While constructing a new scale, I have used items from various scales (mixed Likerts, such as 5-point, 6-point, 7-point), and collected data. This was followed by EFA. I have new factors now. Can this scale (with new items) be administered using a uniform Likert (say 7-point) at my discretion, before doing CFA?
I do not think there is any problem if it goes according to your theoretical framework and you consider your sample.
As i know, both type scale ( 5,7- point ) can be used for CFA & SEM, therefore why u need to uniform the items?
Of course you can use a new range for responses.
One thing that would be worth checking (afterwards) is if all response options are used or if categories require collapsing (which you are still allowed to do afterwards in any case, before doing a CFA). Some items might suffer from a lack of monotonicity when you have too many response categories and participants can no longer be accurately discriminated from each other due to overlap of response categories (i.e. when you have a scale with "almost never" and "barely ever" in the same 7-point response scale). As long as your response categories are well-defined this should not be a problem.
Best,
Michiel
Thank you Cristian Ramos-Vera Alireza shabani shojaei Michiel AJ Luijten your answers have been very helpful. Truly appreciate.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey