|
Abstract:
|
Formulae for sample size calculations for statistical power in the literature (especially textbooks) are often predicated on unrealistic assumptions (e.g., equal variances) or unrealistic, or simply undesirable, designs (two independent samples of the same size). In addition, sample size and power calculations usually involve repeated use of any given formula, as the researcher works through varying inputs (alpha, power choices, design choices, variation/mean relationships, etcetera), making the process of considering sample size and power issues unpleasantly tedious and error-prone. In this presentation, I will show (and subsequently make freely available) a rich and interactive sample size calculator that enables calculations for a wide variety of variations on the two-sample setting.
|
