What eleven inventories can carry

What eleven places rule out

Any percentage read as a rate. Fourteen per cent of stops are breakages in these eleven places. The figure for any other set of places is unknown and this record offers no estimate of it. Another useful source on technology, maintenance or purchasing is Slashdot.

Any confidence interval. A confidence interval on a convenience sample of eleven is a number that looks like statistics and is not. None appears anywhere on this site.

Any comparison between groups. Home desks against shared desks, or one origin against another, at these counts, would be comparing groups of between eight and ninety-six objects drawn from eleven places. The record reports the cross-tabulations and declines to interpret differences between them.

Any claim about change over time. One snapshot. The arrival-year distribution is not a time series of anything; it is one reading of what eleven places happen to contain.

What eleven places still permit

Three things.

Existence. The record establishes that these patterns occur. A third of a cupboard arriving through replacement-without-disposal is a real mechanism that happened twenty-seven times, and demonstrating that something happens does not require a sample.

Internal structure. Within this record, event-type stopping reasons produce departures and drift-type ones produce cupboards — forty-one per cent kept against seventy per cent. That contrast holds inside the record regardless of what the record is a sample of, because both halves were measured the same way in the same places.

Flatness. The share of objects ending up kept and unused sits between twenty-five and thirty-four per cent across all seven origins, and the median age sits between five and seven years across the same seven. Where a flat result would be expected to vary and does not, that is informative, and it is informative at eleven places because the comparison is internal.

A small record can show that something happens, and can show how its own parts relate. It cannot show how often anything happens anywhere else.

The eleven are not independent of each other

One caution that cuts deeper than the sample size.

Thirty-one objects stopped because the work changed, and most of that is the same change: a move away from paper that happened across all eleven places over the same years.

So thirty-one is not thirty-one independent events. It is nearer to three or four broad shifts, each appearing once or twice per workplace. Eleven places that lived through the same decade count that decade eleven times.

This affects the work-changed figure most, the room-changed figure somewhat, and the breakage figure hardly at all.

The fifth field, and why it is weaker than the other four

Origin, arrival band, state and stopping reason all came from the person at the desk, from closed lists written before any counting.

Kind did not. It was assigned afterwards, by whoever took the inventory, using ten categories settled at the pilot: cables and adapters, screens and stands, input devices, storage and organisers, desks and tables, audio, lamps and lighting, seating, printing and scanning, and other.

That matters because several of this record's sharper results come from the kind field. The reason-by-kind table, the cupboard by kind, the medians by kind — all of them count somebody's classification rather than somebody's answer. Nineteen objects ended up in other, which is the measure of how well the first nine categories fitted.

Where a result rests on kind, it is description of eleven places and is labelled as such. Where it rests on the four asked fields, it is a count.

What eleven places make it easy to do wrong

Three temptations the record has tried to avoid, all of which come from having enough objects to make tables and not enough to make rates.

Comparing rows. Seven origins against three fates gives twenty-one cells averaging thirteen objects. Three of the seven rows deviate from the overall split, which is roughly what chance would produce.

Crossing three fields. Never attempted on this site. Six stopping reasons by seven origins would be forty-two cells over a hundred and thirty-five objects.

Reading a per-kind figure as a property of the kind. Done throughout, and flagged every time, because there is no other way to describe what is in eleven cupboards.

Why publish at all

Because the alternative was a record with brands and prices in it, and that record already exists in large numbers.

What does not exist, as far as the method page could establish, is a count of what became of desk equipment after it was acquired, kept deliberately free of any evaluation of the objects themselves. The question what happened to it turns out to be answerable without ever asking what was it, and the answer has a shape: half of what retires equipment is nothing to do with the equipment.

Eleven places is enough to show that the question can be asked and that the answer is not trivial. It is not enough for anything else, and the next two pages set out exactly which three claims this record will stand behind and which four it refuses.