condisc assesses convergent and discriminant validity of latent constructs expressed by congeneric (not loading on more than one factor) indicators making up a confirmatory factor model estimated using sem.
condisc(x)Convergent validity is the extent to which a set of indicators reflecting the same construct are positively correlated. Convergent validity of a construct can be claimed to be demonstrated when the construct can explain an average amount of 50 per cent variance of its indicators. This value is commonly referred to as average variance extracted (AVE) in the litterature.
AVE = Sum of squared standardised loadings/Number of indicators
Discriminant validity can be exhibited when the average variance extracted (AVE) value of a latent construct is larger than its squared correlation (SC) with any other latent construct in the model, showing that each latent construct shares more variance with its associated indicators than with any other latent variable expressed by different sets of indicators in the model.
When both convergent and discriminant validity are established, a latent construct can be considered valid.
condisc applies a strict criterion supporting discriminant validity only when all of the AVE values are larger than all of the SC values.