Validation of the Self-assessment of Communication Competence (SACCom) in Brazilian Portuguese Through Item Response Theory

Abstract

To evaluate the validity of the Self-assessment of Communication Competence - SACCom by analyzing the communication competency of adult individuals using the item response theory (IRT). Overall, 778 individuals who responded to SACCom participated in this study. The initial version consisted of 20 polytomous items regarding communication skills. Polytomous answer keys to each item were ordered as follows: 1 - no, 2 - more or less, and 3 - yes. The IRT and the gradual response model, a 2-parameter logistic model for polytomous items, were used for validation. Due to problems in the preliminary analysis using the item characteristic curve (ICC), answer keys to SACCom items were reorganized in an ordinal dichotomous way. Response 1 was coded as 0 - no, while responses 2 and 3 were coded as 1 - yes. Item 9 of the ICC was problematic and did not add information to the instrument; hence, this item was excluded. SACCom, in its dichotomous format with 19 items, was unidimensional according to the eigenvalues graph and tetrachoric correlation analysis. The Cronbach's alpha coefficient showed an internal consistency value of 0.711. All item-total correlations were greater than 0.284. Item discrimination parameters ranged from a=0.563 to a=1.505, and the difficulty of the items ranged between b=-2.725 and b=0.612. The likelihood ratio test showed that the model without restrictions best fit the data. ICC gleaned more information for individuals with lower communication skills. The 19-item, dichotomous SACCom is valid for the analysis of communication skills according to the IRT.