A PERMA-framework analysis of what DAU design students say brings them joy — distilled from seven raw survey categories into five well-being pillars, cross-checked demographically, and corrected end-to-end from an earlier flawed pass.
This is a corrected, deepened re-analysis of a 364-respondent survey of DAU IUxD students. A silent bug in the original notebook dropped 12.6% of responses from every earlier chart; once fixed, the headline story mostly survives — but a demographic pattern the earlier draft leaned on does not, and this version adds the significance testing the original never ran.
Accomplishment is the single largest PERMA pillar at 33.5% (122 of 364), with Relationships close behind at 28.3%. Three themes, not one, carry this dataset.
Every mapping in the earlier notebook keyed “Enjoy the Moment” with a capital M. The real answer in the data was lowercase — so pandas quietly dropped it, no error, no warning.
Both chi-square tests come back statistically non-significant with negligible effect sizes. The bar-chart gap the earlier draft told a story about is noise at this sample size.
98.6% of responses landed across four days, mostly in-class. The sample is heavily skewed (74.7% male, 81.3% undergraduate) and doesn’t generalize beyond it.
This analysis explores the diverse sources of happiness among survey respondents. By examining responses across emotion categories, intensity levels, PERMA pillars, and demographic dimensions, it identifies patterns in what respondents say actually makes them happy — and, this time, tests whether those patterns hold up statistically before calling them findings.
| Dimension | Count | Share |
|---|---|---|
| Total survey responses analyzed | 364 | 100% |
| Male respondents | 272 | 74.7% |
| Female respondents | 89 | 24.5% |
| Preferred not to say (gender) | 3 | 0.8% |
| Undergraduate participants | 296 | 81.3% |
| Postgraduate participants | 68 | 18.7% |
| Raw happiness-source categories identified | 7 | — |
| PERMA well-being components mapped | 5 | — |
Respondents answered one forced-choice question: recall a specific event that made you happy, then classify it as one of seven preset categories — Achievement, Affection, Bonding, Enjoy the moment, Exercise, Leisure, or Nature. No free text, no validated psychometric scale — every framework below is a researcher-assigned reading laid on top of that single choice.
Respondents were DAU IUxD undergraduate and postgraduate students: 296 undergraduate (81.3%), 68 postgraduate (18.7%); 272 male (74.7%), 89 female (24.5%), 3 preferred not to say (0.8%). Timestamps show 359 of 364 responses (98.6%) landing across four consecutive days — October 14–17, 2025, with October 15 alone accounting for 140 responses, many seconds apart — a signature of in-class group administration rather than an open, self-paced survey. One response arrived two weeks later, on October 30.
Column names were standardized, missing or inconsistent values handled, text responses normalized into the seven fixed categories, and demographic fields extracted and coded. Every mapping now carries an explicit assertion check — if a future edit introduces an unmatched category, the notebook fails loudly instead of silently dropping rows.
Three parallel lenses are applied to the same seven raw categories: a 3-way Emotion Category split (Growth/Self-Development, Relational/Social, Calm/Presence/Restoration) for a headline stat, a 5-pillar PERMA classification for a citable theoretical backbone, and an Intensity mapping (High/Medium/Low) as an illustrative, researcher-assigned proxy. Personas are then synthesized from the resulting distribution patterns.
The dataset and every number on this page come from a revised notebook that fixed ten issues in the original analysis. The four that changed actual findings:
A casing typo silently dropped 46 responses. Every mapping dictionary keyed “Enjoy the Moment” (capital M); the raw data used lowercase. Pandas returns a blank for an unmatched key instead of erroring, so those rows vanished from every chart with no warning.
An undefined variable meant the notebook couldn’t run top to bottom. It only appeared to work in the original because the defining cell had been deleted after one live session, leaving the value cached in memory.
Together, the above meant every earlier statistic was computed on n=318, not the true n=364. Every chart in this analysis is now recomputed on the full, correctly-mapped sample — which is also why several numbers on this page (intensity distribution, PERMA shares, the emotion-category split) differ from the live site’s previous version.
Zero significance testing existed anywhere in the original. Demographic comparisons were stated as fact with no test behind them. Section 05 below adds chi-square testing and effect sizes — and reverses the earlier conclusion.
Six further fixes were technical hygiene — a non-portable file path, non-linear cell order, a deprecated chart argument, a duplicated cell, a misleading column name (“Confidence Score” renamed to “Intensity Score” since nothing about it was ever a confidence rating), and a missing validation check — none of which change a finding on this page, all of which make the notebook reproducible from a fresh kernel.
Descriptive statistics and frequency distributions; cross-tabulation for demographic patterns; chi-square tests of independence with Cramér’s V effect sizes (new); non-parametric bootstrap confidence intervals, 10,000 resamples (new); and three D3.js visualizations — a zoomable sunburst, a beeswarm plot, and a grouped bar chart — built to make the underlying counts inspectable rather than only summarized.
Achievement and personal development are the most frequently cited source of happiness (representing accomplishments and skill acquisition). Strong social connections, bonding, and affection form a close second. Nature, unhurried presence, and leisure together form the third, corrected theme — larger than the buggy original showed.
| Category | Responses | Share | 95% Bootstrap CI |
|---|---|---|---|
| Growth / Self-Development | 145 | 39.8% | 34.9–45.1% |
| Relational / Social | 103 | 28.3% | 23.6–33.0% |
| Calm / Presence / Restoration | 116 | 31.9% | 27.2–36.5% |
Bootstrap 95% confidence intervals (10,000 resamples). Growth’s and Calm’s intervals overlap — this sample alone can’t rule out Calm being just as common in the underlying population. Growth vs. Relational/Social does not overlap — that gap looks more solid. Report “Growth leads, Relational trails” with confidence; report “Growth clearly beats Calm” only with a caveat.
High Intensity now sits at 61.8% (225 responses) — still the majority, but well below the earlier site’s inflated 70.8%, because that figure excluded the 46 “Enjoy the moment” responses entirely. Low Intensity rises to 31.9% (116) once those responses are correctly counted as calm, low-arousal experiences. Medium Intensity (Exercise only) stays a clear minority at 6.3% (23). Read this as a proxy for arousal by category, not a self-report — see the construct-validity note in Section 03.
The seven raw categories are deliberately read through two independent lenses, not one. A simple 3-way Emotion Category split works as a headline stat; the 5-pillar PERMA model gives the analysis a citable theoretical backbone. Keeping both is a framework choice, not accidental duplication — the 3-way split is the simpler number to lead with, PERMA is the number to defend in a review.
| Happiness Source | Responses | Primary Intensity | Share |
|---|---|---|---|
| Growth / Self-Development | 145 | High | 39.8% |
| Relational / Social | 103 | High | 28.3% |
| Calm / Presence / Restoration | 116 | Low–Medium | 31.9% |
The same 364 responses, rolled up as the 5-pillar PERMA model or the simpler 3-way emotion split. Click a ring segment to zoom in; click the center to zoom back out.
Achievement→Accomplishment and Affection/Bonding→Relationships are clean fits. Nature→Meaning is the most debatable call — Nature could just as defensibly map to Positive Emotion (aesthetic pleasure) as to Meaning (purpose). And “emotion intensity” (High/Medium/Low) is a researcher-assigned proxy by category, not something respondents actually rated — treat it as illustrative, not measured. Because the instrument is a single forced-choice item and not a validated multi-item scale, none of these mappings can be empirically checked against the data itself; they are theory-driven interpretations laid on top of the categories, and are presented here as exactly that.
A rounded percentage can hide how thin or skewed a sample really is. Every dot below is one respondent, positioned by their emotion category and colored the same way as every other chart on this page — a more honest picture of sample size than a pie chart offers.
Hover a dot for that respondent’s exact answer, gender, and course.
| Gender | n | Growth-Oriented | Bonding-Oriented | Calm | Avg. Intensity Score |
|---|---|---|---|---|---|
| Male | 272 | 41.9% | 27.9% | 30.1% | 2.324 |
| Female | 89 | 33.7% | 30.3% | 36.0% | 2.247 |
| Preferred not to say | 3 | 33.3% | 0.0% | 66.7% | 1.667 |
| Level | n | Share | Avg. Intensity Score |
|---|---|---|---|
| Undergraduate | 296 | 81.3% | 2.301 |
| Postgraduate | 68 | 18.7% | 2.294 |
Both academic levels prioritize achievement-based happiness at nearly identical intensity — 2.301 vs 2.294 is not a meaningful gap. The tables above are descriptive only; whether the gender pattern is anything more than sampling noise is answered properly in Section 05.
The earlier draft described a “males lean toward growth, females toward bonding” pattern as fact, with zero statistical testing behind it — the original site had no significance testing anywhere. This version runs a chi-square test of independence with Cramér’s V effect size on both demographic splits shown in Section 04.
Hover any bar for exact counts and percentages.
Figure 5.1 — Percent of each gender falling into each emotion category (n=361, excluding 3 respondents who preferred not to say). The visual gap is real in this sample; the test below says it’s indistinguishable from chance.
Both p-values sit well above the .05 threshold, and both effect sizes (Cramér’s V < 0.1, the negligible band) are negligible. Happiness-source category does not detectably differ by gender or course level in this sample. Descriptive gaps in survey counts often vanish under a real test — and revising the story when the data doesn’t support it is a stronger research signal than the uncorrected claim would have been.
Accomplishment leads: goal achievement, personal development, and mastery are the primary drivers of reported happiness. Relationships follow closely — deep social connections, affection, and bonding create profound well-being experiences on their own. Positive Emotion, once correctly counted, is a real third pillar rather than a rounding error; Meaning and Engagement round out the model as smaller but distinct themes.
| PERMA Component | Count | Share | Associated Factors |
|---|---|---|---|
| Accomplishment | 122 | 33.5% | Achievement, personal growth |
| Relationships | 103 | 28.3% | Affection, bonding, social connection |
| Positive Emotion | 69 | 19.0% | Enjoying the moment, leisure |
| Meaning | 47 | 12.9% | Nature connection |
| Engagement | 23 | 6.3% | Exercise, flow, activity |
Accomplishment (33.5%, 122) and Relationships (28.3%, 103) are unchanged — those categories weren’t affected by the casing bug. What changes substantially is Positive Emotion, which rises from an undercounted 23 responses (6.3%) to its true 69 responses (19.0%) once “Enjoy the moment” is correctly included alongside Leisure. Positive Emotion is now the third-largest PERMA component, ahead of Meaning (12.9%) — a materially different picture from the original chart, where it tied for last place.
Core driver: Achievement (122 responses, 33.5% of the full sample) → PERMA: Accomplishment. High intensity.
Core drivers: Affection (51) + Bonding (52) → PERMA: Relationships. High intensity.
Core drivers: Nature (47), Enjoy the moment (46), Leisure (23) → PERMA: Meaning + Positive Emotion. Low–medium intensity.
A fourth, Engagement-driven persona was considered and dropped: at 23 responses (6.3%) it’s too thin a slice to characterize on its own, and it already lives inside the Achiever’s Growth umbrella.
With 39.8% of responses centered on Growth/Self-Development, 31.9% on Calm/Presence/Restoration, and 28.3% on Relational/Social, respondents show a genuinely three-way split rather than one dominant driver. Achievement leads, but not by a landslide.
61.8% of happiness sources are high-intensity (Achievement + Relationships), while 31.9% relate to calm, low-arousal experiences. Respondents seek both impactful moments and quiet contentment — and the corrected split shows more room for the latter than the original notebook did.
Relationships account for 28.3% of happiness sources, within five points of Accomplishment’s 33.5%. Strong social bonds are nearly as central as personal achievement to this population.
The earlier draft claimed males lean toward achievement and females toward bonding and calm. Tested properly, that pattern is statistically indistinguishable from chance (gender: p=.37, V=0.07; course: p=.33, V=0.08). This is a revision, not a footnote: any prior recommendation built on a gender-based split should be retired.
At 31.9% (116 responses), Calm/Presence/Restoration is barely behind Growth/Self-Development and ahead of Relational/Social. The earlier, buggy analysis undercounted it at roughly 26%, making achievement look more dominant than the corrected data supports.
This is a student-centric sample drawn from one design program at one university, skewed male (74.7%) and undergraduate (81.3%). Any program built on these findings should treat them as a starting hypothesis for this population, not a general claim about “what makes people happy.” The Engagement pillar (6.3%, tied to Exercise) is the clearest opportunity gap in the data — either genuinely underrepresented in this population’s happiness sources, or underrepresented because the seven-category instrument gave it only one route in.
All respondents are DAU students. Findings don’t generalize beyond that population.
74.7% male, 81.3% undergraduate. Subgroup comparisons are likely underpowered to detect a real difference even if one exists.
Not a validated scale like PANAS or SWLS. PERMA and intensity mappings are researcher interpretations of one choice, not independently measured constructs.
The prompt asks respondents to recall a past happy event; recency and social-desirability effects are both plausible.
Timestamp clustering suggests in-class group administration, which reduces independence between responses.
Deterministically assigned per category, not something any respondent actually rated. Illustrative only.
Accomplishment (33.5%) is the largest single PERMA component, indicating that tangible achievement and measurable progress are core to reported happiness — but at barely a third of responses, it is one pillar among several, not the whole picture.
Relationships (28.3%) sits close behind Accomplishment, revealing that while individual achievement matters profoundly, it is nearly matched by the importance of social bonds.
The 61.8% concentration on high-intensity emotions reveals a population that leans toward “peak experiences” over steady-state contentment — though at 31.9%, low-intensity, restorative experiences are far from marginal.
Calm/Presence/Restoration (31.9%) is the corrected figure, not the undercounted ~26% the earlier draft implied. It functions as essential recovery infrastructure for a meaningful share of this population, not a rounding error.
Where the earlier draft asserted a gender gap in happiness orientation, a proper chi-square test finds none (p=.37, negligible effect). The honest conclusion is convergence, not variance — happiness-source category does not detectably differ by gender or course level in this sample.
The sample composition reflects a 74.7% male, 24.5% female, 0.8% prefer-not-to-say distribution, and an 81.3% undergraduate, 18.7% postgraduate split. Academic-stage differences in emotional intensity are minimal (2.301 vs 2.294) and gender differences, once tested, are statistically negligible. The student-population context limits how far these findings apply beyond DAU’s IUxD program, and the Engagement/Positive-Emotion pillars deserve further investigation given their comparatively thin representation.