Sequential Accuracy in Parameter Estimation: Accurate Estimates with a Smaller Sample Size – Ken Kelley
Sequential estimation is a well-recognized approach to inference in statistical theory. In sequential estimation, the sample size for a study is not prespecified before the study begins, rather the study itself informs when researchers should stop sampling, namely when a stopping rule has been satisfied. In particular, data in sequential estimation procedures are collected in stages, whereby at each stage the estimated population values are updated and the stopping rule evaluated. In this talk, I discuss a general theory for sequential estimation procedure for constructing a narrow confidence interval for a general class of effect sizes with a specified level of confidence (e.g., 95%) and a specified upper bound on the confidence interval width. This approach extends the accuracy in parameter estimation framework for planning sample size, in which the necessary sample size depends on the goals of the researcher and on unknown parameter values, to an approach in which no prespecified parameter values is necessary. The sequential approach discussed not only makes it easier to plan sample size for accurate estimates, the sample size from the sequential perspective is often smaller than necessary from a pre-planned (fixed n) perspective, holding everything else constant. Methods have been developed for a variety of effect sizes and can be implemented in software via the MBESS R package.
University of Notre Dame