The factor structure of the eating disorder examination in clinical and community samples

Abstract

Overall, there was more support for a one-factor model of EDE data than for a multi-factorial model. It may be more appropriate to use Global EDE scores than individual subscale scores for research purposes.

Are common measures of dietary restraint and disinhibited eating reliable and valid in obese persons?

“…Only the Restraint subscale from the EDE-Q was used in the present analyses. Previous research indicates that the EDE-Q demonstrates acceptable internal consistency (Cronbach's alpha ranging from .77 to .84) (Byrne, Allen, Lampard, Dove, & Fursland, 2010), and good convergent validity with the EDE interview (Black & Wilson, 1996;Fairburn & Beglin, 2008) and criterionrelated validity for distinguishing obese individuals who engage in binge eating from individuals who do not binge eat (Wilson, Nonas, & Rosenblum, 1993).…”

Psychometric Evaluation of Disordered Eating Measures in Bariatric Surgery Candidates

“…combining the subscales) were not made using CFA. For the three-factor model presented by Byrne et al [23], fit statistics indicated that the model did not fit the data well and standardised factor loading estimates revealed that not all indicators were strongly related to the latent factors (range=0.214 to 0.782).…”

“…Given the poor original factor structure of the EDE and EDE-Q evident in this study and reported in previous studies [18][19][20]23], a three-factor model was tested as outlined by Byrne et al [20] was interpreted as restraint, eating concern and weight and shape concern. An alternative four-factor model proposed by Hrabosky et al [18] was not tested as it included additional behavioural frequency items that were not part of the original scales.…”

“…A seven-item three-factor model proposed by Grilo et al [19] was not tested due to the substantial item reduction and loss of one of the underlying constructs. An eight-item one-factor model proposed by Wade and colleagues [57] that has been previously tested [20,23] was not assessed as the reduction to one factor substantially altered the measure to the extent that underlying constructs were not distinguishable.…”

“…For example, factor analyses of the EDE interview in clinical eating disorder, obese and non-eating disorder community samples have failed to replicate the original four-factor structure [23][24][25]. For the 51-item TFEQ, two studies have reported EFA in obese and non-obese community samples, and both were unable to validate the original three factors (cognitive restraint, disinhibition and hunger) [26,27].…”