Ecological Momentary Assessment: Measuring Anxiety as It Happens

When researchers measure anxiety, they almost always ask people to remember. The GAD-7 asks about the past two weeks. The Beck Anxiety Inventory asks about the past week. Diagnostic interviews ask about the past months. In every case, the data is a reconstruction — the person’s summary of their experience after the fact.

Ecological momentary assessment takes a different approach: it asks in real time.

How it works

EMA uses technology — typically a smartphone — to prompt participants at random or scheduled intervals throughout the day. At each prompt, the person answers a brief set of questions: How anxious are you right now? What are you doing? Who are you with? Where are you? What are you thinking about?

Responses are time-stamped and geocoded, creating a fine-grained record of anxiety as it fluctuates through daily life [1]. A typical EMA protocol might prompt participants five to eight times daily for one to two weeks, producing 50 to 100 data points per person.

Why it matters: retrospective recall is biased

The core motivation for EMA is that memory is not a neutral recorder. Research on affective recall has shown systematic biases:

Peak-end bias. People’s retrospective evaluations of an emotional episode are disproportionately influenced by the peak intensity and the end of the episode, neglecting duration [3]. A bad week that ended with one terrible evening will be remembered as worse than a consistently unpleasant week that ended ordinarily.

Recency bias. The most recent experiences weigh more in retrospective reports. How you felt today colors your answer about how you felt this week.

Global versus momentary. When people with anxiety disorders rate their anxiety retrospectively, their ratings tend to be higher than the average of their momentary ratings. They overestimate their average anxiety, probably because anxious moments are more salient and memorable [2].

This means that the standard outcome measures in anxiety research — the ones that treatment effects are calculated from — are reconstructions biased toward the worst moments.

What EMA has revealed about anxiety

A systematic review of EMA studies in anxiety disorders found several patterns that standard measures miss [2]:

High within-person variability. Anxiety is not stable within a day or across days. People with anxiety disorders show substantial fluctuations, with some parts of the day consistently worse than others. Morning anxiety is common in generalized anxiety; situational spikes characterize social anxiety.

Situational specificity. EMA captures which situations trigger anxiety and which provide relief. Being alone, anticipating a social event, or encountering physical sensations may consistently precede anxiety spikes in ways that a weekly questionnaire cannot resolve.

Temporal dynamics. EMA can reveal that avoidance tends to follow anxiety rather than precede it (challenging simple avoidance models), that poor sleep predicts next-day anxiety with a lag, and that social contact can both increase and decrease anxiety depending on context.

Variability as a signal. Some researchers have proposed that anxiety variability itself — how much a person’s anxiety fluctuates day to day — may be a clinically meaningful indicator, separate from average severity. High variability might signal emotional instability, while low variability at a high level might signal chronic anxiety [6].

EMA and treatment monitoring

EMA could transform treatment monitoring. Rather than asking a patient once a week whether they feel better, a therapist could see a real-time record of anxiety across contexts. This could reveal:

• Whether improvement happens uniformly or only in certain situations
• Whether medication reduces average anxiety but increases variability (or vice versa)
• Whether side effects are constant or tied to specific times of day [4]
• Whether homework compliance correlates with between-session improvement

These are clinically important questions that session-by-session questionnaires cannot answer with the same resolution.

Limitations

EMA is powerful but not without problems:

Burden. Answering prompts five to eight times daily for weeks is demanding. Compliance typically starts high and declines. People who are most impaired may be least able to respond.

Reactivity. The act of monitoring anxiety may change it. Repeatedly rating your anxiety might make you more or less aware of it, altering the very experience being measured. Studies of reactivity show mixed results: most find small or no effects, but they are hard to rule out.

Sampling bias. Prompts that arrive during panic attacks, dissociative episodes or acute distress may go unanswered, creating missing data at exactly the most clinically relevant moments. Rare but important events, like panic attacks, are poorly captured by random sampling.

Analytic complexity. EMA produces nested, time-series data that requires multilevel modeling and careful handling of missing data. The analytical demands are substantial, and the methods are still evolving [5].

What this means for reading research

When you encounter a study that measured anxiety, ask:

1. Was measurement retrospective or momentary? Retrospective measures overestimate average severity.
2. If EMA was used, what was the compliance rate? Studies with less than 70% response rates may have biased results.
3. Were EMA findings replicated with standard measures? Discrepancies are informative, not errors.
4. Was variability reported alongside means? Average anxiety and anxiety variability capture different aspects of the condition.

For a discussion of what standard questionnaires measure and miss, see what does an anxiety study measure?. For the specific strengths and limitations of the GAD-7, see GAD-7 scores: what they mean.

When to look for more support. If your anxiety fluctuates unpredictably or is tied to specific situations, this is not unusual. A therapist who uses structured monitoring can help identify patterns and tailor treatment. Talk to your GP. If you are in crisis, call or text 988 (US) or Samaritans on 116 123 (UK and Ireland).