Before you print the exam: twelve checks

An annotated exam page: where each check sits

The figure shows the first page of a short Grade 6 test on ratio and proportion, written for this post to explain the checks — it is not a full test. The numbers are the twelve checks: eight you can see on the paper itself, and four behind it, in the blueprint and the marking scheme. Students never see those four, but they decide whether a student's mark says anything about what was learned.

The twelve checks, each with its reason in one line

Print one copy, pick up a pen, and tick each check the paper passes. Whatever fails can usually be fixed before the paper reaches the copier.

What each check protects

Content: the test measures only what you taught

Helenrose Fives and Nicole DiDonato-Barnes open “Classroom Test Construction: The Power of a Table of Specifications” (Practical Assessment, Research & Evaluation, February 2013) with students complaining about a test that asked about something covered for “like 2 seconds” in class. They write that what was discussed longer or in more detail should appear in greater proportion on the test: an objective that took 10% of instructional time should count for 10% of the assessment. They also write that the thinking level of the items needs to match the thinking level of instruction if teachers are to make valid judgments about their students' understanding. The unit test blueprint has a completed table for a 30-mark unit to run the first three checks against.

Clarity: students know the task before they start

The guide “Improving Your Test Questions”, from the Center for Innovation in Teaching & Learning at the University of Illinois, says to express a single idea in each item, to phrase each item so the student's task is clearly indicated, and to give each essay item a point value and an estimated time limit. For problem-solving items it asks that the directions state whether students must show their work for full or partial credit, and that the parts of an item be clearly separated with their point values.

Fairness: students get it wrong on the subject, not on the paper

The same guide says to use negatively stated stems sparingly and, when you do, to underline or capitalise the negative word. It says to avoid unfamiliar vocabulary, and to avoid sweeping words such as “always” and “never” in true-false items, because they let a test-wise but unprepared student answer correctly.

On colour, the W3C accessibility guidelines, WCAG 2.2, say colour should not be the only visual means of conveying information, and note that people using limited-colour or monochrome displays cannot get information carried by colour alone. A paper run through a black-and-white copier is in the same position — which is why Question 3 above tells boys from girls with fill and hatching, not colour.

On time, Fives and DiDonato-Barnes describe the ideal test as one students can complete in the time allotted, with enough time to brainstorm any writing and to check their answers before handing in. Our advice: solve the paper yourself, time it, and multiply your time by about three to estimate a student's.

Layout and marking: the same mark whoever marks it

The Illinois guide notes that different readers can give the same essay answer different marks, and that the same reader can judge the same paper differently over time. It describes analytical scoring — each answer compared with an ideal answer, and points given for the necessary elements — and recommends, when possible, asking another instructor to read and grade your students' answers, and working through each problem yourself before the test is given.

Page numbers and answer space are our advice, not a source's: an unnumbered back page can come back blank, and a cramped space hides the working you want to give marks for. The good worksheet: eight checks before you print draws the same space rule on a sheet before and after review.

In what order: from the blueprint to the copier

1. 1

   Start with the blueprint, not the paper

   Checks 1, 2 and 3. On your copy, write each question's objective and level beside it, then total the marks per objective and per level and compare with the table. No table? Write the unit's objectives on a sheet and start from there.

2. 2

   Solve the whole paper in pen, in its own boxes

   Checks 9 and 11. Time yourself. If the boxes cannot hold your working, they will not hold a student's.

3. 3

   Read every question as a slow reader would

   Checks 4, 5, 6 and 7. Is the task clear from the words alone? Does the question ask for one thing? Can you see the negative? Are the marks there?

4. 4

   Copy one page in black and white

   Checks 8 and 10. Did a drawing, a shade or a word disappear? Are the page number and “continued” in place?

5. 5

   Give a colleague two questions and their scheme

   Check 12. Pick the two longest questions and give them one student's answers, or an answer you write with a deliberate mistake. If their mark differs from yours, add to the scheme the line that was in your head.

What this checklist will not catch

The checklist examines the paper, not the quality of each question. A question can pass every check and still be easier than you meant, or have wrong options so far off that nobody picks them. Write wrong options from mistakes your students actually make, as in hinge questions, and raise a question's level the way depth-of-knowledge questions describe. If your school requires an official header or exam form, use it: these checks are run inside it, not instead of it.

Once the paper is printed, there is still review week. Seven exam review games gives you ways for every student in a class of 40 to recall what they learned, with no equipment.

Written with String's editorial team: we use AI for research and first drafts, and a person checks every fact and every source before anything is published. How we write