Working Out an Attendance or Absence Rate: the Denominator Decides

What the tool computes — and three places it misleads

Attended divided by held, times a hundred. Then, if you are at or above your target, it works out the largest number of further absences you could take and still be at or above it — attended times a hundred, minus target times held, all divided by the target, rounded down. If you are below, it works out the smallest number of consecutive attendances that would bring you back — target times held minus attended times a hundred, divided by a hundred minus the target, rounded up. Both formulas are right; I checked them by stepping one class either side of every answer they gave.

The first place it misleads is silent. If the classes attended exceed the classes held — one digit typed wrong, a paste of the wrong column — the component returns nothing at all. No error, no red border: the entire result block disappears and the page shows four input boxes and blank space below. If you have entered numbers and there is no percentage, that is why, and the fix is to check that the first box is not larger than the second.

The second and third places are the same bug seen twice. The algebra above only runs when the target is strictly between 0 and 100, but the message that displays it runs regardless. Set the target to 100 while you are below it and the amber box reads "Below target. You must attend 0 classes", which is not merely unhelpful but false: once you have missed a single class, 100 % is unreachable for the rest of the count, and no number of future attendances fixes it. Set the target to 0 and the green box reads "Target met. You can still miss 0 classes", when in fact you could miss every remaining class and still be at or above zero. Both are boundary cases you will only hit by typing a round number, and both say something confidently wrong when you do.

The denominator decides the answer

Absences are a fact; a rate is a construction. Three denominators are in common use and they are not interchangeable. Scheduled days — the days you were actually timetabled to be somewhere — is the one almost every school and employment policy means, because it is the only one that measures against an obligation. Working days, meaning the weekdays in a period whether or not they were scheduled, flatters anyone whose timetable is lighter than a full week. Calendar days flatters everyone, and appears mostly in figures quoted by whoever benefits from the smaller number.

There is a fourth choice hiding inside the first, and it is the one that catches people out: the unit. In England, schools take a register twice a day, once in the morning session and once in the afternoon, and attendance is counted in sessions, not days — so a single afternoon off is one missed session, half a day, not a whole one. France counts the same way for the absenteeism threshold: the Code de l'éducation makes the trigger four unjustified half-days in a month. Choosing sessions rather than days does not change the percentage when every absence is a whole day — 342 of 360 is the same 95.0 % as 171 of 180 — but it changes it as soon as anybody leaves at lunchtime.

The tool has one box for all of this. "Total classes held" is whatever you put in it, and the percentage it returns is exactly as meaningful as your choice. That is the right design for a calculator — it cannot know your institution — but it does mean the number on screen is not portable. Before you quote a rate to anyone, write down what you divided by, and check that it is what they divide by too.

Part-days, and the projection that is really a best case

Half-days work. The input boxes are marked numeric but the parser accepts decimals and a comma as the decimal mark, so 59.5 and "59,5" both go in and both return 74.4 %. That is enough to model an afternoon here and a morning there without converting everything to sessions. It is also the only concession the tool makes to reality: there is no field for authorised versus unauthorised, none for lateness, and no way to weight a missed practical differently from a missed lecture.

The optional fourth box, "total term classes", produces a line labelled "projected final attendance", and the label is generous. Enter 60 of 80 so far with 120 in the term and it returns 83.3 %. That figure assumes you attend every single one of the remaining forty. It is not a projection from your current rate — a projection at 75 % would land you back at 75 % — it is the ceiling, the number you get if nothing goes wrong again. Read it as "the best I can still do" and it is genuinely useful; read it as "where I am heading" and it will disappoint you in nine weeks.

The Bradford Factor: what it is for, and what it is not

A percentage treats every absent day the same. The Bradford Factor deliberately does not: it multiplies the number of separate absence spells, squared, by the total number of days absent, usually over a rolling twelve months. Ten days off in a single stretch scores 1 × 1 × 10 = 10. The same ten days taken as ten separate single days scores 10 × 10 × 10 = 1,000, a hundred times higher. Five two-day absences score 250; two five-day absences score 40. The purpose is explicit: to make frequent short absences visible in a way a percentage never will, on the argument that unplanned single days are more disruptive to a rota than one long, notified absence.

Its statistical basis is weak, and that has to be said plainly. The formula is folk-attributed to the Bradford University School of Management in the 1980s, an attribution the university has never confirmed; no derivation of the squared term has been published, and there is no peer-reviewed evidence that using the score reduces absence or improves anything. The choice to square the spell count rather than cube it, or to weight it linearly, is not justified anywhere — it is a shape someone liked. That is a very thin foundation for a number that routinely triggers a formal meeting.

The predictable consequences follow. Conditions that produce recurring short absences — epilepsy, asthma, migraine, an autoimmune flare, chemotherapy cycles, a caring responsibility — score catastrophically against a single long illness with the same total days, which is a discrimination risk before it is a measurement problem. And because the score rewards not being absent rather than being well, a low trigger point teaches people to come in ill, which spreads what they have and produces exactly the absence the policy was meant to reduce. If a Bradford score is being quoted at you, the first question is not whether the arithmetic is right — it will be — but which spells were counted and whether any of them should have been excluded.

Frequently asked questions

Which denominator should I use?

The one your policy uses, which is almost always scheduled sessions — the ones you were required to attend — and almost never calendar days. Two checks settle it. Ask whether the count is in days or in half-days: a school that registers twice a day counts an afternoon off as half an absence, and a policy written in days will not. Then ask whether the denominator moves with your timetable: a part-timer measured against their own scheduled days will have a worse-looking rate than a full-timer with the same number of absent days, which is arithmetically correct and worth saying out loud before someone reads it as a difference in commitment.

I entered numbers and nothing appeared. What happened?

Almost certainly the first box is larger than the second. If classes attended exceed classes held, the component returns nothing and the whole result area vanishes without an error message. The same silence happens if the total is zero or either box is empty. Check the two figures against each other — the attended count can never exceed the held count — and the result will come back.

Why does it say I must attend 0 classes when my target is 100 %?

That is a bug, and the honest answer is the one the tool cannot give. The algebra that produces the number only runs when the target sits strictly between 0 and 100; at exactly 100 it is skipped and the counter stays at its initial value of zero, while the message that prints it runs anyway. The true answer is that 100 % is unreachable the moment you have missed one class, because past absences stay in the denominator forever. Set a target you can still reach — 95, or whatever your policy actually requires — and the number becomes meaningful again.

Can I enter half-days?

Yes. The boxes look like whole-number fields but the parser accepts decimals and treats a comma as a decimal mark, so 59.5 and "59,5" both work and both give 74.4 % out of 80. If more than a couple of your absences are part-days, the cleaner move is to switch the whole calculation to half-day units — double both numbers and count each morning and each afternoon separately, which is how English schools and the French absenteeism threshold both work. The percentage is identical when every absence is a whole day, so nothing is lost by switching.

Does the calculator work out a Bradford Factor?

No. It has four inputs — attended, held, target and an optional term total — and none of them is a count of separate absence spells, which is the piece the Bradford Factor needs. Work it out by hand: count the number of distinct occasions on which you were absent, square that, and multiply by the total days absent over the same rolling period. Ten single days score 1,000 against 10 for the same ten days taken in one stretch. Whether that difference should carry any consequences is a policy question, and one worth pressing, because the formula has no published derivation and no peer-reviewed evidence behind it.