Here are two numbers from the same ONS table that cannot both mean what they appear to mean. Between 2012 and 2022, eleven full years, Britain's life-satisfaction total moved by a grand span of 2.4 points: from a low of 859.5 to a high of 861.9. Then 2023 arrived at 429.6, a fall of 430.7 in a single year (Source: ONS, Personal well-being).

So this measure was flatter than still water for over a decade, then supposedly halved. One of those two facts is telling you about the country. The other is telling you about the file.

First, the unit. These are not scores out of ten and they are not people. The ONS aggregation note says the values are summed across every category in the dataset: the estimate itself, plus the lower and upper confidence limits, plus quarterly breakdowns and seasonal-adjustment variants. In other words, it is a running total. A total like that moves when the underlying wellbeing moves, and it also moves when the number of rows being added up changes. That distinction is the whole story here.

Look at how little the wellbeing signal actually shifts. Covid, lockdowns, schools shut, funerals over video call: the total went from 861.9 in 2019 to 859.8 in 2020, a dip of 2.1 points. Then it recovered to 860.0 in 2021 and 860.3 in 2022. The 2023 movement is more than 200 times larger than the pandemic movement. Nothing happened in 2023 that was two hundred times the pandemic.

Now the arithmetic that makes the case. Double the 2023 figure and you get 859.2, which sits 1.1 points below 2022's 860.3, comfortably inside that 2.4-point band the series wobbles around anyway. Do something similar at the other end of the table: 2011 comes in at 644.6, which is exactly 75.0% of 2012's 859.5. Multiply it by four thirds and you land on 859.5 on the nose. The most likely reading is that both the first and last years of this series contain fewer periods of data than the middle years, and that the totals scale almost perfectly with coverage.

That is a hypothesis, not a published fact, and it deserves the honest counter-argument. I cannot see the underlying quarters from this table, so I cannot prove the 2023 figure is partial rather than genuinely lower. It is possible the aggregation dropped a category rather than a time period, or that a definition changed. And two data points landing on tidy fractions can be coincidence. What I can say plainly is that a 50% one-year fall in a series with an eleven-year range of 2.4 points is not a behavioural change. Something structural moved.

Here is my opinion, and it is the only one in this piece. Wellbeing data is meant to be the humane counterweight to GDP, the thing councils and health commissioners point to when they argue for services that do not show up in output figures. A published total that reads as a halving of national life satisfaction, when it most plausibly reflects an incomplete year, undermines exactly that argument. Label the partial years on the face of the table. The number is too useful to be misread.