ABSTRACT
BACKGROUND: This study examined the psychometric value of the Center for Epidemiologic Studies Depression Scale (CES-D) translated for use with an Armenian population. METHODS: Using data obtained from a country-wide health survey of 2310 households involving female and male respondents aged 18 and over, we investigated the response pattern to the CES-D items, the factor structure, internal consistency, inter-item correlations of the total scale and its negatively and positively formulated subscales. We used logistic regression analysis to relate the constructs measured by the CES-D and its subscales to known determinants of depression. RESULTS: Armenian respondents of both genders significantly suppressed their positive emotions, thus over-endorsing positively formulated (reverse-coded) items, therefore producing artificially high depression scores. Factor analysis of the scale yielded a three-factor structure (combined Depressed/Somatic, Positive Affect, and Interpersonal). The Positive Affect factor correlated weakly with the other two factors, and its inclusion reduced the internal consistency of the whole scale. Unlike the 16-item subscale of negatively formulated items, Positive Affect was not related to several known determinants of depression and did not reflect known depression-specific differences between genders. The set of determinants of Positive Affect included mainly lifestyle and attitudinal variables. LIMITATIONS: This study did not assess the concurrent and discriminate validity of the Armenian CES-D. CONCLUSIONS: For Armenians, the construct measured by the four Positive Affect items of CES-D is not related to depressive symptoms as measured by the other items. It introduces ethnical/cultural response bias in CES-D score and reduces the cross-cultural comparability of the latter.