Three questions every result report should answer
After an online test, result analytics should help you answer three questions. How did the group do as a whole? Where are the class and each student weak? Which questions did not work as intended? A rank list on its own answers none of these. This article shows how to read each layer of the data and what to do with it.
It is meant for teachers, academic coordinators and coaching faculty who already run online tests and now want results to guide teaching, not just fill a report card. The examples use made-up round numbers to explain the ideas; they are not data from any institute.
Look at the spread of scores before the topper
The average score is the figure most people check first, but on its own it can mislead. Two classes can both average 50 out of 100: in one, most students scored between 40 and 60; in the other, half scored near 80 and half near 20. Those two classes need completely different responses from the teacher.
Group the scores into bands and count the students in each. An illustrative class of 40 students on a 100-mark test might look like this:
| Score band | Students | What it suggests |
|---|---|---|
| 81–100 | 4 | A small group ready for harder material. |
| 61–80 | 9 | Comfortable with most of the syllabus. |
| 41–60 | 15 | The middle of the class; topic-wise figures will show where marks were lost. |
| 21–40 | 8 | Gaps in basics; check whether the same topics keep recurring. |
| 0–20 | 4 | Needs individual attention; check attendance and whether they completed the test. |
Common shapes, and what they usually mean:
- Most scores bunched near the top: the paper was too easy to separate students, or questions had leaked.
- Most scores bunched near the bottom: the paper was too hard, the time was too short, or the topic had not been fully taught.
- Two separate clusters: two groups learned differently. Check whether the clusters match batches, teachers or medium of instruction.
- A long tail of very low scores: look at those students individually, including whether they faced technical problems during the test.
Also check the median, which is the score of the middle student. When a few very high or very low scores pull the average, the median gives a fairer picture of a typical student.
Marks, rank and percentile: what each one means
Online exam systems commonly report three figures for each student. They are related but answer different questions, and parents often confuse them.
- Marks: how much of this paper the student answered correctly, after any negative marking.
- Rank: the student's position within the group that took the paper. Rank 5 in a batch of 30 and rank 5 in a batch of 300 mean very different things.
- Percentile: roughly the percentage of the group that scored below the student. A 90th percentile means the student did better than about 90 percent of the group, which is not the same as scoring 90 marks.
Rank and percentile only make sense within the same paper and the same group. Comparing ranks across batches that took different papers, or across months when difficulty changed, gives a false picture of progress. To track one student over time, compare their topic-wise accuracy instead.
Topic-wise results: finding weak areas for the class and for each student
Topic-wise performance is where analytics starts to change teaching. It only works if each question is tagged by chapter and topic; if the tagging is patchy, so is the report. Our guide to building a question bank for online exams explains how to set tags up.
Class level
Sort topics by the percentage of correct answers across the class. The two or three weakest topics are candidates for reteaching. Before reteaching, confirm that the topic was actually covered before the test, and that the weak figure does not come from one confusing question rather than the whole topic.
Student level
For each student, compare their topic scores with the class figure for the same topic. A student who is below the class in one topic needs targeted practice. A student who is below the class in every topic needs a conversation about attendance, study habits or something happening outside class.
Accuracy and attempt rate
When a test has negative marking, two more figures help: how many questions a student attempted, and what share of those were correct (accuracy). A student who attempts few questions with high accuracy may be too cautious and leaving marks unearned. A student who attempts nearly everything with low accuracy is probably guessing and losing marks to negative marking. The advice for each is the opposite of the other.
Question difficulty and discrimination, explained simply
Item analysis means checking each question rather than each student. Two simple measures do most of the work.
Difficulty, sometimes called the facility value, is the share of candidates who answered the question correctly. If 32 out of 40 got it right, its value is 0.8, which marks an easy question. If 6 out of 40 got it right, the value is 0.15, which marks a hard one.
Discrimination tells you whether a question separates students who know the subject from those who do not. A simple way to estimate it: rank students by total score, take the top third and the bottom third, and compare the share of each group that got the question right. Subtract the bottom group's share from the top group's. A clearly positive result means the question is doing its job. A result near zero means it does not tell strong and weak students apart. A negative result means weaker students did better than stronger ones, which usually points to a fault in the question.
| Question | Top group correct | Bottom group correct | Reading |
|---|---|---|---|
| Q4 | 12 of 13 | 5 of 13 | Works well: strong students get it, weaker ones mostly do not. |
| Q9 | 13 of 13 | 13 of 13 | Too easy to tell anyone apart. Fine as a warm-up, not useful for ranking. |
| Q15 | 3 of 13 | 7 of 13 | Weaker students did better. Check the answer key and the wording. |
| Q21 | 1 of 13 | 0 of 13 | Almost nobody got it. Check whether it was taught, or whether the key is wrong. |
Look at the wrong options too. If one wrong option attracts most of the top group, it may actually be correct or at least arguable. If a wrong option is never chosen by anyone, it is not doing anything and should be rewritten.
Fix wrong answer keys and re-evaluate before releasing results. Changing marks after students have seen them causes far more trouble than a short delay.
Student and parent reports that lead to action
A report that shows only marks and rank tells a student where they stand but not what to do next. A useful individual report includes:
- Marks, rank and percentile within the batch, with the batch size shown so the rank has context.
- Topic-wise scores compared with the class figure, with the weakest topics highlighted.
- Attempt rate and accuracy wherever negative marking applies.
- Access to solutions or the answer key, so the student can see why an answer was wrong.
- Two or three specific next steps, such as a practice set on a named topic, chosen by the teacher.
Think about how rank is shown to younger students. A public rank list for a class of ten-year-olds can discourage more than it motivates, and many schools share only the student's own position. In coaching mock series for competitive exams, students usually expect ranks. Results and short summaries can be sent to students or parents by SMS or WhatsApp; our article on SMS use cases for schools and hospitals covers wording and timing.
A routine for acting on each test
Analytics only pays off if someone acts on it within days. A short routine after every test keeps the work manageable and makes sure findings reach the classroom.
| Finding | Action | Who |
|---|---|---|
| A question with negative discrimination, or almost no correct answers | Check the key and wording; correct and re-evaluate, or drop the question for everyone. | Subject teacher, before results are released |
| Two or three topics weak across the class | Reteach in the next few classes, then include fresh questions on them in the next test. | Subject teacher |
| A group of students weak in the same topic | A small remedial group or an extra practice set on that topic. | Class teacher or batch coordinator |
| A student with low scores across all topics | A one-to-one conversation; check attendance and, where relevant, involve parents. | Class teacher or counsellor |
| Questions that performed well | Mark them as reliable in the bank and correct their difficulty tag if needed. | Question bank coordinator |
Then compare the next test with this one, topic by topic, to see whether the reteaching worked. Over a term, this loop matters more than any single report.
Mistakes that make exam data misleading
Careful analysis can still lead to wrong decisions if the underlying data is distorted. Watch for these before drawing conclusions:
- Counting absentees as zero: this drags the average down and hides how those who sat the test actually did. Report absentees separately.
- Reading too much into small groups: in a batch of eight, one student moves every percentage. Treat small-group figures as hints.
- Comparing different papers as if they were equal: a fall in the average may simply mean a harder paper.
- Ignoring technical problems: a student who lost time to a frozen system should not be judged on that score without a note.
- Mixing unchecked subjective marks with objective scores: publish combined analytics only after subjective evaluation is complete and reviewed.
- Changing the marking scheme mid-term: negative marking in some tests and not others makes trends meaningless.
What MATROP's results module provides, and where a spreadsheet still helps
In MATROP's online exam software, objective questions are evaluated automatically when a test is submitted, and teachers mark subjective answers through a separate evaluation workflow. The results and analytics section shows marks, rank and percentile, section-wise analysis, accuracy and topic-wise performance, and results can be exported to PDF or Excel. Showing answer keys and solutions to students, and a re-evaluation process, are available as options. Results can also reach students or parents through SMS or WhatsApp notifications.
Discrimination and option-wise analysis are not on our standard module list, so ask during a demo what the current reports show. Check too whether the export includes question-wise responses; if it does, the calculations in this article can be done in a spreadsheet with simple formulas.
If you only need a marks list for a handful of tests, a spreadsheet may be all you need. A results module earns its place when you run frequent tests, need topic-wise reports for many students, or want reports to reach parents without manual work. Institutes preparing for their first online tests can start with our checklist for running a CBT.
Questions people ask
What difficulty mix should a graded test have?
There is no single correct mix, but a paper made mostly of medium questions, with some easy and a few hard ones, usually separates students best. Questions that almost everyone gets right, or almost everyone gets wrong, tell you little about differences between students.
How many students do we need for question analysis to be reliable?
There is no fixed cut-off, but in very small groups one student changes the figures a lot. For small batches, combine results from batches that took the same paper, or treat the figures as hints to check rather than conclusions.
Can subjective answers be analysed the same way as MCQs?
Partly. You can compare average marks per question and per topic, but discrimination is less exact because marking involves judgement. A shared marking scheme across evaluators makes the comparison more meaningful.
Should students see the full class rank list?
It depends on age and purpose. For younger students, showing only their own position and topic scores is often kinder and just as useful; competitive exam mock series usually publish ranks because students expect them.
