The construct validity of the Restraint Scale among mainland Chinese women

• The Chinese version of Restraint Scale has good factor structure among Chinese females.
• It also has acceptable internal consistency and satisfactory convergent reliability.
• This scale can identify restraint eaters and evaluate the level of restraint eating.

This study aimed to test the construct validity of the newly developed Chinese version of the Revised Restraint Scale (RRS) in two separate female samples from Chinese universities (n1 = 171; n2 = 158). The RRS is the most widely used measure of restricted eating for the purpose of achieving or maintaining a desired weight. Results showed that the 10-item Chinese RRS comprised of two subscales of concern dieting (CD) and weight fluctuation (WF), which together accounted for 53.01% of the total variance. Confirmative factor analysis supported this two-factor model as an acceptable model (χ2/df = 1.545, p < 0.001; RMSEA = 0.059, GFI = 0.938, and CFI = 0.935). Cronbach's alpha coefficients were 0.77 for RRS, 0.71 for CD and 0.66 for WF. Significant correlations were shown between RRS and its subscales (rRRS-CD = 0.894, p < 0.01; rRRS-WF = 0.850, p < 0.01; rCD-WF = 0.523, p < 0.01). Moreover, RS significantly correlated with the Three-Factor Eating Questionnaire-R (TFEQ-R), the Dutch Eating Behaviour Questionnaire-R (DEBQ-R), the Fatness Concern (FC) and the body measure index (BMI) (r = 0.548, p < 0.01; r = 0.631, p < 0.01; r = 0.620, p < 0.01; r = 0.351, p < 0.01, respectively). In sum, the RS has stable factor structure, acceptable internal consistency and satisfactory convergent reliability among female Chinese college students. The adapted scale may serve as a useful tool for identifying restrained and unrestrained eaters among Chinese women and may be particularly useful in assessing eating restraint before and after prevention and intervention programs.