DATA ANALYSIS

[priti sharma]

Dear Members,

If you have any query or seek help regarding STRUCTURAL EQUATION MODELLING TECHNIQUE (for both measurement as well as structural), feel free to contact.
With Regards

[Preeti Jain]

hello priti

Hwz u.... i want 2 learn SEM from u. Would u give me ur time in department.
Preeti jain

--
Preeti Jain

[priti sharma]

Sure dear.......contact me on any working day.

--

Priti Sharma,
Email:priti...@gmail.com

[Preeti Jain]

Any time as per ur convenience............

--
Preeti Jain

[khagesh agarwal]

Priti,

Can you send across some study material for learning SEM using AMOS?

Thank You,
Khagesh Agarwal
------------------------------------------------------------------------
View my research on my SSRN Author page:
http://ssrn.com/author=1532073
------------------------------------------------------------------------
Linkedin: http://in.linkedin.com/in/khageshagarwal
------------------------------------------------------------------------

[priti sharma]

Hello Khagesh,
Please find an attachment and along with must read MARKETING RESEARCH: AN APPLIED INTRODUCTION By Naresh Malhotra and Satyabhushan Dass. I too gt familiarity with the help of this book. this book explains various techniques using SPSS, AMOS, SAS.
With regards

[Dr Neeraj Kaushik]

Dear Priti

Thanks for the wonderful book.
SEM is the need for the hour and every researcher want to apply it.
I've written very basics of SEM in 2 posts but didnt mention how to read the output.

Can u plz write on How to evaluate AMOS output. There're so many tables, which one to refer?

Regards
Neeraj

[priti sharma]

Respected Sir,

Sorry for the delay.

An SEM model consists of two models: the measurement model and structural model. The measurement model depicts how measured or observed or manifest variables represent construct or latent variables, uses Confirmatory Factor Analysis (CFA) in which researcher specifies which variables or measures define which construct or latent variables or factor. In other words, CFA used to test the hypothesis that a relationship between observed variables and underlying latent construct exists. The researcher uses knowledge of the theory, empirical research, or both; postulates the relationship pattern a priori; and then tests the hypothesis statistically.  In structural model, relationship between constructs exists or not is evaluated. First, you will be acquainted with Measurement model.

In Measurement Model as already told relationship between construct and the manifest variables is tested. Firstly, the item-to-total correlation is estimated of all the manifest variables of a particular construct, then the variable having item-to-total correlation less than .35, should not be included in the study (this is the first condition). Secondly, the loadings of manifest variables should be more than .7 (A loading of 0.7 or higher indicates that the construct is explaining 50 percent or more of the variation in the observed variable since, (0.71)² = 0.5)  but if it is more .5 than also be accepted.

After this Indices for the model fit is checked, in this basically the value of Cronbach alpha equal to or greater than .7   is expected, GFI (goodness of fit) equal to or greater than .9 , CFI (Comparative fit index) equal to or greater than .9, and RMSEA (Root mean square error of approximation) equal to or less than .08.

After checking model fit, the reliability and validity of the model is assessed through composite reliability and convergent validity (in case of measurement model and dscriminant validity in case of structural model).

Comopsite reliability =      (sum of standardised loadings)²_____________________________

                                    (Sum of standardised loadigs)  + sum of indicator measurement error

 

 

For Convergent validity= Basically the size of the factor loadings provides evidence of convergent validity. Higher factor loadings indicate that the manifest variables converge on the same construct. At a minimum, all factor loadings should be statistically significant and higher than 0.5, ideally higher than the 0.7. A loading of 0.7 or higher indicates that the construct is explaining 50 percent or more of the variation in the observed variable since, (0.71)² = 0.5. Sometimes, a cut off level of 0.6 is used. Besides this, another measure used to assess the convergent validity is the average variance extracted (AVE), which means the variance in the indicators or manifest variables that, is explained by the latent construct.

AVE is calculated:

 

AVE = _______________________Sum of the squared standardised loadings____________

             Sum of the squared standardised loadings + Sum of the indicator measurement error  

 

The value of AVE varies from 0 to 1, and an AVE of 0.5 or more indicates satisfactory convergent validity, which means that latent construct accounts for 50 percent or more of the variance in the observed or manifest variables.

After this, if convergent validity is satisfactory then first drop the item having loading less than .5, drop the variable having standardised residual more than 4, do modification indices by checking that in output box, and lastly trial and error method to fit the model.

But it should be noted that, the dropping should not exceed more than ten percent.

Feel free to ask further, Structural model i will put later on

With regards

 

[priti sharma]

Sorry there is a mistake in last para. that "if covergent validity is satisfactory" plz read it as if convergent validity is not satisfactory.

With Regards

[drsa...@gmail.com]

Dear priti mam,

i'm form Nepal. and pursuing doctoral degree. ...i've seen Gasking Videos and Dr neeraj Videos uploaded by Avinash Pal.  In Nepal , It's new and very are applied in the research field. so , i need you cooperation for SEM. 

regards,  

sajeeb shrestha

faculty, Shanker Dev Campus

Tribhuvan University

Nepal.

[Priti Sharma]

Dear Sajeeb sir,
Im always ready to help researchers. I will help you with the knowledge im having pertaining to SEM but i must tell you one thing that Dr. Kaushik knows SEM better thn me.

Regards

--
You received this message because you are subscribed to the Google Groups "DataAnalysis" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dataanalysistrai...@googlegroups.com.
To post to this group, send email to dataanalys...@googlegroups.com.
Visit this group at http://groups.google.com/group/dataanalysistraining.
For more options, visit https://groups.google.com/d/optout.

[Sajeeb Shrestha]

Dear priti mam,

I'm delighted from getting your reply. i want to add into this circle. 

dr neeraj has mentioned you in his SEM videos on You Tube uploaded by Avinash Pal.. 

How to contact dr neeraj sir. can you pls send me his mail id. i'll contact him.

One question mam,

what to do when only 3 indicators are on one Constructs. how to get standardized loadings from CFA to check AFVE.

Thanks form your great cooperation.

regards,

sajeeb shrestha

faculty, tribhuvan university

Nepal.

--
You received this message because you are subscribed to a topic in the Google Groups "DataAnalysis" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/dataanalysistraining/V5HD_-X5WaI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to dataanalysistrai...@googlegroups.com.

[Priti Sharma]

Dear Sajeeb Sir,

You are already in conversation with Dr. Kaushik because this dataanalysistrainning google group belong to him and all the keen researchers. So, if you have any query you may ask by writing new post on this group.

Regarding your question: If you are thinking of applying CFA thn there should be atleast three manifest variables for a construct (Below three will be problem). In your case, you can easily apply cfa as there are three manifest variables are thre

 By default AMOS displayed unstandardized loadings, for getting standardized loadings you have to choose instead to display standardised laodings.

Regards

[Sajeeb Shrestha]

Thanks mam. a great help from your side.

--

[Sajeeb Shrestha]

Hello Dr Neeraj kausik Sir,

Good After noon, 

I'm a new comer in SEM. I've a question to you. From your Video, you have taught, if there is AVE less than 0.5 we should delete the indicator item. if there is two indicator items having less than 0.5 , which one to remove. 

regards,

sajeeb shrestha

faculty on marketing

tribhuvan university

nepal

[Neeraj Kaushik]

Dear Sajeeb

For the AVE to be > 0.5 all the items must average to 0.71 or higher.

If AVE<0.5 then remove the item with min std beta so that average go higher.

Best wishes
Neeraj

[Sajeeb Shrestha]

Respected neeraj Kaushik sir,

thanks for our support.

in your videos, you have told in EFA , PCA method was done. .. Gaskin is telling Maximum likelihood method. what to do sir.

regards,

[Neeraj Kaushik]

Regression is of many types.

EFA make use of OLS regression while CFA uses MLE regression method.

Whats the doubt?

[Sajeeb Shrestha]

Dear dr neeraj  kaushik sir,

my queries is what to do first in EFA.  and how to get factors.

regards,

[Neeraj Kaushik]

Plz watch https://www.youtube.com/watch?v=yzVeeVPGKIY&list=UUopzXUS1wFc3YndE-KhsWvw

& ask for ur doubts

[Sajeeb Shrestha]

Watching videos. Thanks sir. We need more videos on cfa and sem from u.
Regarda.

[pardee...@lpu.co.in]

Is there any way in excel to highlight the cells containing similar information? Please suggest.

[Anuja Mishra]

Home-> highlight cell rules-> equal to->  fill required details

On 20-Jun-2016 5:41 pm, <pardee...@lpu.co.in> wrote:

Is there any way in excel to highlight the cells containing similar information? Please suggest.

--
Protocols of this Group:

1. Plz search previous post in group before posing the question.
2. Don't write query in someone's post. Always use the option of New topic for the new question. You can do this by writing to dataanalys...@googlegroups.com
3. Its better to give a proper subject to your post/query. It'll help others while searching.
4. Never write Open ended queries. This group intend to help research scholars NOT FOR WORK THEM.
5. Never write words like URGENT in ur posts. People will help them when they are free.
6. Never upload any info about National Seminars/Conferences. Send such info on personal emails. And feel free to share any RESEARCH related info.
7. No Happy New Year, Happy Diwali, Happy Holi, Happy B'day, Happy Anniversary etc. allowed on this group.
8. Few months back there was a facility for asking & sharing the Research Papers. Now there is no provision of asking for the research paper here.

Let’s make a better research environment.
---

You received this message because you are subscribed to the Google Groups "DataAnalysis" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dataanalysistrai...@googlegroups.com.
To post to this group, send email to dataanalys...@googlegroups.com.

Visit this group at https://groups.google.com/group/dataanalysistraining.

[Neeraj Kaushik]

Home-> Conditional Fomrating->Highlight cell rules-> Duplicate values

[mickpan...@gmail.com]

Hi . I am working on a model when I see the model fit it is fine but estimates are showing that out of 4 hypothesis only one is significant. I am sure there are no missing value, cronbacks alpha is fine