A Revalidation of the Weight Related Behaviours Questionnaire within an Australian Pregnancy Cohort

Translation of the Weight-Related Behaviours Questionnaire into a Short-Form Psychosocial Assessment Tool for the Detection of Women at Risk of Excessive Gestational Weight Gain

“…Demographic and weight gain characteristics of the WATCH sample have been published previously [ 21 , 44 ]. Briefly, participants’ mean age was 28.9 years (SD 5.64), 71% had an education level at or above high school completion, 61% were married, and 55% were multiparous.…”

“…Exploratory factor analysis (EFA) with principal axis factoring and varimax rotation was performed for all WRB-Q items listed under the 6 psychosocial subscales, to examine their overall performance (i.e., construct validity and internal consistency) within the WATCH cohort. The results of this analysis have been reported elsewhere [ 44 ]. Briefly, the EFA conducted amongst the WATCH sample indicated that the weight locus of control, self-efficacy and body image scales demonstrated consistent construct validity, retaining the same item factor structure to the original analysis conducted by Kendall, Olson and Frongillo [ 27 ].…”

“…Briefly, the EFA conducted amongst the WATCH sample indicated that the weight locus of control, self-efficacy and body image scales demonstrated consistent construct validity, retaining the same item factor structure to the original analysis conducted by Kendall, Olson and Frongillo [ 27 ]. All 6 psychosocial subscales demonstrated acceptable internal consistency (Cronbach’s alphas α > 0.70) [ 45 ], when tested amongst the WATCH cohort with the exception of the weight locus of contro l scale (Cronbach’s α 0.49) [ 44 ]. In the current analysis, EFA was a necessary step in the scale reduction process, accounting for the assumption of unidimensionality (i.e., checking that the data are appropriate for the model), for the application of further analysis techniques such as IRT [ 46 ].…”

“…In the current analysis, EFA was a necessary step in the scale reduction process, accounting for the assumption of unidimensionality (i.e., checking that the data are appropriate for the model), for the application of further analysis techniques such as IRT [ 46 ]. During the EFA, strength of scale item correlations was examined using Spearman’s rho ( r s ) coefficient and p values (<0.05) [ 44 ], indicating the presence of probable redundant items. To detect these redundant items, further deleted items modelling analysis was performed on each of the 6 psychosocial scales.…”

“…Where the deletion of scale items improved the internal consistency (Cronbach’s alphas α) of each individual scale, these were considered as redundant items, as displayed in Figure 2 [ 45 ]. These results were the precursor for reducing the WRB-Q into a short form [ 44 ].…”