Evaluating the construct validity of a questionnaire using Multiple Factor Analysis: considerations on item reduction
Keywords:
Validity, Reliability, Virtuality, StatisticsAbstract
Questionnaires are widely used instruments for data collection in educational
research, both in face-to-face and virtual contexts. To ensure meaningful results,
questionnaires must meet fundamental criteria of both validity and reliability. Validity
is established based on evidence related to three aspects: content, criterion, and
construct. Reliability refers to both the consistency and stability of the measurements
obtained. In construct validity analysis, exploratory or confirmatory factor analysis is
commonly used. However, its correct application depends on compliance with
restrictive theoretical assumptions, a limitation for its use in applied studies or real
life educational contexts. This paper proposes the use of Multiple Factor Analysis as
a methodological alternative for assessing construct validity. It is a descriptive
multivariate technique that does not require verification of prior assumptions and also
allows for the identification of relevant items, facilitating questionnaire optimization.
This methodology is applied to the analysis of an instrument used in a virtual course,
presenting useful results to guide the validation of the instrument as well as the
reduction of its length. It is concluded that the presented technique is an effective
tool, directly applicable to the evaluation and improvement of educational
questionnaires.