Assessing Construct Reliability and Convergent Validity using SPSS AMOS
IBM SPSS AMOS Series - 11
In CB-SEM, Loadings, Model Fit, and Modification Indices are critical to model identification. This post will discuss in detail all these concept.
- Once the factor loadings and model fit are assessed. The next step is to check the reliability and validity of the constructs.
- Construct reliability assessment allows the evaluation of the extent to which a variable or set of variables is consistent in what it intends to measure (Straub, Boudreau, & Gefen, 2004).
- The essence of reliability is covered in the word Consistency. For example, if the same measuring instrument produces the same results with the same individuals on different occasions, then the measuring instrument is reliable.
- Construct reliability is usually assessed using composite reliability and Cronbach’s alpha.
- Cronbach’s Alpha can be calculated using SPSS, in AMOS we will focus on Composite Reliability.
- Both values are interpreted using the guidelines offered by Nunnally and Bernstein (1994) who suggest 0.7 as a benchmark for a modest reliability applicable.
Composite Reliability is calculated based on factor loadings. The formula is
- Construct validity is the measure of how well the items selected for the construct actually measure the construct.
- For example, Life Satisfaction in this case is measured using 5 indicators, construct validity will help determine how well these five items measure the latent unobserved construct of Life Satisfaction.
- Construct validity is established through two forms of validities, convergent validity and discriminant validity.
- Convergent validity refers to the degree to which multiple measures of a construct that theoretically should be related, are in fact related (Gefen, Straub & Boudreau, 2000).
- Hence, the multiple indicators measuring the same concept through convergent validity are assessed to whether these indicators converge to measure the underlying construct.
- This will ensure uni-dimensionality of the multiple-item constructs and will help in eliminating any unreliable indicators (Bollen, 1989).
- Convergent validity is assessed using Average Variance Extracted (AVE). The AVE indicates how much of the indicators’ variance can be explained by the latent unobserved variable.
- An AVE greater than 0.50 provides empirical evidence for convergent validity (Bagozzi & Yi, 1988), as the corresponding latent variable explains more than half of the variance in the belonging indicators.
- AVE is calculated by taken sum of squares of the factor loadings and dividing it by the no. of items in the unobserved latent variable.
Supporting Argument for Low Average Variance Extracted
According to Hair et al. (2016), if the average variance extracted is greater than 0.4 and composite reliability is higher than 0.6, the convergent validity of the construct is still acceptable (Fornell and Larcker, 1981; Lam, 2012).
Fornell, C. and Larcker, D.F. (1981), “Evaluating structural equation models with unobservable variables and measurement error”, Journal of Marketing Research, Vol. 18 No. 1, pp. 39-50.
Hair, J.F., Jr., Hult, G.T.M., Ringle, C. and Sarstedt, M. (2016), A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM), Sage publications.
Lam, L.W. (2012), “Impact of competitiveness on salespeople’s commitment and performance”, Journal of Business Research, Vol. 65 No. 9, pp. 1328-1334.