Hypothesis

Statement

A bounded procedural-documentation comparison is a better first pilot than an exhaustive style taxonomy or a persuasion study because it offers controlled inputs, observable user goals, low-risk synthetic tasks, and multiple outcome measures.

Mechanism

Procedures map communication to visible actions. This makes failures, errors, effort, and successful completion observable while still exercising wording, sequencing, hierarchy, examples, and audience expertise.

Predictions

• Two materially different instruction variants will produce detectable differences in at least one of task success, error count, time, or confidence calibration.
• Preference and task performance will sometimes diverge.
• Effects will vary with participant expertise and task complexity.

Evidence that would support it

A preregistered pilot can recruit representative users, execute safely with synthetic material, and yield reliable outcome differences.

Evidence that would contradict it

Recruitment is infeasible; realistic tasks require sensitive data; ceiling effects dominate; or outcomes cannot discriminate variants at feasible sample sizes.

Tests performed

Comparative desk assessment against the mission criteria and established usability measurement frameworks. No participant experiment has run.

Results

Provisionally supported for selecting the next research project, not yet validated as a product use case.

Falsification attempts

Considered exhaustive taxonomy, generic readability scoring, persuasion, and high-stakes health/legal instructions. Taxonomy lacks a near-term behavioral criterion; readability is an inadequate outcome proxy; persuasion introduces ethical/confounding risks; high-stakes domains raise avoidable safety risks.

Current assessment

Medium confidence. The decisive missing evidence is feasibility and stakeholder value for a specific procedural task.

Next experiment

Compare two versions of a synthetic, multi-step software setup or data-transformation instruction with representative participants.