A. Leplege et al., PRELIMINARY PSYCHOMETRIC ANALYSIS OF THE FRENCH VERSION OF AN INTERNATIONAL QUALITY-OF-LIFE QUESTIONNAIRE - THE MOS SF-36 (VERSION-1.1), Revue d'epidemiologie et de sante publique, 43(4), 1995, pp. 371-379
The International Quality Of Life Assessment (IQOLA) Project is a 4 ye
ar project initiated in 1991 to translate and adapt the Medical Outcom
e Study Short Form 36 item Health Survey (SF-36) in at least 15 countr
ies. This paper reports on the preliminary psychometric assessment of
the SF-36 in French (version 1.1). The validation data come from two s
tudies: a phase IV study of 121 patients with arthritis and a phase Ty
study of 159 patients with angina. In both cases, the patients were s
urveyed using the SF-36 and a disease specific module. The main object
ive of this analysis was to determine how well the scaling assumptions
(summated rating or Likert type scaling construction) of the SF-36 we
re satisfied. Item convergent validity was supported as items-scale co
rrelation range from 0.47 - 0.87. Item discriminant validity was suppo
rted as all items were more correlated with their hypothesised scales
than with scales measuring other concepts. Our data support the assump
tion that the items measuring the same concept had approximately equal
variance. Items in a given scale contained about the same proportion
of information about the concept being measured as most items of any g
iven dimension had approximately the same correlation with that dimens
ion. The Cronbach alpha coefficient ranged between 0.79 and 0.95. The
correlation between two scales was less than the reliability coefficie
nt for those scales, and these correlations adjusted for attenuation w
ere less than 1. These preliminary results are encouraging. They indic
ate that the items are linearly related to the underlying concept bein
g measured. There is good evidence of the existence of a unique and re
liable variance which is measured by each scale score and it seems rea
sonable to suggest that scales scores are reproducible and interpretab
le. All this supports the claim that the scaling properties of this Fr
ench version of the SF-36 are respected and that health outcomes may b
e reliably assessed using this version.