Developing a Korean Version of the Experiences in Close Relationships Scale Using Quantitative Methods for Verifying Semantic Equivalence

The aim of this study was to verify semantic equivalence of a Korean version of ECRS using quantitative methods. Data were collected from 204 bilingual Koreans. The criterion sample included the 399 American college students surveyed by Mallinckrodt and Wang (2004). After translating the ECRS from English to Korean, five analyses for semantic equivalence were conducted: mean differences, DLSH reliability, internal reliability, test-retest reliability, and construct validity. First, the English items were equivalent to the Korean items except for the Avoidance subscale from Form B when comparing means differences. In addition, the English and Korean versions of each subscale showed good model fit except for the Avoidance subscale from Form B in the second administration. Second, regarding DLSH reliability, correlations in the bilingual sample between the English and Korean items were not equivalent to the correlations in the criterion sample between the English-only items. Third, ECRS-K exhibited high internal consistency. However, compared to the criterion sample, the internal reliability of the DLSH version for the Korean bilingual participants was not equivalent to the internal reliability of the English version for the American participants, except for the Anxiety subscale from Form A. Fourth, the bilingual sample showed appropriate temporal stability across a twoweek interval between the English and Korean subsets. Lastly, a measure of construct validity, Fear of Intimacy, demonstrated an expected positive relationship with the Avoidance subscale. Unexpectedly, Fear of Intimacy also exhibited a positive correlation with the Anxiety subscale. Furthermore, another measure of construct validity, Social Self-Efficacy, showed an unexpected positive correlation with the Anxiety subscale. Overall, ECRS-K has good internal reliability and retest reliability alone; however, ECRS-K was not parallel to the English version in terms of reliability and validity. Unexpected findings, their probable causes, and implications are discussed.