Hi,

Either of the following link will help you a lot:

https://books.google.co.th/books?hl=en&lr=&id=A86GBAAAQBAJ&oi=fnd&pg=PR5&dq=What+are+the+prerequisites+before+running+an+Exploratory+Factor+Analysis%3F&ots=c1mehVq2FE&sig=Ye8_GSyEDsPvf6gfptPkeC6k-jc&redir_esc=y#v=onepage&q=What%20are%20the%20prerequisites%20before%20running%20an%20Exploratory%20Factor%20Analysis%3F&f=false

https://www.youtube.com/watch?v=x4GFIzKzf2E

Dear Marcel,

Before you start you should also consider multicollinearity. It is recommended that you begin by calculating the bivariate correlation between all pairs of items and maybe setting a cut off value, such as 0.8. However, there is no way of knowing which of a pair of items should be removed as this is a qualitative decision.

By the way, the Gaussian distribution requirement is not too strict (according to Tabachnick and Fidell).

There are several other things you will need to decide as you go along. Have a look at my guide:

https://www.researchgate.net/publication/304490328_Advice_on_Exploratory_Factor_Analysis

Working Paper Advice on Exploratory Factor Analysis

The pre-requisites you mentioned are all sound. Multivariate normality is usually violated if you have hundreds of items. Before running EFA, make sure that you have standardized all the items based on a common scale (e.g., z-scores, student's t-scores) if the item scores are randomly distributed across different scales (e.g., some are based on 5-point Likert scales while others are based on 7-point Likert scales). Another thing to watch out for is to consider whether there are any theoretical model or framework to back up your factors. If so, you may want to conduct confirmatory factor analysis based on maximum likelihood (ML). If you designed an entirely new cluster of survey/questionnaire items with inadequate theoretical backing, EFA will be the way. However, if you constructed them based on time-tested theoretical principles, you should conduct CFA. I trust that your approach met the requirements for EFA. 

Thank you all for your contributions, I appreciate them greatly and will discuss them with my fellow researchers.

@Jimmy Y Zhong: The CFA is up next.

Thanks again!

Thank you Chalamalla, very helpful articles indeed!

Dear Marcel,

Please pay attention to:

1)       N of the sample:

-          minimum 100 subjects; ratio of 5 subjects per variable. Good: ratio of 5 subjects per variable ((Bryman & Cramer, 1993; Gorsuch, 1983)

-          rule of thumb: at least 300 cases

2)  correlation matrix (in SPSS: Analyze / Data reduction / fator;  Descriptives: Correlation matrix: coefficients significance levels). You should have at least some correlations higher than .30. (minimum)

3)  KMO ( Kaiser-Meyer-Olkin Measure of Sampling Adequacy). (in SPSS: Analyze / Data reduction / fator;  Descriptives: KMO and Bartlett’s test of sphericity). You should have a value greather than .70 (over than .80 = good sample acdequacy).

4)  Bartlett's Testo f sphericity. You should find a chi-square with a p-value < .05 (= correlation matrix differ from identity matrix)

5)  anti-image matrix (in SPSS: Analyze / Data reduction / fator;  Descriptives: Anti-image). You should obtain high values in the diagonal of the matrix and low values out of the diagonal.

If all are ok you can proceed with you research!

good luck with you research!

 Dear Lisete,

I will add your feedback directly into my guidlines for the EFA. Thank you as well for adding the steps in SPSS. 

I will keep you posted through the updates on my project.

thank you!

best wishes and go ahead with your data analysis!

To examine whether the data set was appropriate for factor analysis, the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity need to be computed.

The KMO statistic, which can vary from 0 to 1, indicates the degree to which each variable in a set is predicted without error by the other variables. A value of 0 indicates that the sum of partial correlations is large relative to the sum correlations, indicating factor analysis is likely to be inappropriate . A KMO value close to 1 indicates that the sum of partial correlations is not large relative to the sum of correlations and so factor analysis should yield distinct and reliable factors. values between 0.5 and 0.7 are mediocre, and values between 0.7 and 0.8 are good.

Bartlett’s test of sphericity is a statistical test for the presence of correlations among variables, providing the statistical probability that the correlation matrix has significant correlations among at least some of variables . For factor analysis to work some relationships between variables are needed. Thus, a significant Bartlett’s test of sphericity is required, say p