A quit-app study randomizes moments: count the decisions correctly
Read decision occasions, availability and near-term outcomes in a micro-randomized trial without treating each moment as a new participant.
No Vape · Updated

Find the decision that could recur
A regular trial might assign a person to one package at enrollment. A micro-randomized design can repeatedly assign options for a time-varying component. Read the component, its alternatives and the scheduled or context-defined occasions. A repeated questionnaire alone is not evidence of repeated randomization.
Harvard’s original methods FAQ describes decision points, availability and proximal outcomes. The UK Office for Health Improvement and Disparities likewise distinguishes component evaluation from whether a whole package works. These are general methods sources, not evidence of a No Vape experiment or a reason to turn your personal routine into one.
Harvard University — Susan Murphy methods resource: Micro-Randomized Trial FAQ · UK Office for Health Improvement and Disparities: Micro-randomised trial (2021)
Check availability before interpreting the denominator
The report may define circumstances in which randomization should not occur, including safety or burden restrictions. Write down what availability means in that study. Unavailable, assigned no message, message not delivered and participant not responding are different states; they should not be collapsed into one unsuccessful intervention category.
Imagine a fictional planning app with a decision point during an approved break. It checks a stated availability condition before assigning a reminder or no reminder. The reader draws one branch for unavailable occasions and another for randomized occasions. This invented diagram contains no actual app workflow or guarantee that sensors determine availability accurately.
Keep the nearby outcome beside its time window
A proximal outcome describes the near-term effect the component is intended to influence. A distal outcome concerns a longer-term goal. Locate the exact measurement and timing for each reported result rather than treating any immediate response as a completed quitting outcome. A message being opened also differs from a behavioral outcome.
For an original reading card, label a row person, occasion, available status, assigned option and defined outcome window. You are not collecting private participant data; these are fields to identify in the public methods account. Keep the study’s handling of repeated observations visible, since the same person can contribute many occasions.
Describe component evidence without expanding the promise
A careful conclusion identifies the randomized component and context to which the estimate applies. It does not say every possible message, time or user benefits. If a headline announces that a complete app is proven because one prompt affected a short-term measure, ask for the separate evidence supporting the broader claim.
This exercise does not appraise a named trial, select notification frequency or suggest self-randomizing nicotine use. Research safety rules are also different from ordinary device settings you control. Finish with a decision-point map and one scope question: what does the reported near-term comparison establish about the longer-term claim, and what remains untested?
Sources and scope
Sources inform the guidance. The worksheets and examples are No Vape editorial suggestions, not tested treatments.
Written with AI assistance and checked against the cited sources. No clinical review is claimed. This guide covers practical planning, not individual medical treatment. How we research and write