I am doing an EFA on a 20-item Likert scaled survey instrument. When I initially did EFA, five factors were extracted, but I could not reasonably fit the survey items into those 5 factor with a variance of 56.75%. So, my professor asked me to try to control the EFA to extract 3 factors. When I did this, all survey items worked well with those 3 factors. However, my variance dropped to 45.63%. So, I know the "recommended" variance percentage is 60%. So, what do I do now? Is there a way to explain why this is low? Do I need to do additional tests or check other things? I am on the final part of my disseratation, and my professor just questioned this...HELP!
Hello Brandi,
let me begin by saying that there is no abstract number of what percentage of explained variances constitutes a good model. If I am not mistaken, we have to thank Hair et al. (2010, p. 109) for those ominous 60 %. True, values of that magnitude are often achieved in the Social Sciences, but I would strongly advise against both ruling out any factor solution below that benchmark and also arbitrarily constraining your EFA to three factors if that goes against the underlying theory/assumptions.
I would kindly ask you to provide us with more details on how you ran the EFA (EFA or PCA, criteria for factor retention, rotation method, factor loadings etc.) Preferably, you have an output at hand you can post as a jpeg or some similar format.
Best
Marcel
---
Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate data analysis: A global perspective (7. Ed.). Upper Saddle River, NJ: Pearson.
Dear Brandi Todd ,
As per my knowledge and experience the total variance should be 50% at least.
For further detail you me watch the following videos
Explanatory Factor Analysis (EFA), Theoretical guidelines and practical application using SPSS
The audience is required to Click on the required video link and view the video. You are also requested to "Subscribe" the channel.
The following short video will elaborate the concept of exploratory factor analysis in easy wording and enhance the understanding of new users of EFA.
https://youtu.be/cnB9B2KP7jc
Following is the second part of EFA clearing the concept of Exploratory factor analysis with the threshold level of different tests.
https://youtu.be/-NVurWGG5I8
#spss #efa #factoranalysis #multivariateanalysis #dataanalysis #varimax #rotation #maximumlikelihood #oblique #promax #factoring, #spss
Following is the third video explaining the steps involved in running the EFA using SPSS.
https://youtu.be/y1PSiYTE7Bw
Following is the fourth video in the series of EFA for understanding the procedure using SPSS.
https://youtu.be/m_gcgedzq0s
Following is a practical Demonstration of EFA using different procedures through SPSS.
https://youtu.be/Kfw-xWG25dk
Alternatively, you may just read the original document by its author, Peter Samuels - no credits given -, instead of listening to someone's translation into Hindi, which is clearly of no help whatsoever.
Working Paper Advice on Exploratory Factor Analysis
Hi! the threshold for variance percentage varies across disciplines. It is also necessary to see whether its a newly developed scale or established scale. Sometimes restricting the number of factors by the researchers themselves while performing the factor analysis also create different in the variance percentage. Thanks
Marcel Grieger Here are some of my charts. I have recently gone in and taken off some of the factors that were interfering with each other with loading. It reduced the 20 Likert items to 10, and it changed my variance from 45% to 59%. See if you think this would work! Thank you so much for your guidance.
- Badges
- Science topic
- Similar topics
- Philosophy
- Philosophy Of Science
- Reasoning