There are two questions you can ask someone about their well-being, and they are not the same question. The first asks for a verdict: all things considered, how does your life measure up? The second asks for a report: how did yesterday actually feel — and when did you last spend an evening with a friend? For most of survey history these two answers moved together closely enough that researchers could be sloppy about which one they meant. They have stopped moving together. The verdict is holding. The report is not.
That divergence is the spine of this book, and it shows up in every major source we have for tracking how humanity feels about itself. We will call the verdict the Evaluative upper blade — the considered rating of a life as a whole — and the report the Experiential lower blade: the texture of daily emotion and the frequency of human contact. Across the Gallup World Poll, the U.S. General Social Survey, and the World Values Survey, the upper blade is resilient and the lower blade is falling. Drawn together, the two lines open like a pair of scissors.
The scissors, on one ruler
The Gallup World Poll’s two well-being measures since 2006, each standardized against its own 2006–2025 spread so an evaluative ladder and an experiential affect share can share a panel honestly. The upper blade is the population-weighted world life rating; the lower blade is daily negative emotion, drawn inverted so that down means worse for both. The widening violet wedge between them is the gap this book measures.
Read the upper blade in its own units and it is a story of remarkable steadiness. The world’s population-weighted rating of its own life sat at 5.31 on the 0–10 ladder in 2006 and 5.57 in 2025 — a gain of +0.26 points across two decades that included a financial crisis, a pandemic, and a global inflation shock. The ladder dipped here and there but never broke. By 2025 it stood at its highest point in the series.
The lower blade tells the opposite story. The share of people worldwide reporting a lot of worry, sadness, stress, anger, or physical pain “yesterday” — the negative-affect index — climbed from 0.242 in 2006 to a peak of 0.341 in 2022 before easing slightly to 0.315 in 2025, a net rise of +0.073. These are different constructs measured on different scales: the ladder is a considered judgment scored 0 to 10; negative affect is a share of people recalling specific bad feelings from a single day. They are never plotted on a shared raw axis here. But standardize each against its own variation, as the figure above does, and the shape is unmistakable — one blade lifting, one blade sinking, the wedge between them widening through the 2010s and never closing after 2020.
The same shape, in one country, across fifty years
Skeptics of global indices have a fair objection: maybe the divergence is an artifact of which countries enter the sample, or of comparing a Western 2006 to a more global 2025. So turn to the longest continuous well-being record in the world — the U.S. General Social Survey, running since 1972 with the same core questions — and look for the same scissors in a single, stable population.
It is there. The upper blade barely moves. The share of Americans calling themselves “very happy” held in the low thirties for decades: 34.1% in the 1970s, 33.4% in the 1980s, 32.3% in the 1990s, 32.5% in the 2000s, 30.7% in the 2010s. A drift, not a drop. The lower blade is another matter. A composite of how often people spend a social evening with friends, with relatives, with neighbors, and at a bar or tavern fell from 3.79 in 1974 to 3.44 in 2024 — a decline of −0.351 on its roughly 1-to-7 frequency scale, the slow thinning of in-person life that Robert Putnam first traced in Bowling Alone.
The American echo: flat verdict, falling connection
Two GSS series since 1974, each on its own axis because they are different things. Evaluative the share calling themselves “very happy” (teal, left axis). Relational the social-evenings composite (amber, right axis). The shaded band marks the 2021–22 web-mode switch.
happy == 3 (very happy), evaluative. Right axis: weighted mean of four social-evening items (socfrend, socrel, socommun, socbar), relational. GSS post-stratified weight wtssps; the two axes are deliberately independent so the constructs are never conflated. The right (relational) axis is zoomed and does not start at zero; tick labels show the true levels. Years 2021–22 (shaded) cross the push-to-web mode change.It lands hardest on the young
Aggregate flatness can hide a generational split, and here it does. Among Americans under 30, the share saying they are “not too happy” — the bottom rung — sat at just 9.1% in the 1987–91 window. By 2022–24 it had reached 25.6%, nearly tripling. The young, who once reported the least unhappiness of any age group, now report the most. The lower blade is not falling evenly across the population; it is falling fastest where the experiential and relational life is lived most intensely.
Under-30s saying they are “not too happy”
Share of GSS respondents under 30 on the bottom rung of the three-point happiness scale, pooled in five-year bins to keep cells stable. Evaluative measure — but among the young it has come unmoored from the national flatness.
wtssps, effective N suppressed below 50. Bins from 2017 onward (lighter, hatched) cross the 2021–22 mode change; the rise begins in the 2017–21 bin and holds in 2022–24. “Not too happy” is happy == 1, the bottom of the three-point scale.And the verdict holds worldwide
One more witness, to make sure the resilient upper blade is not a quirk of two surveys. The Integrated EVS–WVS asks a tenth-point question — all things considered, how satisfied are you with your life these days, 1 to 10? — in dozens of countries across seven waves since 1981. On a fixed set of 86 countries with repeated coverage, the equal-country average of life satisfaction was 7.25 in the earliest wave and 7.11 in the most recent — a change of just −0.141 over four decades. More to the point, across the recent, more globally representative waves the evaluative measure is flat-to-rising: 6.85 in wave 5, 6.86 in wave 6, 7.11 in wave 7. People’s considered verdict on their lives is, if anything, edging up.
The WVS deserves a caveat that earns its place in the argument. Its early waves were heavily Western — 19 countries in wave 1 against 80 in wave 7 — so the small four-decade dip partly reflects a changing roster, not a changing world. That is exactly why we lean on the WVS only as recent-wave evaluative-resilience support, not as a clean 40-year level trend. On that narrower, fairer claim it agrees with the other two sources: the upper blade does not fall.
World life satisfaction, by WVS wave
Equal-country average of life satisfaction (1–10) across the seven Integrated EVS–WVS waves, on a fixed set of 86 countries. Read the recent waves (5–7, brighter) as the evaluative-resilience signal; the early waves (1–4, dimmed) carry a Western country-set bias and a different, expanding roster.
satis_your_life, 1–10). The y-axis is zoomed (it does not start at zero) to resolve the small wave-to-wave movement; tick labels show the true levels.What the scissors does and doesn’t show
It is two constructs, never one. Every figure in this prologue keeps the evaluative and the experiential apart — on separate axes, or standardized and labeled — because the whole finding is that they have come apart. A ladder point and an affect share are different units and are never averaged together. Where the scissors figure puts them on a shared standardized ruler, that ruler measures each line against its own history, not against the other’s.
The mode change is real and bounded. The GSS push-to-web in 2021–22 and the Gallup face-to-face→phone shift in 2020 can move measured levels on their own. So the cleanest evaluative comparisons here stop at 2018, and the relational and global-affect declines that carry the argument are long secular trends that began decades before any mode switch. The divergence does not depend on the pandemic-era data points.
This is description, not diagnosis. The scissors says nothing yet about why the daily and relational life is thinning while the considered verdict holds — whether people are recalibrating their expectations, narrating resilience they don’t feel, or genuinely valuing lives that are emotionally harder than they used to be. That question is the rest of the book. The prologue only establishes the shape: an upper blade that endures, a lower blade that falls, and a gap that has not closed. Everything after this is an attempt to explain it.
Notes & data
- Sources. Gallup World Poll cleaned well-being extract, 2006–2025; U.S. General Social Survey cumulative file, 1972–2024; Integrated EVS–WVS, 1981–2022 (waves 1–7). Every figure in the text is recomputed and traced in
analysis/the-scissors/claims.json. - Gallup World (the blades). Within-country weighted means (
wgt) per wave, then population-weighted across countries byctry_pop_millions— never a plain average of country means. A country enters a wave only with a substantial interview count (several hundred); a wave enters at ≥ 20 countries with population data. 2005 is excluded (only 27 countries). Coverage runs 96–146 countries per wave. - GSS (the echo). Post-stratified weight
wtssps(full 1972–2024 coverage). Upper blade: weighted share withhappy == 3(very happy), direction-fixed higher = happier. Lower blade: mean of four social-evening items (socfrend,socrel,socommun,socbar), direction-fixed higher = more frequent, items suppressed below n = 50. Under-30 series pooled in five-year bins, effective N < 50 suppressed. - WVS (the corroboration).
satis_your_life(1–10), within-country weighted means averaged across countries with equal weight (no population variable in the extract). Fixed set of 86 countries present in ≥ 2 waves with ≥ 50 observations. Early waves are Western-biased by an expanding roster; used as recent-wave evaluative-resilience support, not a 40-year level trend. - Constructs. Evaluative = Cantril ladder, “very happy”, and WVS life satisfaction (a considered rating). Experiential / relational = Gallup’s “yesterday” negative-affect index and the GSS social-evenings composite. They are labeled everywhere and never plotted on a shared raw axis.
- Known limits. GSS 2021–22 push-to-web and Gallup 2020 face-to-face→phone mode changes can shift levels; the carrying trends predate them and the clean evaluative window runs to 2018. Cross-country ladder gaps under 0.2 points are not definitive. Population weighting in the Gallup blades uses a static ~2018-vintage population figure.
- Prior art. Robert Putnam, Bowling Alone (the long thinning of American social life); the World Happiness Report series (Cantril-ladder resilience alongside rising negative affect); Ed Diener and colleagues on the evaluative–experiential distinction. This prologue re-derives the divergence across three independent backbones and sets the book’s visual grammar.