I have a data which contains Three Level Items (YES-NO-NotSure). Is it technically right to transform data into numeric type and perform EFA (Exploratory Factor Analysis) to extract factor scores to use in subsequent analysis?
Dear Saeed Shah
setting aside the obvious problem of interpreting Likert-type scales as continous, three response options is a really low threshold to still reasonably assume your respondents perceive the intervals between the categories as equidistant. To the best of my knowledge, four to five categories are needed for the calculations you desire, but I cannot give you a source for that off the top of my head.
Also, I guess the "Not sure" option is placed between "Yes" and "No"?
Best
Marcel
Dear @Saeed refer to the link below. I guess it would help your cause.
https://www.researchgate.net/post/Is_it_possible_to_perform_factor_analysis_on_categorical_data
Hello Saeed,
This is certainly a popular question in the last day or two (as well as many times in the past on RGate).
Here's my response to a similar one:
https://www.researchgate.net/post/CFA_on_7-point_Likert_scale_items
Good luck with your work.
Dear David Morse ,
I read your response to the given link but the question seemed different (7-point Likert scale) to mine (3-point scale) or may be I could not comprehend.
For example :
I have 16 items which have responses like (Yes-Not Sure-No) which are coded as 3-2-1 respectively and transformed to numeric data type.
For these items, I would like to run Exploratory Factor Analysis to extract factors to be used as Independent Variables for subsequent analysis (logistic regression).
So my question is that, is it technically right to do EFA on this kind of data, if not what is the possible solution?
There are latent variable models for categorical observed variables. Look up the nominal IRT model if you want continuous latent variables. This is implemented in mirt (Article Mirt: A Multidimensional Item Response Theory Package for th...
). If you want categorical latent variables google latent class models.
Hi Saeed Shah
David Morse has explained the issue of the "Don't know" categorie in this thread:
https://www.researchgate.net/post/EFA_and_CFA_of_a_mix_of_ordinal_questions
Best
Marcel
Dear Saeed Shah Please have a look at it.
Yes=3
No=2
Not sure=1.
Transform accordingly and get an SPSS data set and then produce a CSV file and then run in PLS.
Good luck.
- Badges
- Science topic
- Similar topics
- Biological Science
- Histology
- Sectioning