SEM with AMOS: From Zero to Hero (18: Model fit assessment)
HTML-код
- Опубликовано: 16 мар 2021
- Learn everything you need to know to apply structural equation modeling (SEM) using AMOS in your research!
Video 18: Model fit assessment
Dr. Saeed Pahlevansharif
Download all files that I used in my SEM workshop from this link:
drive.google.com/drive/folder...
I have designed this hands-on course to provide you an intense yet enjoyable instructional experience that focuses on a large number of both introductory and advanced topics in SEM using the latest version of one of the most well-known software package, IBM AMOS (Analysis of Moment Structures). In this course I will provide the materials and instruction that you need to both develop a good understanding of SEM and to be able to thoughtfully apply a variety of basic and advanced models to your data. More specifically, this course will cover a broad range of topics such as
• data preparation,
• model development,
• model fit assessment using various indexes,
• construct reliability and validity assessment using various methods,
• different methods to improve your model,
• structural model assessment,
• hypothesis testing,
• mediation using Baron and Kenny approach
• mediation using Sobel test
• mediation using bootstrapping method
• moderation,
• and multi-group analysis
I know I already commented yesterday, but I just cannot express how thankful I am. This is the greatest series on how to use AMOS on all of RUclips. Thank you so much.
You're welcome! I'm so glad to hear that my videos have been helpful for you. It's always rewarding to know that my work is making a positive impact on others. All the best!
Simply amazing and to the point. Hats off!
life saver!!! thank you!!!!
You're welcome!
Best Explanation of CFA I have ever observed on youTube ...Great Effort Sir.Thanks for the Sharing
Thank you for your comment 😊
Best explanation so far! thank you sir!
One of the best video on CFA and its complications. Hoping it will help me too Insha Allah.
Thanks for the feedback. All the best
@@saeedsharif thank you
good teacher Dr. Saeed
appricate
This tutorial is absolutely superb. A real life saver, a huge thank you Dr Sharif
Thank you
You're most welcome!
Thank you so much for this vodeo i thinking for many days to how to improve my model and i find the solution here. Thanks
Thank you for your comment 😊
Hi Prof Sharif. Thank you very much for sharing your videos. They are really helpful. In this video, for modification indices, I wonder, why only one covariance way is accepted and the others are not. Do you have reference for that? I saw also in some related videos, they can covariance error terms from one constructs with ones from others and I am confused.
Thanks for your help.
Thank you so much, Dr, for the series on using AMOS; helpful and insightful. I have a question concerning improving the model using the residuals table; in my model, I don't have any value equal to or more than 4.00; however, how can I decide on removing an item based on the absolute values?
thank you, a lot for your presentation. can I take your ppt file to benefit it in my article? I need to support my result with formal resources
❤️❤️
Hi Professor, I have created 3 groups using cluster analysis on my first study variable (consists of 4 observed variables). Should i include this variable(4 items), in the CFA model??
Hi, sir thank you for the video it is great! I just had a question. How can we justify the correlation of error terms within the same construct in our paper? I have read that it is not allowed to do simply to improve model fit? Thanks!
Thank you very much for the helpful content. But I faced serious problem with model fit. I tried all the methods but my model is not even close to the assumptions of model fit. Where is the problem. Thank you Dr. Sharif.
Do you have a reference list of the authors cited on the video
Sir I am getting 2 of my factor loadings as 1.09 and 1.06..how to deal with it?
respected sir,
in CFA can CMIN/df value and RMR be zero?
also sir can GFI AND CFI can be 1.00 value?
aslo sir in my one construct CMIN/df value was nil, RMR VALUE was .000 and GFI and CFI value was 1.000 but my RMSEA value was .569 so, how can I improve my model fit? since RMSEA value is above .08
Prof, if I have a large sample size (n>500), can I accept lower loading values at around 0.3? I have found some literature that allows it but I'm still nervous. Thank you for the informative video!
Well.... I don't recommend it.. :) I understand there may be some studies that accepted lower loadings. However, it won't be easy to convince the examiners/reviewers.
Thank you for this video. It truly helped my instrument development and validation. But, why do the required points for CFI and TLI differ from other literature? Some say CFI and TLI would be a good fit if it’s greater than 0.95. Sir, can you clarify this matter? Anyway, thank you for your time.
Yes. There are different opinions. It depends how strict you want to be. You may refer to any of them as long as you acknowledge the limitations of each.
@@saeedsharif I understand it now, Sir. Thank you and please continue giving us worth-watching Statistics video lessons.
@@makathleenadona9957 All the best
Professor could you please help me with the SEM?
Hi Dr., I have problem with discriminant validity. The model is fit and convergent validity is met. I did SD and EFA but still has problem...
Do you have email? Thank you so much
You provided the authors for the recommended fit indices, but I can't find them anywhere online. I need some more information to be able to reference it. The citation alone is not enough.
You can use this book as your reference:
Pahlevan Sharif, S., & Sharif Nia, H. (2018). Structural equation modeling with AMOS. Tehran: Artin Teb.
@@saeedsharif Thank you, appreciate the help
If the NFi, GFI Values are less than 0.9 , we refuse the model ? ( while all the other values are ✅)
Not necessarily. You may refer to other indexes. Also, you may find references to support for example NFI greater than .85
Did you get the ans to this question? because i'm facing the same issue,
@@khifzarehman1929 See the answer above
@@khifzarehman1929 yes, the values less than 0.9 are not accepted. So u need to modify the model based on the modification indices and pattern matrix
Sir i m very worried my model fit is poor and discriminant validity too
In these series of videos I have provided some solutions to improve the model fit and discriminant validity. You may need to covariate some error terms, remove weak factor loadings, etc. For discriminant validity, you need to find the item with cross loading and remove it. You may perform an EFA on the problematic constructs and find the item with cross loading.
@@saeedsharif Sir, for my one variable average variance extraction is 0.395 and after deletion of one item it goes upto 0.421 and it is greater than msv. Composite reliability is 0.88 for the same variable, can i use this value with reference to Fornell Locker reference that less than 0.5 is permissible if composite reliability is above 0.6?
@@farigreactions8202 Sometimes AVE is less than 0.5 as AVE is a bit conservative. In some of my papers I have reported AVE of less than 0.5 too. If you had an AVE of less than 0.5, refer to CR for convergent validity. This is the reference to support it: Sharif, S. P., Mostafiz, I., & Guptan, V. (2018). A systematic review of structural equation modelling in nursing research. Nurse Researcher, 26(2), 28-31.
@@farigreactions8202 For reliability measures, to be safe they should be greater than 0.7. However, I think in some of my papers I had some with CR > 0.6 too.
@@saeedsharif sir i m facing issue for GFI it is very poor, 0.84 i m increasing sample size still not improving much infact validity then gets affected which is fine now for variables considered your points regarding AVE.but my CFI, chsquare,RMSEA,RMR are fine.