Factor Analysis of the Nowicki-Strickland Locus of Control Scale: Why is Replication So Difficult?

Abstract

This study represented an attempt to replicate the factor structure obtained by Lindal and Venables (1983) for the Nowicki-Strickland Locus of Control Scale (NSLCS) for Children. Although the same factor analytic techniques were used on a similarly-constituted sample, a somewhat different solution was obtained. Comrey (1973) suggested that scales with dichotomous items cannot be factor analyzed reliably. As a more rigorous test of this theory, the factor analyses performed on two randomly-selected samples from one study were compared. However, this replication attempt also failed. The authors concluded that the dichotomous nature of the NSLCS Scale is the reason that replication of its factor structure has been so elusive.

Locus of Control revisited: Development of a new bi-dimensional measure

“…We used a 5-point scale, since the best estimations of psychological parameters are obtained with between 4 and 6 response categories (Lozano, García-Cueto, & Muñiz, 2008). The Likert-type format was chosen with the aim of overcoming the limitations of forced-choice questionnaires (Ferrando et al, 2011) and dichotomous scales (Watters, et al, 1990). In constructing the instrument we followed the international recommendations that can be found in the psychometric literature (AERA, APA, NCME, 2014; Downing, 2006).…”

“…To this end we designed a new measurement instrument. There are several reasons that justify the development of a new measurement instrument, among them: a) that was not context-dependent, thus permitting greater generalization of the results beyond a particular domain, be it academic, clinical, work-related, or other (Lefcourt, 1991;Little et al, 2002;Wang et al, 2010), b) the Likert-type format was chosen with the aim of overcoming the limitations of forced-choice questionnaires (Ferrando et al, 2011) and dichotomous scales (Watters, Thomas, & Streiner, 1990), y c) it proposed a bifactorial structure in order to reflect parsimoniously the construct measured. As a second goal we analyzed differences in LOC according to sex and age.…”

Bifactorial structure of Locus of Control cross-culturally invariant across Spain, Chile and United Kingdom

“…Of the total, 10 items evaluate internal LOC and 13 evaluate external LOC. This instrument was chosen for three fundamental reasons: a) it offers appropriate psychometric properties (α internal = .87, α external = .85; CFI = .90; RMSEA = .04; χ2 /df = 1.58; validity evidence based on relations to other variables; Suárez-Álvarez et al, in press); b) the Likert scale overcomes the limitations of forced-choice questionnaires (Ferrando et al, 2011) and dichotomous scales (Watters, Thomas & Streiner, 1990) and in addition, the five categories of response optimise the psychometric properties of the scales (Lozano, García-Cueto & Muñiz, 2008); c) the test makes no reference to any specific area and is therefore able to be applied independently of any contexts of application such as educational, organisational or clinical, in such a way as to eliminate this form of contextual specificity which can make the comparison of results difficult (Wang, Bowling & Eschleman, 2010). The data from the samples used in this work are presented in the results section.…”

The effects of transition intervention on students with mild disabilities.

“…This is a 40-item dichotomous response scale to measure perceived control in affiliation, achievement, and dependency . Watters and Thomas (1990) . These research findings were upheld by more recent research by Choi (2003, p. 479) who reported that the internal consistency was satisfactory: "The Self-efficacy General subscale tended to be consistently higher (mid to high .80s) than the reliability of the Social Self-efficacy scores (far lower than .80).…”