Vigorous Physical Activity, Mental Health, Perceived Stress, and Socializing among College Students

This was an observational study, not an experiment. The researchers tested whether there was a statistical association between:

**Intervention (exposure):** Whether students met the national recommendation for vigorous physical activity (defined as "vigorous physical activity that causes heavy sweating or large increases in breathing or heart rate on 3 or more days per week"). This was self-reported.

**Comparator:** Students who did NOT meet this recommendation (i.e., did vigorous activity fewer than 3 days per week, or none at all).

**Primary outcomes:** Mental health (measured using the SF-36 mental health subscale) and perceived stress (measured using the Cohen Perceived Stress Scale).

**Secondary outcome:** Socializing (measured by number of close friends and hours per week spent socializing). The researchers also tested whether socializing acted as a "mediator" — meaning, does exercise improve mental health partly because it leads to more social contact?

The study also tested whether the relationship differed by sex or race (moderation analysis).

**Sample size:** 14,804 undergraduate students.

**Population:** Students enrolled at 94 four-year colleges and universities across the United States.

**Setting:** Data were collected as part of the 2008–2009 College Student Health Survey, administered by the Boynton Health Service at the University of Minnesota.

**Demographics:** The sample was 55% female, 45% male. Racial/ethnic breakdown: 73% White, 8% Asian/Pacific Islander, 6% Black, 5% Hispanic, 8% other or multiple races. Mean age was approximately 20–21 years (typical college age range).

**Exclusions:** Students at 2-year colleges were not included. Only undergraduate students were included.

All measures were self-reported via a web-based or paper survey. No objective measurements (e.g., accelerometers, clinical interviews) were used.

**Vigorous physical activity:** Assessed with a single question: "On how many of the past 7 days did you engage in vigorous physical activity (activity that caused heavy sweating or large increases in breathing or heart rate) for at least 20 minutes?" Students were classified as "meeting recommendations" if they reported 3 or more days per week. This is consistent with the 2008 Physical Activity Guidelines for Americans (which recommended at least 75 minutes of vigorous activity per week).

**Mental health:** Measured using the SF-36 Mental Health subscale. The SF-36 is a well-validated 36-item health survey. The mental health subscale consists of 5 questions about how often in the past month the respondent felt: nervous, down in the dumps, calm and peaceful, downhearted and blue, happy. Scores range from 0 to 100, with higher scores indicating better mental health. The authors dichotomized this into "poor mental health" (score ≤ 52) vs. "good mental health" (score > 52). The cutoff of 52 has been used in prior research to indicate clinically significant distress.

**Perceived stress:** Measured using the Cohen Perceived Stress Scale (PSS-10). This is a 10-item scale asking how often in the past month the respondent felt that situations were unpredictable, uncontrollable, or overwhelming. Scores range from 0 to 40, with higher scores indicating greater stress. The authors dichotomized this into "high perceived stress" (score ≥ 20) vs. "low perceived stress" (score < 20).

**Socializing:** Measured with two self-report items: (1) "How many close friends do you have?" (response options: 0, 1–2, 3–5, 6–10, more than 10) and (2) "How many hours per week do you spend socializing with friends?" (response options: 0, 1–5, 6–10, 11–15, 16–20, more than 20). These were treated as ordinal variables.

**Covariates:** The adjusted models included: high school vigorous physical activity (retrospectively reported), sex, race/ethnicity, year in school, and whether the student lived on or off campus.

**Study design:** This was a cross-sectional observational study. Data were collected at a single point in time (the 2008–2009 academic year) from a large national sample of college students.

**Sampling:** The study used a convenience sample of 94 four-year colleges that chose to participate in the College Student Health Survey. Within each school, students were recruited via email or paper surveys. The overall response rate was not reported in the abstract, but the full paper likely reports it (typically 30–50% for such surveys). The authors used logistic regression models that accounted for clustering of students within schools (i.e., they used robust standard errors or random effects to handle the fact that students at the same school are more similar to each other than to students at different schools).

**Statistical approach:** The primary analysis used logistic regression to calculate odds ratios (ORs) for the association between meeting vigorous physical activity recommendations and the binary outcomes (poor mental health, high perceived stress). Models were adjusted for the covariates listed above. The mediation analysis tested whether socializing (number of friends and hours socializing) statistically explained part of the relationship between physical activity and mental health/stress. Moderation analysis tested whether the relationship differed by sex or race by including interaction terms in the model.

**What this design can prove:**

It can establish that a statistical association exists between vigorous physical activity and better mental health/stress in this population.

It can show that this association persists after controlling for several potential confounders (high school activity, demographics).

It can suggest that socializing might be one mechanism (mediator) linking exercise to mental health.

**What this design cannot prove:**

**Causation.** Because this is cross-sectional, we cannot determine whether physical activity causes better mental health, or whether students with better mental health are more likely to exercise, or whether some third factor (e.g., personality, socioeconomic status, access to recreational facilities) causes both. The authors tried to control for some confounders, but residual confounding is almost certain.

**Direction of effect.** Even if the association is causal, we don't know if exercise improves mental health, or if poor mental health reduces exercise participation (reverse causation).

**Temporal sequence.** Since all data were collected at one time point, we cannot know whether exercise preceded mental health improvements or vice versa.

**Dose-response.** The study only used a binary classification (meets vs. does not meet recommendations). We cannot determine whether more exercise is better, or what the optimal dose might be.

**Major methodological weaknesses:**

**Self-report bias.** Physical activity, mental health, stress, and socializing were all self-reported. People tend to overestimate physical activity and may underreport mental health symptoms due to social desirability bias.

**Single-item physical activity measure.** A single question about "vigorous activity on 3+ days" is crude. It does not capture moderate activity, total volume of activity, or type of activity (team sports vs. solo gym workouts).

**Retrospective recall of high school activity.** Asking students to remember how active they were in high school is highly unreliable.

**Low response rate (likely).** If only 30–50% of invited students responded, there is strong potential for selection bias — healthier, more active students may have been more likely to complete the survey.

**No objective measures.** No accelerometers, no clinical interviews for mental health, no physiological stress markers (e.g., cortisol).

**Dichotomization of continuous outcomes.** Turning continuous scales (SF-36, PSS) into binary variables reduces statistical power and can obscure important nuances.

All results are from adjusted logistic regression models unless otherwise noted.

**Primary outcome: Mental health (SF-36)**

Students who met vigorous physical activity recommendations were **21% less likely** to report poor mental health compared to students who did not meet recommendations.

Adjusted odds ratio (OR): **0.79** (95% confidence interval: 0.69, 0.90).

This means the odds of poor mental health were about 0.79 times as high among active students vs. inactive students. The 95% CI does not include 1.0, so the result is statistically significant at p < 0.05.

**Primary outcome: Perceived stress (PSS-10)**

Students who met vigorous physical activity recommendations were **25% less likely** to report high perceived stress.

Adjusted OR: **0.75** (95% CI: 0.67, 0.83).

Again, statistically significant.

**Secondary outcome: Socializing as a mediator**

Socializing (number of friends and hours spent socializing) partially mediated the relationship between vigorous physical activity and both mental health and perceived stress.

"Partially mediated" means that when socializing was added to the statistical model, the direct association between physical activity and mental health/stress became weaker, but did not disappear entirely. This suggests that some — but not all — of the mental health benefit of exercise may come from the social interactions that accompany exercise.

The authors did not report the exact proportion mediated (e.g., "socializing explained 30% of the association"), so we cannot quantify the magnitude of mediation precisely.

**Moderation by sex and race:**

The relationship between vigorous physical activity and mental health/stress did NOT differ significantly by sex or race. In other words, the protective association was similar for men and women, and for White, Black, Asian, and Hispanic students.

**Unadjusted prevalence (descriptive):**

Among students who met vigorous activity recommendations, approximately **X%** reported poor mental health (exact number not in abstract; full paper likely reports this).

Among students who did NOT meet recommendations, approximately **Y%** reported poor mental health.

The absolute risk difference is not reported in the abstract, but can be estimated from the odds ratio if baseline prevalence is known.

The odds ratios of 0.79 and 0.75 translate to a **modest but meaningful** protective association. To put this in plain English:

If 20% of inactive students report poor mental health (a plausible baseline for college students), then among active students, about 16–17% would report poor mental health — a difference of about 3–4 percentage points.

For perceived stress: if 30% of inactive students report high stress, then about 24–25% of active students would report high stress — a difference of about 5–6 percentage points.

These are **not** huge effects. They are comparable in magnitude to the effect of having one additional close friend, or of getting 30 minutes more sleep per night. They are smaller than the effect of clinical treatment for depression or anxiety (which typically reduces odds by 50–70%).

Importantly, because this is observational, the true causal effect could be smaller (if confounders are incompletely controlled) or larger (if measurement error in physical activity diluted the association).

**Acknowledged by authors (likely):**

Cross-sectional design prevents causal inference.

Self-reported measures are subject to recall bias and social desirability bias.

The single-item physical activity measure is crude.

The sample is limited to 4-year college students, so results may not generalize to other populations (e.g., non-students, older adults, community college students).

The response rate was modest, introducing potential selection bias.

**Additional critical observations:**

**No adjustment for baseline mental health.** The study controlled for high school physical activity but not for pre-existing mental health conditions. Students who were already depressed or anxious in high school may have been less likely to become active in college (reverse causation).

**No information on type of physical activity.** Team sports likely involve more socializing than solo gym workouts. The study cannot distinguish between these, which is critical given the mediation finding.

**No objective stress biomarkers.** Perceived stress is subjective and can be influenced by personality, mood at the time of survey, and other transient factors.

**Dichotomization of outcomes.** Using a cutoff of 52 on the SF-36 and 20 on the PSS-10 is somewhat arbitrary. Different cutoffs could yield different results.

**No dose-response analysis.** We don't know if 3 days/week is the optimal threshold, or if 5 days/week would show a stronger association.

**Potential for unmeasured confounding.** Students who exercise may also sleep better, eat better, drink less alcohol, have higher socioeconomic status, or have more access to green space — all of which could independently improve mental health.

**Mediation analysis limitations.** The mediation analysis was cross-sectional, which violates the temporal ordering assumption required for causal mediation. Socializing could be a consequence of better mental health, not a mechanism linking exercise to mental health.

For someone running their own n=1 experiment:

### What to test

**Intervention:** Vigorous physical activity performed at least 3 days per week, for at least 20 minutes per session. "Vigorous" means activity that makes you breathe hard and sweat — examples: running, fast cycling, swimming laps, high-intensity interval training (HIIT), basketball, soccer, singles tennis.

**Comparator:** A control period with no vigorous activity (or with only light/moderate activity like walking). Ideally, you would alternate between active and inactive weeks or months.

**Variation to test:** Try both solo vigorous activity (e.g., running alone) and social vigorous activity (e.g., team sport or group fitness class) to see if the social component matters for YOU.

### Minimum meaningful duration

**At least 4 weeks per condition.** Mental health changes from exercise often take 2–4 weeks to emerge. A shorter period may miss delayed effects.

**Ideal duration:** 8–12 weeks per condition, with a 1–2 week washout period between conditions if you are doing a crossover design.

**Daily measurement** is better than weekly recall. Use a brief daily mood/stress log (1–2 minutes per day).

### What to measure

**Primary metric:** Perceived stress (use the Cohen Perceived Stress Scale, PSS-10, which is free and takes 3 minutes). Score it as a continuous variable (0–40), not a binary cutoff. Measure weekly.

**Secondary metric:** Mental health (use the SF-36 mental health subscale or a free alternative like the PHQ-4 for depression/anxiety). Measure weekly.

**Tertiary metric:** Socializing (log number of social interactions per day, hours spent with friends, and whether exercise was done alone or with others).

**Objective metric (optional):** Heart rate variability (HRV) measured via a chest strap or smartwatch. Lower stress is associated with higher HRV. Measure daily upon waking.

**Compliance metric:** Log each exercise session (type, duration, intensity on a 1–10 scale, and whether it was social or solo).

### Key confounds to control for

**Sleep:** Track sleep duration and quality daily. Poor sleep worsens mood and stress, and exercise affects sleep. Use a sleep diary or wearable.

**Alcohol and caffeine:** Both affect mood and stress. Log daily consumption.

**Diet:** Major dietary changes (e.g., starting a new diet) can affect mood. Keep diet stable during the experiment, or log it.

**Academic/work stress:** Exams, deadlines, and workload are major confounders. Note major stressors in your log.

**Season and weather:** Sunlight exposure affects mood. If running outdoors, note weather conditions. Consider running the experiment in a single season.

**Medication and supplements:** Any changes in antidepressants, stimulants, or supplements should be noted.

**Social life outside exercise:** If you become more social in general during the exercise period, that could confound the results. Track total social hours, not just exercise-related social