Since in the result table of SEM (using AMOS) for mediation test, the values of total, direct, and indirect effects should be reported, would you please let me know whether standardized values are preferable or unstandardized ones in this table?
I searched different sources and found different ideas in this regard. Some agreed that standardized values should be reported because in bootstrapping test, the lower and upper bound would be between -1 and +1, while others preferred unstandardized effect to report!!
Thank you for your help..
I suggest you standardised estimate. Please find the attached file regarding the mediation test
Conference Paper ANALYZING THE MEDIATING VARIABLE IN A MODEL Zainudin Awang (...
In analyzing the mediation effects in a model manually, you need to compute the indirect effect and compare it with the direct effect. And in computing the indirect effect, you need to have the product of standardized values. The standardized values are values converted into the same unit (standard deviation).
Even through Bootstrapping also you need to obtain the significance of the standardized indirect effect to confirm mediation, and the standardized direct effects to determine the type of mediation whether full mediation or partial mediation.
Dear Mr. Wan Mohamad and Prof. Zainudin
Thank you very much for your responses.
All the best,
Farimah
Hi, I am attaching 2 articles on mediation and moderation, it is presented them what and how to report. hope this is helful
Dear Prof. Muayyad Ahmad
Thank you very much for the articles. They are very helpful.
All the best,
Farimah
My dear
Using standardized values are preferable because it provide comparability between values but you need unstandardized values when using Sobel test to obtain the significance of indirect effects.
Generally we use standardized values because in many cases the variables haven't the same nature or the same average values
Dear Dr. Amer
Thank you for your reply to my question. I calculated Sobel test using the standardized values. You mean I did a mistake?
I found it in some papers that they used standardized values for Sobel test! and the justification was exactly what Dr. Bessadok said in previous comment.
Would you please let me know why I need to use unstandardized values when using Sobel test ?
Thank you & Best regards
Farimah - you have to do bootstrapping every time you do mediation analysis. The reason is given in Preacher & Hayes (2004, 2008) which mentions that you have 2 paths in your mediation, a and b, a can be normal and b can be normal but the indirect (a*b) cannot be normally distributed so you need to run bootstrapping to correct the standard error. The Sobel test you do online is based on normal distribution.
standardized effects are preferrable. however better to conduct bootstrapping to measure the bias in the data or the results are.
Thank you very much for your response dear Prof. Zainudin.
All the best,
The standard error in the standardized model results section in Mlus7 does not show the t value and p-value. How can I handle this problem?
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS