I need two authentic references for the following;
1. If I collect data from 10 respondents for one item of latent variable.
2. If I collect data from 5 Respondents for one item of latent vairbale.
For latent variable analysis, sample size per item is critical:
- 10 respondents per item is generally acceptable and aligns with common guidelines in factor analysis. Reference: Everitt, B.S. & Hothorn, T. (2011). An Introduction to Applied Multivariate Analysis with R. Springer.
- 5 respondents per item is minimal and usually only suitable for pilot studies. It may lead to unstable estimates. Reference: Nunnally, J.C. & Bernstein, I.H. (1994). Psychometric Theory (3rd ed.). McGraw-Hill.
Tip: For robust SEM or CFA results, larger samples (e.g., ≥10–20 respondents per item or total N ≥ 200) are recommended
What type of analysis are you planning? How many items? If all respondents answer all items, are you telling us that you are planning only 10 respondents in total, or if there are 200 items, then there will be 2,000 respondents who answer all 200 items?
This is a very basic rule and it is not a sufficient to establish the sample size needed. It depends on the type of statistical method you use for the analysis. You should assess the statistical power you can achieve with you sample, with the effects aimed to achieve (f square, R square values). For the power analysis the G*Power software can be used.
- Badges
- Science topic
- Similar topics
- Computer Science
- Research Topics