28. SPSS AMOS | Analyzing, Interpreting, and Reporting Multiple Mediators

[@selwynstanley9066]

Excellent. So easy to follow, Coherent and clear. Sheer genius!

[@researchwithfawad]

Thanks. I am glad you liked it.

[@AmeenKhan-ls6mw]

@@researchwithfawad one IV, one DV and two mediators. The same model as you discuss in the video. The direct path would be single beta for both mediators??

[@Nidz1]

Excellent Professor. Your teaching style is so easy going and cool.

[@researchwithfawad]

Glad you think so!

[@salaheldinkassab1005]

Very useful step by step explanation of difficult concepts. Well done..

[@researchwithfawad]

Glad it was helpful!

[@damithagayan9672]

I totally agree. Easy to understand and step by step.

[@researchwithfawad]

Thanks. I am glad you liked it.

[@Clonerianyambali]

You are doing a great job! This was very helpful! Thank you very much!

[@researchwithfawad]

You're very welcome!

[@meryaamelkaddouri1386]

THANK YOU. PERFECT EXPLANATION💯

[@researchwithfawad]

Thank you. I am glad you liked it.

[@pringalext]

this is so helpful, thank you Research with Fawad

[@researchwithfawad]

Glad it was helpful!

[@talhamansoor7108]

Very informative and comprehensive session thankyou sir

[@researchwithfawad]

Most welcome

[@shakerodmunuhwamunuhwa1648]

Thank you so much @ Research With Fawad, so informative and straight foward. Is this the same way we report mediation results for PHD project?

[@researchwithfawad]

Thanks for your appreciation. I am glad the video was helpful. Sure, you can report in PhD thesis in similar format.

[@madhumitabanerjee4542]

Sir, thank you so much for this wonderful tutorial. I can see that from the results we are able to find the indirect effect of OL on OP via CC, and the indirect effect of OL on OP via Orc. Can you please help me understand how can we assess from the results the direct effect of OL on OP only in the presence of CC or the direct effect of OL on OP in the presence of OrC alone? Without this info, it will be difficult to assess whether it is a case of partial or full mediation in the presence of a single mediator. (The direct effect of OL on OP in presence of both CC and OrC is clearly understood). Thank You very much

[@researchwithfawad]

Thanks for watching.

[@Alhamzah_F_Abbas]

Brilliant work prof. thank you. is it necessary to report t-statistics as well?

[@researchwithfawad]

Thanks. I am glad you liked it. Yes please

[@manoshaoshani7327]

Thank you so much for the series of videos of AMOS, that are really valuble and very clear. It is really helpful. Sir, I have a small Q. Can we keep the corvariances (double arrows) between error terms of indicators of a mediator? As an example, in your model can we draw covariance between e9 and e10?

[@researchwithfawad]

Thanks for watching. Yes we can within the error terms of any construct.

[@manoshaoshani7327]

Thank you for the clarification@@researchwithfawad. I must say again your videos were very much helpful for me to carry out the advanced analysis of my research. Hope to continue watching all your videos to brush up my Statistical knowledge. Thank You Sir!!!

[@sarahguinoubi4199]

Thank you again Sir, I have a question concerning Bootstrap. Could you please explain it to me, why we use it in mediation effect ? when we choose Bootstrap 2000 or 5000 ?

[@researchwithfawad]

Pleasure. I am glad you liked it. Bootstrapping is used in case your data is not normal, recommended is 5000 or above.

[@moumoon000]

thank you so much for the clear and easy to understand explanation, Professor. I have a problem in running the bootstrap said that "An error occurred while attempting to fit the model. During the analysis of a bootstrap sample, an attempt was made to compute the correlation between 2 variables, one of whose estimated variances failed to be positive. The attempt was made because ' Standardized Estimates' in the 'analysis properties' window was checked or because the Standardized method was used." but I still can interpret the result as I follow the video. will this be a problem for any misinterpretation? thank you so much for any feedback

[@researchwithfawad]

Thanks for watching. How much is your sample, plus there are any categorical variables in the model.

[@Arjumanmasuma]

Thank you so much sir ..I followed your mentioned procedure. However, I'm getting an error in the estimand. The error is "Keyword is not valid as indentifier. 'IS' is not a member of 'v'. Keyword is not valid as identifier". Please help me with the same and i have mailed you my model. Thanks in advance

[@researchwithfawad]

Thanks for watching. Change the construct name from IS to something else.

[@rraj3167]

Thank you so much for all your videos!! I have a doubt. if I do CFA on my model and after many modifications, I am able to get Model fit, but when I run SEM or mediation , my model fit indices go down, what am I supposed to do then? Am I supposed to show covariances between the error terms in the SEM /Mediation model? or Should I report the model fit indices from my CFA and not the mediation model? please help

[@researchwithfawad]

Thanks for watching. You may use PLS-SEM, if you are having issues with CFA.

[@Liamzp]

The model that I am calculating includes one dependent variables with three categories. Three mediators, each mediator consisting of three categories, and one independent variable. I take three mediators and one dependent variable as latent variables. I found that the one mediator was statistically significant on the relationship between independent variable, independent variable. But how I can know this relationship was mediated by which specific category involved in this significant mediator ?

[@researchwithfawad]

Do the categories have an order?

[@AmeenKhan-ls6mw]

while discussing the mediation analysis, isn't it compulsory to discuss the model fit?

[@researchwithfawad]

Report fit for the structural model, that includes the mediation in the model.

[@upasanadhanda603]

I had a doubt. we report standardized direct and indirect effects but here you are showcasing user estimands calculations with unstandardized values. Can you please clarify.

[@researchwithfawad]

You can report both. Report STD or Un STD.

[@shaaneemaine966]

Great explanation sir. I have a few queries: 1. When I introduce my first mediator the direct relationship between IV and DV disappears indicating towards full mediation but when I am introducing my second mediator it gets significant p-values in bootstrap results but one of the direct relationship between the mediator(2) and DV doesn't exist. Though if i check the model only with the mediator (2), there exists full mediation there as well. a) should I check for serial mediation as well? b) what values for direct effect between IV and DV needs to be reported? Only the one with mediator (1) or for the full model even if mediator (2) is not contributing to it. Thanks in advance

[@researchwithfawad]

Thanks for your appreciation. 1. When I introduce my first mediator the direct relationship between IV and DV disappears indicating towards full mediation but when I am introducing my second mediator it gets significant p-values in bootstrap results but one of the direct relationship between the mediator(2) and DV doesn't exist. Though if i check the model only with the mediator (2), there exists full mediation there as well. a) should I check for serial mediation as well? Check for Serial mediation if it is part of your research scope, if not, NO, report results as they are. b) what values for direct effect between IV and DV needs to be reported? Only the one with mediator (1) or for the full model even if mediator (2) is not contributing to it. Report the full model with any number of mediators, please.

[@muhammadaamirkhan8182]

If we have some control variables, should we also include them in the mediation analysis?

[@researchwithfawad]

Yes, you can if you would like to assess their role.

[@ibrahimabdullatifibrahim9942]

please sir, how can i report direct and indirect effect after conduct constraints, we have parameters for each group thanks very much for your excellent videos

[@researchwithfawad]

Thank you. The reporting procedure doesn't change. For mediation you still have to report the direct and indirect effects.

[@Yash_k_28]

When I am going to status bar it is showing "create a new estimand" not "define estimand". What should I do?

[@researchwithfawad]

You can use it to create/define an estimand.

[@kekeshiheteki4495]

Question, How to find the p-value for direct effects, indirect effects, and total effects?

[@researchwithfawad]

Please watch kzbin.info/www/bejne/n2WXmHmvpr53h6s

[@neiban.h6806]

Sir in case i have a multiple independent variables while setting parameters do, i need to set for each construct differently or what should I do.. Thank you

[@researchwithfawad]

Yes each should be done separately for multiple mediators

[@neiban.h6806]

@@researchwithfawad Thank you sir

[@mukhtar1319]

In my case p1 and p2 are negative and p3(direct effect) is postive. Multiple of p1 and p2 is mathematical positve and in essence also these two paths are supposed to be negative as per theory and literature. With direct effect positive and indirect effect also positive, when actually the paths to and from mediators are actually negative, are significant at .05. How shall I interpret this result.

[@researchwithfawad]

If they are supposed to be negative and they are negative, interpret and support hypotheses as required. For mediation look for significance of indirect effect instead of signs.

[@swetaagrawal8589]

Sir I tried this model but while creating a new estimate it is showing error it shows that function value doesn't return a value on all code paths in null reference could occur at the runtime when the result is used and then it is writing declaration expected

[@researchwithfawad]

Email the code and screenshot of the model at [e-mail]

[@AmeenKhan-ls6mw]

suppose i am running the same model and i am interested to compare the direct and indirect effect both path i.e. OC -->CC-->OL and OC -->CC -->OP. The indirect effect is very understandable but what about the direct effect of two different path? in that case a single coefficient and P-value will represent the direct effect for both path? please comment

[@researchwithfawad]

If you have different DVs, how can there be a single coefficient.

[@AmeenKhan-ls6mw]

@@researchwithfawad I mean the same model as you discussed in the video. One IV, one DV and 2 mediators. The direct will contains a single coefficient value one both mediators?

[@researchwithfawad]

Yes, in that case there will be on coefficient

[@muhammadaamirkhan8182]

Prof. Latif: Though, I followed your mentioned procedure. However, I'm getting an error while testing/saving the estimand. The error is "Keyword is not valid as indentifier. 'Mod' is not a member of 'v'. Keyword is not valid as identifier". Please advise which keyword(s) should be worked on to remove the error, thanks.

[@researchwithfawad]

Please change the Labels of the Regression Weights. They are keywords. This will solve the issue.

[@muhammadaamirkhan8182]

@@researchwithfawad Actually, I used different combinations of the alphabets. However, every time, I gets the mentioned error and results can't be calculated 😪😭

[@researchwithfawad]

Please email the screenshot on [e-mail] along with screenshot of the model.

[@siroboratsify]

thank you. can we have a copy of the sample data spss you use in the lecture?

[@researchwithfawad]

Thanks for your interest. Not this one, but i do have another sample data. You can email me on [e-mail]

[@Rohit_chauhan097]

when i am running estimands, it shows syntax error. I followed every steps similar to you.

[@researchwithfawad]

What is the error?

[@Rohit_chauhan097]

​@@researchwithfawadit says syntax error.... although i m typing same as you....can i mail you the screenshot anywhere of the same??

[@sarahguinoubi4199]

Thank you so much, i have a question : if CC and OC are both significant, can we say that CC and OC fully mediated the relation between OL and OP ? Many thanks

[@researchwithfawad]

Full or Partial is based on the direct relationship between IV and DV. If the relationship between IV and DV is significant in presence of the mediator, it is partial otherwise full.

[@sarahguinoubi4199]

@@researchwithfawad Thank you Sir

[@talhamansoor7108]

i faced error when i try to run analysis such "the file 11.simpleEstimand contains syntax errors" mean user defined new estimands not properly working would you help me how to fix this issue?

[@researchwithfawad]

Send me a screenshot of the defined estimand.

[@muhammadaamirkhan8182]

Did you solve this issue? I'm also getting the same error though I used the same labels. 😪

[@dr.abdulkadirjeilani320]

@@researchwithfawad Did you solve this issue? I'm also getting the same error

[@Liamzp]

The model that I am calculating includes one dependent variables with three categories. Three mediators, each mediator consisting of three categories, and one independent variable. I take three mediators and one dependent variable as latent variables. I found that the one mediator was statistically significant on the relationship between independent variable, independent variable. But how I can know this relationship was mediated by which specific category involved in this significant mediator ?

[@researchwithfawad]

Do the categories have an order.

[@Liamzp]

no they are just nominal variables.

[@researchwithfawad]

Use multinomial logistic regression when DV is with categories that are nominal.