Improvements in Stress, Affect, and Irritability Following Brief Use of a Mindfulness-based Smartphone App: A Randomized Controlled Trial

The researchers compared two interventions delivered via the same smartphone app (Headspace):

**Experimental condition:** The first 10 sessions of Headspace's "Take 10" introductory mindfulness meditation program. Each session was approximately 10 minutes of guided meditation, including techniques like breath awareness and body scanning, narrated by former Buddhist monk Andy Puddicombe.

**Active control condition:** 10 excerpts from Andy Puddicombe's audiobook *The Headspace Guide to Meditation and Mindfulness*. Each excerpt was approximately 10 minutes, narrated by the same person, and delivered through the identical app interface. The audiobook content was psychoeducational — it explained concepts and background of mindfulness and meditation but contained no guided meditation exercises.

Both groups were told they were participating in a "well-being program." The key difference was that one group actively practiced meditation while the other group learned about meditation without practicing it.

**Outcome measures:**

**Stress** — measured using two subscales of the Perceived Stress Scale (PSS-14):

- "Perceived Helplessness" (stress from external pressure, feeling unable to cope)

- "Perceived Self-Efficacy" (stress from personal vulnerability, feeling inadequate)

**Positive and negative affect** — measured using the Positive and Negative Affect Schedule (PANAS)

**Irritability** — measured using the Irritability subscale of the Buss-Durkee Hostility Inventory

All outcomes were assessed immediately before and after the 10-session intervention period.

**Sample size:** 69 participants completed the study (41 in the mindfulness group, 28 in the audiobook group). Originally 88 started, but 19 withdrew before finishing.

**Population:** Healthy adults recruited from the general population via a third-party participant recruitment service (findparticipants.com).

**Age:** 18–50 years old (26.9% aged 18–24, 24.4% aged 25–29, 26.9% aged 30–39, 22.0% aged 40–49 in the mindfulness group; similar distribution in control).

**Gender:** 63.4% female in mindfulness group, 53.6% female in control group.

**Ethnicity:** Predominantly White (75.6% mindfulness, 64.3% control).

**Education:** Mostly university-educated (63.4% mindfulness, 53.6% control had university degrees).

**Meditation experience:** None in the prior 6 months; no prior experience with Headspace.

**Exclusions:** Anyone with current or previous psychological illness, non-English speakers, those without smartphone and computer access.

All measures were self-report questionnaires completed online via SurveyMonkey:

**Perceived Stress Scale (PSS-14):** 14 items, 5-point Likert scale (0 = never, 4 = very often). Measures two factors:

- Perceived Helplessness (7 items) — e.g., "How often have you felt that you were unable to control the important things in your life?" Higher scores = more stress from external pressure.

- Perceived Self-Efficacy (7 items, reverse-scored) — e.g., "How often have you felt confident about your ability to handle your personal problems?" Higher scores = more stress from personal vulnerability.

**Positive and Negative Affect Schedule (PANAS):** 20 items (10 positive, 10 negative), 5-point scale (1 = very slightly, 5 = extremely). Measures current affect. Higher positive affect = better; higher negative affect = worse.

**Irritability subscale of the Buss-Durkee Hostility Inventory:** Items measuring irritability (e.g., feeling annoyed, losing temper). Higher scores = more irritable.

Participants completed these measures at baseline (before starting any sessions) and after completing all 10 sessions (or within 1 month).

**Study design:** Randomized controlled trial (RCT) with two parallel groups.

**Randomisation:** Participants were randomized using simple randomization via a computer-generated sequence after completing the screening questionnaire. The researchers did not use blocked or stratified randomization because they anticipated high dropout rates after randomization (which occurred — 81 of 171 volunteers withdrew after being randomized but before starting any baseline assessments or sessions). This is a reasonable pragmatic decision for online studies.

**Blinding:** This was an **unblinded** study. Participants knew which condition they were in (they either did guided meditations or listened to audiobook excerpts). The researchers also knew which condition participants were in. This is a major limitation because expectation effects could influence results — participants who believed meditation would help might have reported more improvement. The control condition was designed to be credible (both were presented as "well-being programs" with the same app, same narrator, same duration), but it's unclear whether participants believed they were equally effective.

**Duration:** Participants were given up to 1 month to complete 10 sessions. Each session was approximately 10 minutes, so total intervention time was about 100 minutes over up to 30 days. This is a **very brief** intervention compared to standard 8-week MBSR programs (which involve ~45 minutes of daily practice plus weekly classes).

**Adherence:** No reminders or encouragement were given. Participants self-selected when to complete sessions. The app tracked completion automatically. This mirrors real-world app usage but means adherence varied.

**Statistical approach:** Mixed-design ANOVAs (time × group interactions) were used to test whether changes from pre- to post-intervention differed between groups. Effect sizes were reported as Cohen's d (between-group). The study was powered to detect a small-medium effect (d = 0.40) with 80% power, requiring 52 participants. They achieved 69 completers, so power was adequate.

**What this design can prove:**

Causal inference: Because of random assignment, differences between groups at post-test can be attributed to the intervention (mindfulness meditation vs. audiobook) rather than pre-existing differences.

The active control condition rules out non-specific effects of using an app, listening to audio, or learning about mindfulness — so any differences are likely due to the *practice* of meditation itself.

**What this design cannot prove:**

It cannot prove that mindfulness meditation is better than doing nothing (no inactive control group).

It cannot prove long-term effects — outcomes were measured only immediately after the 10 sessions.

It cannot prove that the effects are due to mindfulness specifically rather than any structured daily practice (e.g., a relaxation exercise might produce similar results).

It cannot prove mechanisms — we don't know *why* meditation helped (e.g., increased attention, reduced rumination, placebo).

The unblinded design means expectation effects are a plausible alternative explanation.

**Major methodological weaknesses:**

**High dropout:** 81 of 171 randomized participants (47%) withdrew before starting any sessions. Another 19 of 88 (22%) who started dropped out before completing. This could introduce selection bias — those who completed might be more motivated or responsive.

**No intention-to-treat analysis reported:** The analysis included only completers (69 of 88 who started). Dropouts were not analyzed, which can inflate effect sizes if dropouts had worse outcomes.

**Unblinded:** Both participants and researchers knew group assignment.

**Self-report only:** No objective measures of stress (e.g., cortisol, heart rate variability).

**Short duration:** 10 sessions over up to 1 month is very brief; effects might not persist.

**Industry funding:** The study was conducted by employees of Headspace Inc. (the company that makes the app). This creates a conflict of interest that could influence study design, analysis, or reporting.

**Primary outcomes (stress):**

**Perceived Helplessness (stress from external pressure):** Significant time × group interaction (F(1,67) = 5.28, p = 0.025, Cohen's d = 0.45). The mindfulness group showed a greater reduction than the audiobook group.

**Perceived Self-Efficacy (stress from personal vulnerability):** No significant time × group interaction (F(1,67) = 0.01, p = 0.92, Cohen's d = 0.01). Both groups improved equally on this measure.

**Secondary outcomes:**

**Positive affect:** Significant time × group interaction (F(1,67) = 5.67, p = 0.020, Cohen's d = 0.47). The mindfulness group showed a greater increase in positive affect than the audiobook group.

**Negative affect:** No significant time × group interaction (F(1,67) = 0.01, p = 0.92, Cohen's d = 0.01). Neither group changed significantly.

**Irritability:** Significant time × group interaction (F(1,67) = 5.01, p = 0.029, Cohen's d = 0.44). The mindfulness group showed a greater reduction in irritability than the audiobook group.

**Within-group changes (pre-post):**

Mindfulness group: Significant improvements in perceived helplessness (p < 0.001), positive affect (p = 0.002), and irritability (p = 0.001).

Audiobook group: Significant improvement only in perceived self-efficacy (p = 0.002) — the measure where both groups improved equally.

**Summary:** Mindfulness meditation outperformed the audiobook control on 3 of 5 outcome measures, with moderate effect sizes (d = 0.44–0.47). Both groups improved on stress from personal vulnerability, suggesting that any structured well-being activity (even learning about mindfulness without practicing) can reduce that type of stress.

The between-group effect sizes (Cohen's d = 0.44–0.47) are considered **moderate** by conventional standards. To translate into plain English:

**Stress from external pressure (Perceived Helplessness):** The mindfulness group's reduction was about 0.45 standard deviations larger than the audiobook group's reduction. If you imagine the average person in the audiobook group, the average person in the mindfulness group moved from the 50th percentile to about the 67th percentile in terms of stress reduction — meaning roughly 17% more improvement than the control.

**Positive affect:** The mindfulness group's increase was about 0.47 standard deviations larger. This is roughly equivalent to the difference in positive affect between someone who just received good news versus someone who had a neutral day.

**Irritability:** The mindfulness group's reduction was about 0.44 standard deviations larger. This is roughly the difference between someone who is mildly annoyed versus someone who is calm and composed.

These effects are comparable to those seen in longer, in-person mindfulness programs (MBSR typically shows d = 0.3–0.6 for stress outcomes) but were achieved with only ~100 minutes of total practice over up to 1 month, compared to ~28+ hours of practice in MBSR.

However, it's important to note that these are **between-group** differences. The within-group improvements in the mindfulness group were larger (e.g., perceived helplessness dropped from ~16 to ~12 on a 0–28 scale), but the audiobook group also improved on some measures, so the unique benefit of meditation is the *additional* improvement beyond what you'd get from just learning about mindfulness.

**Acknowledged by authors:**

High dropout rate (47% before starting, 22% of starters)

No long-term follow-up

Self-report measures only

No measure of mindfulness itself (so can't confirm the intervention actually increased mindfulness)

The control condition, while active, may have been less engaging than the meditation sessions

**Critical reader additions:**

**Industry funding and authorship:** Three of four authors are employees of Headspace Inc. This is a significant conflict of interest. Industry-funded studies are more likely to report positive results. The study design, analysis choices, and interpretation may be biased.

**No intention-to-treat analysis:** The authors analyzed only completers. If dropouts had worse outcomes (which is plausible — they might have found the intervention unhelpful or stressful), the results would be inflated.

**No blinding:** Participants knew which condition they were in. The audiobook condition was described as a "well-being program," but participants may have guessed they were in the control group if they expected to meditate. Expectation effects could account for the entire between-group difference.

**Small sample:** 69 completers is modest. The study was powered for d = 0.40, but the confidence intervals around the effect sizes are wide.

**Short duration:** 10 sessions over up to 1 month is very brief. We don't know if effects persist, grow, or diminish with continued practice.

**Healthy population only:** Results may not generalize to people with clinical anxiety, depression, or chronic stress.

**No dose-response analysis:** The authors didn't report whether completing more sessions or practicing more frequently was associated with larger improvements.

**Single app:** Results may not generalize to other mindfulness apps (Calm, Insight Timer, etc.) which have different content and structure.

For someone running their own n=1 experiment:

### What to test

**Specific intervention:** 10 minutes of daily guided mindfulness meditation using a smartphone app (Headspace or similar). Focus on breath awareness and body scanning techniques.

**Dose:** One 10-minute session per day for 10 days (total ~100 minutes). You could also test longer durations (e.g., 20 minutes) or longer periods (e.g., 30 days) to see if effects increase.

### Minimum meaningful duration

**At least 10 sessions over 10–14 days.** The study found effects after just 10 sessions. However, effects might be larger or more reliable with 20–30 sessions (2–4 weeks). For a self-experiment, run it for at least 2 weeks, ideally 4 weeks.

### What to measure

**Primary metric:** Perceived Stress Scale (PSS-10, the shorter version). Available free online. Score range 0–40, higher = more stress. A meaningful improvement would be a drop of 4–6 points.

**Secondary metrics:**

- Positive and Negative Affect Schedule (PANAS) — 20 items, measures positive and negative affect separately.

- Irritability — track daily on a 1–10 scale ("How irritable did I feel today?").

- Optional: Heart rate variability (HRV) using a wearable device (e.g., Oura Ring, Apple Watch) — lower resting HRV is associated with stress; mindfulness may increase HRV.

**Measure at baseline** (before starting), **after each session** (optional, for dose-response), and **after completing all sessions**.

### Key confounds to control for

**Expectation effects:** If you believe meditation will help, you might report improvements regardless. To control for this:

- Run a blinded experiment if possible (e.g., have someone else randomize you to meditation vs. a sham activity like listening to nature sounds).

- At minimum, track your expectations before starting ("How much do I expect this to help?") and compare actual outcomes.

**Time of day:** Meditate at the same time each day to control for circadian effects on mood.

**Other stressors:** Track major life events (work deadlines, relationship issues, illness) that could confound results.

**Sleep:** Poor sleep increases stress and irritability. Track sleep quality (e.g., using a sleep diary or wearable) and control for it in analysis.

**Exercise and diet:** Both affect mood. Keep these consistent during your experiment.

**Regression to the mean:** If you start the experiment during a particularly stressful period, you'll likely improve regardless of intervention. Measure baseline for at least 3–5 days before starting.

### What a