Introduction

I often use structural equation modeling (SEM) to analyze my data. Although this technique has many advantages over some traditional correlational approaches, there exists a practical problem to researchers to use this advanced multivariate data analysis. SEM requires a generally larger sample size than traditional approaches in order to generate reliable results. Many people like me would ask "how many is enough?". Many reviewers would challenge if your sample generates an adequate statistical power for your model, particularly when your model is really quite complex.

Although there are some studies talking about how we can calculate the statistical power of SEM models (e.g., MacCallllum, Browne, & Sugawara; Muthen & Muthen, 2002), it is somehow difficult for layman student researchers like me to know how to calculate this, or know how to operate the required programs (e.g., SAS and Mplus). I am a poor student who doesn't have access to these statistical programs. Therefore, I wrote a Matlab program based on the calculation procedures of MacCallum and colleagues to compute the statistical power of SEM models.

It is quite user-friendly to use. Please have a look at the working environments, instructions, and installation procedures my program. The original version of this program can only calculate the power achieved in your SEM model. In this latest version, I added a function to calculate your required sample size based on your model fits and desired power. Hope that you will find that useful.

How to cite this software in APA format?

Chan, D. K. C. (2009). PowerSEM 2.0 [Computer software]. Available from www.derwinchan.com/software.

References

MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1(2), 130-149.

Muthen, L. K., & Muthen, B. O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 9(4), 599-620.