ABSTRACT
Questions about sample size tend to focus on two topics: what sample size is sufficient to apply a specific statistical estimation method, and what sample size is needed to obtain a specific power? In multilevel analysis, these problems are made more difficult because there are different sample sizes at the separate levels, and because the model includes a fixed and a random part, where typically the fixed part can be estimated with more precision than the random part. This chapter reviews both the sample size and the power issues. It includes a description of software useful for power and sample size determination.
