Average variance extracted (AVE) values were used to determine whether or not convergence validity and discrimination validity were verified (Fornell &
Larker, 1981).
Larker, Richardson, and Tuna (2007) found no evidence of a link between directors' age and financial performance measured by return on assets (ROA) or future stock returns.
In determining discriminant validity, the squared correlation between two constructs must be less than either of their individual AVEs (Fornell and
Larker, 1981).
This was done using the suggested procedure of Fornell and
Larker (1981).
Get West London photographer Grahame
Larker took stunning photos from Harmondsworth, near Heathrow Airport, and captured the formation while looking towards the direction of London.
Table 1: Discriminant Validity (Fornell and
Larker) Construct ALT KGIV KREC KSE REC REP TRU ALT 0.782 KGIV 0.113 0.718 KREC 0.268 0.405 0.766 KSE 0.427 0.525 0.360 0.747 REC 0.032 0.262 0.080 0.185 0.804 REP 0.342 0.373 0.416 0.534 0.374 0.729 TRU 0.320 0.480 0.175 0.663 0.277 0.504 0.734 Note: Diagonals (in bold) represent the square root of average variance extracted (AVE) while the off diagonals represent the construct correlations.
De acordo com Ringle, Silva e Bido (2014), deve-se usar o criterio de Fornell e
Larker (Henseler et al., 2009), cujo valores de referencia da AVE devem ser iguais ou superiores a 0,50 (AVE's > 0,50).
The cut off values exceeding 0.50 in respect of both AVE and construct reliabilities are considered significant to ensure the convergent validity (Fornell &
Larker, 1981; Bagozzi & Yi, 1991; Hair et al., 2006).
Para evaluar la validez convergente, Fornell &
Larker (1981) sugieren el uso de la varianza media extraida (AVE).
A consistencia interna do instrumento foi avaliada pelo alfa de Cronbach (MESSICKS, 1995) e pela fiabilidade composita (FORNELL;
LARKER, 1981).