For three decades, one of the most cited findings in the economics of happiness had a clean, uncomfortable shape. American women had begun the 1970s rating their lives happier than men did. By the mid-2000s that lead was gone, and on some measures men had pulled ahead — even as women's wages, schooling, and freedoms climbed. Betsey Stevenson and Justin Wolfers gave it a name in 2009: the paradox of declining female happiness.
Their data ran through 2006. The General Social Survey has kept asking the same plain question every couple of years since — "Taken all together, how would you say things are these days?" — and the answers now reach 2024. So we did the obvious thing: re-ran the gap, women minus men, through the whole half-century, and looked at what happened after the famous study stopped.
The decline does not continue. It doesn't reverse into a new female advantage either. The women-minus-men gap, which had narrowed so tidily for thirty years, stops trending around the early 2000s and starts wobbling across the zero line — positive in some years, negative in others, never far from nothing. The 1970s crossing was real. The continuation, the part that made the paradox feel like a trajectory, is the part that doesn't replicate once you give it more rope.
The crossing, and what came after
Women’s happiness minus men’s, by multi-year bin. Above the line, women report more; below it, men do. The shaded band is a weighted bootstrap 95% interval — when it straddles zero, the gap is within noise.
A landmark result, extended
Read the chart left to right. In the first bin, 1972–76, women lead men by about 4.6 points in the share calling themselves “very happy,” and by about 0.055 on the three-point scale. That early lead is the genuine article, and it is exactly the starting point Stevenson and Wolfers worked from.
Then it erodes. By the late 1980s the mean-scale gap has slipped below zero; by the early 2000s the very-happy gap crosses too. On the 1–3 mean the sign first flips negative in the 1988–93 bin; on the very-happy share, in 2002–06 — right where the original analysis ends. If you stopped there, as the literature largely did, the story writes itself: a steady, generation-long decline in women's relative happiness.
But the survey didn't stop. And in the eighteen years past the 2006 endpoint, the line does not keep falling. It jitters. The year-to-year mean gap changes sign 3 times after 2006 and 16 times across the full run — the fingerprint of a quantity hovering at zero, not one moving through it. The 2008–12 bin actually swings back in women's favor, by about 3.0 points on the very-happy share: the Great Recession landed harder on men's mood than on women's. The bins on either side sit close to nothing.
The honest summary isn't “women kept falling.” It's that a clean trend turned into a cloud of small, sign-switching differences — and a cloud is not a continuation.
Does a continued decline survive a trend line?
You can force a straight line through the year-by-year gap, and through 2006 it does slope gently down — about −0.0021 per year on the mean scale. Fit the same line from 2006 onward and the slope is −0.0014: similar in size, but riding a series so noisy that the “trend” is mostly the 2002 reading, an unusually low year for women, doing the pulling. Over the entire half-century the slope is −0.0009 per year — statistically a whisper, practically a flat line that happens to tilt. Change the window and the story changes with it. That fragility is the result: the decline is not robust to being asked to keep going.
Which lives, which decades
The same zero-centered gap, split by life stage and collapsed into three eras. Teal bars mean women in that group reported more happiness; clay bars mean men did. Hover any bar for the gap, both measures, and the sample.
Where the early reversal actually lived
The narrowing wasn't spread evenly. Break the gap apart by life stage and a few groups carried most of it. Among married people, women's early lead of about 0.044 on the mean scale collapses toward 0.028 — a real compression in the largest group in the sample. Among parents, the gap goes from roughly zero in the first era to mildly negative in the middle and recent ones: mothers and fathers report nearly identical happiness now, with fathers a hair ahead. And in the oldest band, 65 and over, men sit ahead of women in every era — the one place a gap has been consistent, and it points the other way.
The groups that resist the convergence are the unattached ones. People who never married and people with no children show a durable female advantage that barely budges across fifty years — childless women still lead childless men by about 0.058 in the most recent era. Whatever pulled married and parenting women's relative happiness down, it didn't touch the women who were outside those institutions.
Put the cuts together and the present looks less like a gap than its absence. Across most groups, in the most recent era, the bars are short and their bootstrap intervals cross zero. By education, the once-wide advantage among the college-educated has flattened to near nothing. By labor-force status, employed men and women report essentially the same happiness. “No consistent gap in most groups today” is not a hedge — it is the finding.
What this does, and doesn't, show
Two cautions keep this honest. The first is size. These are differences of a few percentage points and a few hundredths of a point on a three-rung scale. A 4.6-point lead in 1974 is real but modest; a 0.6-point gap in the 2020s is, for practical purposes, zero. Most of what looks like movement in the raw series is sampling jitter around a small number, which is exactly why the bootstrap band matters: where it crosses the line, there is no gap to explain.
The second is the instrument. This whole article rides on one evaluative question — an overall judgment of how things are. That item is not the same as how people feel moment to moment, and on those experiential measures the picture is less flattering to the headline. In the GSS health modules, women report about experiential 1.2 more bad-mental-health days a month than men — roughly 4.3 against 3.1 — a gap that has not vanished. Surveys of anxiety, depression, loneliness and disturbed sleep tend to find the same: women reporting worse, even in places where the happiness gap is flat or tips their way. A flat evaluative gap does not mean men and women are living the same emotional life. It means the global “all things considered” verdict has converged, which is a narrower and more specific claim.
None of this is a verdict on the women's movement, in either direction. The 1970s lead, the long narrowing, the recent flatness — these are associations in survey responses, read evenhandedly. What the extended record will not support is the tidy extrapolation. The paradox of declining female happiness described something true about the past. As a trajectory carried into the present, it simply doesn't hold.