Blinding during the analysis of research data

Abstract

Blinding in randomized controlled trials (RCTs) is a strategy that is widely endorsed as a method of reducing the biases that can result from people's awareness of study participants' treatment group status. Blinding of participants and interventionists is often impossible in nursing RCTs, but data analysts can almost always be blinded. Yet, such blinding seldom occurs, perhaps because of misperceptions about the objectivity of statistical analysis. Data analysts make many semi-subjective decisions about such issues as handling missing data, transforming variables, undertaking subgroup analysis, and selecting covariates. These decisions ideally should be made without the analyst's knowledge of how treatment groups are coded. Strategies for achieving blinding among data analysts are discussed.