What the entry field could not ask

The entry field accepted one of four answers and nothing else. This page sets out what the closure bought, what it cost, and where it broke.

The four, as they were fixed

nothing            nothing to pay and nothing to buy
one purchase       a business with chairs; one transaction
a purchase an hour the same, with a visible tempo
a fee              a ticket, a hire charge or a membership

The list came out of a pilot afternoon in a twelfth place and was frozen before the first entry. It never grew, which is unusual for a closed vocabulary in this series and is the strongest thing that can be said for it.

Why closed

An open field collects a dozen wordings for the same three situations, and somebody then has to sort them — by which point that somebody has seen the counts and knows which sorting flatters them. Teams coordinating visits, notes and follow-up can check the setup before collecting data, then check whether the resulting evidence actually improves the field process.

Fixing four values in advance puts the arguable decision where a reader meets it before any number does. The four above are what the pilot produced; a reader who would have cut them differently can see exactly which entries move. Additional practical or institutional guidance relevant to the survey can be found at ADA.gov.

The three questions not asked

How much. Deliberately absent, like every price in this record. A field with amounts in it turns a description of terms into a comparison of value, and a comparison of value is a recommendation about where to go.

The cost is real and falls hardest on the ninety-six places that expect a purchase, because for them the price is the terms, and on the fifteen behind a fee for the same reason.

Whether anybody could afford it. Not asked, not inferable, and the single largest thing this record does not measure. A door with a price on it is counted here as a door, and for some people it is a wall.

Who sets the expectation. An owner's habit and a chain's policy are the same entry. So are a verger and a diocese, a shop manager and a landlord.

The three that were argued about

A bookshop with a counter serving coffee at the back. Logged as one purchase. It is two places under one roof and the record has no way to say so. Seven of thirteen bookshops are affected by this reading, and a reader who would separate them should know the figure.

A hall entered during somebody else's booked session. Logged as a fee, on the grounds that the room was hired. The observer sat at the back where a stranger could sit, and whether the hire terms permitted that at another hour is unknown. Eight of the fifteen fee entries rest on this.

A station lounge behind a card reader, inside a concourse that is free. Logged as two places rather than one, because a stranger walking in meets two different doors. Three entries.

All three resolutions are judgement calls and all three are printed so that a reader who disagrees can see which rows move.

What the field gets right

It records what a stranger meets at the door, which is the thing that decides whether they get as far as a chair.

And it separates cleanly the two situations the rest of the record needs separated: places where a transaction mediates the sitting, and places where nothing does. A hundred and twenty against eighty-six. Every later finding is read against that split, and the split is firm in a way the four values individually are not.

What it gets wrong

It measures an expectation as though it were a fact.

Ninety-six entries rest on the proposition that sitting in a café without buying anything would be understood as a kind of theft. Nobody tested that, nobody asked, and no sign anywhere says it. The largest value in the field is the observer's reading of a silence.

The entry field is firmest where places state their terms and softest where they do not — and a hundred and twenty of two hundred and six places do not.

That is the honest summary. The seventy-one free entries and the fifteen fee entries are observations. The ninety-six and the twenty-four are interpretations, and the record would be improved more by asking a hundred and twenty managers one question than by anything else this method could do.

What the four values are worth, one by one

Nothing (71). The firmest value in the field. A place either requires something at the door or it does not, and in seventy-one cases it plainly did not. A second observer would reproduce this column almost exactly.

A fee (15). Equally firm. A ticket, a hire charge or a card reader is a visible fact.

One purchase (96). An interpretation. Nothing in any of the ninety-six says a purchase is required; the coding rests on the proposition that sitting at a café table without buying would be understood as a kind of theft. Nobody tested it.

A purchase an hour (24). An interpretation of a smaller and softer thing: a sign, a word, or a table cleared. Eleven of the twenty-four rest on the observer's reading of a social act performed on somebody else.

firm values      86 places    42%
interpretations 120 places    58%

The majority of this field is interpretation, and that proportion is worth stating because it is the opposite of what a table of four neat counts suggests.

Why the field was kept binary where it could have been graded

A scale from open to closed was drafted and dropped at the pilot.

Two reasons. A scale invites a reader to treat the order as a ranking of hospitality, and the record's own figures refuse that: the fee group limits staying less than the single-purchase group, and the free group has the highest concentration of written notices.

And a scale would have hidden the one distinction that does work: a transaction at a counter against everything else. A hundred and twenty places against eighty-six, and every later finding on this site is read against that line.

The change worth making

One line in the sheet: what does the place say about time, if anything?

The record has it implicitly, scattered across the limit field, and pulling it out as a single yes-or-no at the door would have made the central finding — that almost nobody says anything — directly countable instead of assembled afterwards.