Dear Rama,

here is a paper that illustrates an easy approach to interaction modeling (the residual centering approach). We did this in Lisrel but it is easy to implement it in AMOS.

https://scholar.google.de/scholar?oi=bibs&cluster=9078355602857122328&btnI=1&hl=de

Best,

Holger

Dear Rama,

here is a paper that illustrates an easy approach to interaction modeling (the residual centering approach). We did this in Lisrel but it is easy to implement it in AMOS.

https://scholar.google.de/scholar?oi=bibs&cluster=9078355602857122328&btnI=1&hl=de

Best,

Holger

Hi Rama

See also here: https://www.youtube.com/watch?v=K34sF_AmWio

In addition, you can check here if your moderator is categorical (i.e. do a multi-groups analysis): https://stat.utexas.edu/software-faqs/amos

Best

Simon

There are two techniques to treat the moderating variables on AMOS. The first is running a multi-group analysis as the colleagues said. The second is running the technique applied by Ping (1995) which is less popular than the first but I saw some researchers had applied it successfully.

Ping RA. 1995. A parsimonious estimating technique for

interaction and quadratic latent variables. Journal of

Marketing Research 32(3): 336–347.

I like Ping's approach which is more complicated but more precise I think. 

https://www.researchgate.net/publication/273389587_Demands_or_Resources_The_Relationship_Between_HR_Practices_Employee_Engagement_and_Emotional_Exhaustion_Within_a_Hybrid_Model_of_Employment_Relations?ev=prf_pub

Article Demands or Resources? The Relationship Between HR Practices,...

 Na Fu, we followed your paper for our research. It helped us a lot. Thank you very much for providing such a useful paper which is theoretically and empirically sound.

It is a good question. I am looking forward to see the answers.

looking for the best answer.....

Hi, I like the way Conway, E., Fu, N., Monks, K., Alfes, K., & Bailey, C. (2016) presented, together with guidance on YouTube provided by James Gaskin. Clear and easy for me to follow - a newbie to statistic world. However, I am wondering if we can test different interactions at the same time? For example i would like to test moderating effect of M on 3 different paths: A-B, A-C and B-D. Appreciate your advice and reference to support. Thank you.

https://www.researchgate.net/publication/299441114

Data 11 Chapter 7 Analyzing the moderating variable

Thank you Holger and Yaser for sharing such a wonderful information.

Rgds

Shayo

can someone share a published paper in a good journal which uses multigroup analysis for categorical moderators? Need references..

Hi Diksha,

you can take a look into the literature on "measurement invariance", e.g.,

Byrne, B., Shavelson, R. J., & Muthén, B. (1989). Testing for the equivalence of factor covariance and mean structures: The issue of partial measurement invariance. Psychological Bulletin, 105(3), 456-466.

Steenkamp, J.-B. E. M., & Baumgartner, H. (1998). Assessing measurement invariance in cross-national consumer research. Journal of Consumer Research, 25, 78-90.

Vandenberg, R. J. (2002). Toward a further understanding of and improvement in measurement invariance methods and procedures. Organizational Research Methods, 5(2), 139-158.

Hayduk, L. A. (2016). Improving measurement-invariance assessments: Correcting entrenched testing deficiencies. BMC Medical Research Methodology, 16(1), 130-130. doi:10.1186/s12874-016-0230-3

Steinmetz, H., Schmidt, P., Tina-Booh, A., Wieczorek, S., & Schwartz, S. H. (2009). Testing measurement invariance using multigroup CFA: Differences between educational groups in human values measurement. Quality and Quantity, 43(4), 599-616. doi:10.1007/s11135-007-9143-x

Best,

Holger

measurement invariance is same as moderation with categorical variables? the interpretations will be same? E.g. if there is no significant chi square difference in unconstrained and constrained structural weights models, but an estimate is significant in one group but not in the other what would that mean? there is moderation or no moderation??

Hi Diksha,

almost :D. Testing the moderation hypothesis H0 = B_groupA equals B_groupB is essentially a test of the difference of the two B's. A significant moderation, thus, means that one of the B's is larger or smaller than the other. This can be the case of both B's are significant by themselves (but one is larger/smaller than the other) or one being non-significant and the other beeing significant.

On the other side, a failed test of the moderation effect can occur even when one of the B's is non-significant and the other is significant. We observed this recently in a study, in which we compared the effect of a treatment on some attitude for women and men. The effect was n.s. for women but significant for men. However, when we test for the significance of the *difference* in the effects by means of invariance of the structural effects and chi-square difference testing, the difference was not signifcant! This is not a nice result because it forces to keep the more restricted model in which the effect is equal (and significant) for both sexes. This can ocurr, when the effect in one group is sligthly below the "significance border" and the other is above but the difference itself is not large enough to become significant.

And finally (to close the loop and summary my response to your question): Testing the invariance of a structural effect via chi-square testing is equivalent to conducting a single-sample dummy regression with a product term between the independent variable and the categorical grouping variable.

HTH

Holger

I will send you a paper based on multi-group moderation. Please send me your email id

Dear Rayees Farooq,

Here is my email address please!

[email protected]

Eagerly waiting for your response

Rgds

France

Good question and informative responses. Thanks!

Dear Rama,

I have already shared a movie that I my self created two years ago. However, it is in Persian language. Hopefully, you get the procedures. Additionally, to write the report, you can find the structure in some journals related to multi- group analysis. If you can not find please let me know.

Best regards,

Dear Rama,

Moderating effect (interaction effect) can be done using SPSS And AMOS

and multi-group analysis can be done using AMOS (which I attached in previous answer).I also have attached doing interaction effect using both SPSS and AMOS. I just put the photos that shows the procedures of computing interaction effect using SPSS and AMOS since my book is in Persian Language. This example wanted to show that Income moderates the relationship between Unanimity and Edprogress (educational progress). Hope this helps.

Best regards,

Could someone give pdf book reference about using Moderating effect in AMOS

Hi Abdul

undortunately I don’t know the exact reference but the book by Jaccard & Wan is nice (it is one of the small green SAGE books). But actually all you need is the paper I posted in my first post. Learn how the residual centering approach works. It’s easy. Beyond that there is a R package about latent interactio by Christina Werner (Google for that) and Fan Yang-Wallentin (a former doctoral student of Yöreskog works at a further R package currently. An if course you can buy Mplus who’s applies the LMS approach (with no product term indicators).

HTH

Holger

Please see the appendix about moderation test in AMOS. Hope this helps.

I want to test the gender differences in the effect of one variable to another, How to do that in AMOS. I have found the R2 and regression coefficient in AMOS for male and female, next to know the the effect was significance with respect to gender? please

I am sharing one of my papers based on multi-group moderation analysis using AMOS.

Good question. Following.

Thank you Rayees for your paper

Dear Rama first understand mediation vs moderation than apply it through amos. take multigroup analysis and compare chi square value of constrained and unconstrained model

Good question.

Any suggestion regarding continuous moderator of the latent variable? If anyone has any ref please share. Thanks

The following web site migh be useful as:

AMOS for Beginners

http://core.ecu.edu/ofe/statisticsresearch/SEM%20with%20AMOS%20I%20PDF.pdf