A note on blinded continuous monitoring for continuous outcomes

TL;DR

This paper explores the theoretical properties of blinded continuous monitoring for continuous outcomes, analyzing its asymptotic and finite-sample behavior and comparing blinded and unblinded variance estimators through simulations.

Contribution

It provides the first comprehensive theoretical analysis of blinded continuous monitoring, including asymptotic and finite-sample properties, filling a gap in existing research.

Findings

Blinded monitoring maintains integrity without bias.

Finite-sample performance aligns with theoretical predictions.

Using blinded variance estimators affects monitoring accuracy.

Abstract

Continuous monitoring is becoming more popular due to its significant benefits, including reducing sample sizes and reaching earlier conclusions. In general, it involves monitoring nuisance parameters (e.g., the variance of outcomes) until a specific condition is satisfied. The blinded method, which does not require revealing group assignments, was recommended because it maintains the integrity of the experiment and mitigates potential bias. Although Friede and Miller (2012) investigated the characteristics of blinded continuous monitoring through simulation studies, its theoretical properties are not fully explored. In this paper, we aim to fill this gap by presenting the asymptotic and finite-sample properties of the blinded continuous monitoring for continuous outcomes. Furthermore, we examine the impact of using blinded versus unblinded variance estimators in the context of continuous monitoring. Simulation results are also provided to evaluate finite-sample performance and to support the theoretical findings.