Objective. For reliable and valid assessment and diagnostic categorization of eating disorders, self-report measures have considerable limitations. A semi-structured interview – the SIAB-EX – was developed for a more reliable and valid assessment of eating disorders.

Methods. One study (videotapes of 31 inpatients, seven raters) was made to establish inter-rater reliability; in another study with 80 patients the SIAB-EX was compared to another semi-structured interview designed for comparable purposes (EDE). In a third study data was obtained on 377 eating disorder patients seeking treatment to explore discriminant and convergent (construct) validity using the following self-rating scales: EDI, TFEQ, SCL-90, BDI, and the PERI Demoralization Scale.

Results. Inter-rater reliability of dichotomous ratings was good with mean kappa values of .81 (current) and .85 (past). Comparison of the SIAB-EX with the EDE generally showed quite similar results and higher intercorrelation of the total scale (.77). There are, however, a number of differences between the two scales, which are discussed in detail. Construct validity of the SIAB-EX was established.

Conclusion. Inter-rater reliability was good. Convergent and discriminant (construct) validity of the SIAB-EX was demonstrated. The constructs assessed by the SIAB and its subscales and items are discussed in the context of their correlations with other well-known scales.