Résumé
Objective: theoretical and empirical analysis of items and internal consistency of the Portuguese-language version of Social Phobia and Anxiety Inventory (SPAI-Portuguese). Methods: social phobia experts conducted a 45-item conted analysis of the SPA-Portuguese administered to a sample of 10.14 university students. Item discrimination was evaluated by Student's t test; inter-item, mean and item-to-total correlations, by Pearson coefficient; reliability was estimated by Cronbach´s alpha. Results: there was 100 porcent agreement among experts conserning the 45 items. On the SPAI-Portuguese 43 items were discriminative (p<0,05). A few inter-item correlations between both subscales were below 0,2. The mean inter-item correlation were: 0,41 on social phobia subscale; 0.32 on agoraphobia subscale and 0,32 on the SPAI-Portuguese. Item-to-total correlation werw all higher then 0.3(p<0,001). Cronbach´s alphas were: 0.95 on the SPAI-Portuguese; 0.96 on social phobia subscale; 0.85 on agoraphobia subscale. Conclusion: the 45-item content analysis revealed appropriateness conserning the underlynig construct of the SPAI-Portuguese (social phibia, agoraphobia) with good discriminative capacity on 43 items. The mean inter-item correlations and reliability coeficients demonstrated the SPAI-Portuguese and subscales internal consistency and multdimensionality. No item was suppressed in the SPAI-Portuguese but the authors suggest that a shortened SPAI, in its diferent versions, could be an even more useful tool for research settings in social phobia.