Hi
According to SmartPLS book , the outer loading more than 0.7 show indicator reliability, and only you remove them when your composite reliability and AVE increase.
After doing my algorithm, I needed to remove some low outer-loading but keep some between 6-7 because the composite reliability was good and AVE was already ok all more than 0.5.
but at the other hand less than 7 means we don't have indicator reliability and therefore you can not claim you have validity since reliability is necessary condition for validity.
should I remove all the outer-loading lower than 7 ? if no how we can justify validity when reliability is not perfect?
appreciate your guidance
indicator reliability necessary for validity?
- kamellia.ch
- PLS Junior User
- Posts: 10
- Joined: Thu May 18, 2017 2:47 pm
- Real name and title: kamelia chaichi
-
- SmartPLS Developer
- Posts: 1284
- Joined: Tue Mar 28, 2006 11:09 am
- Real name and title: Dr. Jan-Michael Becker
Re: indicator reliability necessary for validity?
You will never have perfect reliability. All measures (items) will have some sort of (random) variation. That’s why you usually have loadings <1.
More indicators may improve reliability and validity (even if the indicator reliability is not good; i.e. loading of some indicators is smaller than 0.7), because you have more information to estimate the latent construct’s shared variance, which is supposed to be the behavior of your latent variable that you want to know.
For each indicator (that has loading <0.7) you therefore assess if the overall quality (AVE, composite reliability, etc.) improves by deleting or if it is already very good. If it is good or does not improve you can keep the indicator because it adds additional useful information to your model. Generally, the more indicators the better (if they are not too unreliable).
More indicators may improve reliability and validity (even if the indicator reliability is not good; i.e. loading of some indicators is smaller than 0.7), because you have more information to estimate the latent construct’s shared variance, which is supposed to be the behavior of your latent variable that you want to know.
For each indicator (that has loading <0.7) you therefore assess if the overall quality (AVE, composite reliability, etc.) improves by deleting or if it is already very good. If it is good or does not improve you can keep the indicator because it adds additional useful information to your model. Generally, the more indicators the better (if they are not too unreliable).
Dr. Jan-Michael Becker, BI Norwegian Business School, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de