4 Study setup and data collection

4.1 Choosing a platform

Most studies now use participants’ own smartphones, sometimes with loaner phones for people who do not own one (excluding non-owners biases the sample). The software you choose constrains your design, so evaluate it against your protocol.

Evaluation checklist

Supports the sampling scheme you need: stratified random blocks, minimum gaps, participant-specific wake and sleep times, event-contingent entries, morning/evening questionnaires, sensor triggers.
Works reliably on both Android and iOS, including notification delivery when the app is closed and battery optimisation is on.
Exports every timestamp you need: scheduled, notification sent, questionnaire opened, each item answered, submitted. Timestamps include time zone or UTC offset.
Stores answers offline and syncs later; handles app updates during a running study.
Supports branching, randomisation of item subsets, and required responses on sliders.
Offers a researcher dashboard showing compliance per participant in (near) real time.
Meets your legal requirements: data location, encryption, data processing agreements, GDPR etc.
Passive sensing and wearable integration, if needed.
Cost model (per participant, per study, subscription, free, self-hosted) and long-term availability of your data.

Examples of platforms

Features, pricing and availability change often. Treat this list as a starting point and check current documentation before committing.

Platform	Model	Notes
m-Path	Academic (KU Leuven)	ESM and momentary interventions; widely used in clinical and academic research
ESMira	Open source, self-hosted	Researchers run their own server, keeping full control of data (Lewetz & Stieger, 2024)
SEMA3	Academic (University of Melbourne)	Smartphone ESM with flexible scheduling
Samply	Academic, web-based	Schedules notifications that link to web surveys built in other tools
movisensXS	Commercial	Android-focused ESM with integration of movisens physiological sensors
ExpiWell, LifeData, Avicenna, mEMA, MetricWire	Commercial	Full-service ESM platforms with dashboards; several support passive sensing
Beiwe, AWARE	Open source	Primarily passive smartphone sensing, with survey modules
REDCap or Qualtrics + SMS	Institutional survey tools	Text-message links to web surveys; simple daily diaries; less control over momentary timing

Pitfall

Android battery optimisation and iOS Focus modes silently suppress notifications. A participant who “never got the beeps” will look like a non-complier in your data. Walk participants through notification and battery settings at briefing, and send a test prompt before the study starts.

Time zones and clocks

Record whether timestamps are local or UTC. Studies that span a daylight-saving change, or participants who travel, will otherwise produce prompts that appear an hour early or late, days with 5 or 7 prompts, or negative time gaps.

4.2 Ethics, privacy and safety

Burden is an ethical issue. Justify the number of prompts and days; let participants pause (e.g., during exams or funerals) and withdraw without penalty.
Data minimisation. Collect only what the question needs. Raw GPS traces, audio and communication logs are highly identifying; consider computing features on the device or deleting raw data after feature extraction.
Informed consent should explain what is collected, when, how often, what passive data are logged, who can see the data, where data are stored, and what happens with data if someone withdraws.
Legal frameworks. In the EU, GDPR applies to all personal data; sensitive categories like health data need an explicit legal basis and usually a data protection impact assessment.
Data security and retention. Encrypt data in transit and at rest, back it up during collection (a lost or wiped phone can mean lost days), and decide in advance how long raw and processed data are kept and when identifiers are destroyed. Share the de-identified dataset where consent and ethics allow, so results can be reproduced.
Incentives should compensate time without being coercive or pushing people to answer when they should not (e.g., while driving).

4.3 Recruitment and enrolment

Who you recruit sets what you can generalise to, and ESM recruitment has a particular bias. The burden of participation self-selects for people who are conscientious, comfortable with their phone, and free enough to answer several times a day.

If it fits your design, enrol in waves rather than all at once. Staggered (rolling) enrolment keeps briefing, monitoring and technical support manageable, and it lets problems surface on a handful of people before they reach the whole sample. It also breaks the link between study day and calendar date: if everyone starts on the same Monday, a shared external event — an exam period, a heatwave, a news shock — lands on the same study day for all of them and is confounded with time-in-study. Spreading start dates desynchronises the two.

Treat informed consent as a process, not a signature (Section 4.2): people are agreeing to two weeks of interruption and, often, passive data collection, so it is worth checking during briefing that they understand what they signed up to.

4.4 Briefing participants

A good briefing session (in person or by video call) is one of the strongest ways to support good data quality.

Explain the purpose in general terms and why each prompt matters, including the ones that arrive at inconvenient moments.
Install and configure the app together: permissions, notifications, battery settings.
Set expectations: answer as soon as possible; skip if unsafe (driving, operating machinery); a missed prompt is fine, just answer the next one; do not answer retrospectively.
Agree on contact: who to call with technical problems, when you will check in.
Give a written or video summary to take home.

4.5 Monitoring and incentives

Monitor compliance daily from the dashboard. A short message after day 1 or 2 catches technical issues and misunderstandings early.
Define in advance when you contact a participant (e.g., fewer than half of prompts answered on a day) and what you say, so contacts are consistent.
Incentives matter. In the meta-analysis by Wrzus & Neubauer (2023), studies with direct monetary compensation reported higher compliance (about 82%) than studies with other or no incentives (about 76–77%).
Tiered bonuses (e.g., extra payment for > 80% compliance) raise response rates but may push participants to rush through prompts to reach the threshold. Monitor completion times if you use them.
Non-monetary incentives such as a personalised feedback report on one’s own mood patterns are valued by many participants. Deliver them after the study so they do not change behaviour during it.