When the first version of this story was written, 2020 looked like a cliff. The pandemic had pushed daily negative emotion to a spike that Gallup’s interviewers — calling by phone because they could no longer knock on doors — recorded across most of the world. And yet the same surveys found something stranger: people’s considered rating of their own lives, the 0-to-10 Cantril ladder, hardly flinched. The headline was the divergence between the two.
Five more waves of data are now in, and they change the ending. The divergence did not just appear in 2020 and fade. It calcified. Life evaluation recovered, then overshot. Across a fixed panel of 115 countries, population-weighted, the world ladder went from a 2017–19 baseline of 5.142 to 5.365 in 2020 and on to 5.551 by 2024–25 — 0.409 points above where it started. The evaluative blade did not just hold. It climbed.
Daily feeling tells the opposite story. The negative-affect index rose from a baseline of 0.305 to 0.328 in 2020, and by 2024–25 had eased only to 0.316 — still 0.012 above baseline, roughly half the spike still unrecovered. That is the lower blade: it registered the catastrophe in full and never quite came back.
Two questions, two kinds of well-being
Survey researchers treat these as different constructs, and the pandemic arc is the clearest demonstration on record of why. Evaluative The Cantril ladder asks for a considered judgment of your life as a whole — slow, comparative, anchored to expectations. Experiential The “yesterday” items ask whether you felt worry, sadness, stress or anger for much of the previous day — fast, concrete, yes-or-no. The ladder is a points scale; the affect items are shares of people saying yes. They are labeled throughout, because the whole argument lives in the gap between them — and in how differently the two recovered.
Every country’s pandemic arc, against its own 2017–19 baseline
Each thin path is one country across three epochs — its 2017–19 baseline, its 2020, and its pooled 2024–25 — colored by region. The thick dark line is the population-weighted world. Switch measures to watch the evaluative ladder recover and overshoot while daily distress lingers.
The worst year was 2022, not 2020
Pinning the catastrophe on 2020 turns out to be a trick of where the first measurement happened to land. Follow the full annual series and the negative-affect index does not peak in the pandemic’s first year at all. It keeps climbing — through the lockdown fatigue of 2021 and into 2022, the year inflation bit, war returned to Europe, and the index reached its highest reading in this panel’s series. Worry hit its own series high the same year. Only after 2022 does daily distress begin to ebb.
The ladder’s path is almost a mirror image. It dipped through 2019, recovered with the 2020 and 2023 waves, and by 2025 sat well above its pre-pandemic floor. Put the two on one frame — the slow evaluative judgment and the fast experiential feeling — and you can watch them come apart and stay apart.
The two blades, 2006–2025
The world’s life evaluation (teal, left axis) and its negative-affect index (rust, right axis), wave by wave. The axes are independent and labeled; the point is the shape of each line, not a shared level. Earlier waves cover fewer countries and are lighter for it.
year_wave). Country coverage varies before 2017 (shown faded); the fixed 115-country panel anchors 2017 onward. Evaluative ladder is left axis (0–10 points); experiential negative-affect index is right axis (share endorsing the five negative items). The dual axis is a deliberate choice to compare trajectories, never levels.An asymmetric recovery, on one ruler
Points and shares don’t compare directly, so put each measure on a common ruler: how far did the world’s value move from baseline, in standard deviations of the 115-country baseline spread? Read it twice — the 2020 shock, then the 2024–25 settlement. In 2020 the experiential items lurched: stress moved +0.39 SD, anger and sadness around +0.36, worry +0.30, the negative-affect index +0.31. The evaluative ladder moved +0.22 SD — in the opposite-to-expected, upward direction.
By 2024–25 the rulers diverge further. The ladder is now +0.41 SD above baseline — nearly double its 2020 move. But the affect items have mostly drained back toward zero: worry is essentially home at +0.03 SD, the negative-affect index sits at +0.15, sadness at +0.19. Distress receded; it did not vanish. Of the 73 countries whose negative-affect rose in 2020, 40 still carry it above baseline; of the 80 where worry spiked, 39 remain elevated.
How far each measure moved — the shock, then the settlement
World population-weighted change from the 2017–19 baseline, in standard deviations of the cross-country baseline spread. The faint marker is 2020; the solid bar is 2024–25. Rust = experiential (yesterday); teal = evaluative (life as a whole).
Who recovered, and who did not
The world averages hide real motion in both directions. The largest evaluative gains by 2024–25 came from places climbing off low or volatile starts — Vietnam (+1.17 ladder points over baseline), Serbia (+1.05), Georgia (+0.93), China (+0.91). The deepest unrecovered falls cluster in countries still inside their own crises: Lebanon (−0.99), Congo-Brazzaville (−0.99), Benin (−0.84). A handful of rich, stable countries also slipped — Canada (−0.54) and Switzerland (−0.52) rate their lives lower now than before the pandemic.
Worry’s map looks different again. Where war or fresh economic strain set in, daily worry rose and stayed: Ukraine tops the list (+0.262 of share over baseline), with the Czech Republic and Lebanon behind it. The biggest worry declines belong to countries that found stability — Croatia, Vietnam, Saudi Arabia. Across the panel the median country’s ladder is up +0.112 points on baseline while its worry is down a hair, −0.008 of share: the typical country recovered its verdict and roughly cleared its worry, even as a substantial minority did neither.
Life evaluation · largest gains
Life evaluation · still furthest below
Worry · largest rises that stuck
Worry · largest declines
China, and the weight of one country
One country does heavy lifting in the world aggregate. China holds about 20% of the panel’s population weight and its ladder rose +0.91 points over baseline, so it flatters the global recovery. Take China out and the evaluative gain shrinks but survives: the world ladder is still up +0.281 points on baseline by 2024–25, against +0.409 with China in. The experiential picture barely changes — without China, the negative-affect index actually sits a touch below baseline by 2024–25 (−0.009 of share), confirming that the lingering distress is broad rather than a Chinese artifact. The recovery is real with or without its largest member; its size is what China inflates.
What this does and doesn’t show
The mode change still shadows 2020. Gallup’s 2020 fieldwork switched from face-to-face to phone in most countries, with no mode flag in this extract, so part of any 2020 shift is the telephone rather than the times. The recovery story leans on the cleaner comparison: 2024–25 is back to face-to-face in most countries and broadly mode-comparable to the 2017–19 baseline. The asymmetry — ladder up, distress half-back — is a baseline-to-recent contrast that mostly sidesteps the 2020 mode confound, though country mode histories still vary.
115 countries is not the world. The fixed panel — surveyed in the 2020 wave with at least 500 interviews, holding a 2017–19 baseline, and re-surveyed in 2024–25 — is what every “world” number here means, weighted by population. Fieldwork is assigned by survey wave, not calendar year: nine countries’ early-2020 interviews belong to the 2019 wave and stay in the baseline. Taiwan appears in the charts but, lacking a population figure, sits outside the aggregates.
Small ladder moves are soft, but these are not small. Response styles differ across cultures, so ladder gaps under 0.2 points shouldn’t be pressed. The world’s +0.409 recovery and the median country’s gain clear that bar; the ex-China +0.281 figure clears it too. The affect items remain binary yes/no recollections of a single day — shares of people having a bad day, not intensities of feeling.
The first telling of this story asked whether 2020 was the worst year and answered: for daily feeling, nearly — but life evaluation held. The fuller record sharpens it. Distress peaked later, in 2022, and has only half receded. Life evaluation did more than hold: it recovered and overshot. The World Happiness Report read the same survey and found post-pandemic resilience in the ladder; this fixed-panel re-computation finds that resilience compounding, while the experiential floor stays lower than it was. The world rates its life better than before the pandemic. It is still having more bad days.
Notes & data
- Source. Gallup World Poll cleaned well-being extract; this article uses the 2017–2025 survey waves, with a long 2006–2025 series shown for context. Analysis code:
analysis/worst-year/build_data.py; every figure in the text is recomputed there and traced inclaims.json. - Fixed country set. 115 countries with a 2020 wave of at least 500 interviews, a 2017–19 baseline (at least two of the three waves, ≥500 pooled interviews), and a 2024–25 recent window of at least 500 pooled interviews. Interviews are assigned to waves by
year_wave(fieldwork year), not calendar year. Baseline = the equal-weight mean of a country’s available 2017, 2018 and 2019 wave means; recent = pooledyear_wave2024 and 2025. - Weights. Within each country-wave, estimates use Gallup’s within-country sampling weight (
wgt). Cross-country “world” aggregates weight country means by total population (ctry_pop_millions, a static ~2018-vintage attribute), never a plain average of country means. China holds ~20% of that weight; an ex-China robustness figure rides alongside. Taiwan lacks the population figure and is excluded from world aggregates (it appears in all charts). - Constructs. Evaluative = Cantril ladder (0–10 points). Experiential = binary “yesterday” items (worry, sadness, stress, anger, enjoyment), reported as shares; the negative/positive affect indices average five items each. The two are labeled throughout and never plotted on a shared axis — the one dual-axis chart uses independent, labeled scales to compare trajectories, not levels.
- Known limits. 2020 survey mode switched to phone in most countries (no mode flag in this extract); 2021–22 most countries returned to face-to-face, so 2024–25 is broadly mode-comparable to the baseline. Cross-country ladder gaps <0.2 points are not definitive; country-level differences are descriptive. The 2006–2016 series rows have shifting country coverage and are used only for the long-run shape.
- Prior art. World Happiness Report 2021, ch. 2: Happiness, Trust, and Deaths under COVID-19 (ladder resilience plus a negative-emotion rise, on the same survey); Gallup Global Emotions 2021 (2020 as a record year for negative emotions). This piece extends both through 2025 on a fixed panel and adds the recovery arc, finding the distress peak in 2022 and an evaluative overshoot.