Improving the psychometric properties of the dissociative experiences scale (DES-II): a Rasch validation study

Abstract

Background: The Dissociative Experiences Scale-II (DES-II) is a self-report questionnaire that measures dissociative experiences such as derealization, depersonalization, absorption and amnesia. The DES-II has been prevalently used as a screening tool in patients suffering from psychotic disorders or schizophrenia. However, dissociative experiences can also be part of normal psychological life. Despite its popularity, the most problematic aspect of the DES-II is the inconsistency in its factor structure, which is probably due to the tendency to treat ordinal responses as responses on an interval scale, as it is assumed in the Classical Test Theory approach. In order to address issues related to the inconsistency of previous results, the aim of the present study was to collect new psychometric evidence to improve the properties of the DES-II using Rasch analysis, i.e. analyzing the functioning of the response scale. Methods: Data were obtained on a sample composed by 320 Italian participants (122 inmates and 198 communitydwelling individuals) and were analyzed with the Rasch model. This model allows the estimation of participants' level of dissociation, the degree of misfit of each item, the reliability of each item, and their measurement invariance. Moreover, Rasch estimation allows to determine the best response scale, in terms of response modalities number and their discriminant power. Results: Three items of the scale had strong misfit. After their deletion, the resulting scale was composed by 25 items, which had low levels of misfit and high reliability, and showed measurement invariance. Participants tended to select more often lower categories of the response scale. Conclusions: Results provided new knowledge on the DES-II structure and its psychometric properties, contributing to the understanding and measurement of the dissociation construct.

The Indonesian version of the Life Orientation Test-Revised (LOT-R): Psychometric properties based on the Rasch model

“…These findings demonstrated that the psychometric characteristics of the Indonesian LOT-R are an important addition for the Indonesian researchers. In line with a previous research, and from a practical perspective, the findings from the Rasch analysis may allow psychologists and non-specialists to confidently interpret sum scores as good indicators of individuals' optimism (e.g., Saggino et al, 2020). This is also in line with the original scoring method proposed by the LOT-R developers (see Scheier et al, 1994).…”

“…Moreover, while LOT-R is theoretically unidimensional and has a bipolar construct (e.g., Chiesi et al, 2013;Steca et al, 2015) that fulfils the unidimensionality requirement of the Rasch measurement (Wright & Masters, 1982), the debate over its dimensionality remains unsolved (Schou-Bredal et al, 2017). Unidimensional measures, as per the Rasch model, are more appropriate for statistical analyses because the differences between participants' scores are interval-scaled, and the total score is an adequate representation of the construct measured by the scale used (Saggino et al, 2020). Therefore, in this study, we hypothesized a unidimensional model of the Indonesian LOT-R.…”

“…It implies that once the Rasch factor has been extracted, that is, the main scale, there should be no leftover patterns in the residuals (Tennant & Conaghan, 2007). In this research, the standardized residual correlation was used to test this assumption, with its value for each item pair being > 0.30 in a positive direction, indicating violations of local independence (Saggino et al, 2020).…”

“…The expected value for the two was 1.0, with acceptable value ranges of 0.5 to 1.5, meaning that those outside the items did not fit to the Rasch model (Linacre, 2018). Furthermore, the point-measure correlation (PT-measure) value had to be observed to understand the compatibility aspect between the model and the data, with negative values indicating that the item was not functioning adequately (Saggino et al, 2020).…”

“…This allowed overcoming the classical test theory's (CTT's) limitations and provided a different perspective from that of a factor analysis (Ewing et al, 2005). This resulted in person and item parameters being on the same metric of logits for the estimation of ability and item difficulty, permitting a total score use for non-specialists (Saggino et al, 2020). Thus far, no research has been carried out on the evaluation of the LOT-R using the Rasch model.…”