Sixty-two against thirty-one

By dwelling

Nine dwellings, nine ratios above one. The smallest is 1.4 and the largest 2.2.

A pattern that holds in every case in a nine-case record is about as strong as a nine-case record can produce. It still rests on nine cases, and several of the bands within a dwelling have fewer than five sittings in them. The practical question is whether a change improves completed work without creating unnecessary surveillance, which is why reviewers may see how the current workflow is described before choosing a baseline and access policy.

Where the gap is widest

The three largest ratios are Home I (2.2), Home E (2.1) and Home F (2.0). The wider design and work context can also be checked through Cornell University.

Home I is the ground-floor flat with the highest ending rate in the record. Home F is the bungalow with the shortest stretches. Home E is a maisonette in the middle of everything.

The three smallest are Home D (1.4), Home B (1.6) and Home H (1.6) — which are, in order, the first, second and third longest-median dwellings in the study.

That ordering is informative. The dwellings where the stretches are longest are the dwellings where the hour matters least.

What that ordering suggests

Two readings, and the record leans toward the second.

The first is a ceiling effect: a dwelling where stretches run to an hour and a half has less room to fall before it hits the length of the work itself.

The second is that the dwellings with long stretches are the dwellings where the place is private and nobody else's — a bedroom desk, a study — and private places do not change as much across the day. Nobody comes home to them, nothing competes for them, and in two of the three they have small windows.

The dwellings least affected by the hour are the ones where somebody works behind a door.

Three cases. A direction, not a result.

The four places where it reverses

Four of the thirty-one places have a longer median after one than before nine.

Two are armchairs. One is a back room. One is a sideboard used on a single day, which should be ignored.

The two armchairs are the interesting pair: both are in rooms the households described as used by the rest of the dwelling in the mornings and free later. The reversal there is about occupancy, not about the room.

The back room is Home F's, and it is the coldest place in the record. Five of the eighteen temperature endings are there, and all five are before eleven. A room that is too cold to sit in at nine and fine at two will reverse the pattern by itself.

The band the comparison skips

Between the two bands being compared sits 11–13, at forty-four minutes, with forty-one sittings in it.

That band holds nine of the fourteen wanted endings and eight of the twenty-four light endings — the two causes that peak around midday. It is where the fall from sixty-two to thirty-one actually happens, and it is the busiest part of the day for the dwelling intervening.

Comparing the first band with the fourth makes a clean headline. Looking at the middle shows that the decline is not gradual: it is mostly one three-hour stretch in which two separate causes both arrive.

What would make this spurious

If the work people do in the afternoon were intrinsically shorter-form than the work they do before nine, the whole effect would be a fact about jobs.

The record cannot rule this out and it is the most serious objection to this page. What it can offer is the bare-ending figure: if the afternoon were full of short tasks finishing, bare endings would be over-represented after one, and they are very slightly under-represented.

That is evidence, not proof, and it is the only evidence available.

The sittings that are missing from the comparison

Thirty-eight sittings begin in the first band and forty in the fourth, which is close enough to compare directly.

What is not comparable is who contributed them. Six of the nine dwellings supplied at least four sittings to the 07–09 band; three supplied two or fewer. The early band is weighted toward the three households whose working day begins before eight.

Reweighting so that each dwelling contributes equally moves the first band from sixty-two minutes to fifty-nine and the fourth from thirty-one to thirty-two.

The gap narrows from thirty-one minutes to twenty-seven and the finding survives. It is reported because an unweighted figure on nine unequal contributors is the kind of thing that should be checked, and because the check could easily have gone the other way.