Let me say from the outset that I’ve long been sceptical about some of the methods used by Richard Florida, celebrated author of The Rise of the Creative Class. And I’m not the only one – this review of his book by Edward Glaeser is written with a velvet glove but packs an iron fist.

So it’s not surprising I’m unimpressed by Commuting is very bad for you, written by Florida for last month’s issue of The Atlantic. He gets it completely wrong and provides a lesson in the dangers of only seeing what you want to see.

Florida seizes on a survey of 173,581 working Americans which he claims shows that those with longer commutes suffer higher levels of back pain, higher cholesterol and higher obesity. It also shows, he says, that commuting takes a toll on emotional health and happiness – those who commute more worry more, experience less enjoyment and feel less well-rested.

Commuting by car is so bad it’s up there with smoking:

“commuting is a health and psychological hazard, not to mention the carnage and wasted time on our over-clogged roads. It’s time to put commuting right beside smoking and obesity on the list of priorities for improving the health and well-being of Americans”.

The trouble is the data he cites doesn’t support these conclusions. A proper reading of the two tables from his article (I’ve reproduced them above) indicates there’s very little relationship between commute time and health.

His conclusion depends to a large extent on the extreme categories, 0–10 minutes and 90-120 minutes, especially the latter. The mistake he makes is to assume there are equal numbers of commuters in each category.

But this is not the case. The survey he cites found only 19% of the sample spend more than half an hour getting to work and only “3% commute for more than an hour each way”. In fact as this analysis of US Census data shows (Table ES-5), 45% of commutes in US metropolitan areas take less than 20 minutes and only 8% take more than 60 minutes.

To make matters worse, Florida’s dissing of the car for overly-long commutes is misplaced. The really long commutes are in fact made by public transport, especially by train.

In Melbourne for example, the median commute time by all modes is 30 minutes. Walking is shortest at 20 minutes and public transport is longest at 55 minutes. Cars are 30 minutes. The longest median commute time in Melbourne by car is 33 minutes (Wyndham) and the longest by public transport is Mornington Peninsula (95 minutes).

Difference between median commute times in Melbourne by car and public transport (minutes)

The accompanying Figure shows the difference between the median commute time by car and public transport for each of Melbourne’s municipalities, arranged approximately by distance from the CBD.

The much larger penalty for commuting by public transport rather than car from the outer suburbs (75 minutes for Mornington Peninsula) is due primarily to workers travelling to the city centre by train (of course it needs to be remembered that the proportion of outer suburban workers who commute by public transport is very small – about 10%, but undoubtedly even lower in Mornington Peninsula).

Commuting is actually a good thing. If you have a job, then unless you work from home, you’ll need to commute. A short commute is ideal, but if you have a really long commute then you’re in a tiny minority and it’s far more likely you’ll be travelling by train than by car – that sounds like, on balance, it might be a good thing too. And it’s worth noting that if you do have a long commute then there’s a good chance you’re doing it because you’ve got one of those higher-paying CBD jobs.

This whole episode should be taught in class as an example of how people see what they want to see and ignore the data.