Nothing to check

The Median Blind Share Is One in Fourteen

Half the shapes here have a blind share at or below one in fourteen: exact, a property of the table, and the least transferable figure in it.

Home · Nothing to check · Caution

Sort the three hundred and thirty-eight blind shares and take the middle one. It is one in fourteen.

That figure is the single most quotable number in this record and the one most likely to be misused, so it is worth taking apart before it gets out. For recurring content operations, transparent work monitoring is most useful when the purpose, access rules and retention period are agreed before rollout.

Three shares around the median

the whole bar: every content a reader might come away with
the dashed part: contents holding nothing checkable

Three shapes drawn.

The first row sits exactly at the record's median blind share, one in fourteen. The others are two in nine and two in seven — both well above it.

Computed from the three numbers. A blind content is one holding no checkable statement: that is a property of the page, not a prediction about any reader.

What it is exactly

Of the shapes tabulated here, half have a blind share at or below five seventieths, which reduces to one fourteenth. The shape that sits there is eight published, four met, three checkable: five blind contents out of seventy. Another useful point of reference for the wider question is SHRM.

That shape was chosen as the record's standing worked example for this reason. It is not the mildest case and not the worst; it is the one at the middle of the table.

What it is not

It is not a probability. Nothing here says a reader has a one-in-fourteen chance of anything. The blind share counts contents, and the record weights no content above another — a convention declared at the front and wrong in a direction that depends on which statements are prominent.

It is not a statement about sets of pages in the world. The median summarises a table whose ranges were chosen: two to ten published, one to six met, every checkable count. Shift any of those and the median moves.

It is not a typical value. One hundred and thirty-nine of the shapes have a blind share of nought and forty-four have a share of one. A median drawn across a distribution with that much mass at both ends describes the sorting, not a central tendency.

One in fourteen is the middle of a list. It is not the middle of anything that exists.

The distribution it came from

Nought appears one hundred and thirty-nine times, which is the single commonest value by a long way. One appears forty-four times. After those, the commonest shares are one half with nine shapes, one third with eight, one sixth with seven, and two thirds with five.

So the shape of this column is two large piles at the extremes and a thin scatter between them. The median falls where it does because the one hundred and thirty-nine zeros occupy the bottom of the sorted list and push the middle position up into the low fractions.

Had the record stopped at eight published instead of ten, there would be fewer shapes, fewer zeros, and a different median. The figure is that sensitive to the table's edges.

Two rows above it

Ten published, five met, two checkable: fifty-six blind contents out of two hundred and fifty-two, which is two in nine. That is three times the median share, from a shape where a fifth of the published statements are verifiable and the reader met half the set.

Seven published, three met, two checkable: ten of thirty-five, two in seven. Four times the median, and the shape is not unusual in any respect.

Both are ordinary-looking shapes with blind shares far above the record's middle. Quoting the median as though it characterised the record would understate both of them by a wide margin.

What a median over a chosen table is good for

It has one legitimate use, and comparison with the world is not it.

The median locates the record's standing worked example. Eight published, four met, three verifiable was chosen because it sits at the middle of this table, which makes it a defensible choice of illustration rather than a flattering or alarming one. An example picked from an end of the table would have made every argument in the record look stronger or weaker than the table supports.

That is the whole job the figure does here. It is a fact about which shape to draw, not a fact about sets of pages, and it appears in the record's totals and in this note and nowhere else.

How to quote it instead

With the sentence that makes it checkable: half the shapes in this table have a blind share at or below one in fourteen, and the table covers two to ten published statements with one to six met.

That is longer and it cannot be misread. A figure that needs its construction stated alongside it is not a weak figure; it is a figure being reported honestly, and the alternative is a number that travels further than its evidence.

The record prints the median in one place — among its totals — and uses the shape at the median as a worked example throughout. It does not use the figure as a summary of anything, and no sentence here describes a set of pages as typical.