SEM with AMOS: From Zero to Hero (18: Model fit assessment)

Поделиться
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

Комментарии • 50

  • @drey7459
    @drey7459 Год назад +2

    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.

    • @saeedsharif
      @saeedsharif  Год назад

      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!

  • @muahmed2220
    @muahmed2220 2 года назад +2

    Simply amazing and to the point. Hats off!

  • @karnavalos2328
    @karnavalos2328 7 месяцев назад +1

    life saver!!! thank you!!!!

  • @devendralodha7869
    @devendralodha7869 2 года назад +2

    Best Explanation of CFA I have ever observed on youTube ...Great Effort Sir.Thanks for the Sharing

    • @saeedsharif
      @saeedsharif  2 года назад

      Thank you for your comment 😊

  • @tzopati
    @tzopati Год назад

    Best explanation so far! thank you sir!

  • @noorhaider1142
    @noorhaider1142 3 года назад +5

    One of the best video on CFA and its complications. Hoping it will help me too Insha Allah.

    • @saeedsharif
      @saeedsharif  3 года назад +1

      Thanks for the feedback. All the best

    • @noorhaider1142
      @noorhaider1142 3 года назад +1

      @@saeedsharif thank you

  • @husseinalani2684
    @husseinalani2684 2 года назад +1

    good teacher Dr. Saeed
    appricate

  • @hanzodragon1376
    @hanzodragon1376 3 года назад +3

    This tutorial is absolutely superb. A real life saver, a huge thank you Dr Sharif

  • @panagiotiskioulepoglou3635
    @panagiotiskioulepoglou3635 Год назад +1

    Thank you

  • @estellekapche5041
    @estellekapche5041 2 года назад +1

    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

    • @saeedsharif
      @saeedsharif  2 года назад

      Thank you for your comment 😊

  • @abdulkadirjaelani2878
    @abdulkadirjaelani2878 2 года назад +1

    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.

  • @sumry90
    @sumry90 Год назад

    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?

  • @SufyanSalem
    @SufyanSalem 2 года назад +2

    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

  • @simplyram2676
    @simplyram2676 Год назад

    ❤️❤️

  • @chefberrypassionateresearcher
    @chefberrypassionateresearcher 5 месяцев назад

    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??

  • @CarodyCustard
    @CarodyCustard Год назад

    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!

  • @cb2497
    @cb2497 3 года назад +1

    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.

  • @clivechifamba179
    @clivechifamba179 2 года назад +1

    Do you have a reference list of the authors cited on the video

  • @06_chaggargurpreet41
    @06_chaggargurpreet41 Год назад +1

    Sir I am getting 2 of my factor loadings as 1.09 and 1.06..how to deal with it?

  • @tenzindawa3338
    @tenzindawa3338 2 года назад +1

    respected sir,
    in CFA can CMIN/df value and RMR be zero?
    also sir can GFI AND CFI can be 1.00 value?

  • @tenzindawa3338
    @tenzindawa3338 2 года назад +1

    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

  • @riverel
    @riverel 2 года назад +1

    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!

    • @saeedsharif
      @saeedsharif  2 года назад +1

      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.

  • @makathleenadona9957
    @makathleenadona9957 2 года назад +1

    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.

    • @saeedsharif
      @saeedsharif  2 года назад +1

      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.

    • @makathleenadona9957
      @makathleenadona9957 2 года назад +1

      @@saeedsharif I understand it now, Sir. Thank you and please continue giving us worth-watching Statistics video lessons.

    • @saeedsharif
      @saeedsharif  2 года назад

      @@makathleenadona9957 All the best

  • @T.V.6388
    @T.V.6388 4 месяца назад

    Professor could you please help me with the SEM?

  • @simplyram2676
    @simplyram2676 Год назад

    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

  • @zizi5445
    @zizi5445 19 дней назад

    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.

    • @saeedsharif
      @saeedsharif  18 дней назад +1

      You can use this book as your reference:
      Pahlevan Sharif, S., & Sharif Nia, H. (2018). Structural equation modeling with AMOS. Tehran: Artin Teb.

    • @zizi5445
      @zizi5445 18 дней назад

      @@saeedsharif Thank you, appreciate the help

  • @jijyme8052
    @jijyme8052 2 года назад +1

    If the NFi, GFI Values are less than 0.9 , we refuse the model ? ( while all the other values are ✅)

    • @saeedsharif
      @saeedsharif  2 года назад

      Not necessarily. You may refer to other indexes. Also, you may find references to support for example NFI greater than .85

    • @khifzarehman1929
      @khifzarehman1929 2 года назад +1

      Did you get the ans to this question? because i'm facing the same issue,

    • @SciRes
      @SciRes 2 года назад

      @@khifzarehman1929 See the answer above

    • @jijyme8052
      @jijyme8052 2 года назад

      @@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

  • @farigreactions8202
    @farigreactions8202 3 года назад +1

    Sir i m very worried my model fit is poor and discriminant validity too

    • @saeedsharif
      @saeedsharif  3 года назад

      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.

    • @farigreactions8202
      @farigreactions8202 3 года назад +1

      @@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?

    • @saeedsharif
      @saeedsharif  3 года назад

      @@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.

    • @saeedsharif
      @saeedsharif  3 года назад

      @@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.

    • @farigreactions8202
      @farigreactions8202 3 года назад

      @@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.