Testing specific indirect effects in AMOS via User defined Estimands

This video provides a walk-through of how you can compute specific indirect effects (and other effects) via the user-defined estimands option available in the program.
The powerpoint referenced in the video can be downloaded here:
drive.google.com/file/d/17KA7...
A copy of the data can be obtained here: drive.google.com/file/d/1RLAo...
A copy of the AMOS file can be obtained here: drive.google.com/file/d/1PmnU...
Link to previous video in this series: • Introduction to path a...

[@The_Quota_Official]

That’s hilarious, I had to find another video in the morning today to find out how to this, and you just posted it now 😂

[@bbaraiourdigitalclassroom834]

Thank you so much sir your lectures are very simple and easy to understand. I am very new to AMOS, using it for my study (phd) and truly speaking without your demos I couldn't have progressed an inch. 🙏😊🇮🇳 Regards.

[@selcukaslan1279]

Thank you very much, your videos are great for the researchers :)

[@maryamsabzevarzadeh2318]

That was very helpful. Thank you!

[@vincenzocalvo5618]

Thank you very much. Your video is very clear and useful. Kind regards

[@trunguc4911]

Thank you, this is really helpful. I'd like to ask one additional question: How can I calculate each standardized SIE??

[@haerolddeanlayaoen2638]

Dear Mike, Thank you for posting this video. I am really learning a lot from the videos you are posting and we are really thankful that you extend such efforts to share your invaluable knowledge on contemporary academic tools and topics. Going back to this topic, I understand that the default values provided in the "User-defined estimands" in AMOS Output (i.e. estimates, confidence intervals, and p-values) are UNstandardized values. What if I am interested with the Standardized values of specific indirect effects, how am I going to access/display it in AMOS Output? I hope you can provide enlightenment on this regard. Yours sincerely, Haerold

[@RegorzStatistik]

Thank you, that is quite helpful. I'd like to ask one additional question: Could you use the User defined Estimands to run a bootstraped test for contrasts (i.e. to test wheter the specific indirect effect 1 is significantly different form the specific indirect effect 2)?

[@mikecrowson2462]

Yes indeed. I didn't show you that in this presentation, but it is definitely something you could do. As an example, if you have two indirect effects (IE1 and IE2), you could test the difference in the following way (bootstrapping the difference in indirect effect): diff=IE1-IE2 Cheers!

[@RegorzStatistik]

Thank you very much!

[@rashydashy12]

Can someone please explain @Mike Crowson's reply to me? Using AMOS, I am trying to run a pairwise comparisons between two specific indirect effects to see whether one has a statistically stronger effect than the other one. Any input would be greatly appreciated!

[@elianelian221]

Hi Mr. Crowson, Really appreciate! why do I keep getting error messages if I don't check the "Estimate means and intercepts" box? I don't have missing data in my dataset. Also, you said we can't run bootstrapping with missing data. The thing is I can run bootstrapping with user-defined estimand when that box was checked. How could this happen? Thank u

[@myokoaung7466]

Thanks, and want to write syntax by VB in amos 21 for user defined estimand

[@niveditajha890]

The video is very useful. But I have a difficulty in creating estimands. In my Amos status bar it shows create estimands and there is option for 'define estimands'. What could be the reason for this and how can I resolve this?

[@jinzapark]

I have the exactly same issue. Any solution?

[@nsa2058]

I did everything, syntax wass OK but amos didn't run the analysis, no red indicator. could you please help me?

[@selcukaslan1279]

Mr. Crowson, is there any way of getting "standardized" values of user defined estimands for specific indirect effects? Thank you

[@mikecrowson2462]

Hi there. No unfortunately I haven't seen any approach that will do that for you in AMOS (if what you are asking is if you can generate bootstrap confidence intervals for specific standardized indirect effects). About the only solution I can think of is to run your model using standardized variables and then bootstrapping the effects computed via the user-defined estimands (but that would be a pain in the neck). On the other hand, if you request standardized coefficients and use the Direct, Indirect, and Total effects option under analysis properties & bootstrap, you can get the bootstrap confidence intervals for the total standardized indirect effect. Thanks for your question & for visiting!

[@selcukaslan1279]

@@mikecrowson2462 Mr. Crowson, I am grateful for your detailed answer. I may not understand your answer or explain my question clearly. In the video above, the "user defined estimands" you get at the 12:42 is unstandardized. Is it possible to obtain the standardized equivalents of exactly these values? Thank you again.

[@mikecrowson2462]

@@selcukaslan1279 Well that's what I was trying to get at in my long-winded answer. The short answer is, I've not seen how that would be possible via User-Defined Estimands. What I was trying to do was to offer alternatives to the limitation of only being able to use it to generate unstandardized indirect effects. If you are unconcerned with generating confidence intervals for standardized indirect effects - and are only interested in the standardized indirect effects, you could simply multiply the standardized path coefficients in your model to arrive at each specific standardized indirect effect. But if you are wanting the confidence intervals, then the only workarounds I could think of are what I suggested previously. Best wishes!

[@selcukaslan1279]

@@mikecrowson2462 Mr. Crowson I think I understand now much better :)) I apologize for tiring you, best regards :)

[@mikecrowson2462]

@@selcukaslan1279 You didn't tire me :) You have a great day!

[@muti7880]

how is the conditions of a significant relationship? i think you never really explain that thankyou