Construction and Validation of a Scale for the Measurement of Body Image

PURPOSE

To construct and validate a scale for the measurement of body image in nursing practice.

METHODS AND FINDINGS

Construction of the scale items was based on a literature review. A list of 81 items was drawn up and, after careful analysis, reduced to 43 items. Content validity of the 43 items was evaluated by seven experts, who were asked to mark their agreement or disagreement as to whether the items referred to the concept of body image. In order to avoid bias in the answers, 5 items were added to serve as confounding variables. Nine items (including the 5 confounding ones) were eliminated because they did not reach the minimum value of interrater agreement to be retained (>0.80). Test/retest was used to verify the reliability of the measure.

An instrument including the 39 retained items was tested twice with a sample of 24 volunteers. Pearson's correlation coefficient was 0.71. Three items achieving low scores were discarded. To determine construct validity, the 36 remaining items were tested with a sample of 375 students in undergraduate courses at the Health Sciences Center of the Federal University of Para Three analytical procedures–evaluation of the internal consistency of the measure through item-total correlation level, analysis of the contribution of each specific item using Cronbach's alpha coefficient (α), and the verification of the dimensionality of the measure using principal components analysis—were carried out. Four components were extracted, cumulatively explaining 41.2% of the variance in the 36 items of the scale. Following this analysis, 10 items were eliminated. After that exclusion, a second extraction was done, generating a five-factor solution with an eigen value of ≥1.00 and a cumulative explained variance of 54.1%. The 26 items yielded an α= 0.90, indicating a high internal consistency of the scale. Three more items were eliminated because they reduced the previously obtained value of item-total correlation and their removal contributed, although only slightly, to improved performance of the Cronbach's alpha coefficient. A third extraction was done for the 23 retained items. Five components were extracted, cumulatively explaining 58.2% of the variance in the items of the scale (α= 0.91).

Based on the total score obtained by each participant and on the distribution of those scores, two criteria groups were created—a lower group that had 30% of the lowest scores, and an upper group that had 30% of the highest scores in the distribution. Using a Student t-test, all 23 items were able to discriminate among subjects belonging to the upper and lower criteria groups (p<0.0001) (i.e., discriminant validity).

CONCLUSIONS

The scale is a valid and reliable instrument to assess body image. Although it reached a high internal consistency index (α= 0.91), the investigators consider the scale still needs to be further refined and tested in different settings/samples so its clinical validity and reliability can be confirmed. It also still needs to be submitted to the equivalence criterion (external validity), which can be done either by an analysis of its convergence with instruments that measure the same construct, or an analysis of its divergence, with instruments that measure a different construct of body image.