EmotionAppResearch Library
White papers and research papers on emotional fitness: skills you build by practice, measured in reps, and kept non-clinical. Each paper has a free PDF.
White papers
WP01
Reps, Not Feelings
The Emotional Fitness Standard: a non-clinical operating model that turns positive-psychology techniques into countable daily reps for Indian professionals.
WP02
The 4% Problem
Why employee assistance programmes in India go unused, what the unused budget actually buys, and the rep-based model that gets used daily
WP03
The Silent Payroll
Presenteeism, not absenteeism, is India’s largest mental-health cost — an estimated $158 billion a year. How to measure it in your own headcount and shrink it in 90 days.
WP04
Burnout Is Not a Diagnosis
WHO classifies burnout as an occupational phenomenon, not a medical condition. Why that makes it an emotional-fitness problem, and a 12-week team rep protocol built on that fact.
WP05
The English Ceiling
Most Indian wellness benefits speak one language; most Indian employees don't. A framework for mother-tongue emotional fitness across 23 languages — and why translation is not localisation.
WP06
Meet Them Where the Thumb Is
Mental-health apps keep under 4% of users by day 15. A WhatsApp-first delivery architecture for daily emotional-fitness reps — no download, no login, no dead app icon.
WP07
First Responder, Not Therapist
Managers shape most of the variance in team engagement, yet get no script for emotional conversations. A 40-minute, non-clinical protocol with clear hand-off lines.
WP08
Whose Feelings Are These?
Employee emotional data under India’s DPDP Act: what HR may see, why ‘aggregated dashboards’ are still a risk, and a firewall architecture in which the employer sees usage, never content
WP11
THE ZERO-CRINGE SHARE
Why most wellness programmes are unshareable (they expose struggle), and the five-factor rubric — countable rep, sunny output, badge of identity, zero-cringe share, built-in audience — that makes a programme spread on LinkedIn.
WP13
Built Like a Medical Device (Even Though It Isn’t One)
What ISO 13485 discipline — risk files, traceability, change control — buys a non-clinical wellness platform, and a 12-point vendor checklist HR can run in an hour
WP14
Measuring Emotional Fitness Without Screening Anyone
Why clinical questionnaires are the wrong instrument for a workplace dashboard, and how rep counts, streaks and emotional-vocabulary breadth give HR real numbers with no clinical data inside them.
WP15
The 3 A.M. Workforce
India’s 2.36 million GCC employees and its night-shift BPO teams run on inverted clocks. An emotional-fitness rep schedule built around circadian reality, not 9-to-5 wellness.
WP16
The Second Shift Has a Pulse
Women’s labour-force participation in India is rising; retention past thirty is not. What the double shift does to emotional bandwidth, and a seven-minute daily protocol that returns some of it.
WP17
Not Fragile. Fluent.
The first Indian cohort that names emotions out loud — and leaves when employers can't. What Gen Z expects from workplace emotional support, and how to design reps they will actually do.
WP18
The Survivor's Desk
Those who stay carry the cut. A 30-day protocol for managers and remaining teams after layoffs that protects output without pretending nothing happened.
WP19
Rejection Is a Rep
Sales teams hear ‘no’ fifty times a week. Turning explanatory style and comeback stance into countable daily reps, with a programme design that never touches quota time.
WP20
Alone Together on Teams Calls
Workplace loneliness in hybrid India, and the engineered antidote: good-news matching, thank-you ambushes and warm-wishes rounds as a six-week team protocol.
WP21
Founder Mode Needs a Cooldown
Emotional fitness for India’s startup leaders: why the people building wellness for everyone skip it themselves, and a ten-minute daily stack with board-safe metrics
WP22
From Placement to Probation
Emotional fitness for first-year professionals: the first twelve months decide whether a fresher stays, and a structured rep curriculum measured in streaks, not surveys.
WP23
The Paper Advantage
Why physical workbooks still beat apps for habit formation: retention, recall and the 66-day mark, and a blended workbook-plus-WhatsApp model.
WP24
Psychological Safety Is a Team Sport With Individual Reps
Edmondson’s construct is team-level; the behaviours that build it are individual. A rep map linking connection and reframe techniques to team-level safety.
Research papers
RP01
Reps Before Results
Does the number of completed practice sessions predict wellbeing gains better than programme length or format? Across the trials that reported sessions or modules completed, a pooled estimate of change in g per extra completed session could not be computed: mean completed dose was almost never published, and the two largest digital tests disagreed. The largest self-help app trial (Kloos and colleagues, 2022; n = 849) found a weak dose–response (modules finished versus wellbeing change r = 0.21). The best-guided self-help trial (Schotanus-Dijkstra and colleagues, 2017; n = 275) found no meaningful dose–response for adherence, while still producing a medium-to-large wellbeing gain versus wait-list (d = 0.66). Prior reviews that pooled assigned length, not completed reps, report small-to-medium wellbeing effects (Bolier 2013, d = 0.34; Carr 2021, g = 0.39; Hendriks 2020 multi-component programmes, g = 0.34). Eight adult non-clinical or mixed-community randomised trials reported completed sessions or a defined adherence rule and a wellbeing outcome. Narrative synthesis shows three consistent facts: people complete far fewer reps than the protocol assigns; people who keep practising after the assigned week tend to hold their gains longer; and a working plateau, not a pooled plateau, sits around four to six modules in two Dutch app trials. No Indian office-professional trial yet reports completed-session dose against wellbeing. Certainty of the dose–response claim is very low on GRADE. The practical case for counting reps remains stronger than the statistical case for a single slope.
RP02
The Attrition Cliff
Background. Digital wellbeing programmes reach millions of working adults. Most of those adults stop opening them within a fortnight. Baumel and colleagues reported, in 2019, a median 15-day retention of 3.9% and a 30-day retention of 3.3% across 93 mental-health apps. This review asks whether that floor still stands, whether trial dropout is the same construct, and which design features — reminders, human support, visible progress, channel, language — bend the curve for adults using apps, web tools or messaging programmes. Methods. PRISMA-2020 systematic review of PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints, 1 January 2019 to 22 September 2026. Eligible records described adult users of digital wellbeing or mental-wellbeing programmes and reported retention, engagement, attrition or dropout. Real-world open rates and trial assessment-dropout were pre-specified as separate constructs. A new random-effects meta-analysis of day-15 and day-30 same-day open rates was planned if five or more independent studies reported extractable events and denominators on that exact endpoint. They did not. Narrative synthesis was therefore used. Trial-side figures are taken from three published random-effects meta-analyses and are not re-pooled here, to avoid double-counting overlapping trials. Results. The only independent multi-app panel that reports same-day open rates remains Baumel 2019: median day-15 retention 3.9% (IQR 10.3%) and day-30 retention 3.3% (IQR 6.2%) across 93 apps. No second panel study from 2019–2026 overturns that floor. Rolling-retention cohorts sit higher because they count any later use, not an open on that calendar day. Workplace and commercially guided programmes report day-30 retention near 45–53%. Trial assessment-dropout is a different number: 26.2% (95% CI 18.1–36.3) in 18 depression-app trials; 24.1% short-term and 35.5% longer-term across 70 smartphone trials; 18.6% post-test and 28.4% at follow-up across 79 later trials. Reminders and human contact are associated with lower trial dropout. Visible progress counters do not have a verified pooled lift; gamification has been associated with higher trial attrition. No Indian consumer programme has published a day-15 or day-30 same-day open-rate. Conclusions. The day-15 floor of roughly 4% is still the best independent estimate for unguided store apps. It is not universal. A human hand-off and a workplace channel move the curve by an order of magnitude. Streaks do not. Registration. This review was not pre-registered. Search strings, eligibility and the fallback rule were specified in the commissioning protocol before extraction.
RP03
Does Tracking Change the Thing Tracked?
Working adults now log mood the way runners log kilometres. The question is whether the log itself moves the mood. Fewer than five eligible randomised trials compared mood or emotion self-monitoring alone with a no-monitoring control in adults, so a pooled Hedges g cannot yet be computed; the narrative direction across the closest trials is near-null for mean mood and symptoms. We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF from 1995 to 2026 for randomised or controlled studies in which adults completed mood or emotion diaries, ecological momentary assessment or app check-ins with no other active component. Primary outcomes were mood, affect, wellbeing or symptoms at the end of the monitoring period. Pre-specified moderators were logging frequency, prompt valence and whether feedback was shown. Two trials supplied a true no-monitoring contrast. In 184 recently bereaved Dutch adults, five daily grief-reaction check-ins for two weeks did not change group-level depression, trauma or grief scores versus waitlist. In 124 Hong Kong women with mild-to-moderate symptoms, six weeks of experience sampling alone produced a depression-anxiety-stress decline comparable to no extra programme; only the arm that added personalised feedback beat control. Frequency-manipulation studies without a no-monitoring arm mostly found no mean change in pleasant-unpleasant mood. Positive-only happiness prompts were an exception in one student sample: people already high in depressive symptoms or neuroticism felt worse when asked about happiness six times a day. Emotional clarity and emotion differentiation sometimes rose even when mood level did not. This is not the same phenomenon as initial elevation bias, which inflates the first report of a negative state and then settles. Certainty is low. No Indian randomised trial of logging-only versus no-monitoring in working adults was found. For an Indian workplace programme the practical reading is simple. A daily mood rep is a thermometer, not a workout. Pair the log with a next action, keep the prompt valence-neutral, and do not sell the check-in as if it were the coaching.
RP04
Streaks, Badges, Points
Background. Health and wellbeing apps leak users the way a sieve leaks water. Designers reach for streaks, badges, points and leaderboards. The open question is not whether the widgets look busy. It is whether they raise adherence without taxing the outcome the programme was built to move. Findings. Across a conservative random-effects pool of six randomised comparisons (n ≈ 1,789), gamification raised adherence by Hedges g = 0.45 (95% CI 0.23 to 0.67). Heterogeneity was high (I² ≈ 70%). The 95% prediction interval was wide and likely crosses zero, so a new trial could still find a null. Outcomes did not fall. In nine conservative comparisons they rose by g = 0.38 (95% CI 0.30 to 0.46; I² = 0). An independent 2024 review of 36 trials found a trivial-to-small steps lift of 489 steps a day. White-hat mechanics (points and badges earned for showing up) and dark-hat mechanics (endowed points lost for missing) both moved adherence; the dark-hat cell was too small to claim superiority. Dark-hat Penn-style programmes raised steps but did not add weight loss, HbA1c change or medication-taking versus a non-gamified programme. Interpretation. Gamification is a small-to-moderate adherence lever and a small outcome lever. It is not a miracle and it is not a tax. Reward-for-showing-up is the safer default for an Indian workplace programme. Loss-framed streaks raise short-term goal hits and then fade. They have not bought extra clinical change. Registration and limits. This is an update synthesis of prior review corpora plus a 2022–2026 forward search (PubMed, Scopus, Web of Science, PsycINFO, Google Scholar, ClinicalTrials.gov, CTRI, OSF). No Indian CTRI trial isolates gamification from cash in working-age adults. Certainty is moderate for steps and adherence, low for wellbeing scales and clinical endpoints.
RP05
The Half-Life of a Wellness Programme
Every published workplace wellbeing trial reports the post-test. Far fewer report what is left when the last session has ended. Across published follow-up pools, about 70 per cent of the post-test wellbeing effect remains at one to three months; at six months, programmes with continued practice still hold roughly 85 per cent, while universal one-shot programmes hold about a quarter. This PRISMA-2020-structured evidence synthesis asked how much of a workplace wellbeing or stress-management gain survives at one, three and six months, and what slows the fade. Eligible evidence was randomised or controlled trials of employed adults, 2000-2026, with a wellbeing or stress outcome at post-test and at least one later follow-up. Search covered PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints. Decay was modelled as the ratio of follow-up Hedges g (or Cohen's d) to post-test g. Prior syntheses already pooled time-stratified effects: Vanhove and colleagues found resilience-building programmes dropped from d = 0.26 within a month to d = 0.07 later; Vonderlin and Michaelsen found mindfulness-based workplace programmes held most of the wellbeing effect to twelve weeks and lost more of it after that. Primary trials with continued digital access or coaching (Heber 2016; Ebert 2016; Radin 2025; Aikens 2014) held or grew the stress effect at four to six months. Self-guided delivery of a similar programme retained about two-thirds. One large worksite trial found a null at six and twelve months. Certainty is low to moderate: follow-up samples shrink, wait-list arms contaminate later comparisons, heterogeneity is high, and corporate India has almost no decay-tracked randomised trial. The practical implication is not mysterious. A programme is not a day. It is a practice that has to be repeated after the facilitator leaves the building.
RP06
Name It to Tame It, Tested
Stuck on a stalled local train, an analyst types one word into her notes app: “dread”. This paper asks whether naming a feeling does any measurable work on its intensity. Across six extractable laboratory experiments in non-clinical adults (N = 261), affect labelling reduced self-reported intensity of high-arousal negative affect by Hedges g = 0.41 (95% CI 0.29 to 0.54) relative to watching without a label. The random-effects (REML) model showed I² = 0% and τ² = 0 on this restricted set. Leave-one-out estimates ranged from g = 0.39 to 0.48. Physiological outcomes were synthesised separately and were not forced into one number. Skin-conductance and amygdala studies more often showed downregulation than did self-report in applied exposure paradigms. In two analogue-fear trials, the body quietened and the rating did not. Emotional granularity could not be tested as a pooled moderator. No eligible trial crossed a granularity measure with a labelling-versus-control contrast on intensity. First-versus-second language of labelling was also untested. One experiment found a reversal at low intensity: labelling a mild aversive picture increased distress. Field evidence is thin. One large observational Twitter study showed a sharp valence reversal after people wrote “I feel…”. No randomised Indian trial was found. Certainty is low to moderate for laboratory self-report, moderate for the replicated amygdala pattern, and very low for granularity and language moderators. The practical reading for an Indian workplace programme is modest and conditional. A short, specific label is a countable daily rep for strong unpleasant affect. It is not a guarantee, not a session, and not a substitute for other skills. People often predict that naming will make the feeling worse. In the high-arousal laboratory tasks reviewed here, the rating usually moved the other way.
RP07
Emotional Granularity as a Trainable Skill
Emotional granularity is the skill of putting a feeling into a precise word rather than a blunt one. Two eligible randomised trials of vocabulary-based programmes in non-clinical adults were identified between 2010 and 2026; a quantitative meta-analysis was therefore not possible and this paper reports a PRISMA-2020 systematic review with narrative synthesis. The first trial (n = 120) taught twelve emotion concepts over five days of about fifteen minutes each and found a reliable rise in negative-emotion differentiation against a geography control, still visible at one month; positive-emotion differentiation did not move. The second trial (n = 118 analysed) used a two-day online word-learning task and found no direct group effect on negative-emotion differentiation, self-efficacy, or later distress; only participants who wrote highly specific personal memories showed a path from learning to differentiation to lower one-week distress. Adjacent evidence — mindfulness programmes, intensive experience sampling, and brief laboratory labelling tasks — suggests the skill can shift, but those designs did not isolate vocabulary training. Transfer to distress and regulation therefore remains unproven in a pooled estimate. No eligible Indian trial was found. For an Indian workplace programme the honest product claim is modest: a short, digital vocabulary practice can raise the resolution of unpleasant feeling-words in some adults; it is not yet a proven lever for burnout scores. Certainty is low (GRADE). The field needs larger, pre-registered, multi-language trials with a differentiation index and a distress endpoint, including Hindi and other Indian languages.
RP08
Savouring Works — But for How Long?
Savouring is the practised skill of noticing a good moment and staying with it long enough for the feeling to register. Across twenty randomised trials and 4,805 adults, the pooled post-test effect of savouring programmes on emotional outcomes was a moderate Hedges g of 0.51 (95% CI 0.26 to 0.77). That number comes from Chen, Sie and colleagues (2026), the largest PRISMA-compliant synthesis in this lineage. Separate estimates in the same synthesis were g = 0.50 (0.24 to 0.75) for positive psychological states and g = 0.61 (0.31 to 0.91) for negative emotional symptoms. Heterogeneity was high (I² = 86.6%). Small-study tests did not show a clear publication-bias signal (Egger p = .108; trim-and-fill adjusted g = 0.48). The present review asked a narrower question than Chen et al.: what remains at one month and at three months in non-clinical adults, after gratitude-only trials are removed. Fewer than five eligible non-clinical controlled trials report extractable pairwise data at both of those windows. A durability meta-analysis is therefore not yet possible. The one non-clinical adult study that measured both one- and three-month follow-up is a single-arm pilot (Smith and Hanni, 2019). High-fidelity completers improved within person. That is not the same as an effect versus control. No CTRI-registered savouring trial was found. No trial in the pooled set was delivered in an Indian language to Indian working adults. For an Indian workplace programme the honest claim is this: a short daily savouring practice produces a moderate lift at the end of a two-week sprint. Whether that lift is still there at day 45 and day 90 has not been pooled. Measure it. Do not assume it.
RP09
Three Good Things, Twenty Years Later
Lyubomirsky’s weekly-beats-daily claim has been repeated for twenty years and has never been pooled. The pooled well-being effect of gratitude interventions is small: Hedges g = 0.19 (95% CI 0.15 to 0.22), from the largest pre-registered synthesis (k = 145 papers, N = 24,804, 28 countries). The listing, or “counting blessings”, subtype sits slightly lower, at g = 0.17. We asked a narrower question. Does weekly gratitude listing produce larger or more durable gains than daily listing in adults? We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF from 2003 to September 2026. Eligible studies were randomised or controlled trials of gratitude listing or journaling with frequency stated in the methods — never inferred. Primary outcomes were well-being and positive affect. Depressive symptoms were secondary. Twelve listing trials stated frequency exactly and met adult-population rules. Zero published trials randomised daily listing against weekly listing. The famous “weekly beats daily” contrast is a mis-statement of an unpublished student study that compared one session a week with three sessions a week. A prior meta-analysis already tested days of listing as a moderator of well-being and found no effect (p = .233). Daily lists produced small gains in positive affect in some trials (typical published d = 0.15 to 0.38 versus a neutral control; larger versus a hassles control) and a null in the largest listing trial (N = 958). Weekly lists produced small gains in life evaluation in some trials and nulls in others. Twice-weekly listing for four weeks, in Hong Kong hospital staff, reduced depressive symptoms (d = −0.49) and perceived stress (d = −0.95) at three months. Effects appear by week one or two. Stretching a list from seven days to fourteen does not reliably enlarge the gain. Thrice-weekly prompts for three weeks faded by three months. Certainty is moderate for a small listing-versus-control gain, and very low for any claim that weekly beats daily. No high-quality adult listing trial from India was found. For an Indian workplace programme the honest dose is not “daily forever”. It is a short, prompted, language-local list, one to three times a week, with a stop-rule when the practice goes stale.
RP10
The Thank-You Letter Effect
Does delivering gratitude to a person outperform writing it privately? A pooled expressed-minus-recorded Hedges g is not yet possible, so the number we can defend is Kirca, Malouff and Meynadier (2023): expressed gratitude versus a neutral control, g = 0.22, 95% CI [0.11, 0.33], from 25 randomised samples and 6,745 adults. Fewer than five eligible adult trials isolate delivered thanks versus private journaling with extractable effects, so this paper is a PRISMA-informed systematic review with narrative synthesis, not a new meta-analysis. That g = 0.22 estimate mixes delivered letters, unsent letters and mixed-intent writing. Inside the same review, whether the letter had reached the recipient before post-test did not moderate expresser wellbeing (Q(2) = 0.30, p = .86; delivered g = 0.19). Head-to-head trials tell a split story. One Irish trial found that adding a weekly thank-you to a gratitude journal improved affect balance and lowered low-mood scores at three weeks. A United States trial of 916 students found that a private unsent letter already lifted positive affect (partial d = 0.29) and life satisfaction (partial d = 0.17); texting the benefactor added connectedness (partial d = 0.18), not extra happiness. Recipients, measured separately, feel more positive than expressers predict (d = 0.74). Indian and other collectivist samples show smaller or null state-gratitude effects and more indebtedness. For an Indian workplace programme the honest product is a short written thank-you, optionally sent on WhatsApp to a person of the writer’s choice, never mandated up the hierarchy. Certainty is moderate for a small expressed-versus-control effect on expresser wellbeing and very low for any incremental happiness from delivery.
RP11
The Capitalisation Effect
People tell someone about the best thing that happened to them on most days. This systematic review and restricted meta-analysis asked how strongly an active-constructive response to that good news — enthusiasm plus curiosity, not a polite nod — relates to relationship quality, and whether training the response changes team outcomes. Across five independent published samples (k = 5, N = 1,071), perceived active-constructive responding was associated with relationship satisfaction at r = .40 (95% CI .35 to .45; I² ≈ 0%). That is a medium-to-large association, consistent from dating couples in the United States to married couples in Italy and dating adults in Malaysia. Only the active-constructive style is reliably positive; quiet congratulations and both flavours of rain-on-the-parade track with lower satisfaction. A randomised trial of a 20-minute active-constructive practice plus four weeks of daily reps did not change self-reported relationship satisfaction (γ = 0.09, p = .20). It did raise perceived partner satisfaction and gratitude receipt. No eligible randomised trial has tested active-constructive training against team cohesion. No India-resident primary study was found in PubMed, Google Scholar, CTRI or ClinicalTrials.gov. The association is real. The training effect on one’s own satisfaction is unproven. Transfer to Indian workplaces is a hypothesis, not a result. Programmes should treat active-constructive responding as a countable daily rep, measure perceived response rather than self-congratulation, and test it in India before claiming team change.
RP12
Six Breaths a Minute
Slow-paced breathing is a countable daily practice, not a personality change. The last published random-effects pool of breathwork versus control for self-reported stress was Hedges g = -0.35 (95% CI -0.55 to -0.14) from 12 randomised trials and 785 adults. That figure comes from Fincham and colleagues, search closed February 2022. This paper updates that review, and Laborde’s 2022 heart-rate-variability (HRV) review, with trials published through September 2026. The question is narrow. In non-clinical adults, and in employed samples where they exist, how large is the effect of paced breathing at ten breaths a minute or fewer, and what is the shortest session that still works? We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF. Eligibility was restricted to non-clinical adults, paced breathing at <=10 breaths per minute with or without HRV biofeedback, any control, and outcomes of perceived stress, HRV or mood. A de-novo study-level re-pool of only employed, non-clinical samples with complete extractable means and standard deviations is not yet possible without inventing numbers. We therefore report the published pools, then narrate the 2023-2026 trials that change how those pools should be read. The largest subsequent trial, Fincham, Strauss and Cavanagh 2023, randomised 400 adults to coherent breathing at about 5.5 breaths per minute or a matched 12-breath placebo, ten minutes a day for four weeks. The group-by-time effect on stress was etap2 < 0.001. Both arms improved. Against a well-matched paced control, the extra calm of “six breaths a minute” is not demonstrated. Against wait-list or usual care, earlier small-to-medium effects remain. During the practice itself, vagally indexed HRV rises with a medium effect (Laborde g = 0.53 for RMSSD). After multi-week programmes the resting-HRV rise is small and unstable (g = 0.32; several newer trials found no resting change). Five minutes is the floor for a useful session. Ten minutes a day for four weeks is the best-tested daily dose. No published randomised trial of isolated slow-paced breathing in Indian working professionals could be verified. Certainty for questionnaire stress in working adults is low to moderate. A null against an active placebo is a result, not a failure of the file.
RP13
Five Minutes a Day
Working adults do not have a spare half-hour. They have five minutes between meetings, and they want to know whether that sliver does anything. This PRISMA-2020 systematic review asked how large the effect of breathing practices of five minutes or less is on stress and mood, acutely and after daily practice. A quantitative meta-analysis is not yet possible: six eligible controlled datasets (approximate unique n = 340) were found for 2010–2026, but designs, comparators and time-frames were too heterogeneous to pool under a random-effects model without inventing a number the data do not support. The clearest four-week signal remains Balban and colleagues (2023): 108 non-clinical adults practised five minutes a day for 28 days. All arms, including mindfulness, improved state anxiety. Breathwork, especially cyclic sighing, produced a larger daily rise in positive affect (about +1.91 versus +1.22 points) and a larger fall in resting respiratory rate. One-minute cyclic sighing and box breathing, delivered in real threat moments, also cut state anxiety versus a passive control in a 2026 field pilot. A first laboratory session of slow-paced breathing can raise perceived stress in novices, which is the opposite of the brochure and matters for onboarding. Certainty is low to moderate for acute subjective calming, low for a four-week mood advantage over mindfulness, and absent for Indian working adults at this exact dose. No completed CTRI trial of isolated ≤5-minute breathing in non-clinical professionals was identified. The practical reading for an Indian workplace programme is modest and useful: five minutes is a countable rep, not a miracle; the first minute appears to carry much of the acute feel; daily practice for four weeks is the only duration with a head-to-head trial against mindfulness; and local evidence still needs to be built.
RP15
Move for Mood, Not Medals
Singh 2023 and Noetel 2024 put exercise level with psychotherapy for depression. That claim was built on people who already met a clinical cut-off. This umbrella review asks whether the same parity holds in adults who are not depressed — the people who fill Indian offices. In 183 randomised trials of 22,811 adults without a diagnosed condition, physical activity improved wellbeing by a standardised mean difference of 0.42 (95% CI 0.26 to 0.57), statistically indistinguishable from mindfulness (0.44, 0.35 to 0.54) and from yoga and other psychological programmes (0.41 to 0.49). Combined movement-plus-psychological programmes reached 0.73 (0.27 to 1.20), but that node rests on three trials. We searched eight databases and registries (2010 to 22 September 2026) for reviews of physical activity in non-clinical adults; fourteen met the eligibility screen. The large depression umbrellas (Singh, Munro, Rebar) overlap heavily, so they were not treated as independent. The independent head-to-head unit is Wilkie 2026, a network meta-analysis of adults without diagnosed conditions. Against inactive controls, physical activity produces a small-to-moderate lift in wellbeing and a small-to-moderate drop in stress and low mood. Against psychological programmes, the difference is compatible with zero. Certainty is moderate for activity versus sitting still, and low for activity versus coaching. Higher intensity helps mood symptoms more than it helps workplace stress. Leisure-time activity helps; occupational physical labour does not. The practical sentence for an Indian workplace is not “skip the coach and just run.” It is “move and coach are both countable reps.” Count the walk.
RP16
Sleep Is a Regulation Skill
Sleep is sold as a mood upgrade. The experimental record for healthy adults is thinner than the slogan. A pooled Hedges g for next-day emotion regulation was not computed: fewer than five randomised trials shared a common regulation endpoint after a sleep-improving programme in adults without diagnosed insomnia. The best-powered trial that measured both affect and regulation strategies (n = 72) found a small next-day drop in negative emotion (interaction d = 0.20) and no change in reappraisal, suppression or related strategy use. This PRISMA-2020 systematic review searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF (2005–2026) for randomised or controlled trials of sleep extension, sleep hygiene or non-clinical cognitive-behavioural sleep components. Eligible outcomes were emotion-regulation measures, irritability, and positive or negative affect, preferably next day. Diagnosed-insomnia samples were excluded. Nine studies inform the narrative. In short sleepers, adding about 45–90 minutes of sleep opportunity for a week can lift next-day affect by a small amount and can lift flourishing and gratitude. In already-adequate sleepers, pushing time in bed by three hours can worsen mood. Emotion-regulation strategy use — the skill of reappraising a rude email rather than merely feeling less grim — has been tested in one eligible randomised trial and did not move. A large field experiment in Chennai found that 27 extra minutes of fragmented night sleep did not improve well-being; a protected workplace nap did. Prior reviews show the other arrow clearly: sleep loss cuts positive affect and adaptive regulation. Improving sleep quality in mixed and clinical samples improves mental-health symptoms by about half a standard deviation. Those numbers do not licence a claim that healthy Indian professionals will reappraise better tomorrow if they simply go to bed earlier tonight. Certainty is very low for regulation, low for affect, and insufficient for irritability. The honest product for a workplace programme is a sleep-opportunity rep plus a separate regulation rep, not a bundled promise.
RP18
Reappraisal on the Clock
Webb and colleagues pooled laboratory reappraisal in 2012. Nobody has pooled workplace training trials of the same skill. This review asked whether reappraisal or cognitive-restructuring training delivered to employees reduces stress and improves performance ratings. Four strict workplace reappraisal-training trials met eligibility (three between-person randomised trials and one micro-randomised trial). That is one short of the pre-specified minimum of five comparable trials, so a random-effects meta-analysis was not performed. Effects were synthesised narratively. The largest trial, a pre-registered United States field experiment with 4,054 early-education and family-service workers, found a small reduction in stress six months after a brief online reappraisal session versus an active control (d = −0.13, 95% CI −0.21 to −0.05) and higher self-rated job performance. A between-person ecological-momentary trial (n = 176) found that twice-daily reappraisal prompts reduced negative affect, raised positive affect, and improved self-rated performance through positive affect, with no fade-out at one month. A companion micro-randomised trial (n = 94) raised positive affect but did not cut negative affect. A finance-sector mindset workshop improved self-rated interpersonal performance at four weeks against a waitlist. Adjacent cognitive-restructuring programmes produced small self-rated performance gains (one Japanese trial, d = 0.31). Every extractable performance outcome was self-rated. No supervisor-rated performance trial was found. No eligible Indian employee trial was found. Certainty is low for stress and very low for performance. The skill looks transferable. The evidence is not yet poolable. Indian workplaces still lack a dedicated trial.
RP19
Stress-Is-Enhancing, Pooled
Brief stress-mindset practices try to shift a single belief: that stress can enhance performance, health and growth rather than only harm them. Across seven randomised comparisons with convertible effect sizes, the pooled near-term effect on stress mindset was Hedges g = 0.66 (95% CI 0.39 to 0.94; I² = 80%; 95% prediction interval −0.20 to 1.53). The average is medium. The prediction interval crosses zero. That is the honest headline. A three-minute video can move the score today and lose the gain by week four. A metacognitive session, an imagery rehearsal, or daily writing reps hold the shift longer. Downstream effects are smaller and less stable. A prior performance meta-analysis of stress-is-enhancing practices alone found d = 0.18 (95% CI −0.15 to 0.51; k = 6), which was not statistically significant. Symptom and wellbeing results split: some employee and student trials report fewer physical complaints or better affect; two German-speaking adult trials found no wellbeing lift once the mindset change faded. Physiological findings are sparse and should not be pooled. No eligible Indian trial was found in CTRI, ClinicalTrials.gov, or the bibliographic databases. For Indian professionals the implication is practical rather than mystical. The technique is cheap, short, and teachable as a daily rep. It is not a silver bullet. Without rehearsal the belief slips. With rehearsal, and with a coach who does not pretend stress is only sunshine, it is a reasonable module in a workplace programme. Certainty is moderate for immediate mindset change and low for durability, symptoms, performance and physiology.
RP20
The Attribution Advantage
Sales work is a rejection sport: the story told about the last “no” often decides the next call. Fewer than five eligible workplace trials of explanatory-style training were found, so effects were not pooled. Seligman and Schulman’s 1986 observational studies of life-insurance agents linked an optimistic explanatory style to higher sales and lower quitting; the open question is whether that style can be trained, and whether training moves persistence, wellbeing and sales numbers. We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI (India), ClinicalTrials.gov and OSF preprints (1985–2026) for randomised or controlled trials of explanatory-style training or attributional retraining in adults, with priority for sales and customer-facing employees. The pre-specified random-effects plan (REML, Hedges g, prediction intervals, Egger, trim-and-fill, leave-one-out) was parked. We report a PRISMA-2020 systematic review with narrative synthesis, Risk-of-Bias 2 judgements and GRADE certainty. Two workplace training studies met tight eligibility: Proudfoot and colleagues (2009), a United Kingdom wait-list randomised trial in financial-services sales (n = 166); and Wang (1994), a Chinese industrial work-group programme (n = 110). One further workplace programme focused on learned optimism in Swedish teams (Harty and colleagues, 2015). One online resilience course for Australian sales managers, which included optimism and causal analysis, found no effect on distress or work performance (Abbott and colleagues, 2009). In the only sales randomised trial, seven weekly three-hour group sessions shifted attributional style, cut resignations from 12% to 4% over eight months, and later saw more agents sitting at or near division-average productivity. Observational MetLife data remain the cleanest sales-number signal: agents in the optimistic half sold 37% more insurance over two years and remained in post at twice the rate. No Indian workplace trial was registered or published. Explanatory style is trainable in a sales workforce when the dose is high, the practice is face-to-face, and the content targets how people explain setbacks. The sales-performance evidence for training itself is still thin. India has the MetLife problem — roughly three in ten new life-insurance agents leave in year one — and none of the MetLife trial. Daily attributional reps on a WhatsApp-first coach are a testable next step, not a settled claim.
RP21
‘Yet’ Is Not Enough
Growth-mindset programmes tell working adults that ability is not a fixed rank. A quality-adjusted reading of the adult-work trials is consistent with a null-to-small effect on performance and wellbeing, in the same band as Macnamara and Burgoyne’s high-quality student estimate of d = 0.02. This PRISMA-2020 systematic review asked what growth-mindset interventions do to adult learning, wellbeing and performance at work. We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints from 2007 to September 2026. Eligible designs were randomised or controlled trials in employed adults or vocational trainees. Four published reports describing five adult-work samples met inclusion for narrative synthesis. A confirmatory random-effects meta-analysis was not possible: fewer than five independent trials offered a clean intervention-versus-control contrast on a pre-specified primary outcome. Proximal belief change was the only replicable signal. In two East-Asian employee samples the immediate shift in growth-mindset scores was about g = 0.47 to 0.50. Distal job satisfaction, flourishing and rated performance were null or confined to a dual-focus design that treated both the self and the job as malleable. A standalone slogan of ‘not yet’ did not produce lasting work outcomes. No eligible trial was conducted in India. Indian workplace programmes should treat the word ‘yet’ as a first rep, not a result. Pair the belief with a skill, a job-crafting plan and a countable practice. Certainty of evidence is low for belief change and very low for performance and wellbeing.
RP22
Be Your Own Yaar
Ferrari 2019 pooled all comers. In employed adults, self-compassion practices reduced stress by Hedges g = 0.45 (95% CI 0.22 to 0.68; k = 4) and reduced burnout by g = 0.44 (95% CI 0.16 to 0.71) in two digital wait-list trials; adding a teacher trial made the burnout interval include no effect. Self-compassion itself rose by g = 0.63 (95% CI 0.43 to 0.83; k = 5; I² ~ 0%). Self-criticism could not be pooled. We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints (2010 to 22 September 2026) for randomised or controlled trials of self-compassion practices — self-kind rewriting, compassionate letters, brief daily exercises — in employed adults, with any control. Effects were Hedges g from pre–post change over the pooled pre-intervention standard deviation, pooled with random-effects REML. Risk of bias used RoB 2. Certainty used GRADE. Practices are described generically. About 1,840 records were identified and 78 full texts assessed. Eleven trials entered the qualitative synthesis. Five randomised trials contributed extractable self-compassion effects. No eligible Indian employed-adult randomised trial entered the pool. Certainty is moderate for self-compassion, low–moderate for stress, low for burnout and very low for self-criticism. The worker-only stress effect is smaller than Ferrari’s all-comers g = 0.67. Brief daily reps of 10 to 15 minutes for four to eight weeks did the work. A 59% burnout workforce in India still has no eligible trial. This review was not pre-registered.
RP23
Write It Out, Forty Years On
Frattaroli’s 2006 random-effects average for experimental disclosure was r = .075. Two decades later the best available synthesis for depression, anxiety and stress in healthy and subclinical adults is r = .06 (Hedges g = −0.12, 95% CI −0.21 to −0.04), from 31 randomised trials and 4,012 participants. The effect has not grown. It has found a niche. This PRISMA-2020 systematic review asked what expressive writing does for wellbeing, affect and health-care use in healthy adults from 2006 to 2026, and for whom. Population: healthy adults. Practice: timed private writing, usually 15–20 minutes across three to four sessions. Comparator: neutral writing or no writing. Designs: randomised and controlled trials. Trauma-focused writing programmes sit outside this wellness lane. Searches of PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF, plus backward citation from prior reviews, identified eight subsequent syntheses and twenty verified healthy-adult trials. A new primary-study REML pool was not computed: cell-level means and standard deviations were not extractable for five or more eligible trials with usable confidence intervals. The quantitative backbone is therefore the published healthy-adult syntheses, converted and compared with Frattaroli. Effects on symptoms are small and delayed. Sessions one to three days apart outperform weekly spacing. In head-to-head trials, positive-focused writing beats classic disclosure for mood in the general population (mean difference 0.45, 95% CI 0.18 to 0.71). Best-possible-self and gratitude prompts are the most consistent positive techniques. Classic disclosure can raise anxiety in people who rarely express emotion, and raised stress in one online pandemic trial. An Asian-sample synthesis was null (g = 0.05). No CTRI-registered trial has tested the practice in India-resident working adults. Certainty of evidence is low. For an Indian workplace programme, treat writing as a short, private, countable rep — not a trauma exercise and not a stand-alone fix.
RP24
The Sky, Not the Weather
Observer-stance techniques ask a person to watch a thought rather than become it. Across four extractable controlled trials in non-clinical adults (N = 313), brief decentering or defusion practices reduced immediate distress by Hedges g = 0.72 (95% CI 0.32 to 1.12; I² = 62%; 95% prediction interval −0.48 to 1.91) versus inactive, sham, or waitlist comparators. That is a medium effect. It arrives in minutes, sometimes in seconds. A prior component analysis of laboratory defusion studies reported g = 0.74 against inactive conditions. A later synthesis of self-distanced versus self-immersed reflection reported a smaller g = 0.19, which is the fair comparison when the control group is also analysing the same event from inside the experience. Speed is the cleanest finding. In parametric vocal-repetition work, emotional discomfort bottomed out after three to ten seconds of rapid repetition; believability of the thought took twenty to thirty seconds. Most laboratory protocols last five to fifteen minutes. A one-hour computer session plus daily use buffered vulnerable adults against rumination and negative affect over ten subsequent days. Durability is thinner. Distanced analysis of a sad memory still buffered depressed affect at one day and seven days, after a matched distraction task had faded. A three-session vocalising study (n = 22) held gains at one month. At three and six months, an observer-stance training and a relaxation training both beat no instruction on depressive symptoms in vulnerable adults; the two active practices did not separate. GRADE certainty is low for the immediate distress pool and very low for durability and for generalisation to Indian working adults. Zero CTRI-registered and zero published Indian randomised trials of these brief techniques were found in non-clinical adults. The number is usable as a workplace starting estimate. It is not a local guarantee. Daily countable reps, not a one-off course, are the honest translation.
RP25
Use Your Strengths, Pooled
Schutte and Malouff’s 2019 meta-analysis found modest gains from signature-strengths programmes. Pooled random-effects estimates remain small: wellbeing g = 0.31 at work and job performance g = 0.28 immediately after the programme. This review updates that picture for working adults. We asked what identify-and-use-strengths programmes produce for wellbeing, engagement and performance, and what an Indian employer can reasonably buy. Searches of PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF, plus the bibliographies of Schutte and Malouff (2019), Ghielen, van Woerkom and Meyers (2018) and Vîrgă and colleagues (2023), identified more than twenty adult controlled trials. The organisational backbone is Vîrgă, Rusu, Pap, Maricuțoiu and Tisu’s pre-registered 2023 meta-analysis of 21 workplace controlled trials. Workplace wellbeing rose by Cohen’s d = 0.31 (95% CI 0.21 to 0.41) at post-test and d = 0.27 at follow-up. General wellbeing rose by d = 0.24 at post-test. Performance rose by d = 0.28 (95% CI 0.11 to 0.45) at post-test; the follow-up interval crossed zero. A 2025 workplace meta-analysis of character-strengths programmes found engagement d = 0.37 and performance d = 0.27. Identification alone is a weak lever. Using a top strength in a new way, or turning that use into a habit, is the active ingredient. Longer, coached programmes beat one-week self-guided emails, especially for follow-up. Risk of bias is typical of positive-psychology field trials: wait-lists, self-report and unblinded participants. Certainty is moderate for wellbeing and low-to-moderate for performance. Indian employed-adult evidence is still thin: one pilot among rural community health workers in Madhya Pradesh (d = 0.55 on happiness, n = 61 completers) and one fully powered trial of 330 workers whose results were not yet public on 22 September 2026. A null on late performance is publishable. Strengths use is not a miracle. It is a small, real, countable practice.
RP26
Best Possible Self, Best Available Evidence
Carrillo and colleagues pooled the best-possible-self exercise in 2019. Across updated randomised and controlled trials, best-possible-self practice produces a small-to-moderate effect on wellbeing (Carrillo d+ = 0.33), optimism (d+ = 0.33; Heekerens g = 0.21) and positive affect (d+ = 0.51; Heekerens g = 0.28); number of sittings is not a confirmed multiplier of that effect. This review asked a narrower question than Carrillo asked. Does one sitting change the effect relative to three or four sittings? Adults who wrote or imagined a best possible future self were compared with any control. Primary outcomes were optimism, positive affect and wellbeing. Searches covered PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints from 2001 to 22 September 2026. Carrillo included 29 studies in 26 articles (N = 2,909). Heekerens and Eid later pooled 34 randomised trials (N = 3,675). An update to 2026 added further adult trials that isolate the exercise, including one-sitting laboratory and online tests, four-week writing packages, an app trial and a sittings-randomised anxiety study. Random-effects synthesis of the prior pools, plus sitting-coded narrative of the update, was planned a priori. One sitting reliably lifts same-day positive affect and positive future expectancies. Those mood effects fade within about a week. Three-to-four sittings is Laura King’s original wellbeing protocol. It is associated with wellbeing at weeks, not with a proven larger g than one sitting on the same outcome at the same hour. Writing and imagery are equivalent in the one sitting that tested them head to head. Follow-up effects are small or absent. Certainty is moderate for immediate affect and expectancies, low for trait optimism and later wellbeing, and low for the sittings contrast itself. No adult Indian trial was found. For a workplace programme the honest product is one 15-to-20-minute rep for a same-day lift, and three further reps across a week if the aim is a wellbeing habit rather than a larger first-hour number.
RP27
Purpose at Work Pays
Can meaning at work be trained, and does the training move engagement and turnover intention? An exploratory random-effects pool of five meaning-specific controlled samples found a small effect on work engagement (Hedges g = 0.26, 95% CI 0.13 to 0.40; I² = 0%; 95% prediction interval 0.07 to 0.46). A pooled effect on turnover intention is not yet possible: no eligible trial measured that outcome with a standard scale. We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF (2005–2026) for randomised and controlled trials of job crafting for meaning, purpose reflection, beneficiary contact and values-at-work exercises among employed adults. Ten controlled studies met the meaning-focused eligibility bar. Brief writing and reflection practices raised measured meaning by about g = 0.26 (four samples). The largest field trial — a purpose-discovery workshop randomised among 2,976 white-collar staff in 14 countries, including India — raised a meaning index by 0.11 standard deviations and raised the annual exit rate by 2.8 percentage points (21% relative to a 13.2% control rate). Stayers improved their performance. Certainty is low. Trials are small, unblinded and short, except the one large administrative-data experiment. Correlational work still shows a large link between meaning and engagement (r ≈ 0.70) and a moderate inverse link with withdrawal intentions (r ≈ −0.49); those associations are not intervention effects and are kept separate. For an Indian workplace programme the practical reading is narrow: meaning can be practised as a daily rep; engagement may rise a little; purpose work can also help the misaligned leave. That is a selection effect, not a failure of the practice.
RP28
Hope Is a Plan
Hope, in Snyder’s model, is not a mood. It is a plan with fuel. The last comprehensive pool, Weis and Speridakos (2011), found a small effect on hope of d = 0.22 across 27 studies and 2,154 participants. This review asked what fifteen further years of trials add, and whether pathways training outperforms agency training in adults. We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints from 2000 to September 2026 for randomised and controlled trials of Snyder hope-theory programmes (goals, pathways, agency) in adults. Primary outcomes were hope, well-being and goal attainment. After screening, we identified a thin but real evidence base: the 2011 pool remains the only broad quantitative synthesis; post-2011 Snyder-protocol adult trials are few, heterogeneous, often medical, and rarely report extractable Snyder-scale means that can be pooled without guesswork. Chan, Wong, Leung and Lee (2022) added a hybrid four-session randomised trial in 72 adults with advanced kidney disease: between-group hope was not significant. Feldman and Dreher (2012) showed that a single 90-minute goal-pursuit session can raise state hope and later goal progress in students, but the hope lift was not held at one month. Cheavens and colleagues (2006) found that an eight-session community programme moved agency more reliably than pathways. Berg and colleagues (2025), in the largest digital hope-coaching randomised trial we could verify (n = 155 young adult cancer survivors), found no overall effect on hope or quality of life. No completed head-to-head trial pits pathways training against agency training. No completed Snyder hope-theory randomised trial in Indian working adults was found on CTRI or in the published record. Certainty of evidence for a workplace hope programme in India is low to very low on GRADE. The honest sentence is this: hope can be practised, the average historical effect is small, brief sessions can move a goal this week, and India still does not have its own trial. EmotionApp should treat hope as countable reps — name a goal, list two routes, take one step — not as a feeling to be installed.
RP29
Psychological Capital, Trained
Psychological-capital (PsyCap) programmes train four state-like resources — hope, efficacy, resilience and optimism — in employed adults. A pooled estimate for business outcomes is not yet possible: fewer than five eligible controlled trials report extractable, comparable effects on the same outcome, so this paper is a systematic review with narrative synthesis rather than a new meta-analytic pool. Prior reviews found small-to-medium effects on the PsyCap construct itself (Lupșa and colleagues, 2020: d = 0.34 across 41 trials, N = 3,911; Loghman and colleagues, 2025: post-intervention d = 0.46 across 40 studies, N = 4,207). Those reviews pooled wellbeing and the construct. They did not settle performance, absenteeism or turnover. This review asked a narrower question for working adults: do PsyCap programmes move business outcomes, not only the construct and wellbeing? Searches covered 2006–2026 across PubMed, PsycINFO-style records, Scopus-indexed sources, Web of Science-indexed sources, Google Scholar, CTRI, ClinicalTrials.gov and OSF, plus citation chasing of the two prior reviews. Eligible designs were randomised or controlled trials in employed adults. Primary outcomes were performance (self-rated separately from other-rated or objective), absenteeism, turnover or turnover intention, and PsyCap. Findings are uneven. PsyCap itself moves. That claim is supported by two independent meta-analyses and by later trials in China, the United States, Spain, Bulgaria and India. Self-rated performance rose in one Chinese controlled reading-materials programme (Zhang and colleagues, 2014; ANCOVA partial η² = 0.20) and in uncontrolled pre–post manager samples. Other-rated and objective performance evidence is thin and mixed. Turnover-intention evidence rests on two Chinese trials: one fragile (Da and colleagues, 2020) and one large but single-site and unblinded (Sun, Gong and Yuan, 2026). Actual turnover was not found as an outcome. Absenteeism was not found as an outcome in any eligible PsyCap-intervention trial. One Indian PCI trial exists; it measured stress and job insecurity, not performance, absenteeism or turnover. Certainty is moderate for the construct, low for self-rated performance and turnover intention, and very low or empty for other-rated performance, actual turnover and absenteeism. Indian workplaces should treat PsyCap practice as a trainable personal resource with a plausible, still unproven, path to business metrics. The India gap is explicit: no verified Indian randomised trial of a PsyCap programme has yet reported performance, absenteeism or turnover.
RP30
66 Days Is a Median, Not a Rule
Background. The 21-day habit claim is folklore. Lally and colleagues (2010) modelled daily automaticity and reported a median of 66 days to 95 per cent of a personal plateau, with a range of 18 to 254 days. This review asks what the distribution looks like when every later automaticity-over-time study is set beside that paper, and which factors shorten the wait. Findings. A pooled estimate of days to automaticity is not yet possible. Three studies that fitted individual automaticity curves put the median at 59 to 66 days among people whose data actually reached a plateau, with individual estimates from 4 to 335 days. Only about one in four participants in the largest randomised diary study crossed the authors’ formation threshold. Morning practice reached a modelled plateau faster than evening practice (106 versus 154 days). Context-stable cues help, but repeating the plan on the day it was supposed to happen helps more. Methods. PRISMA-2020 systematic review of adult studies, 2010–2026, that measured automaticity over time with a self-report habit or automaticity scale. Searches covered PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF. Primary estimand: days to 95 per cent of an individual automaticity asymptote. Pre-specified rule: fewer than five eligible trials means narrative synthesis, not a random-effects pool. Interpretation. Sixty-six is a median of a selected subsample, not a finish line. Working adults in India should plan for two to five months of context-stable daily reps, treat a missed day as a dip rather than a reset, and stop selling 21-day challenges as automaticity. Registration. This review was conducted as an update and re-analysis of the published Singh et al. (2024) evidence map; no new PROSPERO record was filed. Primary sources only.
RP31
Two-Minute Starts
Working adults do not fail for want of goals. They fail at the first step. Across 642 independent tests, if-then planning improved outcomes by d = 0.36 (95% CI 0.33 to 0.40); the 2006 initiation slice remains d = 0.61, but no trial yet isolates whether shrinking the first step adds to that plan. Gollwitzer and Sheeran’s 2006 synthesis of 94 tests reported a medium-to-large effect on goal attainment (d = 0.65). Later, larger, more field-heavy reviews revised that figure downward. Behaviour-change effects now sit nearer d = 0.27. Workplace samples sit at d = 0.26. This paper asked a narrower question. Does making the first move tiny add anything to an if-then plan? We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints from 2006 to 2026. We looked for randomised and controlled trials in adults that compared implementation intentions, tiny-step techniques, or both, against goal intention or control. Primary outcomes were initiation rate and goal attainment. A planned random-effects meta-analysis of the additive contrast could not be performed. We found no factorial trial that held the if-clause constant and manipulated only the size of the then-clause. Tiny-step trials exist, but they are few, measure the wrong outcomes, or have not been published as results papers. A pooled increment is not yet possible. What can be said is more modest and more useful. If-then plans still help people start. Specificity of the then-clause matters. Rehearsal and boosters matter. Digital delivery is weaker than face-to-face coaching. No Indian trial of either technique on task initiation in working adults has been registered or published. For an Indian workplace programme the honest design is: write one if-then sentence, shrink the first rep to two minutes, and do not claim an extra d that the literature does not contain.
RP32
Warm Wishes, Measured
Zeng’s 2015 review asked whether loving-kindness practice lifts positive emotion. This update asks a sharper question for people who work: after another decade of trials, what happens to connection, compassion and stress? In working adults the best available random-effects estimates are Hedges g = 0.38 for interpersonal connection (12 employee samples) and g = 0.54 for stress (10 employee samples). Both figures come from Wang and colleagues’ workplace synthesis, which closed in March 2022 and remains the most defensible pooled number for employed samples. Trials published since then, including a 2025 four-week, three-minute-a-day video programme in full-time employees, point the same way against no-practice controls and shrink against active practices such as light exercise or mindfulness. Compassion against passive controls in Petrovic and colleagues’ 2024 adult review was small and significant (g = 0.36, three trials). Dose in minutes and weeks did not reliably change the size of the effect. Live teachers were not required. Audio and short video were enough. Certainty is moderate for workplace stress and interpersonal outcomes, lower for positive affect against active controls, and very low for Indian professionals: no randomised trial of this practice was found on CTRI, PubMed or Google Scholar through September 2026. A null against another active practice is a real result, not a failed study. For an Indian workplace programme the honest offer is a short daily warm-wishes rep, counted, with a human hand-off when the number is not the point.
RP33
Co-Regulation Is Real
Does a calm other person measurably lower your arousal? In the largest nonevaluative partner-contact trial, a 10-minute handhold and hug halved the systolic rise to a speech (11 versus 22 mmHg; N = 183). That is the number this paper will not hide behind a slogan. A 1999 meta-analysis of 22 laboratory experiments (N = 1,167) reported medium-to-large attenuation of heart-rate and systolic-pressure reactivity (d = 0.61) and a large attenuation of cortisol (d = 0.83). Those figures are the historical ceiling from small student samples, not a new fitted model. An updated reading of adult trials, 1990–2026, finds the direction intact and the claim bounded. In men, ten minutes of friend support before a speech cut the mean cortisol rise from 15.1 to 3.6 nmol/L. In women, the same verbal partner support did not help; brief partner touch cut the rise from 12.0 to 4.3 nmol/L, and heart-rate rise from 30.6 to 18.8 beats per minute. A “supportive” evaluative audience raised peak cortisol versus no audience (g = +0.63) in 183 adults. Recovery is thinner: positive partner coping after a speech sped the cortisol decline in 123 couples, but fewer than five adult trials report a clean recovery slope, so a pooled recovery g is not yet possible. A 2023 registered-report meta-analysis of broader “emotional social support” found a naïve g = −0.14 that did not survive bias correction. Our narrower question — a known, nonevaluative other during or after a laboratory stressor, measured on cortisol or the heart — still has laboratory support. Certainty is low to moderate. No Indian adult randomised biomarker trial exists. For working adults the practical translation is a connection rep after load, not a watching panel during evaluation.
RP34
Peer versus Professional
Outside the clinic, working adults carry stress that never reaches a hospital door. An exploratory random-effects pool of five waitlist- or usual-care-controlled trials gave Hedges g = −0.48 (95% CI −0.64 to −0.33; I² = 0%; 95% prediction interval −0.70 to −0.27) on distress. That pool uses converted effects from mixed adult samples and is GRADE low. Head-to-head with professionally led programmes the published difference remains g = 0.10 (95% CI −0.20 to 0.39) in Pfeiffer and colleagues (2011) and d = 0.08 (95% CI −0.15 to 0.32) in Bryan and Arkowitz (2015); a 2025 Indian campus trial against supportive counselling was also null. Pfeiffer’s older usual-care pool was g = −0.59. Yeo and colleagues (2025) found digital peer support in people without a clinical condition improved mental health by SMD 0.53. No eligible trial reported cost per outcome. In Indian private markets a professional session typically costs five to fifteen times a peer or listener session. India still has no published randomised trial of structured peer support for subclinical workplace distress in working professionals. Peer programmes are a high-volume first line, not a substitute when risk is high. Train the peer. Count the reps. Hand off when the score is no longer subclinical.
RP35
Lonely at Work
Workplace loneliness is common, costly, and strangely under-tested. Four controlled occupational trials measured loneliness after a workplace programme; a meta-analysis was not performed. This PRISMA-2020 systematic review asked what has been tested against workplace loneliness in employed adults from 2010 to 2026, and what worked. Searches covered PubMed, PsycINFO-equivalent sources, Scopus and Web of Science via public interfaces, Google Scholar, CTRI, ClinicalTrials.gov, UMIN-CTR and OSF. Eligible designs were randomised or otherwise controlled trials of any programme targeting loneliness or connection at work, with loneliness, belonging or social connection as outcomes. Four completed trials met the loneliness-outcome bar. All four were United States military samples. Two short leader-support coaching programmes (about 60–90 minutes) produced small reductions in employee loneliness against usual practice. One eight-hour social-resilience package produced a small decrease relative to an active control. One condensed unit-connection package of about three and a half hours did not change loneliness. Certainty of evidence is low to very low: few trials, high indirectness to civilian and Indian knowledge work, unblinded self-report, and attrition. No completed, peer-reviewed controlled trial of a workplace loneliness programme was found in India. The India-registered garment-worker pairing study remains unpublished. Until more trials exist, the honest claim is narrow. Brief leader coaching that teaches supportive behaviours has a replicated, small loneliness signal. A pooled effect size is not yet possible. Indian programmes should treat that signal as a starting hypothesis, not as product-market fit.
RP36
The Guidance Premium
Working adults do not fail digital wellbeing programmes because the techniques are weak. They fail them because the programme sits unopened, like a gym membership bought in January. This systematic review asked how much a human coach adds when the same digital programme is offered with and without guidance, in adults who are not a clinical sample. In the largest matched occupational trial, adding reminders and on-demand feedback raised full-module completion from 53 per cent to 72 per cent and did not change the stress score. Three further contemporaneous trials of the same idea in students and healthy adults found the same pattern on scores. A formal random-effects meta-analysis was not performed: the pre-specified minimum of five strictly matched trials was not met. Searches of PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints covered 2010 to September 2026. Eligible designs were randomised or controlled trials of non-clinical adults. The intervention was a guided digital programme. The comparator was an unguided version of the same or a closely similar programme. Primary outcomes were wellbeing, stress and adherence. Prior clinical reviews had put the guidance premium at about 0.17 to 0.27 of a standard deviation on symptoms. Those samples were more unwell. In occupational-stress and student samples the score gap shrinks toward zero. The remaining premium is behavioural: people finish more modules when a light-touch coach sends reminders and answers on demand. Coach qualification did not matter in earlier work. Minutes per week did, up to a point. Less than 15 minutes a week was enough. Certainty is low for scores and moderate for adherence. No Indian head-to-head trial was found. The next useful study is a WhatsApp-first, multilingual randomised comparison in Indian working adults, with completion as a co-primary outcome.
RP37
No Symptoms to Reduce
Background. Prior pooled estimates show that conversational agents reduce depressive symptoms in mixed samples (g = 0.31) but not in non-clinical adults (g = 0.07). Whether the same agents raise wellbeing, skill use or behaviour in people who are not ill has not been settled. Methods. We conducted a PRISMA-2020 systematic review and random-effects meta-analysis of randomised and controlled trials published from 2017 to 22 September 2026. Eligible trials enrolled non-clinical adults, delivered a rule-based or generative conversational agent for wellbeing, and reported wellbeing, positive affect, skill use, behaviour change or engagement. Searches covered PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF preprints. Effects were Hedges g under restricted maximum likelihood (REML). Certainty was graded with GRADE. Risk of bias was assessed with RoB 2. Results. On wellbeing scales alone (three trials, N = 1,658) the pooled effect was g = 0.07 (95% CI −0.06 to 0.19). Adding the one trial that reported positive affect as the extractable endpoint (four trials, N = 2,144) raised the estimate to g = 0.17 (95% CI −0.01 to 0.35; I² = 68%; tau² = 0.022; 95% prediction interval −0.40 to 0.73). Both intervals include or approach zero. Versus wait-list or care-as-usual the effect was g = 0.28 (95% CI 0.14 to 0.42; I² = 0%). Versus active web-based wellness resources the effect was g = 0.02 (95% CI −0.10 to 0.14). Self-care behaviour in one trial was d = 0.36 at ten days and was not sustained at one month. Engagement was low: in the largest trial users chatted on a mean of 4.3 of 30 days. No completed adult non-clinical Indian trial was found. Most trials did not pre-specify or systematically report adverse events. Conclusions. Measured on wellbeing rather than symptom scales, AI conversational agents deliver a small and uncertain benefit in non-clinical adults. The benefit is visible against wait-list and not visible against good digital resources. Skill and behaviour outcomes remain sparsely measured. An Indian workplace evidence base does not yet exist. Registration. This review was not prospectively registered.
RP38
Can You Bond With a Bot?
Working adults in India are already talking to software about stress, sleep and self-talk. The open question is whether that conversation forms a working alliance — agreement on goals and tasks, plus a felt bond — and whether the alliance predicts change the way it does with a human coach. Across large observational cohorts of adult users of conversational agents and unguided apps, mean scores on a 12-item working-alliance short form sat between 3.36 and 3.75 on a 1–5 scale within five days of first use, close to published human outpatient means of about 3.8. Bond scores often arrived first and highest; task agreement lagged. Four independent samples linked later alliance to better outcomes, with small coefficients (standardised betas about −0.13 to −0.27 for distress; one partial r of −0.18). The human-coach benchmark remains r = 0.28. Alliance measured in week one did not predict change; alliance measured in weeks three or four did. No CTRI-registered trial has yet measured software-alliance in Indian working adults. Industry-affiliated samples dominate the largest means. Certainty is low for formation and speed, and very low for the outcome link. For an Indian workplace programme the practical reading is this: a bond with an agent can appear in days, not months; it is not a substitute for a human hand-off; and task-and-goal agreement, not warmth alone, is the piece most worth designing for.
RP39
Mother Tongue, Bigger Effect?
Do wellbeing programmes work better when they arrive in a participant’s first language? Between-study comparisons put the first-language advantage at about g = 0.28, but fewer than five eligible trials compared the same programme in a first versus a second language, so a new pooled estimate is not yet possible; the number of Indian head-to-head trials is zero. This PRISMA-2020 systematic review asked whether psychological and wellbeing programmes delivered in a multilingual adult’s first language outperform the same kind of programme delivered in a second language. Searches of PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF (2000–2026) found no randomised trial that assigned the same programme to first-language versus second-language delivery in multilingual adults. The best available numbers therefore come from language-matching subgroups inside cultural-adaptation meta-analyses. Griner and Smith (2006), across 76 studies and 25,225 participants, reported that programmes in a client’s native language (if other than English) were twice as effective as programmes with no language match described (d = 0.49 versus d = 0.21). Soto and colleagues (2018) updated that picture across 99 studies: overall d = 0.50, falling to 0.35 after publication-bias correction; preferred-language delivery and translated measures both enlarged effects. Hall and colleagues (2016) found culturally adapted programmes beat unadapted versions of the same programme (g = 0.52) but did not find language itself a reliable moderator. Interpreter-mediated comparisons are mixed. Indian programmes of real quality — including the Healthy Activity Programme in Goa — have been delivered in local languages, yet none randomised language of delivery. Certainty that first-language delivery causes better outcomes is low. The practical implication for an Indian workplace programme is simpler than the causal claim: English is the mother tongue of 0.02 per cent of Indians and the office language of almost every professional this platform serves. Matching the language of the rep remains the low-regret design choice while the clean experiment is missing.
RP40
Adapted Beats Imported
Hendriks and colleagues reported that positive-psychology interventions (PPIs) run outside the West produced a moderate effect on subjective wellbeing (Hedges g = 0.48). No trial directly compared an adapted PPI with an unadapted version of the same programme in a South Asian adult sample, so a pooled estimate of the adaptation increment is not possible. This PRISMA-2020 systematic review asked how much cultural adaptation adds to PPI effects on wellbeing and depressive symptoms among South Asian adults (India, Pakistan, Bangladesh, Sri Lanka, Nepal, and diaspora), 2005–2026. Databases and registries searched were PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF. Eligible designs were randomised and controlled trials of gratitude, strengths, savouring or multi-component PPI practices versus control or versus an unadapted version. Zero trials isolated the adaptation increment. Nine adult PPI-versus-control studies entered a narrative synthesis (India seven, Pakistan two, Bangladesh zero, Sri Lanka zero core PPI, Nepal zero, diaspora zero). Locally built strengths programmes produced small-to-moderate wellbeing gains against waitlist or usual supervision: the only fully reported between-arm standardised difference was Bondre 2025, Cohen’s d = 0.55 (95% CI 0.04 to 1.06) for happiness in rural Indian women health workers. Imported gratitude scripts often failed to raise state gratitude in Indian adults. A 2025 cross-cultural meta-analysis found no significant gratitude effect on wellbeing in India. Study quality was low. Certainty for the adaptation-increment question is very low because the comparison does not yet exist. The practical implication is not “PPIs do not work in South Asia.” It is that a worksheet trained on California undergraduates is a poorly specified model for a Mumbai working week. Domain shift is not a slogan. It is the finding.
RP41
The India Evidence Gap
Indian workplaces buy wellbeing programmes the way a start-up buys ads: quickly, hopefully, and often without a randomised receipt. Seven eligible Indian randomised trials of non-clinical emotional-wellbeing programmes were found; four could be pooled, giving a random-effects Hedges g of 0.89 (95% CI 0.62 to 1.16) on stress or wellbeing. Heterogeneity was modest (I² = 20%; τ² = 0.016). The 95% prediction interval was 0.52 to 1.26. Certainty is low. Every trial was open-label. Most used a wait-list. Yoga and breath practices supplied five of the seven trials. One rural trial tested character-strengths coaching (g = 0.54). Zero eligible worker trials tested a mindfulness-only programme, a gratitude programme, or a WhatsApp-first or AI-coach programme. Only one trial delivered the programme in Hindi. Resilience as a primary outcome is almost empty. Prior maps already said this. Pandya and colleagues (2022) found thirty workplace records and almost no trials. A 2025 review of thirty-eight workplace mental-health papers still did not pool effects. This review asks a plainer question: how many Indian randomised trials exist, what do they pool to, and where are the holes? Adults in India, non-clinical samples, psychological or behavioural wellbeing programmes, any control, outcomes of wellbeing, stress or resilience, years 2000–2026. Searches covered PubMed, Google Scholar, Scopus (via published reviews), Web of Science equivalents, CTRI, ClinicalTrials.gov and OSF. Random-effects REML, Hedges g, RoB 2, GRADE. The number is not zero. It is seven. That is a pilot batch, not a production run. An Indian workplace programme can cite a large average effect and still be flying on instruments built for a different language, a different channel, and a different follow-up window.
RP42
Messaging as the Medium
Messaging channels have been pooled for health behaviour. They have not been pooled for emotional wellbeing. This PRISMA-2020 systematic review and random-effects meta-analysis asked what SMS-, WhatsApp- and similar-messaging programmes deliver for adult wellbeing and stress, and what retention they hold. Ten randomised or controlled trials (about 5,225 adults, 2012–2024) met criteria and yielded an extractable effect size. The pooled Hedges g was 0.18 (95% CI 0.08 to 0.27). Heterogeneity was substantial (I² = 64%; τ² = 0.012). The 95% prediction interval ran from −0.10 to 0.46 and included no effect. A wellbeing-only subset was smaller still (g = 0.10). Stress and distress effects were larger (g = 0.26) but driven in part by two small clinical SMS pilots. One-way scheduled messages and two-way chat or group formats did not differ cleanly. High-income and low- and middle-income trials produced similar averages on sparse data. Retention is the clearer number. Median completion at the primary endpoint was about 80%. One-way daily or near-daily broadcasts held 85–93% of starters. Two-way conversational agents held as few as 42%. No completed Indian adult trial delivered a wellbeing or stress programme primarily by SMS or WhatsApp. Certainty is low for wellbeing, low for stress and distress, and moderate for one-way retention. A null workplace LINE programme of 1,021 employees sits inside the pool and is part of the result, not an inconvenience. Messaging is a distribution system, not magic. It keeps people in the room. What they practise in the room still has to be specified, counted and tested in India.
RP43
Single Session, Lasting Change?
Schleider and Weisz showed that one sitting can help young people. This review asks the adult question: how much does one sitting buy, and for how long? Across adult randomised and controlled trials of single-session or 15-minute-or-shorter practices published from 2010 to 2026, a single sitting produced a small follow-up effect of g = 0.15 (95% CI 0.05 to 0.22) on distress or wellbeing at four to eight weeks; most 10-minute digital programmes did not outlast the month. A conservative all-comers pool that kept the null digital programmes in the average found g = 0.08 (95% CI 0.00 to 0.16). Immediate state hope, gratitude and agency moved more, with a short-horizon pooled estimate near g = 0.48 (95% CI 0.18 to 0.79), but those lifts were often gone by four weeks. The largest adult trial assigned 7,505 people with elevated low mood to one of twelve 10-minute digital programmes or a no-content control; only two programmes still beat control on depression at four weeks (d = 0.14 and 0.15). Longer scripts were not better. An 8-minute loneliness practice outperformed a 23-minute version of the same material at eight weeks (d = 0.22). Heterogeneity at follow-up was low once the megastudy dominated the weights. Small live-session trials against waitlists produced larger effects and higher risk of bias. Egger’s test in mixed pools suggested small-study inflation. Certainty is moderate for same-day state change and low-to-moderate for follow-up. No eligible adult trial was conducted in India. One sitting is a first countable rep, not a course. For Indian workplaces the number is useful precisely because it is small and honest: a 10-minute WhatsApp-first practice can open a door. It does not replace a human hand-off.
RP44
Who Should Change?
This updated systematic review and second-order synthesis asks which lever moves burnout more in employed adults — individual skills or organisational change — and whether the two add. Across physician controlled comparisons, organisation-directed programmes reduced burnout by g = 0.45 (95% CI 0.28 to 0.62) while individual-directed programmes reduced it by g = 0.18 (95% CI 0.03 to 0.32). That head-to-head, from 20 comparisons in 1,550 physicians, remains the cleanest published split. Across occupations, organisational and combined programmes reduced exhaustion by g = 0.30 (95% CI 0.18 to 0.42) in 13 studies; the two combined trials in that pool reached g = 0.54. Individual programmes, which dominate the trial base, reliably move exhaustion by about g = 0.17 to 0.25 and often leave cynicism and efficacy untouched. A 2025 cluster-randomised trial in 370 intensive-care units across 60 countries found burnout prevalence of 52.2% after a four-week positive-communication programme versus 63.3% in usual care (adjusted odds ratio 0.56). Post-2020 work added many individual digital and coaching trials and almost no Indian organisational trials. Certainty is moderate for a small individual effect on exhaustion, low for a larger organisational effect, and very low for combined programmes. The practical reading for an Indian workplace is not a slogan. Teach people countable skills. Change the work. Do both. A mindfulness rep on a 70-hour week is a good brake and a bad steering wheel.
RP45
Train the Manager, Move the Team
Gayed and colleagues asked, in 2018, whether training workplace managers changed anything that employees could feel. Manager knowledge moved (SMD 0.73). Employee psychological symptoms did not: SMD −0.08 (95 per cent CI −0.21 to 0.06; p = 0.28; I² = 0 per cent). A new pooled Hedges g for employee outcomes is not yet possible: six independent controlled programmes from 2010 to 2026 measured different employee constructs, two were pilots, and comparable wellbeing or distress estimates number fewer than five. The last official pooled figure that includes later trials is the World Health Organization 2022 update of Gayed: standardised mean difference −0.15 (95 per cent CI −0.29 to −0.02) across six randomised trials, high certainty in the WHO text, most of them older than our year window. Our 2010–2026 living update finds one Australian cluster trial in which a four-hour face-to-face programme reduced work-related sick leave by 6.45 hours per employee over six months; one Canadian wait-list trial in which employees of trained managers used more workplace resources; and a consistent null or trivial signal on employee wellbeing scales in British and Australian trials. Employee-rated manager support leans positive when it is measured. Certainty of evidence for every employee-level domain is low by GRADE. No Indian randomised trial of workplace manager mental-health training with employee outcomes was found on CTRI or in the international databases. For an Indian workplace programme the honest reading is this: train the manager as a high-leverage practice, measure the team, and do not advertise a wellbeing effect the trials have not yet earned. The manager is the fine-tuning layer. The team is the held-out test set. Until Indian trials exist, the test set is still overseas.
RP47
EI Can Be Trained
Emotional intelligence can be practised. That sentence used to be a slogan. It is now a result. Across workplace programmes that train emotional competencies, the most recent comprehensive controlled-trial pool finds a moderate effect of SMD = 0.46 (95% CI 0.30 to 0.63) in 27 trials; the 2018–2026 adult update does not overturn that number. Self-report scores for managing and understanding emotion still move by about g = 0.43 (0.26 to 0.60) in the best-reported new online trials. Performance tests of ability move less: g = 0.14 (−0.04 to 0.33), a confidence interval that includes zero. Perceiving emotion in others is the stubborn branch. Job-performance endpoints after training remain too few to pool. This paper updates Hodzic and colleagues (2018) and Mattingly and Kraiger (2019) against Mehler and colleagues (2024), the current workplace anchor. We searched PubMed, PsycINFO, Scopus, Web of Science, Google Scholar, CTRI, ClinicalTrials.gov and OSF from 2018 to 22 September 2026. Adults. Emotional-intelligence training. Any control. Outcomes: ability or trait EI; job performance; wellbeing. Random-effects models, Hedges g, RoB 2, GRADE. Twenty-two controlled trials entered the narrative synthesis. Eight had extractable between-group statistics on an EI score. A restricted random-effects model of the two largest placebo- or waitlist-controlled online trials confirmed a moderate self-report effect and a small, uncertain ability-test effect. Follow-up at two to twelve months often held for practised skills and faded for global trait scores. No adequate Indian corporate randomised trial with job endpoints was found. Certainty is moderate for self-report managing and understanding, low for ability tests, and very low for job performance. For Indian professionals the practical reading is simple. Treat EI as a countable weekly practice, not a personality transplant. Train managing and understanding first. Deliver in the language the team actually thinks in. Measure a real conversation, not only a questionnaire.
RP50
Coaching versus Counselling for Subclinical Distress
This PRISMA-2020 systematic review asked whether coaching or counselling helps more for non-clinical adults with subclinical distress, and what that means for the coaching–clinical line. Across randomised trials of coaching versus inactive control, the best modern RCT-only syntheses report Hedges g = 0.43 (95% CI 0.35 to 0.51) and g = 0.59 (0.45 to 0.74); Theeboom’s 2014 overall figure of g = 0.66 remains the most-cited hook but shrinks to g = 0.39 when only controlled designs are kept. No randomised trial compared coaching with counselling head-to-head in this population between 2000 and 2026. A pooled g difference cannot be computed. Indirect benchmarks for brief psychological programmes in subclinical depression sit near g = 0.35 (95% CI 0.23 to 0.47). The two inactive-control estimates overlap. Session count, digital versus face-to-face format, and programme length did not reliably moderate coaching effects. Publication bias is present; trim-and-fill pulls the RCT-only coaching estimate downward. Certainty is moderate for coaching versus waitlist on wellbeing and goal attainment, and very low for any claim that one practice beats the other. India contributes one pilot of character-strengths coaching in rural health workers (d = 0.55) and no trial in urban professionals. Coaching remains a developmental practice. It is not a substitute for clinical care. The regulatory line is purpose, population and non-diagnosis — not a superiority contest between two effect sizes.
RP51
Presenteeism Priced
Mental ill-health does not only empty chairs. It fills them with people who are physically present and economically half-there. Across verified cost-of-illness and productivity-loss studies published between 2000 and 2026, presenteeism costs attributable to mental ill-health ranged from about US$700 to US$8,000 per affected employee per year in 2024 prices (roughly ₹60,000 to ₹6.7 lakh), and from about US$90 to US$1,350 per employee when the same bill is spread across the whole workforce (roughly ₹7,500 to ₹1.1 lakh). That is a converted range, not a pooled mean. A conventional random-effects meta-analysis of Hedges g was not performed. The eligible literature is observational cost-of-illness work, not randomised trials, and the estimates are not combinable. The same person, priced with two questionnaires, two wage bases, or two costing methods, can generate a threefold swing. Presenteeism commonly costs five to ten times as much as sick leave among people with depression; published ratios run from 1.4 in Japan to 14.2 in the United States. Seventeen sources met eligibility for narrative synthesis. Zero eligible randomised trials reported a per-employee presenteeism cost as a primary outcome. Instrument type, denominator (affected staff versus the whole payroll), human-capital versus friction-cost valuation, and country wage levels explained most of the variance. India still lacks a peer-reviewed, professional-wage, per-employee presenteeism price. The widely cited Indian total — ₹51,000 crore of presenteeism inside a ₹1.1 lakh crore mental-health bill — comes from a 2022 consultancy model that imputed presenteeism as three to four times absence, rather than measuring on-the-job performance. For working adults in India the practical implication is not a single invoice. It is that programmes which only count sick days will miss the larger share of the bill. Certainty that presenteeism exceeds absenteeism is moderate. Certainty around any one global rupee figure is very low.
RP52
The Mindfulness App Plateau
Mindfulness apps sell an eight-week story. This paper asks what is left of that story at three and six months. A PRISMA-2020 systematic review of randomised and controlled trials (2012–2026) found too few comparable wait-list trials sharing the same outcome at the same horizon to justify a pooled Hedges g; the durability number is therefore a decay ratio, not a single pooled effect. In the three wait-list trials that measured both post-programme stress or distress and a three-to-four-month follow-up, decay ratios were 0.62, 0.84 and 0.94 (unweighted mean 0.80; sample-size-weighted mean 0.76). About three-quarters to four-fifths of the immediate effect was still visible. The largest employee trial (n = 1,458) moved from d = 0.85 at eight weeks to d = 0.71 at four months from randomisation. A four-week wellbeing app in school employees retained 62 per cent of its distress effect three months after the programme ended (d = 0.33). Two app-only trials in public-sector and general-population adults found no change in perceived stress at three or six months. Six-month app-only evidence is too thin and too mixed to pool. Continued use during the programme tracks larger gains; use collapses for most people once the structured weeks end. Certainty of evidence is low for three-to-four-month stress and very low for six-month stress and for wellbeing. No eligible mindfulness-app randomised trial with a three-to-six-month follow-up has been run in Indian working adults. For an Indian workplace programme the practical reading is blunt: an eight-week download is a seed, not a harvest. Daily reps after week eight are the root system.