Dear all,
I hope you're doing well. I need your help as I have a doubt regarding the next step in my thesis.
I have developed and used several nstruments in my study:
In my pilot study, I collected data from 100 participants and have already calculated Cronbach’s alpha for reliability.
At this stage, I plan to reduce items, especially for the newly developed and long scales. I am unsure whether I should proceed with Exploratory Factor Analysis (EFA) or Confirmatory Factor Analysis (CFA). Could you kindly advise which approach is more suitable at this stage?
Thanks
A sample size of 100 cases seems rather low for both EFA and CFA given the number of items that you're looking to analyze.
Use Exploratory Factor Analysis (EFA) if:
You're not sure about the underlying factor structure of your self-developed scales (e.g., the Perceived Parenting Styles scale).
Your goal is to identify and reduce items based on how they naturally group together.
You don’t yet have a strong theoretical basis or previous studies specifying the exact number or nature of the factors.
Use Confirmatory Factor Analysis (CFA) if:
You already have a clear hypothesis or model about the factor structure (e.g., MFQ-FF, which is a standardized scale with established dimensions).
You're testing whether the data fits a predefined model.
You’ve already refined the scale (usually after doing EFA in earlier stages).
So, for MFQ-FF and QCL, if they are based on existing theory or validated instruments, and you have a hypothesized structure (e.g., three dimensions of mood in MFQ-FF), then CFA is more appropriate.
If you proceed with EFA, make sure your sample size is large enough. For factor analysis, a minimum of 5–10 participants per item is generally recommended — so you'll need to increase your sample beyond the 100 from the pilot for better stability and generalizability.
Hi,
I would run an EFA in this stage. For the CFA you ll need a larger sample.
The determination of sample size is important because on the one hand you want enouth power, and on the other hand gathering data is a lot of work and sometimes also costy.
Thee exiss several suggested rules for determining sample size. However, you will get a more precise estimate of the needed sample size if you uae a procedure in the program Mplus.
With thie mthod you also will see that differet statistical estimates in your study may require different sample sizes.
See: Linda K. Muthén & Bengt O. Muthén (2002) How to Use a Monte Carlo Study to Decide on Sample Size and Determine Power, Structural Equation Modeling: A Multidisciplinary Journal, 9:4, 599-620,
DOI: 10.1207/S15328007SEM0904_8
- Badges
- Science topic
- Similar topics
- Statistics
- Survey