Information Quality

4InfoQ at the study design stage

4.1 Introduction

Statistical methodology includes study design approaches aimed at generating data with high quality of analysis method, f, and implicitly of high information quality (InfoQ). For example, the field of design of experiments (DoE or DoX) focuses on designing experiments that produce data with sufficient power to detect causal effects of interest, within resource constraints. The domain of clinical trials employs study designs that address ethical and other human subject constraints. And survey methodology offers sampling plans aimed at producing survey data with high InfoQ. In this chapter we review several statistical approaches for increasing InfoQ at the study design stage. In particular, we look at approaches and methodologies for increasing InfoQ prior to the collection of data. Despite the data unavailability at this planning phase, there are various factors that can affect InfoQ even at this stage.

It is useful to distinguish between causes affecting data quality and InfoQ a priori (or ex ante) and a posteriori (or ex post). A priori causes are known during the study design stage and before data collection. They result, for example, from known limitations of resources (e.g., the sample size), ethical, legal, and safety considerations (e.g., an inability to test a certain drug on certain people in a clinical trial) and constraints on factor‐level combinations in experimental designs. A posteriori issues (the focus of