and other disorders and non-clinical adults.
Abstract
Purpose
The Czech version of the Eysenck Personality Questionnaire-Revised Short Scale (EPQ-R/S) has not been formally validated. This study examined its psychometric properties across two clinical cohorts and non-clinical subjects.
Methods
The total sample consisted of 2 560 participants: non-clinical adults (n = 382), patients with neurological and other disorders (PWND; n = 576), and patients with epilepsy (PWE; n = 1 602), aged 18-86 years. Internal consistency, item-level performance, and factorial validity were examined. Confirmatory factor analysis (CFA) and exploratory structural equation modelling (ESEM) with target rotation were performed.
Results
Extraversion (E) and Neuroticism (N) showed acceptable to good reliability across cohorts (α = 0.79-.85; ω = 0.89-.93). The Lie (L) scale showed lower but mostly acceptable reliability, whereas Psychoticism (P) showed weak item coherence and poor internal consistency, particularly in non-clinical adults and people with epilepsy. The theoretical four-factor CFA model showed poor fit, χ2(1074) = 15005.28, CFI = 0.643, TLI = 0.625, RMSEA = 0.076. A target-rotated ESEM solution was more interpretable but showed mixed fit, χ2(942) = 5198.66, TLI = 0.838, RMSEA = 0.042, RMSR = 0.030. A reduced two-factor CFA model for E and N showed a more coherent structure, χ2(251) = 3296.72, CFI = 0.900, TLI = 0.890, RMSEA = 0.072. Measurement invariance analyses across cohorts and gender supported threshold and loading invariance of this model based on changes in approximate fit indices. Women scored higher on N in both clinical subsamples, whereas no significant gender difference was found in the non-clinical sample. Among people with epilepsy, higher seizure frequency was associated with higher N, longer epilepsy duration with lower E, and monotherapy with lower N compared with polytherapy. N correlated strongly and positively with depressive symptoms and negatively with quality of life, supporting concurrent validity.
Conclusions
The Czech EPQ-R/S E and N scales showed adequate psychometric properties and appear suitable for research and clinical use in neurological and epilepsy populations.