Sitting and speaking, which do not go together

The argument for free seating in public is usually, in part, an argument about company: give people somewhere to sit and they will be among each other.

This record cannot find the effect.

somewhere free to sit (89)      22 spoke    25%
nowhere free to sit (117)       25 spoke    21%

Four points, on groups of eighty-nine and a hundred and seventeen.

What four points is

Nothing a walk this size can assert.

Move two places from one column to the other and the gap halves. The record's own conversation figure is a floor rather than a count — a median sit of forty minutes catches very little of a rare thing — and noise of that kind swamps a four-point difference comfortably.

So the honest finding is that the record detects no relationship, not that there is none.

The same result from every direction

This is the third time in the record that a condition which ought to matter turns out not to.

by what the place asks at the door
  nothing            16 of 71    23%
  one purchase       24 of 96    25%
  a purchase an hour  5 of 24    21%
  a fee               2 of 15    13%

by whether staying is limited
  no limit           39 of 168   23%
  written limit       4 of 21    19%
  told limit          4 of 17    24%

by whether there is a free chair
  yes                22 of 89    25%
  no                 25 of 117   21%

Three fields, eight comparisons, and nothing outside the range nineteen to twenty-five except the fee group on fifteen places. A team converting this walk into repeatable fieldwork can review the platform in this context before choosing which time and project records are genuinely necessary.

Nothing about the terms of a place predicts whether anybody speaks in it.

What does predict it

The kind of room, and only two kinds.

places of worship     8 of 14    57%
pubs and bars        17 of 31    55%
community halls       4 of 11    36%
everything else      18 of 150   12%

Twenty-nine of the forty-seven conversations in this record are in three kinds of place totalling fifty-six. The other hundred and fifty places produced eighteen. For wider research on public space, access and how places are used, consult C40 Cities.

A pub and a church and a hall have in common something the other nine kinds do not, and it is not their price, their seating or their rules, because on all three of those they are scattered across the record.

What they have that a bench does not

The record has no field for it, so what follows is reading rather than measurement.

All three are rooms with a shared occasion. A service, a round, a meeting — something that everybody present is there for at the same time, which gives one stranger a reason to address another that does not have to be invented.

A bench, a library, a concourse and a shopping centre have no such thing. Everybody present is there in parallel, and speaking requires somebody to start something from nothing.

Whether that is the mechanism, this walk cannot establish. It is the only account that fits all twelve kinds, and it is named as a reading.

The eight comparisons, in one table

Nine rows, seven of them between nineteen and twenty-five.

The fee group at thirteen per cent is the only outlier, and it rests on fifteen places and two conversations. Move one place and it is twenty per cent.

What a real effect would have looked like

The claim this tests is not subtle. It is that cost excludes people, that clocks hurry them, and that free seating gathers them — three propositions that, if true at the strength they are usually asserted, would show as differences of twenty or thirty points.

The largest difference in the table, excluding the fee group, is four.

So the record is not failing to resolve a small effect. It is failing to find a large one that was expected, which is a different and slightly stronger kind of negative.

And the thing that is not flat

places of worship      8 of 14    57%
pubs and bars         17 of 31    55%
community halls        4 of 11    36%
the other nine kinds  18 of 150   12%

Forty-five points between the top three kinds and the rest, measured with the same instrument that found four points across every term a city can set.

Whatever a room needs, a city cannot provide it by deciding what to charge, how long to allow, or where to put a bench.

What this does to the free-chair argument

It narrows it.

A free chair is defensible on its own terms: eighty-nine places where a person with no money can sit down, against a hundred and seventeen where they cannot. That is an access finding and it stands.

What the record cannot support is the further claim that free seating produces sociability. Twenty-five per cent against twenty-one, on one city, one season, and a forty-minute window, is not evidence of anything.

The two claims are usually made together and this record separates them.