If you can go for CFA then why do you take EFA as an extra burden? CFA is an integrated approach to describe the whole scenario as a whole. Reliability and validity are two different issues, please mind it.

you can do efa on your training data set, and cfa on testing data set. efa is used in factor reduction while cfa can be used in theory testing.

Hi Gersha,

this issue has been debated on RG a dozen times, for instance

https://www.researchgate.net/post/Can_I_use_EFA_before_SEM_analysis

and

https://www.researchgate.net/post/Can_we_do_exploratory_and_confirmatory_factor_analysis_in_the_same_data_set2#view=5486170ad4c118047c8b477f

Here are some dangers presented of relying too much on the factor model at all:

https://www.researchgate.net/post/How_to_deal_with_Single_and_Multiple_Item_latent_variables_in_Structural_Equation_Modeling#view=5e849fe521137522f2332355

HTH

--Holger

Thank you Holger Steinmetz , Niyungeko Antoine and Muhammed Ashraful Alam for your responses. This was also my view but I wanted to see if there had been any recent shift researchers position on this issue.

Thank you again

Hi @Gersha,

For a scale validation process, you must use EFA fisrt and CFA after. These technics are very important.

For further information, see Churchill (1979) and Hsu, M.-S. & Yu-You, T., 2012. A Factorial Validation of Internship Perception Struc-

ture : Second-Order Confirmatory Factor Analysis.

Dear Cisagara Bisimwa

perhaps you may be motivated to take a look into the links that I provided to threads where this has been debated intensively.

As you will see, I have some doubts (or see no reason) for a "must" of an EFA and would prefer to conduct a CFA whenever I have at least a glimpse of an idea what my indicators measure. Further, I argue that a successful CFA after an explored EFA provides no real evidence on the causal structure implied by a factor model as systematically wrong models can be successfully replicated.

Perhaps you or the paper by Churchill and Hsu and Yu-You bring in some thoughts I don't see, though.

All the best

--Holger

I am curious about the answer to the same question. I would be glad if you can add your reference sources to your ideas.