Abdreheman Seid Abdella It is difficult to say in general how large the sample size needs to be in CFA/EFA as this depends on many factors (e.g., the number of items/variables, the loadings/reliabilities of the indicators, the amount of missing data, etc.). The best way to find out for a specific application is by means of Monte Carlo simulations, which allow you to examine the performance of a specific CFA model under the specific conditions that you expect to encounter in your application. I provide a brief conceptual introduction to the idea of Monte Carlo simulations in the following Youtube video:

https://www.youtube.com/watch?v=hjAJSPGMQ4E&t=1s

I also offer a free mini-course on sample size planning through simulation with Mplus:

https://www.goquantfish.com/courses/sample-size-planning-in-mplus

Abdreheman Seid Abdella Sample size for CFA also depends on the measurement model (number of items you have, model fit, etc.).

You may refer to the following methods to calculate your sample size:

1: RMSEA: http://www.quantpsy.org/rmsea/rmsea.htm

2: Number of items and Number of constructs: https://www.danielsoper.com/statcalc/calculator.aspx?id=89

In order to identify the sample, first you need to think about the population first. Then use Slovin's formula or Cochran (1977) to arrive at the sample/subjects

Usually, it is recommended to be more than 250 samples as the more the sample the less err, however, 250 samples give 8% of error or less which is generally accepted in social since please refer to:

Reyes, E. M., & Ghosh, S. K. (2013). Bayesian Average Error-Based Approach to Sample Size Calculations for Hypothesis Testing. Journal of Biopharmaceutical Statistics, 23(3), 569-588.

and see the attached graph