Ask someone about their anxiety on Monday morning and again on Friday night. Both answers can be honest and different. How can we study those changes as they happen, instead of later asking for a summary of what they felt throughout the week?

01

The difference between a memory and a record

At the end of a difficult month, someone says that work stress always made evenings with their partner worse. Their memory may capture an important pattern, but it smooths over the days that did not fit. What if they had recorded each day's stress after work and their feelings about the relationship later that evening? Jean-Philippe Laurenceau, Ana M. DiGiovanni and Niall Bolger examine how such repeated observations bring psychological research closer to life as it unfolds.

The family of methods includes daily diaries, brief phone questions several times a day, experience sampling and measurements from wearable devices. Their defining feature is not merely a large spreadsheet. Each person is observed often enough to trace a change process within that person's days or weeks. A conventional survey can compare people who usually feel more or less stressed. These methods can ask whether one person tends to feel different after their own unusually stressful day.

The advantage is a closer match between the question and the time scale of the experience. A memory questionnaire asks for a summary after the fact. A daily report can capture an event and a response nearer to when they occur. Neither is perfect: answering repeatedly may itself be tiring, and a phone prompt can miss the crucial moment. The method helps only if its schedule fits the process being studied.

02

First describe what changed

The authors distinguish three questions. The first is what happened over time: when a feeling rose, fell or remained steady. They call this kinematics. A graph of each person's reports is often more revealing than one group average. In their example, three people record work stress and relationship dissatisfaction over 21 days. One reports consistently low dissatisfaction; another has a higher average; a third has especially large swings.

A study should display the time course of both the possible cause and the outcome before applying a complicated model. Otherwise an average relationship might conceal that the events happened in the wrong order, only during a brief period or only for some participants. Plotting can also reveal missing reports and impossible values. These are not decorative charts; they are a check that the study's account of daily life resembles the observations it collected.

The time scale matters. If an emotional reaction lasts half an hour, one survey at the end of each week may miss it. If a change takes months, several prompts in one afternoon may provide detail without addressing the real process. More frequent measurement is not automatically better. The best schedule follows a theory about how quickly the relevant experience can change and how long its effects might last.

03

Then ask why it changed

The second question is dynamic: does a daily event help explain a later change? Suppose dissatisfaction tends to rise on evenings after an unusually stressful workday. That pattern is more informative than finding that people with stressful jobs are, on average, less satisfied with their relationships. The two comparisons answer different questions. One follows variation inside a person; the other compares different people who may differ in many additional ways.

Even a within-person pattern does not automatically establish cause. Work stress and an evening argument might share another cause, or an unsettled relationship might influence how the day at work is experienced. Careful timing helps but cannot remove all hidden explanations. The authors discuss the need for explicit causal assumptions and for designs that place a possible cause before the outcome. When an intervention can be randomized at moments during the day, stronger causal questions become possible.

A practical danger is to speak about ‘the effect of stress’ as if all stressors were interchangeable and the path always went one way. Some days contain several events, and people may anticipate tomorrow's demands. Diaries can test narrower hypotheses: whether a specified kind of work event precedes a change over a specified interval. The gain is precision about the sequence, not a magical escape from the difficulties of causal inference.

04

The average may fit almost no one

The third question is how much the pattern differs across people. In the authors' example, work stress has little relation to evening dissatisfaction for some participants and a much stronger relation for others. A group average alone can hide that range. The review calls it heterogeneity and asks researchers to report the variation in effects, not merely the effect estimated for a hypothetical typical person.

A person may also vary in how much their own feelings fluctuate, independent of their average level. Two people with the same average mood can have very different sequences of calm and difficult days. Statistical models can describe differences in both the average level and the variability around it. The numbers matter when they test a real question, such as who is more sensitive to a daily stressor, rather than when complexity is added for its own sake.

This does not mean every participant needs a personalized explanation based on a few observations. Individual estimates can be uncertain, especially when reports are sparse or days are very similar. The authors recommend displaying the observed data and the modeled range of effects so readers can see the evidence behind an average. A claim that ‘people react this way’ should survive a look at how many people actually did.

05

A phone can collect more data and more problems

Phones and wearables make it easier to ask brief questions or passively record movement and location during ordinary life. They also increase the risk of collecting intimate information that a study does not need. A screen capture every few seconds could reveal far more than the psychological process under investigation. The review treats sensor choice and ethics as part of design, not as technical afterthoughts.

Repeated prompts can interrupt work, relationships and rest. Long questionnaires can lead people to skip responses, leave the study or answer carelessly. More data points on paper do not guarantee better data if the schedule becomes burdensome. Missing reports may be especially consequential if someone is least likely to answer during the difficult moments the researcher wants to understand. Statistical ways of handling missing data depend on assumptions that the observed data cannot fully verify.

Time between reports also varies in many studies. A question answered after ten minutes is not interchangeable with one answered after three hours when researchers estimate how one state predicts the next. The review discusses methods that account for unequal intervals and sensitivity checks for missing data. These are reasons to plan measurement before collecting it. A sophisticated model applied after the fact cannot fully repair a schedule that missed the phenomenon.

06

When observation becomes an intervention

The same tools can deliver help at a moment when it may be useful. The review discusses interventions that adapt a phone prompt to the person's recent activity and situation. In one physical-activity study, suggestions sent at selected moments were followed by more steps in the next half hour than moments without a prompt. The effect varied by location and weakened as people spent longer in the program.

That result has more structure than the promise that a clever app will change a life. Researchers randomized whether to send prompts at repeated decision points, which makes the short-term comparison more credible. They measured steps soon after the suggestion and distinguished this immediate outcome from long-term activity. They also learned that repeated prompts may lose force or become irritating. Timing can help, but constant interruption can defeat the point.

Intensive longitudinal data can therefore support both description and experiments, provided the questions remain separate. A pattern in a diary shows what tends to precede what; a randomized momentary suggestion tests what follows that suggestion under particular conditions. Neither tells us automatically whether a practice will work for everyone, whether the benefit lasts or whether the data collection is worth its burden. The review treats these as design questions, not details to smooth over.

07

The point is a better account of a day

Return to the person who says that work always ruined the evening. A month of reports might show that the association was strongest on days with a certain kind of demand, absent on other days and different for their partner. That is a more useful account than either dismissing the memory or treating it as a law. It can also reveal how much uncertainty remains when only a few stressful days were observed.

The authors urge researchers to show individual trajectories, estimates for the typical participant and the range of person-to-person differences. They also ask for enough detail about timing, sample and model to let others understand and challenge the conclusion. Different defensible analytic choices can yield different answers from the same intensive data. Clear reporting makes that uncertainty visible instead of hiding it behind a precise-looking coefficient.

The method's promise lies in matching a theory of change to the moments when change occurs. Its danger is mistaking a dense stream of observations for understanding. A record becomes informative when the researcher can say what changed, what may have driven it, for whom the pattern held and what alternative explanations remain. That is how repeated questions can illuminate daily life without pretending that a phone captured all of it.

What to keep

  1. Daily life is a scientific setting.
  2. Within-person change differs from between-person differences.
  3. Design and timing are inseparable from analysis.
?

The nuance

What this story does not let us claim

Frequent measurement can burden participants and create complex missing-data and timing problems.

Go deeper: concepts and optional reading practice

Reversible cards

Test the concept

1 / 1

Topic assessment

Check whether you can explain the idea, not merely recognize it.

Every answer is grounded in the explanation above. Some questions introduce new cases to test whether you can apply and analyze the ideas, not merely recall them.

7 questions4 options per questionOne correct answer
01Which statement best captures the review’s central thesis?
02Which is the most appropriate operational definition of “Experience sampling”?
03Which case correctly applies the central concept?
04If two studies obtain different results, which interpretation best preserves the logic of this guide?
05Which complementary conclusion belongs to the paper’s core argument?
06Which formulation best distinguishes “Experience sampling” from a superficial label?
07Which limit should accompany any application of the paper to a new case?
Answer all seven questions.

Source & scope

Intensive Longitudinal Methods: Toward a Psychological Science of Daily Life. Jean-Philippe Laurenceau, Ana M. DiGiovanni & Niall Bolger. Annual Review of Psychology, vol. 77 (2026).

This editorial guide is an independent explanation. Consult the original publication for the studies, references and full context.

Open the original review ↗