What Happens When Individuals Answer Questionnaires in Two Different Languages.

Clara Paz,Carlos Hermosa-Bosano,Chris Evans

doi:10.3389/fpsyg.2021.688397

Abstract

The aim of the present study was to compare scores from the English and the Spanish versions of two well-known measures of psychological distress using a within-subject approach. This method involved bilingual participants completing both measures in four conditions. For two groups of people, measures were offered in the same language both times and for the other two groups, each language version was offered, the order differing between the groups. The measures were the Clinical Outcomes in Routine Evaluation-Outcome Measure and the Schwartz Outcome Scale-10, both originally created in English and then translated to Spanish. In total, 109 bilingual participants (69.7% women) completed the measures in two occasions and were randomly allocated to the four conditions (English-English, English-Spanish, Spanish-English and Spanish-Spanish). Linear mixed effects models were performed to provide a formal null hypothesis test of the effect of language, order of completion and their interaction for each measure. The results indicate that for the total score of the Clinical Outcomes in Routine Evaluation-Outcome Measure just language had a significant effect, but no significant effects were found for completion order or the language by order interaction. For the Schwartz Outcome Scale-10 scores, none of these effects were statistically significant. This method offers some clear advantages over the more prevalent psychometric methods of testing score comparability across measure translations.

Highlights

Translation of psychological measures is a common practice worldwide and there is steadily growing interest in comparing findings across countries and languages
This study reports the use of the four group method with the Clinical Outcomes in Routine EvaluationOutcome Measure (CORE-OM; Evans et al, 2000, 2002) and the Schwartz Outcome Scale-10 (SOS-10; Blais et al, 1999), both originally created in English and translated to Spanish
Total scores of each measure were included as the dependent variables and gender, language, completion order and the interaction of language and order of completion were included as fixed effects, while the subjects were placed as random effects for each model

Summary

Introduction

Translation of psychological measures is a common practice worldwide and there is steadily growing interest in comparing findings across countries and languages. With two different models, these methods test the covariance of items across the individuals between the language samples This is highly appropriate for measures largely designed to compare individuals’ scores at a single completion, e.g., to determine school or university entry or to measure personality traits. This is tangential to the aims of measures of within individual change, i.e., to the typical aim of measures used in psychotherapy to assess change over time (Tarescavage and Ben-Porath, 2014). This situation creates a need for other approaches to explore the equivalence of measures across translations

Objectives

Results

Conclusion