effect size

mazi · Post by **mazi** » Fri May 13, 2011 10:50 am

Hi every one
i run my model and i compute the effect size for my dependet constructs. for one of the variables the effect size is more than one and for two variables the effect size is computed az zero.
does these values provide problem for my results?
Thank you
Best regards
Nazila Babakhani

Diogenes · Post by **Diogenes** » Sat May 14, 2011 10:49 am

If what you naming "effect size" is the:
- Standardized path coefficient (Data metric = mean = 0, s.d. = 1) --> the model has a problem of multicollinearity
- Unstandardized path coefficient or factor loading (Data metric = Original) --> ok, but we, usually, do not use because the interpretation of these results.

Best regards,
Bido

mazi · Post by **mazi** » Sat May 14, 2011 1:45 pm

i think it is standardized paths, you mean that for effect size of above one there is a problem of multicollinearity? so how i can remove it
and what if the effect size equal to zero?
in some where i read that effect size of above one create no problem it can be even zero
thank you
best regards

mazi · Post by **mazi** » Wed May 25, 2011 4:46 am

what should i do to refine this proble, if it is
Thank you