Method at a glance
- Modality: multimodal (self-report on recalled episodes)
- Measures: Emotional experience across a person’s actual day — intensity, valence, and (given enough emotion terms and episodes) granularity — reconstructed episode by episode the following day.
- Procedure: Participants recall the previous day as a sequence of discrete episodes (commonly up to five each from morning, afternoon and evening). For each episode they report when it occurred, what they were doing, where and with whom, and rate a fixed list of emotion terms. Usually repeated across two or more days to accumulate episodes per person.
- Also called: DRM, day reconstruction
Strengths
- Yields many within-person observations of self-experienced emotional events, so it supports individual-difference indices that need repeated measures (granularity, variability, co-occurrence) without the burden of live sampling.
- The events are the participant’s own, so it evokes specific emotions that standardized laboratory stimuli often fail to evoke — in Ventura-Bort et al. (2021) every effect appeared here and none in the matched picture-viewing task.
- Cheaper and far less intrusive than Ecological Momentary Assessment (EMA): one sitting per day rather than ten prompts, and no equipment.
- The episode scaffold (time, activity, place, company) is designed to cue reconstruction concretely rather than inviting a global ‘how was yesterday’ judgment.
Limitations
- It is recall, not sampling. Kahneman et al. designed the scaffold to reduce retrospective bias, not to abolish it; peak-end effects, mood-congruent recall and reconstruction from beliefs about one’s typical day all remain live.
- The number of episodes retrieved varies by person and is itself confounded with what is being measured — Ventura-Bort et al. had to enter episode count as a covariate, where it predicted intensity at trend level.
- Emotion is measured only as ratings on a supplied term list; granularity computed from it is bounded by which terms were supplied.
- Purely observational. Nothing is manipulated, so like Ecological Momentary Assessment (EMA) it establishes correspondence and cannot test causation.
- No physiological measurement of any kind, in any use the wiki holds.
Connections
Introduced by Kahneman, Krueger, Schkade, Schwarz & Stone (2004) as a compromise between the global life-satisfaction survey and the expense of Ecological Momentary Assessment (EMA): reconstruct yesterday as a sequence of episodes, then rate each one. The wiki met it first in MacCormack et al. (2021) Study 2 without giving it a page; it earns one with Ventura-Bort et al. (2021), where it carries every result in the paper.
Where it sits between the neighbouring methods
| when rated | what is rated | bias risk | burden | |
|---|---|---|---|---|
| Ecological Momentary Assessment (EMA) | now | the present moment | low (memory-free) | high |
| DRM | next day | discrete recalled episodes | moderate | low |
| retrospective questionnaire | any time | one’s life in general | high | minimal |
The DRM’s design bet is that reconstructing specific bounded episodes with contextual cues recruits episodic memory rather than semantic self-belief — which is the same bet the Property association task declines to make and Ecological Momentary Assessment (EMA) does not need to make.
The use that matters here: granularity
Any repeated-measures emotion instrument can in principle yield Emotional granularity, which requires many ratings of many terms per person. The DRM’s contribution is that it can supply ~30 episodes × ~19 emotion terms from two sittings.
Ventura-Bort et al. ran it alongside a matched laboratory instrument — the Emotion differentiation (ED) task, 40 IAPS pictures with 8 emotion labels — and the contrast is the strongest methodological result of that paper:
- DRM: granularity and intensity both predicted by individual differences in interoceptive/emotional self-report.
- Picture task: nothing predicted by anything, in 127 participants.
The authors attribute the difference to two things: standardized pictures modulate valence and arousal but “may not evoke specific emotions,” and eight labels (four positive, four negative) may not span the evoked state where the DRM offered nineteen. Both explanations favour idiographic material over normed material for anything that depends on emotional specificity rather than emotional strength.
Recorded with the caution that this is a single-paper contrast, and the DRM had a within-person advantage the picture task did not (thirty episodes vs one session), so material type and design are confounded.