Work out construct validity estimate instantly with clear inputs, formula shown and shareable results.
Average variance extracted is the mean squared loading, and convergent validity requires it above 0.50 - so 0.62 passes. The Fornell-Larcker criterion for discriminant validity requires the square root of AVE, 0.787 here, to exceed the correlation with any other construct, and 0.55 is comfortably below it.
Average variance extracted
AVE = mean of squared standardised loadings
Fornell-Larcker criterion
sqrt(AVE) must exceed the inter-construct correlation
Because below that the construct explains less of its indicators' variance than measurement error does, so the latent variable is poorly defined.
It remains widely reported, though the HTMT ratio is now often preferred because it detects discriminant validity problems more sensitively.