Experience sampling (ESM / EMA)

Ecological momentary assessment — also called experience sampling — asks participants to report on mood, behaviour, or context at multiple moments across days or weeks in their natural environment. Done right it yields ecologically-valid data a lab paradigm cannot match. Done wrong it yields 30 % compliance, a mountain of missing data, and a reviewer comment that tanks your revision.

The four levers

1. Schedule type. Fixed (every day at 9 / 13 / 18), random within windows (three prompts between 8 and 22), or signal-contingent (triggered by a GPS enter/exit, a calendar event, or a partner app's webhook). Signal-contingent captures rare events but requires the participant to run a client; fixed is the most tolerant to participant drift.

2. Prompt channel. PWA push is free and instant but fails silently if the participant hasn't installed the app. Telegram and WhatsApp bots reach participants who won't install anything but need a one-time setup step. Email is universal but buried. SciBLIND ships all four and lets the participant pick their preferred channel during enrolment.

3. Response window. The window between prompt fire and auto-expiry. Five minutes captures in-the-moment data but pushes compliance down. Four hours maximises responses but blurs the »momentary« in EMA. The standard compromise is 30 minutes with one gentle nudge at the 20-minute mark.

4. Participant compensation. Pay per completed ping, not per enrolment. Completion codes for Prolific / MTurk / Bakker issue automatically when the study's minimum-ping threshold is met. SciBLIND's PayoutMode lets you set per-ping + completion-bonus so participants stay engaged across the full two-week window.

Compliance benchmarks

Academic ESM studies typically report 50–80 % compliance. Below 50 % the data is unusable; above 85 % is exceptional. Levers that move compliance:

• Shorter prompts (≤ 90 seconds per fill) — each additional minute drops compliance ~8 %.
• Channel diversity — participants who can pick their channel respond ~15 % more.
• One gentle nudge at 20 minutes (snoozable from the notification itself) — +5 %.
• Real-time feedback — a »you filled 14 of 21 so far« progress card — +10 %.

What SciBLIND ships

• Builder wizard step for schedule type (fixed / random window / signal-contingent), prompt count per day, and response window.
• Multi-channel delivery out of the box. Participants choose at enrolment; you do no integration work.
• Longitudinal per-participant invite tokens so the same person is tracked across weeks without re-enrolment.
• Compliance dashboard showing per-participant response rates and time-to-respond histograms.
• Completion codes with PayoutMode configurable (fixed / per-ping / per-ping-plus-bonus).
• Snooze from notification on PWA push (tap »Snooze 15 min« without opening the app).

When ESM is the wrong tool

• You need high-frequency sampling (< 10 min between pings). Use physiological sensors instead.
• The phenomenon is stable over weeks. A one-shot survey is cheaper and equally informative.
• Your target population cannot use a smartphone app. Paper-diary is still acceptable in clinical geriatric and low-resource settings.

References

• Shiffman, S., Stone, A. A., & Hufford, M. R. (2008). Ecological momentary assessment. Annual Review of Clinical Psychology.
• Trull, T. J., & Ebner-Priemer, U. (2013). Ambulatory assessment. Annual Review of Clinical Psychology.