[Confirmatory factor analysis. Its role on the validation of health related questionnaires]

Researchers who use questionnaires in the health sciences tacitly base themselves (often inadvertently) in Classical Test Theory, the suppositions of which are unrealistic and frequently violated, leading to defective evaluation of the reliability and validity of the instrument.The present article emphasizes the need for precise definition of essential terms of measurement (reliability and validity) by clarifying the deficiencies that traditional methodology incurs in their use. The limitations of evaluation of reliability through Chronbach's alpha and laxity in quantitative evaluation of validity through exploratory factor analysis are described. As an alternative, a sequential and integrated approach to validity and reliability within the framework of confirmatory factor analysis models is proposed. These models provide the appropriate statistical framework to evaluate the validity and reliability of each item, instead of carrying out overall evaluations only. The confirmatory approach guides researchers to optimize the process of designing or adapting a questionnaire, freeing them from the largely unfounded ritual laid down by classical methodology.