L'objectif de ce travail est d'étudier la validité interne et la structure factorielle de la TAS 20 sur un groupe de 192 femmes obèses en demande d'amaigrissement. Méthode – Les méthodes de travail sont les études de la cohérence interne de l'échelle par le calcul du coefficient a de Cronbach et du nombre de facteurs par analyse en composantes principales (ACP). Résultats – La structure tridimensionnelle est globalement confirmée (a = 0,74) mais la consistance interne de la troisième sous-dimension («pensées tournées vers l'extérieur ») est mauvaise (a = 0,33). L'ACP suggère une solution à 5 facteurs expliquant 53,6 % de la variance totale. Le facteur principal correspond à la 1 ^re dimension « difficulté à identifier ses sentiments », le quatrième facteur rappelle la 2 ^e dimension « difficulté à décrire ses sentiments». Les trois autres facteurs séparent la 3 ^e dimension initialement proposée : nous les avons appelés «évitement émotionnel», « désintérêt pour la compréhension psychologique» et « évitement cognitif ». Discussion–L'utilisation de la TAS 20 pour évaluer l'alexithymie et ses différentes dimensions chez les femmes obèses souhaitant maigrir suggère quelques précautions d'interprétation en rapport avec les particularités de cette population. Si la première dimension « difficulté à identifier ses sentiments » peut être retenue, la deuxième dimension « difficulté à décrire ses sentiments » nous paraît devoir être analysée dans la problématique de ces patientes. Nous proposons de séparer la troisième dimension, « pensées tournées vers l'extérieur », en 3 facteurs différents pour obtenir une interprétation plus pertinente.

The term « alexithymia » is derived from the Greek and means « no words for feeling » and is used to describe a personality trait characterized by the inability to experience and express emotion. Alexithymia is a relevant feature in subjects with eating disorders. The Toronto Alexithymia Scale (TAS) is the more used scale to assess alexithymia. The TAS 20 represents the latest revised and short version of the TAS, with psychometric properties clearly superior to the other alexithymia scales. The TAS 20 is a 20-items self-administred questionnaire including three subscales: ability to identify feelings, ability to describe feelings and externally oriented thinking. The TAS 20 was sparsely used in obese people and its validity was not studied at the present time in this specific population. The aim of this study was to assess the internal consistency and the factorial structure of the TAS 20 in a group of obese women. Method–The TAS 20 was administred to 192 obese women willing to loose weight, aged from 18 to 60 years. Their body mass index (BMI) was ranged from 27.5 to 55,0 kg/m2. The patients also completed a depression scale (Beck Depression Inventory 13). Cronbach's alpha coefficient was used to investigate the internal consistency of the scale. A confirmatory factorial analysis was performed to examine the validity of the initial three-factor structure in this population. The confirmatory factor analysis was conducted using STATISTICA5.0 ^® . Results–An exploratory factorial analysis using the principal components method was performed to search for a more relevant factorial structure. The global internal consistency value (a) was 0,74 for the full scale. The a coefficients were 0.78 and 0.67 respectively, for the two first subscales, (ability to identify feelings and ability to describe feelings), but was low (a = 0.33) for the last subscale (externally oriented thinking). This result questionned the initial three-factor solution and conducted us to perform a exploratory factorial analysis. This analysis yielded a six factors solution, explaining 58.7 % of the total variance, with the sole item 3 forming the sixth factor. The factor analysis yielding a five factors solution was retained. This solution explained 53.6 % of the total variance. The first factor (22.3 % of the variance) was constituted by the majority of the first subscale's items (items 1, 3, 6, 7, 9, 11, 13, 14) and corresponded to the ability to identify feelings. Four items out of five of the second dimension « ability to describe feelings » (items 2, 4, 12, 17) constituted the fourth factor (7.1 % of the variance) of the analysis. The third subscale (externally oriented thinking) was divided in three factors explaining respectively 10.7 %, 7.5 % and 6.0 % of the variance. The first of them included the items 5, 10 and 18, the second one included the items 15, 16 and 20, and the last one included the items 8 and 19. We called these three factors « emotional avoidance », « lack of interest in psychological comprehension » and « cognitive avoidance », respectively. The parameter estimates of the confirmatory factorial analysis for this five-factor solution were : c ² goodness-of-fit = 287.50 (df = 141 ; p < 0.001), goodness-of-fit-index = 0.87, adjusted goodness-of-fit index = 0.82, root-mean-square-residual = 0.08. The different models were compared by testing the differences in the chi-squares. The five-factor solution was better than the other tested models. Discussion–The results of the present study indicate that the use of the TAS 20 to assess alexithymia in obese subjects willing to loose weight, requires a careful interpretation related to the features of this population. Exploratory factorial analysis suggests a five factors solution statistically and clinically more relevant than the three factors solution. The first dimension (ability to identify feelings) is retained. The second dimension (ability to describe feelings) could rather be linked to the common wishes of obese women to keep their own feelings to themselves than to a dimension of alexithymia. Our data suggest to divide the last initial dimension « externally oriented thinking » in three independant factors which we named : « emotional avoidance », « cognitive avoidance » and « lack of interest in psychological comprehension ». Emotional and cognitive avoidance could be linked to the fact that our subjects wanted to loose weight and exhibited restrain eating. A positive correlation between restrain eating and alexithymia was previously reported. The last factor « lack of interest in psychological comprehension » would constitute an actual dimension of alexithymia : former studies found a negative correlation between alexithymia and psychologically oriented thinking.