1988
DOI: 10.2307/3172650
|View full text |Cite
|
Sign up to set email alerts
|

An Updated Paradigm for Scale Development Incorporating Unidimensionality and Its Assessment

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

22
1,146
1
50

Year Published

1999
1999
2019
2019

Publication Types

Select...
6
3

Relationship

0
9

Authors

Journals

citations
Cited by 2,879 publications
(1,219 citation statements)
references
References 0 publications
22
1,146
1
50
Order By: Relevance
“…The CFA indicated that several items had shared residual variance, and, accordingly these were dropped in order to achieve unidimensionality and good ®t indexes in LISREL, too (the dropped items are marked as such in Appendix A). It is not unusual that items are dropped during a CFA because a CFA also examines unidimensionality and that the residual variance of the items do not signi®cantly overlap, none of these are examined by factor analysis or by Cronbach's Alpha [14]. The resulting CFA showed that all the remaining items loaded signi®cantly and highly on their assigned constructs.…”
Section: Additional Data Collectionmentioning
confidence: 99%
See 1 more Smart Citation
“…The CFA indicated that several items had shared residual variance, and, accordingly these were dropped in order to achieve unidimensionality and good ®t indexes in LISREL, too (the dropped items are marked as such in Appendix A). It is not unusual that items are dropped during a CFA because a CFA also examines unidimensionality and that the residual variance of the items do not signi®cantly overlap, none of these are examined by factor analysis or by Cronbach's Alpha [14]. The resulting CFA showed that all the remaining items loaded signi®cantly and highly on their assigned constructs.…”
Section: Additional Data Collectionmentioning
confidence: 99%
“…The data show that all the measurement items loaded signi®cantly on their respective latent constructs and did not share signi®cant residual variance with each other. The insigni®cant w 2 at 80.53 with 69 degrees of freedom ( p = 0.16) shows very good model ®t [21], and indicates that the constructs are distinct and unidimensional [14]. Construct reliability coe cients, presented in Table 2, are also above the 0.80 threshold [34].…”
Section: Hypotheses Testingmentioning
confidence: 99%
“…All factor loadings exceeded 0.65 and t-values exceeded 10.59, providing evidence of convergent validity. We assessed discriminant validity through model comparisons with  freed versus fixed at 1 for all construct pairs (Gerbing & Anderson 1988). In all cases, the chi-square differences were significant at the .05 level.…”
Section: Measurement Modelmentioning
confidence: 99%
“…Para su realización se utilizó el méto-do de extracción de ejes principales, y rotación oblicua oblimin. Acorde con la elaboración de escalas para evaluar constructos 18,19 se procedió a desarrollar un análisis alfa de Cronbach y de capacidad discriminativa de los ítems. Finalmente, se procedió a relacionar las dimensiones del CIO con el CPP usando Rho de Spearman.…”
Section: Análisis De Datosunclassified