Thanks for the great video. It was incredibly helpful. Here's an updated citation from the Marsh et al (2004) authors that offers more guidance on creating product indicators when you have an unequal number of items: Wu Y., Wen, Z., Marsh, H. W., & Hau, K.-T. (2013). A Comparison of Strategies for Forming Product Indicators for Unequal Numbers of Items in Structural Equation Models of Latent Interactions, Structural Equation Modeling: A Multidisciplinary Journal, 20(4), 551-567. The 2004 article deals mostly with interactions between measures with an equal number of items, and only briefly discusses what you *might* do if the number of items is unequal.
great MR. Collier. It will be wonderful if you describe the way of standardizing and creating interaction variables with some subconstructs before imputing AMOS.
For sure. I have a few other videos lined up first. To create those standardized values you need to do this in SPSS before brining it in to AMOS. You will go to the "Analysis" drop down and then "descriptive statistics" and then "descriptives". One you pull over the items you are concerned with you need to check a box called "save standardized values as variables" and hit ok. This will standardize or mean center all your selected variables. I have a pretty lengthy discussion in my book about it if you are looking for more detailed information.
@@joelcollier9387 Hi. Would you describe how can we include the moderator of Education level as a non-continuous variable? How should we include it in the model using AMOS?
@@rahelehh1147 You are going to have to run the moderator as a two group analysis. I have a video that breaks it down called "Two group analysis in SEM using AMOS". That should help fix your continuous variable issue.
Sir I have been trying to build a model where we have a second-order factor that moderates relationships between three IVs and a DV. I am fine with first-order moderators but really wondering how the interaction would work if the moderator itself is a second-order factor. Will it be okay if I collapse the lower level factors? Can you suggest any material that could help us here?
Thanks. The data is this video was made up for the book Applied Structural Equation Modeling using AMOS. I needed a clean data set to work with. Saying that, the original research using these variables was published in the Journal of Business Research. The title was: Idiosyncratic service experiences: When customers desire the extraordinary in a service encounter
@@joelcollier9387 I apologize for any confusion. I would like to cite your method for testing the moderation effect in my paper, so I need a scientific reference for your method. I don't require a reference for the data. Thank you! :)
@@yl3355 You can cite the book but you can also cite Marsh, Herbert W. Zhongline Wen, and Kit-Tai Hau. (2008). "Structural Equation Models of Latent Interaction and Quadratic Effects", in Gregory Hancock and Ralphy D. Mueller (eds) Structural Equation Modeling: A Second Course. Greenwich, CT: Information Age
Too many thanks for your great presentation, Mr. Joel Collier. I notice that you generate a component variable for the interaction. I wonder if I need to conduct a confirmatory factor analysis (CFA) for the model or conduct a full SEM with reporting CFA results, how many factors would be included? In your presentation, there seem to be five factors (friendliness, Adaptive behavior, Interaction_A_X_F, customer_Delight, and Word_of_Mouth). However, there actually are four factors. I am a little confused about the number of factors for conducting CFA for an SEM with moderating effect. Looking forward to your reply.
Performing a CFA before assessing structural relationships is always advisable. You do not need a run an interaction term in a CFA. You only need to include the unobservable variables that have observables attached that are capturing the concept. So, in this example I would only run the CFA with friendliness, adaptive behavior, customer delight, and word of mouth. Hope that helps.
Great! Can we apply the same approach to examine mediation (not for the independent variable) and determine whether the moderator moderates the effects of the mediator on the outcome?
Yes, you absolutely can do that. You will just have an interaction term with the mediator and moderator to the DV along with a moderator with a relationship to the DV.
Thanks! What would be the indirect effect of ID on DV? I mean do I need to have a direct effect from ID to the moderator and interaction term as well?@@joelcollier9387
Hi Dr Collier, Thanks for the great video! I already ordered a copy of your book, can't wait for it to arrive!! Please could you clarify if it is okay to use the "groups" feature in AMOS when I have multiple categorical moderators? For example, I have gender (male,female) and residency (resident, non-resident). Can I create those two groups as moderators in the same model or do I have to run separate models, one with gender as moderator, and a second with residency as moderator? If I can use one model, then can I interpret results for only one (of four) group at a time or is it possible to interpret for 2 groups each, e.g. male AND resident? Thanks a lot!
It is the preferred way to test categorical variables that dichotomous. It is not ideal with multicategorical moderators. The reason why is when you form groups and look at a chi-square difference test across the groups, this works great with two groups....its a problem with more than two. The reason is the groups function will try to assess differences across all the groups as a whole instead of individual comparisons. You can do multiple two group analysis of categories. It is not ideal but it is an option. P.S. hope you enjoy the book.
Thanks for your reply Dr Collier! I'm still awaiting delivery of the textbook so I'm grateful for your help in the meantime. Does the book cover in detail how to use AMOS SEM with a categorical (e.g. gender) DV? I couldn't find any resources on this besides an IBM support page recommending to use Bayesian rather than maximum likelihood estimation. I seem to recall a chapter on advanced SEM in your book but I don't know if it goes into detail about analysing and reporting on models with categorical DVs. I would much appreciate if you could direct me to additional resources on this form of SEM in AMOS. Thanks again, Omo
Please help! Can you run interactions in a SEM with second-order latent variables (i.e., a latent variable explained by other latent variables). I want to run an interaction between 2 predicting latent variables, but one of them is a second order LV, with 4 sub-constructs. Is is a case of running interactions between each sub-construct and the other predicting LV?
It does make things way more complicated with a second order construct. The good news is you can run it. You will need to have a unobservable interaction that is made up of interactions with all first order dimensions and the moderator. I would use a matched pairs method for each first order dimension
hi professor, thank you so much for your video. However, I have one question regarding the interaction term. I can understand that the number of indicators depends on the moderator or iv whichever has fewer items. and then selects an equal number of the scale items based on their factor loading. but then, how do you decide how to interact these items. eg. in your video, friend1_x_adapt1, friend2_x_adapt2, and friend3_x_adapt4. can't it be eg friend1_x_adapt2, friend2_x_adapt4, and friend3_x_adapt1 ?
You choose the interactions based on the which ones has the strongest loading on the unobservable interaction construct that is created. Yes, you can use a lot of different combinations but you want to use the one with the strongest loading on the interaction construct. ...hope that helps.
@@joelcollier9387 thank you so much for replying. just got a further question. how to check the factor loading for the interaction construct? do I need to create all the possible combinations and then run efa to check the factor loading?
@@joelcollier9387 Thank You. But can we use Mean centering for either IV, Moderator, and Interaction? and Beside Mean centering, can we use unstandardized residual for interaction?
@@neardey_near1976 If you are going to use mean centering for the IV then you need to mean center the moderator too. In that instance, the IV, moderator, and interaction are mean centered. As for the interaction, I have never used unstandardized residuals to form the interaction term. I am not sure if that is a valid way to approach it.
You are my absoloute hero! As I did not really got grip on the Marsh et al.'s paper, this video safed my thesis! Many thanks!
Great Presentation and I gained much knowledge from your Video. Thanks Mr. Joel Collier
Thank You, great job - short, without blablabla... And specially thanks for materials attached
Many thanks Mr. Collier. I have already read your book. It is very helpful.
Thanks for the great video. It was incredibly helpful. Here's an updated citation from the Marsh et al (2004) authors that offers more guidance on creating product indicators when you have an unequal number of items:
Wu Y., Wen, Z., Marsh, H. W., & Hau, K.-T. (2013). A Comparison of Strategies for Forming Product Indicators for Unequal Numbers of Items in Structural Equation Models of Latent Interactions, Structural Equation Modeling: A Multidisciplinary Journal, 20(4), 551-567.
The 2004 article deals mostly with interactions between measures with an equal number of items, and only briefly discusses what you *might* do if the number of items is unequal.
Thanks. I appreciate it. Glad the video helped
great MR. Collier. It will be wonderful if you describe the way of standardizing and creating interaction variables with some subconstructs before imputing AMOS.
For sure. I have a few other videos lined up first. To create those standardized values you need to do this in SPSS before brining it in to AMOS. You will go to the "Analysis" drop down and then "descriptive statistics" and then "descriptives". One you pull over the items you are concerned with you need to check a box called "save standardized values as variables" and hit ok. This will standardize or mean center all your selected variables. I have a pretty lengthy discussion in my book about it if you are looking for more detailed information.
@@joelcollier9387 Hi. Would you describe how can we include the moderator of Education level as a non-continuous variable? How should we include it in the model using AMOS?
@@rahelehh1147 You are going to have to run the moderator as a two group analysis. I have a video that breaks it down called "Two group analysis in SEM using AMOS". That should help fix your continuous variable issue.
Sir
I have been trying to build a model where we have a second-order factor that moderates relationships between three IVs and a DV. I am fine with first-order moderators but really wondering how the interaction would work if the moderator itself is a second-order factor. Will it be okay if I collapse the lower level factors? Can you suggest any material that could help us here?
dear sir, my path model GFI and AGFI value dont seee.
Thank you for a great video! Could you please specify any scientific literature you used to create this demo as a reference?
Thanks. The data is this video was made up for the book Applied Structural Equation Modeling using AMOS. I needed a clean data set to work with. Saying that, the original research using these variables was published in the Journal of Business Research. The title was: Idiosyncratic service experiences: When customers desire the extraordinary in a service encounter
@@joelcollier9387 I apologize for any confusion. I would like to cite your method for testing the moderation effect in my paper, so I need a scientific reference for your method. I don't require a reference for the data. Thank you! :)
@@yl3355 You can cite the book but you can also cite Marsh, Herbert W. Zhongline Wen, and Kit-Tai Hau. (2008). "Structural Equation Models of Latent Interaction and Quadratic Effects", in Gregory Hancock and Ralphy D. Mueller (eds) Structural Equation Modeling: A Second Course. Greenwich, CT: Information Age
Much appreciated!!
my path model GFI and AGFI value dont see, can you help me pelease,
Too many thanks for your great presentation, Mr. Joel Collier. I notice that you generate a component variable for the interaction. I wonder if I need to conduct a confirmatory factor analysis (CFA) for the model or conduct a full SEM with reporting CFA results, how many factors would be included? In your presentation, there seem to be five factors (friendliness, Adaptive behavior, Interaction_A_X_F, customer_Delight, and Word_of_Mouth). However, there actually are four factors.
I am a little confused about the number of factors for conducting CFA for an SEM with moderating effect.
Looking forward to your reply.
Performing a CFA before assessing structural relationships is always advisable. You do not need a run an interaction term in a CFA. You only need to include the unobservable variables that have observables attached that are capturing the concept. So, in this example I would only run the CFA with friendliness, adaptive behavior, customer delight, and word of mouth. Hope that helps.
Dear sir, is path analysis moderation procedure better or full structural moderation model is best.
Great! Can we apply the same approach to examine mediation (not for the independent variable) and determine whether the moderator moderates the effects of the mediator on the outcome?
Yes, you absolutely can do that. You will just have an interaction term with the mediator and moderator to the DV along with a moderator with a relationship to the DV.
Thanks! What would be the indirect effect of ID on DV? I mean do I need to have a direct effect from ID to the moderator and interaction term as well?@@joelcollier9387
Is it possible to test moderated moderation in SEM?
Hi Dr Collier,
Thanks for the great video! I already ordered a copy of your book, can't wait for it to arrive!!
Please could you clarify if it is okay to use the "groups" feature in AMOS when I have multiple categorical moderators?
For example, I have gender (male,female) and residency (resident, non-resident). Can I create those two groups as moderators in the same model or do I have to run separate models, one with gender as moderator, and a second with residency as moderator?
If I can use one model, then can I interpret results for only one (of four) group at a time or is it possible to interpret for 2 groups each, e.g. male AND resident?
Thanks a lot!
It is the preferred way to test categorical variables that dichotomous. It is not ideal with multicategorical moderators. The reason why is when you form groups and look at a chi-square difference test across the groups, this works great with two groups....its a problem with more than two. The reason is the groups function will try to assess differences across all the groups as a whole instead of individual comparisons. You can do multiple two group analysis of categories. It is not ideal but it is an option. P.S. hope you enjoy the book.
Thanks for your reply Dr Collier!
I'm still awaiting delivery of the textbook so I'm grateful for your help in the meantime.
Does the book cover in detail how to use AMOS SEM with a categorical (e.g. gender) DV? I couldn't find any resources on this besides an IBM support page recommending to use Bayesian rather than maximum likelihood estimation.
I seem to recall a chapter on advanced SEM in your book but I don't know if it goes into detail about analysing and reporting on models with categorical DVs.
I would much appreciate if you could direct me to additional resources on this form of SEM in AMOS.
Thanks again,
Omo
@@omot9983 Yes. There is a whole chapter on how to use categorical variables in SEM using AMOS. That should really help you out.
Please help! Can you run interactions in a SEM with second-order latent variables (i.e., a latent variable explained by other latent variables). I want to run an interaction between 2 predicting latent variables, but one of them is a second order LV, with 4 sub-constructs. Is is a case of running interactions between each sub-construct and the other predicting LV?
It does make things way more complicated with a second order construct. The good news is you can run it. You will need to have a unobservable interaction that is made up of interactions with all first order dimensions and the moderator. I would use a matched pairs method for each first order dimension
hi professor, thank you so much for your video. However, I have one question regarding the interaction term. I can understand that the number of indicators depends on the moderator or iv whichever has fewer items. and then selects an equal number of the scale items based on their factor loading. but then, how do you decide how to interact these items. eg. in your video, friend1_x_adapt1, friend2_x_adapt2, and friend3_x_adapt4. can't it be eg friend1_x_adapt2, friend2_x_adapt4, and friend3_x_adapt1 ?
You choose the interactions based on the which ones has the strongest loading on the unobservable interaction construct that is created. Yes, you can use a lot of different combinations but you want to use the one with the strongest loading on the interaction construct. ...hope that helps.
@@joelcollier9387 thank you so much for replying. just got a further question. how to check the factor loading for the interaction construct? do I need to create all the possible combinations and then run efa to check the factor loading?
@@zheli9483 You will create all possible combinations and then run a CFA (not an EFA) to check the loadings.
Excuse me sir, Did you use mean centering for Moderator ?
You really only need the mean centering for the interaction. The IV and the moderator can be original values.
@@joelcollier9387 Thank You. But can we use Mean centering for either IV, Moderator, and Interaction? and Beside Mean centering, can we use unstandardized residual for interaction?
@@neardey_near1976 If you are going to use mean centering for the IV then you need to mean center the moderator too. In that instance, the IV, moderator, and interaction are mean centered. As for the interaction, I have never used unstandardized residuals to form the interaction term. I am not sure if that is a valid way to approach it.
@@joelcollier9387 Thank You sir.