How to Investigate Within-Subject Associations between Physical Activity and Momentary Affective States in Everyday Life: A Position Statement Based on a Literature Overview

This is not a single experiment but a methodological position statement combined with a systematic literature overview. The authors tested three core methodological requirements for studying how physical activity and mood fluctuate together within a person across the day:

**Repeated assessments** (not just one pre/post measure) to capture the dynamic, moment-to-moment relationship.

**Objective measurement of physical activity** (accelerometers) because everyday activity is often spontaneous and unconscious—people cannot accurately recall how many steps they took or how intensely they moved.

**Real-time assessment of affective states** (electronic diaries, not end-of-day recall) because memory for mood is systematically distorted by peak-end effects and current mood at recall.

The "intervention" being evaluated across the 22 studies is simply *naturally occurring physical activity* (not a prescribed exercise program). The comparator is each person's own baseline or less-active moments. The outcome is momentary affective state—typically measured as positive affect (e.g., alert, happy, energetic) and negative affect (e.g., stressed, anxious, tired).

The 22 studies reviewed included a total of approximately 1,200 participants across various populations:

**Healthy adults** (university students, community volunteers) – most common

**Clinical populations** (chronic pain patients, individuals with depression, obese individuals)

**Age range:** 18–65 years, with a few studies including older adults (65+)

**Settings:** All studies were conducted in participants' natural daily environments (home, work, commuting, leisure)

**Exclusion criteria varied** but typically excluded people with mobility impairments, severe psychiatric conditions, or those taking medications affecting mood or activity

The authors note that most studies had small sample sizes (median ~40–60 participants) and were predominantly conducted in Western, educated, industrialised populations.

The authors recommend a specific measurement toolkit, which they call **ambulatory assessment**:

**Physical activity:** Triaxial accelerometers worn on the hip or thigh. These devices record acceleration in three planes (vertical, anterior-posterior, medio-lateral) and convert raw data into activity counts per minute or METs (metabolic equivalents). Some studies also used heart rate monitors or combined accelerometry with GPS. The key metric was *momentary activity intensity* (sedentary, light, moderate, vigorous) averaged over 5–30 minute windows before each mood assessment.

**Affective states:** Electronic diaries (PDAs or early smartphones) programmed to beep participants at random intervals (typically 6–10 times per day) or at fixed times. Participants rated their current mood on validated scales:

- **Positive affect:** Items like "happy," "energetic," "alert," "calm" on 5- or 7-point Likert scales

- **Negative affect:** Items like "stressed," "anxious," "irritated," "tired," "sad"

- Some studies used the **Positive and Negative Affect Schedule (PANAS)** or the **Multidimensional Mood Questionnaire (MDMQ)**

**Context variables:** Participants also reported where they were (home, work, outdoors), who they were with (alone, with friends, with colleagues), and what they were doing (working, socialising, commuting). This allowed statistical control for confounding by social context.

**Timing:** The critical design feature was that mood was assessed *during* or *immediately after* activity bouts, not hours later. The typical window was 5–30 minutes post-activity.

### Study Design

This is a **position statement with a systematic literature overview**—not a meta-analysis with pooled effect sizes. The authors searched PubMed, PsycINFO, and Web of Science for studies published between 1980 and 2012 that used ambulatory assessment to examine within-subject associations between physical activity and momentary affective states. They identified 22 eligible studies.

### Design Features of Included Studies

All 22 studies used **within-subject (intensive longitudinal) designs**:

**Repeated measures:** Each participant contributed 20–100+ observations (e.g., 6 beeps/day × 7 days = 42 data points per person)

**No randomisation:** Because this is observational (not experimental), participants were not assigned to activity conditions. The "exposure" was naturally occurring variation in physical activity.

**No blinding:** Participants knew they were being monitored for activity and mood.

**Duration:** Most studies lasted 5–14 consecutive days of monitoring.

**Statistical approach:** Multilevel modelling (hierarchical linear models) was used to separate within-person effects (e.g., "when I am more active than usual, do I feel better?") from between-person effects (e.g., "do people who are generally more active also have better mood?"). This is critical because the two can give opposite results.

### What This Design Can and Cannot Prove

**Can prove:**

That within a given person, moments of higher-than-usual physical activity are associated with better or worse momentary mood.

That this association holds after controlling for time of day, day of week, social context, and location.

That the direction of the association is likely bidirectional (activity → mood and mood → activity), though lagged analyses can partially disentangle temporal precedence.

**Cannot prove:**

Causation. Without random assignment to activity conditions, unmeasured confounders (e.g., "I went for a walk because I was already in a good mood") could explain the association.

That the effect is due to physical activity per se rather than the context in which activity occurs (e.g., being outdoors, socialising).

Long-term effects. Most studies only capture 1–2 weeks of daily life.

Generalisability to prescribed exercise programs (e.g., a structured gym workout may differ from spontaneous walking).

### Major Methodological Weaknesses Noted by the Authors

1. **Small sample sizes** (many studies had <50 participants), limiting statistical power for detecting within-person effects.

2. **Heterogeneous measurement** of both activity (steps vs. METs vs. heart rate) and mood (different scales, different time windows).

3. **Low compliance** with electronic diaries (some studies reported only 60–70% response rates to beeps).

4. **Reactivity** – people may change their activity or mood reporting simply because they are being monitored.

5. **Lack of standardised time windows** – some studies looked at activity in the 5 minutes before mood, others in the 30 minutes before, making comparisons difficult.

6. **Publication bias** – studies finding null or negative associations may be less likely to be published.

The authors summarise results across the 22 studies. Because this is not a formal meta-analysis, they report patterns rather than pooled effect sizes:

### Primary finding: Positive association between physical activity and momentary positive affect

**18 of 22 studies** (82%) found a statistically significant positive within-subject association between higher-than-usual physical activity and higher momentary positive affect (e.g., feeling more energetic, alert, happy).

**Effect sizes varied widely:** Standardised regression coefficients (β) ranged from 0.08 to 0.45 across studies. In plain terms, a one-standard-deviation increase in activity (e.g., going from sitting to brisk walking) was associated with a 0.1 to 0.5 standard-deviation increase in positive affect.

**The association was stronger in studies using objective accelerometry** (β range 0.20–0.45) compared to those using self-reported activity (β range 0.08–0.25).

### Secondary finding: Negative association with negative affect

**12 of 22 studies** found a significant negative association between physical activity and momentary negative affect (e.g., feeling less stressed, less anxious, less irritable).

**Effect sizes were smaller:** β ranged from −0.05 to −0.30. The association was more consistent for "tense" or "stressed" than for "sad" or "depressed."

**Several studies found no association** between activity and negative affect, suggesting the mood-enhancing effect may be more about boosting positive feelings than reducing negative ones.

### Temporal dynamics

**Immediate effects:** The positive association was strongest when mood was measured within 5–15 minutes after activity. Effects decayed after 30–60 minutes.

**Bidirectional effects:** 6 studies examined whether mood predicts subsequent activity. They found that higher positive affect at one beep predicted more activity at the next beep (2–3 hours later), suggesting a reciprocal loop.

**No evidence of a "rebound" effect:** Activity did not lead to later decreases in positive affect or increases in negative affect (i.e., no crash after exertion).

### Moderation by study quality

Studies that met all three methodological criteria (repeated assessments, objective activity, real-time mood) showed **larger and more consistent effects** than studies that used retrospective recall or self-reported activity.

Studies with higher compliance rates (>80% of beeps answered) also showed stronger effects.

### Context effects

The activity–mood association was **stronger when activity occurred outdoors** versus indoors.

**Social context mattered:** Activity with others was associated with larger mood improvements than activity alone.

**Time of day:** Effects were stronger in the morning and early afternoon than in the evening.

Because this is a literature overview rather than a single experiment, the authors do not provide a single effect size. However, translating the range of findings into plain English:

**Typical effect:** On a 1–7 scale of positive affect (where 4 is neutral), going from sitting (sedentary) to a brisk 10-minute walk (moderate activity) was associated with an average increase of **0.3 to 1.0 points** in positive affect. This is roughly equivalent to the mood boost from receiving an unexpected compliment or eating a favourite snack.

**For negative affect:** The same activity was associated with a **0.2 to 0.5 point decrease** in feelings of stress or tension—comparable to the relief of finishing a difficult task.

**Duration of effect:** The mood boost typically lasted **15–45 minutes** after activity stopped, then returned to baseline.

**Dose-response:** Light activity (slow walking, light housework) showed smaller effects (0.1–0.3 point increase) than moderate activity (brisk walking, cycling) which showed 0.3–0.8 point increase. Vigorous activity (jogging, fast cycling) showed similar or slightly smaller effects than moderate, possibly due to discomfort during exertion.

### Acknowledged by the authors

1. **Observational design:** Cannot establish causality. The association could be driven by third variables (e.g., being outdoors, social interaction, time of day).

2. **Heterogeneity across studies:** Different activity measures, mood scales, time windows, and populations make synthesis difficult.

3. **Small number of studies:** Only 22 studies met inclusion criteria, and many had small samples.

4. **Short monitoring periods:** Most studies lasted 5–14 days; longer-term patterns are unknown.

5. **Lack of clinical populations:** Most studies were in healthy adults; findings may not generalise to people with depression, anxiety disorders, or chronic pain.

6. **Reactivity to monitoring:** People may change their behaviour when wearing an accelerometer or carrying a diary.

7. **Publication bias:** Studies with null results may be underrepresented.

### Additional critical observations

8. **No blinding:** Participants knew the study hypothesis (activity improves mood), which could influence reporting.

9. **Self-selection bias:** People who volunteer for these studies may be more health-conscious or more interested in activity–mood links.

10. **Accelerometer limitations:** Hip-worn accelerometers cannot capture upper-body activity (e.g., weightlifting, carrying objects) or cycling accurately.

11. **Mood measurement:** Brief single-item mood ratings (e.g., "How happy are you right now?") have lower reliability than multi-item scales.

12. **Missing data:** If people skip beeps when they are in a bad mood or very active, the data may be biased toward positive associations.

13. **No control for sleep:** Poor sleep the night before could both reduce activity and worsen mood, creating a spurious association.

14. **Industry funding:** Not reported, but some accelerometer manufacturers may have funded studies.

For someone running their own n=1 experiment:

### What to test

**Specific intervention:** Compare 10-minute bouts of moderate-intensity walking (brisk pace, slightly out of breath) versus 10 minutes of sitting quietly, versus 10 minutes of light activity (slow walking, stretching). Do this 3–4 times per day for 14 days.

**Dose-response:** Test different durations (5 min vs. 15 min vs. 30 min) and intensities (light vs. moderate vs. vigorous) on different days.

**Context variation:** Compare walking outdoors vs. indoors, alone vs. with a friend, morning vs. afternoon vs. evening.

### Minimum meaningful duration

**At least 7 days** of monitoring to capture enough variation in both activity and mood. 14 days is better to account for weekday/weekend differences.

**At least 5–6 assessments per day** (e.g., every 2–3 hours while awake) to capture the temporal dynamics.

**Activity bouts should be at least 5 minutes** to produce a measurable mood effect.

### What to measure (specific metrics)

**Activity:** Use a step counter or smartphone accelerometer to record steps per 5-minute window. Alternatively, use a heart rate monitor (target: 50–70% of max heart rate for moderate activity). Record start time, end time, and perceived exertion (1–10 scale).

**Mood:** Use a 1–7 scale for each of these dimensions immediately after each activity bout:

- Positive affect: "How energetic do you feel?" (1 = completely drained, 7 = fully energised)

- Positive affect: "How happy/cheerful do you feel?" (1 = very unhappy, 7 = very happy)

- Negative affect: "How stressed/anxious do you feel?" (1 = completely relaxed, 7 = extremely stressed)

- Negative affect: "How irritable do you feel?" (1 = completely calm, 7 = extremely irritable)

**Context:** Record location (indoors/outdoors), social context (alone/with others), time of day, and what you were doing before the activity.

### Key confounds to control for

1. **Time of day:** Mood naturally fluctuates (often lower in early morning and late evening). Compare activity effects at the same time of day.

2. **Social context:** Being with others boosts mood independently of activity. Record who you are with.

3. **Location:** Being outdoors improves mood. Record indoor vs. outdoor.

4. **Prior mood:** If you were already in a good mood, you might choose to be active. Measure mood *before* the activity bout as a baseline.

5. **Sleep quality:** Poor sleep reduces both activity and mood. Track sleep duration and quality each morning.

6. **Caffeine, alcohol, and food:** These affect both energy and mood. Record consumption in the 2 hours before each activity.

7. **Weather:** Rain, extreme heat, or cold may affect both willingness to be active and mood.

8. **Day of week:** Weekends vs. weekdays have different patterns. Analyse separately or include as a covariate.

### What a positive result would look like

**Within a single day:** On days when you take more steps or do more moderate activity, your average positive affect is 0.5–1.0 points higher on the 1–7 scale compared to less active days.

**Within a single bout:** Within 5–15 minutes of starting a brisk 10-minute walk, your energy rating increases by