thank you so much. we can use the following path to display the names of the variables instead of their labels. View → Interface Properties → Misc → untick Display Variable Labels
I have a path model with more than two variables: several independent variables, one dependent variable and mediator. I understood I have to do factor analysis for each model that goes into SEM before I actually do SEM. My question is: do I have to create these relationships between all the variables (independents, mediator and dependent)? I mean connect them with double-sided arrow. For example independent variable 1 to independent variable 2, independent variable 1 to independent variable 3, Independent variable 2 to independent variable 3, independent variable 1 to mediator and so fourth.
Hello! Is the dataset mentioned in the tutorial freely available? If yes, where may I download a copy to follow through the procedure? Thank you by the way for this well curated discussion on CFA!
Hello, I'm trying to run (calculate estimates) the model and it says:'no working data has been specified. Please select data file from the file menu and specify a data file.'
Sir, do we need to put covariance in our independent variables in structural model even though in our hypothesis/conceptual model we do not correlate those?
Section 1 and Section 2 have observed data, together they are defining Latent Variable "Listening 1" , my question is that latent "Listening 1" is sum of Section 1 and section 2 or average of section 1 and section 2 or they have individual impact/ role on latent variable.. what is the data inside Latent variable ?
The test items (Sections) measure the latent variable. The latent variable itself is the weighted sum of the items. see Thompson's (2005) book on Confirmatory factor analysis for further information, please.
Hey Dr, suppose I have three (3) latent variables as follows: Listening1, Listening2 and Listening3. In case my theoretical framework support that listening 1 is correlated to Listening2, but does not support that Listening1 is correlated to Listening3, should I still covariate Listening1 and Listening3 together to conduct the CFA analysis? or I should respect the theoretical framework hypothesised in my thesis, and which it was extensively explored and validated by other research? thank you.
Hi Dr I would like to ask you : When we estimate measurement model by using CFA, what kind of test theory that used e.g IRT or G theory..? Could you please explain and if there is reference to read more about this issue?
Hi, Thanks for the video. It was very helpful. I am trying to perform SEM analysis on my collected data. Do I need to run PCA or CFA before performing SEM. Please advise
Hi, thank you for your videos! My data is not normally distributed, its z-value is not in the normal area and the kolmogorov and wilk tests are significant... so what should I do, when having not normally distributed data and wanting to create a SEM? Does it matter? Kind regards.
@@VahidAryadoust Thank you for your fast reply, I havent watched the second at this point, so now i know. Another question: I am using a Likert-Scale in my survey and now i am in a dilemma... there are paper which say one shouldnt interpret ordinal data as continuous and others which argue vice versa. What is your opinion on that? Can I transform my ordinal data from the likert-scales into continuous data by just manipulating the data set and adding numbers? Or should I stick to my ordinal data and do bayesian estimation?
@@normanschmitt6303 I do not recommend that transformation. Just enter the Likert scale data into the analysis directly. For Likert scale data you won't need Bayesian estimation.
@@VahidAryadoust thank you! so you would suggest to just add the ordinal data into the model and then, in case of SEM, (because my data is not normally distributed due to the various categories of the 5-point and 7-point-likert scale) perform a ULS, SFLS or ADF analysis? Many thanks, Mr. Aryadoust, you help a lot with my thesis!
thank you so much. we can use the following path to display the names of the variables instead of their labels. View → Interface Properties → Misc → untick Display Variable Labels
Good info! You keep inspiring me, Dr. Vahid.
Nice to see CFA series! CFA is important tool, but not that easy to understand and/or draw in AMOS. With this series, it can be worked out smoothly.
I have a path model with more than two variables: several independent variables, one dependent variable and mediator. I understood I have to do factor analysis for each model that goes into SEM before I actually do SEM. My question is: do I have to create these relationships between all the variables (independents, mediator and dependent)? I mean connect them with double-sided arrow. For example independent variable 1 to independent variable 2, independent variable 1 to independent variable 3, Independent variable 2 to independent variable 3, independent variable 1 to mediator and so fourth.
why you do not use Jamovi insted of SPSS?
Hello! Is the dataset mentioned in the tutorial freely available? If yes, where may I download a copy to follow through the procedure? Thank you by the way for this well curated discussion on CFA!
Hello, I'm trying to run (calculate estimates) the model and it says:'no working data has been specified. Please select data file from the file menu and specify a data file.'
Really helpful and nice demonstration..many thanks
Keep working Good like all your videos.
Sir, do we need to put covariance in our independent variables in structural model even though in our hypothesis/conceptual model we do not correlate those?
Section 1 and Section 2 have observed data, together they are defining Latent Variable "Listening 1" , my question is that latent "Listening 1" is sum of Section 1 and section 2 or average of section 1 and section 2 or they have individual impact/ role on latent variable.. what is the data inside Latent variable ?
The test items (Sections) measure the latent variable. The latent variable itself is the weighted sum of the items. see Thompson's (2005) book on Confirmatory factor analysis for further information, please.
Very informative videos! Have a nice day :)
Thank you very much :)😍
Hey Dr,
suppose I have three (3) latent variables as follows: Listening1, Listening2 and Listening3. In case my theoretical framework support that listening 1 is correlated to Listening2, but does not support that Listening1 is correlated to Listening3, should I still covariate Listening1 and Listening3 together to conduct the CFA analysis? or I should respect the theoretical framework hypothesised in my thesis, and which it was extensively explored and validated by other research? thank you.
I would try two models, one correlating and one not correlating the factors and examine their fit. Whichever has a better fit is preferred.
hi Dr can I have the data showed in the video to run ?
Hi Dr
I would like to ask you :
When we estimate measurement model by using CFA, what kind of test theory that used e.g IRT or G theory..? Could you please explain and if there is reference to read more about this issue?
It is neither; CFA is a simple form of structural equation modeling (SEM). So, you would use principles of SEM in using CFA.
Thanks a lot!
Hi, Thanks for the video. It was very helpful. I am trying to perform SEM analysis on my collected data. Do I need to run PCA or CFA before performing SEM. Please advise
If the structure of the instruments (questionnaires) is not clear, you should confirm it first by using EFA / PCA.
@@VahidAryadoust Thanks for quick response
Thanks Dr
do we need to chick the normality in CFA? and way?
Yes, please check for normality.
Here is how: th-cam.com/video/v5P4iUj_eoM/w-d-xo.html
Hi, thank you for your videos! My data is not normally distributed, its z-value is not in the normal area and the kolmogorov and wilk tests are significant... so what should I do, when having not normally distributed data and wanting to create a SEM? Does it matter?
Kind regards.
Use other methods of parameter estimation as discussed in the video.
@@VahidAryadoust Thank you for your fast reply, I havent watched the second at this point, so now i know. Another question: I am using a Likert-Scale in my survey and now i am in a dilemma... there are paper which say one shouldnt interpret ordinal data as continuous and others which argue vice versa. What is your opinion on that? Can I transform my ordinal data from the likert-scales into continuous data by just manipulating the data set and adding numbers? Or should I stick to my ordinal data and do bayesian estimation?
@@normanschmitt6303 I do not recommend that transformation. Just enter the Likert scale data into the analysis directly. For Likert scale data you won't need Bayesian estimation.
@@VahidAryadoust thank you! so you would suggest to just add the ordinal data into the model and then, in case of SEM, (because my data is not normally distributed due to the various categories of the 5-point and 7-point-likert scale) perform a ULS, SFLS or ADF analysis? Many thanks, Mr. Aryadoust, you help a lot with my thesis!
@@normanschmitt6303 That sounds like a possible solution.
Good work! Can you suggest me free software to conduct a full SEM analysis? Thank you
Amos and Jamovi.
@@VahidAryadoust thank you for your reply. Do you mean the SPSS Amos?
@@emanuelemarino5164 Yes, I do. If ou like, you can watch my videos on doing SEM using AMOS. Please browse the channel for relevantvideos.
Hi Dr thank you for your videos! Are you able to show SEM in MPlus as well? :)
I will include that in my to-do list. Thanks for your interest in the channel.