Doubts regarding EFA

[adya mishra]

Respected Group members,

I am constructing a questionnaire as my PhD work. I was running EFA on my data, in the output window of SPSS I saw some cross loaded variables so I ran EFA again after excluding cross loaded variables. Then in the output window there were no cross loaded variables, however there were variables that were not loaded under any factor. 

My doubt is that shall I run EFA again after excluding those variables that are not loaded under any factor?

As per my knowledge one should run EFA till there are no cross loaded variables and no variables that are not  loaded under any factor. Please correct me if I am wrong and clarify my doubt. 

Regards

Adya Mishra

[Prof. Arunkumar Dubey]

Every item in EFA loads on every other factor but it will be loaded heavily on its parent factor. In SPSS if you give cutoff values then it appears that there is no cross loading but still there are cross loadings which is not visible in the output. 

Performing EFA post deletion of items will amount to loss of vital data. Remove items only if the difference in cross loadings for an item loading in two factors is less than 0.30. Try retaining those items where the difference is more. 

Items not loading on any factor itself suggest that they are a factor in themselves. 

Please check the number of factors extracted in the total variance explained output. Preferably total variance explained should be greater than 50%. 

Neeraj Sir has beautifully explained the concept of EFA in one of its video. Please check that as well. 

Regards,

Arun

--
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 view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAFVXDOzcZSFJ455YTGfe_hdT28YdJ1pNHU6zXY4XcOx0F2rbiQ%40mail.gmail.com.

[adya mishra]

Respected Professor Arun,

Thank you for clarifying my doubts. Can you please tell me how to set the cut off value so that no cross loading appear in SPSS output. 

Regards 

Adya

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAGqXa90rWOiJ50BTJEv8fSB2qq8Q9_piGiZPJ6%3DRpkH9zkUPxg%40mail.gmail.com.

[adya mishra]

Respected Group Members,

 I am constructing a questionnaire for my PhD work. I did EFA on my data. In the output window there were cross loadings. I want to know how to set the cut off value so that cross loadings do not appear in the output window of SPSS.

If the difference in cross loadings for an Item loading in two factors is more than 0.30 then shall I consider that the Item is under that factor which has a high loading value?

My next doubt is that if  I see some Items that are not loaded under any factor so shall I create another factor for such Items? What should be its basis? or I shall drop such Items for further analysis?

Should I check  the reliability of each factor after EFA or after CFA?

Kindly Clarify my doubts

Regards

Adya

[Prof. Arunkumar Dubey]

Dear Adya, 

You should consider the item to load in that factor where the loading is greater. In case the difference is greater than 0.30 you may consider the item to load on that factor where the loading is more. However, you can also delete that item where it is cross loading on two factors. As a researcher the call should be yours. Please do not be confused. I suggested you to retain items where the difference is more than 0.30 so that you do not miss on the vital data that you have collected. Therefore, before going for data collection, it is highly advisable to get the content validity established based on literature review and expert opinion. 

Again, I suggest you to watch Neeraj Sir's video on EFA on how to fix cutoff values in SPSS. The link is as follows. 

https://www.youtube.com/watch?v=uhWKd1vA8X8&list=PLzUJUtTJcj8Tub9LHkIBBScx0N6DyBMWL&index=2

https://www.youtube.com/watch?v=zInaszuQI2o&list=PLzUJUtTJcj8Tub9LHkIBBScx0N6DyBMWL&index=3 

Coming to your question of a single item factor, usually a factor is composed of 2 or more items. You can cite some papers where an item is composed of one factor and go ahead with your analysis if you wish to retain the item in your analysis. However, you cannot check Cronbach Alpha reliability of one item factor. 

EFA is a dimension reduction technique. It is in no way used to establish validity of the constructs. In order to assess the validity of the constructs (convergent and discriminant validity), one should go for CFA. CFA is also known as the measurement model which is a technique available under Structural Equation Modelling. 

I personally advise you to go for CFA as it is a PhD thesis. Who will take your measures seriously if the validity is not established? 

You can present the reliability figures of each dimension in EFA. However, post CFA, some items may have to be removed due to lower loadings, then the reliability might change. So when you calculate the discriminant validity using Fornell and Larcker Criterion, you may present your analysis in that table along with Cronbach Alpha reliability, Composite relaibility, AVE, Correlation figures etc.  

Please find some papers attached for your reference which will solve most of your query. I request other members as well to share their opinion on the question of Adya. 

Regards, 

Prof. Arunkumar Dubey

Email: prof.ar...@gmail.com

Cell: 950 359 1321

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAFVXDOx5P-d0MjbOCjGRpBrcYi1Kg-4G00LvRO767e40%2BBJPyg%40mail.gmail.com.

[adya mishra]

Respected Professor Arun, 

Thank you sir for clarifying my doubts. I I have one more doubt, it is about those Items which are not loaded on any factor. Shall I drop such Items as SPSS has not created further factors for such Items? Or shall i  retain such Items and create new factors for such Items. 

Regards 

Adya

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAGqXa922tdjiH3j2tSG2_%2BKt2yVGJuN7EPSG%2BbwCT79ZQMqo%2Bg%40mail.gmail.com.

[Prof. Arunkumar Dubey]

Usually, we decide factors based on Eigen Values where it is greater than 1. Please do not create factor on your own. Some statistics is required to support any factor in your study. You may consider dropping it. 

Best wishes. 

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAFVXDOzVMWJoNSeR0v%3DX4SvAcPRccuBJ-QhNtWSKMSNMu8QfKw%40mail.gmail.com.

[adya mishra]

Respected Professor Arun,

Thank you so much for clarifying my doubts 

Regards

Adya

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAGqXa90pdCFoz-1t-hht1XFxftfSYntD4uYN%3DRjXjgwSLDfccQ%40mail.gmail.com.