SOFTWARE, INSTRUMENTACIÓN Y METODOLOGÍA Medición de constructos bipolares mediante escalas tipo Likert: ¿por qué aparecen factores monopolares?

Abstract

Measuring bipolar constructs but finding monopolar factors. The structure of bipolar constructs is often studied using counterbalanced Likert-type scales. Answers obtained by means of these scales have been submitted to factor analysis, and contrary to expectations, the results frequently have shown that the obtained factor structure is not bipolar, but made up by two monopolar factors. Recently, Van Shuur and colleagues (1994, 1995) proposed an explanation for this phenomenon, arguing that the emergence of monopolar factors is due to the fact that this kind of data fits to the unfolding model, not to the factor analysis model. This paper presents and tests a new explanation: the reformulation of the bipolar model under the "two halves hypothesis". Under this hypothesis the bipolar dimensions split into two monopolar factors because the items of counterbalanced Likert-type scales do not measure the intended bipolar dimension. On the contrary, they only are indicators of one half of the dimension. From this point of view the emergence of monopolar factors barely related do not invalidate the bipolarity of the underlying dimension. The empirical evidence presented here has been obtained in a sample of 298 subjects who responded to the Warr's (1990) scales of affective well-being. This evidence supports the explanation offered in this paper.