Limits...
An Examination of a New Psychometric Method for Optimizing Multi-Faceted Assessment Instruments in the Context of Trait Emotional Intelligence.

Siegling AB, Petrides KV, Martskvishvili K - Eur J Pers (2014)

Bottom Line: The analyses revealed five facets, which did not occupy unique construct variance in any of the six samples.As expected, a composite of the remaining 10 facets consistently showed greater construct validity than the original 15-facet composite.European Journal of Personality published by John Wiley & Sons Ltd on behalf of European Association of Personality Psychology.

View Article: PubMed Central - PubMed

Affiliation: London Psychometric Laboratory, University College London London, UK.

ABSTRACT

Driven by the challenge of representing and measuring psychological attributes, this article outlines a psychometric method aimed at identifying problem facets. The method, which integrates theoretical and empirical steps, is applied in the context of the construct of trait emotional intelligence (trait EI), using data from six different samples (N = 1284) collected across Europe. Alternative representations of the trait EI variance, derived from the outcome variables used in previous validation studies of the Trait Emotional Intelligence Questionnaire, were regressed on the 15 trait EI facets using the stepwise method. The analyses revealed five facets, which did not occupy unique construct variance in any of the six samples. As expected, a composite of the remaining 10 facets consistently showed greater construct validity than the original 15-facet composite. Implications for construct and scale development are discussed, and directions for further validation of the method and for its application to other constructs are provided. © 2014 The Authors. European Journal of Personality published by John Wiley & Sons Ltd on behalf of European Association of Personality Psychology.

No MeSH data available.


Illustration of redundant and extraneous facets with respect to their component (i.e. common and specific) variance.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4359043&req=5

fig01: Illustration of redundant and extraneous facets with respect to their component (i.e. common and specific) variance.

Mentions: As regards the second and third criteria earlier, two types of problem facets can be operationally defined. We refer to them as extraneous and redundant facets (hereafter abbreviated as ET and RD facets, respectively). The best way to describe these facets is with respect to their component variance, as graphically illustrated in Figure1. Facets can have two types of variance: reliable common variance, which is due to the target construct and shared with the other facets, and reliable specific variance, which is unrelated to the target construct (Smith et al., 2003). ET facets have no common variance at all (i.e. variance due to the target construct); their variance is due to dimensions other than the one reflecting the target construct, thus likely violating the second criterion. As indicated, however, ET facets may still share variance with valid facets, because of measurement bias or dimensions other than the target construct. Although RD facets have common (construct) variance, this variance is more efficiently covered by at least one other. Therefore, RD facets do not occupy ‘unique common variance’ and do not add to the comprehensive representation of the construct (Criterion 3).


An Examination of a New Psychometric Method for Optimizing Multi-Faceted Assessment Instruments in the Context of Trait Emotional Intelligence.

Siegling AB, Petrides KV, Martskvishvili K - Eur J Pers (2014)

Illustration of redundant and extraneous facets with respect to their component (i.e. common and specific) variance.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4359043&req=5

fig01: Illustration of redundant and extraneous facets with respect to their component (i.e. common and specific) variance.
Mentions: As regards the second and third criteria earlier, two types of problem facets can be operationally defined. We refer to them as extraneous and redundant facets (hereafter abbreviated as ET and RD facets, respectively). The best way to describe these facets is with respect to their component variance, as graphically illustrated in Figure1. Facets can have two types of variance: reliable common variance, which is due to the target construct and shared with the other facets, and reliable specific variance, which is unrelated to the target construct (Smith et al., 2003). ET facets have no common variance at all (i.e. variance due to the target construct); their variance is due to dimensions other than the one reflecting the target construct, thus likely violating the second criterion. As indicated, however, ET facets may still share variance with valid facets, because of measurement bias or dimensions other than the target construct. Although RD facets have common (construct) variance, this variance is more efficiently covered by at least one other. Therefore, RD facets do not occupy ‘unique common variance’ and do not add to the comprehensive representation of the construct (Criterion 3).

Bottom Line: The analyses revealed five facets, which did not occupy unique construct variance in any of the six samples.As expected, a composite of the remaining 10 facets consistently showed greater construct validity than the original 15-facet composite.European Journal of Personality published by John Wiley & Sons Ltd on behalf of European Association of Personality Psychology.

View Article: PubMed Central - PubMed

Affiliation: London Psychometric Laboratory, University College London London, UK.

ABSTRACT

Driven by the challenge of representing and measuring psychological attributes, this article outlines a psychometric method aimed at identifying problem facets. The method, which integrates theoretical and empirical steps, is applied in the context of the construct of trait emotional intelligence (trait EI), using data from six different samples (N = 1284) collected across Europe. Alternative representations of the trait EI variance, derived from the outcome variables used in previous validation studies of the Trait Emotional Intelligence Questionnaire, were regressed on the 15 trait EI facets using the stepwise method. The analyses revealed five facets, which did not occupy unique construct variance in any of the six samples. As expected, a composite of the remaining 10 facets consistently showed greater construct validity than the original 15-facet composite. Implications for construct and scale development are discussed, and directions for further validation of the method and for its application to other constructs are provided. © 2014 The Authors. European Journal of Personality published by John Wiley & Sons Ltd on behalf of European Association of Personality Psychology.

No MeSH data available.