Ecological Momentary Assessment (EMA)
In brief: A psychological assessment method based on repeated sampling of a subject's behaviors, affects, and cognitions in real time, in their natural environment, via smartphone micro-questionnaires. The goal: capture experience as it occurs, minimizing retrospective recall bias.
Why this concept is useful
In session, we largely depend on the patient's recollection of their week. Yet this memory is reconstructed, biased by current mood, salient events, and habitual narrative patterns. EMA offers an alternative: capturing snapshots of experience as it happens.
An increasing number of mental health applications integrate EMA-inspired features (mood notifications, daily journals, micro-check-ins). Understanding the underlying methodology allows clinicians to distinguish rigorous approaches from marketing gimmicks, and to evaluate what these data actually contribute to therapeutic follow-up.
The 3 Sampling Protocols
1. Signal-Contingent (Random Alerts)
The smartphone sends notifications at random moments throughout the day, prompting the patient to answer a brief questionnaire ("How are you feeling right now?", "What are you doing?"). This protocol captures a representative sample of daily life, without the patient being able to anticipate or prepare their responses.
2. Event-Contingent (Triggered by an Event)
The patient fills out a questionnaire each time a target event occurs (a panic attack, an interpersonal conflict, a craving). This protocol is particularly useful for addictions and anxiety disorders, as it captures the precise context of problematic episodes.
3. Time-Contingent (Fixed Schedule)
The questionnaire is sent at predefined times (morning, noon, evening). Less ecological than signal-contingent, this protocol facilitates integration into the patient's routine and allows tracking of circadian variations in mood or symptoms.
Average compliance rate: approximately 79% response rate, with 10% dropout. Questionnaire brevity (1 to 3 items) is essential to maintain engagement.
Beyond Assessment: Therapeutic Extensions
EMI (Ecological Momentary Intervention)
EMI extends EMA toward care: instead of merely assessing, the application offers a context-sensitive therapeutic micro-intervention (a breathing exercise during an anxiety peak, cognitive restructuring when facing an automatic thought).
JITAI (Just-in-Time Adaptive Interventions)
JITAIs combine EMA data, passive sensors, and machine learning to trigger tailored interventions at the right moment. While promising in theory, their effects remain modest in current studies (effect size g = 0.15), raising questions about the cost-benefit ratio compared to standard therapeutic follow-up.
Illustrative Clinical Case
Sophie, 28, followed for generalized anxiety disorder, reports in session: "My week was terrible, I was anxious all the time." Her therapist had suggested she use an EMA app for two weeks (3 random notifications per day, 2 questions: anxiety level and current activity).
The data reveals a more nuanced picture: out of 42 measurements, Sophie reports high anxiety in only 12 cases (29%), mostly concentrated on weekday mornings and moments preceding work meetings. Weekends and evenings show low to moderate anxiety levels.
Reading with EMA: the data here enables work on retrospective recall bias — Sophie reconstructs her week from the most intense moments, overlooking periods of calm. The clinician can use this data as a support for cognitive restructuring, while being careful not to reduce Sophie's subjective experience to numbers: if she says it "was terrible," that perception deserves to be acknowledged, even if the data suggests otherwise.
In Practice for the Clinician
- Use as a collaborative tool: EMA data does not replace the patient's narrative, it enriches it. Present it as a support for joint exploration, never as "proof" against subjective feelings.
- Choose the right protocol: signal-contingent suits general state assessment; event-contingent is preferable for addictions or crises. Adjust frequency to avoid fatigue.
- Anticipate reactivity: the simple act of being assessed multiple times a day can modify the patient's emotional experience. Discuss this openly.
- Set a limited timeframe: EMA is a focused exploration tool, not permanent monitoring. Setting a clear time frame (1 to 4 weeks) preserves engagement and clinical meaning.
Points of Caution
EMA is NOT:
- A digital diary: it is a structured protocol with standardized items
- A substitute for clinical assessment: micro-questionnaires (1-3 items) sacrifice psychological richness for frequency
- A universal tool: older, economically precarious, or tech-reluctant populations are underrepresented
Limitations and risks:
- Reactivity: being assessed repeatedly may modify emotional experience itself
- Event under-reporting: in event-contingent protocols, the patient may not perceive or report certain episodes
- Reductionism: ultra-brief items do not capture the complexity of lived experience
- Analytical demands: rigorous interpretation of EMA data requires advanced statistical skills (multilevel models, time series)
- Assessment/surveillance boundary: continuous data collection (especially combined with GPS) can blur the line between care and control
Key Researchers
| Researcher | Contribution |
|---|---|
| Arthur A. Stone | Co-creator of the EMA term and methodological framework (1994) |
| Saul Shiffman | Co-creator of EMA, specialist in addiction applications |
| John Torous | Integration of EMA into digital phenotyping via the mindLAMP platform |
To Learn More
- Foundational article: Stone, A. A. & Shiffman, S. (1994). Ecological momentary assessment (EMA) in behavioral medicine. Annals of Behavioral Medicine, 16(3), 199-202.
- Key review: Shiffman, S., Stone, A. A. & Hufford, M. R. (2008). Ecological Momentary Assessment. Annual Review of Clinical Psychology, 4, 1-32.
- Clinical application: The mindLAMP platform (Harvard/BIDMC) integrates configurable EMA protocols in an open-source framework.
Related concepts: Digital Phenotyping · Quantified Self (glossary)
Resource updated: January 2026