Could someone kindly explain briefly the basis of the Composite Reliability measure ?
And the same for AVE please.
[Sorry if these are stupid questions.]
Scott
Composite reliability
-
Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi,
please, see viewtopic.php?t=293&highlight=ave
and viewtopic.php?t=288&highlight=ave
or search function here in the forum.
A classical paper is FORNELL, C.; LARCKER, D. F. Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research. v.18, p.39-50, fev. 1981.
Trying to explain AVE...
1) The factor loading is the correlation between LV and its indicator (MV);
2) The square of factor loading (like R2) could be undertood as % of the variance of the indicator (MV) that is explained by LV. If we have just one LV linked with one MV, this is the communality of this MV.
In factor analysis one MV is linked with all extracted LV (factors), then, in this case, communality is equal the sum of all squared factor loadings (between one MV and all extracted factors) and its meaning is: % of the variance of MV that is explained by all extracted factors (LV).
Then each MV has its communality.
3) AVE is computed for each LV like the average communality, the reason is: if a LV has 5 MV, in standardized mode we have 5 units of variance (100% of the information), but the communality is the % of the variance explained, then the sum of the 5 communalities is the total of variance explained by this LV, for example 3,97. Then AVE will be 3,97/5 = 0,794 = 79,4%, this LV explain 79,4% of the variance of these 5 MV.
The composite reliability is quite different (see the article..)
Best regards.
Bido
please, see viewtopic.php?t=293&highlight=ave
and viewtopic.php?t=288&highlight=ave
or search function here in the forum.
A classical paper is FORNELL, C.; LARCKER, D. F. Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research. v.18, p.39-50, fev. 1981.
Trying to explain AVE...
1) The factor loading is the correlation between LV and its indicator (MV);
2) The square of factor loading (like R2) could be undertood as % of the variance of the indicator (MV) that is explained by LV. If we have just one LV linked with one MV, this is the communality of this MV.
In factor analysis one MV is linked with all extracted LV (factors), then, in this case, communality is equal the sum of all squared factor loadings (between one MV and all extracted factors) and its meaning is: % of the variance of MV that is explained by all extracted factors (LV).
Then each MV has its communality.
3) AVE is computed for each LV like the average communality, the reason is: if a LV has 5 MV, in standardized mode we have 5 units of variance (100% of the information), but the communality is the % of the variance explained, then the sum of the 5 communalities is the total of variance explained by this LV, for example 3,97. Then AVE will be 3,97/5 = 0,794 = 79,4%, this LV explain 79,4% of the variance of these 5 MV.
The composite reliability is quite different (see the article..)
Best regards.
Bido
-
viswadatta
- PLS Expert User
- Posts: 139
- Joined: Wed Jul 05, 2006 1:43 pm
- Real name and title:
- Location: Coimbatore, India
ave vs composite reliability
Ave values are used to test convergent validity.
Composite reliability is a measure of reliabity (referred to as rho) in addition to cronbach's alpha
Composite reliability is a measure of reliabity (referred to as rho) in addition to cronbach's alpha