Hi,

See the following links:

http://www.stats.ox.ac.uk/~snijders/PowerSampleSizeMultilevel.pdf

http://www.esourceresearch.org/eSourceBook/MultilevelModeling/17PowerandSampleSize/tabid/353/Default.aspx

For Syntax:

http://www.statmodel.com/discussion/messages/12/3679.html?1444636773

For general orientation on this - see

http://www.bristol.ac.uk/cmm/software/mlwin/mlwin-resources.html#datadesign

You can use the following software to do the power calculations - it uses a simulation based approach so that you can use it as a 'sandpit' for the specific model you are interested in 

http://www.bristol.ac.uk/cmm/software/mlpowsim/

the manual is a very good primer.

This is also very useful especially if time is involved:

Optimal Design for Longitudinal and Multilevel Research Documentation for the “Optimal Design” Software

http://www.stat.cmu.edu/~brian/463/week14/od/od-manual-20111016-v300.pdfhttp://sitemaker.umich.edu/group-based/optimal_design_software

Hi Ibrahim,

I often use a formula for calculating effective sample size based on number of observations and intra-class correlation. I then plug the effective sample size into a power analysis program (I use PASS) to derive standard power estimates. For instance, this is what I wrote for a recent grant proposal:

"For the EMA analyses, the power of the statistical test depends on the total effective sample size (ESS) and the statistical model used. The total ESS is the number of statistically independent observations available for this study. The number of statistically independent observations is the total number of observations (number of participants x number of data collections) adjusted for within-individual correlations. As the within-individual, or intra-class, correlation (ICC) increases, the ESS decreases. The following formula illustrates this relationship, whereby the effective sample size is equal to nm/(1+(m-1)ρ), where n = number of participants; m = number of repeated measures for each participant; and ρ = ICC (Diggle, Liang, & Zeger, 1994).

I hope this helps,

Paul

Hello Ibrahim,

Please see the attached PDF. 

Best wishes

Dear colleages, 

WebPower an online free software is for statistic power analysis. It also works for multilevel model and other SEM models. It does not require you writing any equations, very friendly! Try it! 

http://webpower.psychstat.org/wiki/models/index

Best, 

Thanks a lot everyone, they are all great resources. 

Thanks 

Stephen Liu's book "Statistical Power Analysis for the Social and Behavioral Sciences: Basic and Advanced Techniques" has R, SPSS and SAS code.

The 'Statistical Power' chapter in Bolger and Laurenceau's book includes code for Mplus :

Bolger, N., & Laurenceau, J.-P. (2013). Methodology in the social sciences. Intensive longitudinal methods: An introduction to diary and experience sampling research. New York, NY, US: Guilford Press.