Skip to content

Reports

See who understood and who slipped, before the next lesson

One paper is enough: a map of the class skill by skill, the wrong idea your students share, and a written report for every one of them.

One student's report headed «Karam al-Zoubi»: a switch between the teacher's version and the parent's, a result of 18 out of 20 at 90% described as «excellent», three tiles for the time taken, his own average this term and the class average, a distribution strip with a marker where his paper falls, a written paragraph summarising how he did, then a list headed «Skills, weakest first» carrying for each skill its name, its curriculum code, a bar of the marks it won, a state, and how many questions that state rests on.Sample data
The engine is real and the student in it is generated: this name and these marks were made for the screenshot, and a real student's work does not go on a public page. Notice what the report does when the evidence is thin — a full mark resting on one question still reads «not enough evidence».

By the numbers

  • 12

    sections in one student's report

  • 13

    sections in the class report

  • 5

    reasons a lost mark is filed under

  • 9

    places it says less rather than guess

What you know before you walk into the next lesson

Four things a column of marks will not tell you, however long you stare at it.

  • The whole class, one skill at a time

    A row per student, a column per skill, and four states in every cell: mastered, partly there, not yet, or not enough evidence. Each state has a shape and a name and not only a colour, because this gets printed in black and white.

  • The wrong answer they agreed on

    Students are grouped by the wrong option they picked, not by what they scored. Then it states the idea rather than the statistic — «15 students are reading the hollow circle as a filled one», never «15 picked option B», because nobody can teach an option letter.

  • The question that runs against your paper

    Every question carries a difficulty and a discrimination figure. When the students who did best on the rest of the paper do worst on one question, it is handed to you to look at: it might be the key, it might be the wording, it might be something taught sideways. You know which. The report does not.

  • A written report for every student

    Twelve sections opening on the result and the short version, then the skills sorted weakest first: each carries its curriculum code, its marks out of the marks available, its state, and the number of questions that state rests on — and where the evidence is a single question it says «not enough evidence» even when the mark is full. Then where the marks went, the biggest blocker, question by question, and what happens next. It has a second version written for a parent, one switch away.

From the paper to Sunday's decision

Four steps, and not one of them is typing marks into a spreadsheet.

  1. 1

    The class sits the paper

    The exam you wrote, in the form you set it in.

  2. 2

    Closed questions mark themselves

    The ones that need a human are marked as waiting; you open the student's paper and set the mark yourself. Your mark always beats the automatic one.

  3. 3

    Both reports build from the papers themselves

    One for the class and one for each student, built from the answers rather than from a summary of them — which is why the report moves when you change a question or a mark.

  4. 4

    You decide what to reteach, and to whom

    Every section ends in something to do: a working group, a period spent whole on one skill, or a question to rewrite before the next paper.

About the class in the screenshot, and about your students' work

The engine above is built and running; the class inside it was generated for the picture, because a real class's marks do not belong on a page anyone can open. And your students' work does not train our models, is not sold, never reaches a data broker, and carries no advertising. Those are practices, written down here plainly — String holds no compliance certification of any kind, and will not display a badge it does not have.

The exam editor with the paper open in it: a masthead carrying the school's crest and name, the grade and the section, the title «Mathematics exam — calculus and statistics», a total of 28 marks and a time of 60 minutes, blank lines for the student's name and the date, and question 1 carrying a graph inside it — labelled axes, numbered ticks, and the region between the curve and the x-axis shaded and marked R.
This paper is where the report comes from — change a question on it and what it says about your class changes with it, including the skill that question was the only evidence for.

The report reads the paper you wrote

There is no second test, no parallel question bank, and no ready-made template your class has to be poured into. The exam you set is the source of every figure in both reports.

  • 23 question types in the system, and three of them — long writing, comparison and drawing — are never machine-marked at all; they wait for you to read them.
  • Difficulty, discrimination and the paper's reliability coefficient are computed over the machine-marked questions alone, so the reliability figure does not shift every time you mark a drawing.
  • And what the paper did not measure is named: a whole section lists the skills this exam never reached.

What the two reports say, and what they refuse to say

The third group is not a list of gaps. Those are decisions, taken deliberately, and they stay.

One student's report — 12 sections

  • The result
  • The short version
  • Skills, weakest first
  • Where the marks went
  • The full analysis
  • The real blocker
  • Question by question
  • How the paper was worked
  • Against the class
  • What changed
  • What happens next
  • About this report

The class report — 13 sections

  • The exam in one line
  • The short version
  • The skill map
  • What to reteach first
  • What the class has wrong
  • Where to work in groups
  • Worth a look
  • The paper itself
  • What the paper covered
  • The exam against the platform
  • How the class has moved
  • Your actions
  • About this report

Things this report will not do

  • It never gives a student a rank in the class or a percentile among their peers, and the distribution strip under the result shows how widely the class spread rather than where the student sits in it.
  • It never names another student in anything a student or a parent reads.
  • It never asserts cheating. A fast right answer is a question for you, not a verdict from it.
  • No group is named for an ability. Groups are named for what they need, and there is no «top group» in it.
  • It does not correct your answer key. It shows you the question and leaves the judgement to you.
  • It will not build a class report on fewer than 5 papers, and will not print a reliability figure for a paper under 8 questions. It says why in both cases.
  • And when the exam disagrees with what the platform recorded, the reading that embarrasses String — that our own bar is too loose — is put before the one that blames the class.

Common questions

No. There is no teacher score anywhere in String, and these reports do not compare one class or one teacher with another. A report opens from your paper, addresses you, and ends every section in a decision you make — that is the shape it was built in from the start.

Your students' work does not train our models, is not sold, is not handed to a data broker, and carries no advertising. String holds no compliance certification and displays no compliance logo — what we give you is stated practice you can hold us to, not a badge.

Closed questions are marked against the key you wrote, so they are exactly as accurate as your key. Long writing, comparison and drawing are not machine-marked at all: they are flagged as waiting, you set those marks yourself, and your mark overrides any automatic one. The item analysis runs over the machine-marked questions alone, so what the machine measured is never mixed with what you measured.

Yes, today. A report is built from papers your students sat inside String, and there is no door for typing in the marks from a paper you graded by hand. You can still print that same paper to A4 and hand it out — but what is marked on paper does not reach the report.

Every student report has a second version written for a parent, and you decide whether to share it. That version carries no curriculum codes, no mastery jargon, no rank and no other child's name, and its one chart is their own paper today against their own last one. The class report has no parent version at all, because it is a teacher's document.

Run your first report

One paper your class sits is enough to give you a map of it.

Start freeNo credit card