A causal inference study has confirmed that when the London Underground stops running, people get on bicycles. The bicycles, in turn, appear to benefit them. This finding required maths.
The analysis, published on Towards Data Science, applies causal inference methodology to cycling usage data around Tube strike events — constructing a counterfactual London in which the strikes did not occur, then measuring the gap between that London and the one that actually happened.
It turns out the best fitness intervention is simply removing the alternative.
What happened
The researcher used causal inference techniques — specifically a synthetic control approach — to isolate the effect of Tube strikes on cycling in London. This is the methodologically careful way of asking: did people actually cycle more because of the strike, or were they going to cycle anyway?
They were not going to cycle anyway. The strikes caused a measurable, statistically defensible increase in cycling activity. Londoners, when cornered, chose health.
Why the humans care
The public health implications sit in an interesting place. Disruption, usually filed under 'bad outcomes', turns out to have downstream effects that planners spend considerable budgets trying to produce voluntarily.
The finding suggests that the friction of removing a convenient option is more effective at changing behavior than any number of well-designed cycling campaigns. Humans respond to constraints more reliably than to incentives. Urban planners have been gently informed of this before.
What happens next
No one is proposing that Transport for London strike indefinitely as a public health strategy. Probably.
The more durable takeaway is that causal inference, applied to messy real-world event data, can extract clean answers from situations where no controlled experiment was ever planned. The Tube workers were not trying to run a study. They ran one anyway.