Creating groups with similar expected behavioural response in randomized controlled trials: a fuzzy cognitive map...

Area of Science:

• Behavioral Science
• Computational Methods
• Biostatistics

Background:

• Controlling bias is crucial for randomized controlled trials (RCTs) in behavior change research.
• Participant allocation methods aim to balance prognostic factors for comparable baseline outcomes.
• Existing methods struggle with complex interactions among prognostic factors, potentially causing bias.

Purpose of the Study:

• To present a novel computational approach for participant allocation in RCTs.
• To address limitations in current methods for balancing complex prognostic factors.
• To reduce accidental bias introduced by unbalanced groups at baseline.

Main Methods:

• A computational approach models participants' experiences to infer expected behavioral changes.
• Allocation is based on predicted behavioral shifts rather than solely on prognostic factors.
• The method accounts for interactions among prognostic factors automatically.

Main Results:

• The approach was evaluated using two real-world datasets (n=430, n=187) and synthetic data.
• Results demonstrated the approach's ability to create groups with similar expected behavioral changes.
• The method shows potential for improving group comparability in behavior change studies.

Conclusions:

• This computational approach complements existing statistical methods for complex behaviors.
• It is particularly useful when quantitative data for modeling relationships is scarce.
• Freely available software aids practitioners in study design and allocation comparison.