The diagnosis inside grading, and why it disappears

Work out your own number first. Sections per class, students per section, unit tests plus terms plus a pre-board. For a secondary science teacher with several sections, the annual total of answer sheets runs into the thousands, and it is the same thousands every year.

That workload is not really the story. Grading is part of the job. The more interesting thing is the second task running underneath it, the one nobody is trained for, timetabled for, or credited with.

Diagnosis.

The invisible job inside grading

Marking a paper is two jobs at once. The first is mechanical: read the answer, compare it to the marking scheme, assign the marks. Tedious, but straightforward.

The second is the real work. A student balances the equation correctly but treats an exothermic reaction as endothermic, and the teacher reading it concludes something quite precise: this student can balance equations and does not understand energy transfer. That is a diagnosis. It identifies the concept, the likely cause, and what should come next.

It happens silently, hundreds of times per exam, while also decoding handwriting. And the difficulty is not that teachers cannot do it. It is that nobody does it as well on the hundred and fiftieth paper as on the tenth, and by the last stretch the goal has quietly become finishing.

What happens to the diagnosis afterwards

Almost nothing. Papers go back. The class is told that many of them made mistakes in acid-base reactions and should revise Chapter 2. “Revise Chapter 2” is the educational equivalent of a doctor saying “you feel unwell, try being healthier.”

The specific insights stay in the teacher's head. There is no field anywhere for “this student confuses mass and weight”, and no time to invent one, because the next unit test is in three weeks and Chapter 5 is still untaught. A few hundred individual micro-assessments are performed and then discarded.

Why the spreadsheet does not rescue it

The standard attempt is a sheet with students down the side and concepts across the top, filled in after each test. It is a good instinct and it usually survives about one exam cycle.

So the sheet gets abandoned and the old method returns: grade, hand back, say revise Chapter 2, move on.

What concept-tagged grading changes

The change is upstream of the marking. Each question carries tags before the paper is set: the concept it tests, the capability it demands, and the context it is set in. Question 3 is not “4 marks”; it is balancing chemical equations, at the application level, in a laboratory context.

Then marking that question records more than a number. It records that this student scored three of four on balancing equations at application level, and where the marks went. Repeat that across every question, student and test, and the result is a per-concept profile rather than a percentage.

The difference is between “72% in Chemistry” and “has stoichiometry and bonding; loses energy-in-reactions; confuses physical and chemical change when it appears inside an application question”. The second is the diagnosis the teacher was already making. The only change is that it survives the week.

What that makes possible

Before class. Instead of leafing back through answer sheets trying to recall what stood out, the teacher can see which concept the section actually lost, and how many of those students also lack its prerequisite. That converts “Chapter 2 revision” into fifteen specific minutes on conservation of mass, aimed at the students who need it.

At a parent meeting. “He needs to study harder” becomes something a parent can act on: the formulas are secure in direct questions and fall apart when the same concept is buried in a paragraph, and here are the specific areas being worked on. That is a two-minute conversation that ends with the parent knowing what to do.

In a remediation period. One worksheet for everybody becomes three groups: those who need the prerequisite, those who have the prerequisite but not the concept, and those who have the concept but cannot apply it. Same room, same period, three different fifteen-minute activities.

The three tasks worth handing to software

1. Identifying which concept each student missed. Teachers do this in their heads, well at first and less reliably as the pile goes down. Tagging makes it consistent across the whole pile, because the tagging happened when the paper was written rather than at eleven at night.

2. Tracking repeat mistakes across tests. Whether a student missed the same concept in July is not something human memory answers reliably across thousands of papers. Stored concept-level results answer it immediately, including whether the prerequisite was missed earlier.

3. Generating the remediation list. Which students need practice on this concept, which need to go back one step first, and what to give each group. This list gets assembled mentally after most tests and forgotten within days.

Why a percentage cannot carry this

The grading system was built to produce one output: a number that goes on a report card. But two students on the same percentage can have opposite knowledge profiles, one strong on structure and weak on organic, the other the reverse. Their remediation needs are entirely different and the report card makes them look identical.

A teacher can see that difference in the answers. The marks register, the report card and the parent portal then collapse it into a single figure. Concept-tagged assessment keeps the distinction, because it stores the diagnosis rather than only the score.

What to take from this if you teach

The most valuable thing happening while you grade is not the marking. It is the diagnosis. It is worth more than the grade, and if it lives only in your head it is gone by Thursday, and your students get the same generic instruction everyone gets: study harder, revise the chapter, do more practice.

So the thing worth changing is not how fast you mark. It is whether every question is linked to a concept and every result is stored against that concept. That can be a platform like Unio's assessments, or a deliberately small version you run yourself for one subject and one class.

The goal is modest and worth stating plainly: that a year of papers should produce something more durable than a year of percentages.