What Is a Table of Specifications (Test Blueprint)? How to Build One
Build a table of specifications with checked topic weights, cognitive levels and marks. Use a 50-mark example and understand what a blueprint cannot prove.
A table of specifications, also called a test blueprint, maps the content and thinking you intend to assess to question counts or marks. It makes your coverage plan visible before you assemble a paper. It supports review, but does not by itself prove validity, reliability, fairness or equivalent difficulty between versions.
This guide explains how to build and check the grid, with an illustrative 50-mark Biology plan. Use the downloadable workbook above as a planning aid and check its output against your actual paper. Neither the example nor the workbook is an official board specification or a validated assessment instrument.
What belongs in a table of specifications?
List the relevant topics or outcomes in rows and the categories of thinking or performance in columns. Decide whether cells record questions or marks and label the unit clearly. Use marks when question values differ; use counts when that is the planning requirement. Do not mix the two without an explicit conversion.
The University of Michigan’s exam-design guidance recommends a blueprint to plan topic and skill coverage. The important step is comparing the completed paper with that plan, not treating a filled grid as a quality certificate.
A blueprint and an exam pattern are complementary. The blueprint plans the evidence and coverage; the pattern describes sections, formats, required counts, marks and instructions.
Build the grid in six steps
- Define the purpose and scope. Identify the intended score use, taught outcomes and any applicable qualification, subject and paper specification.
- Choose useful rows. Make topics specific enough to reveal gaps without creating distinctions the paper cannot meaningfully sample.
- Choose columns that fit the outcomes. Bloom’s categories can help, but no fixed set of four or six columns is required for every assessment.
- Allocate weights. Consider outcome importance, the specification and opportunities to learn. Teaching time can inform the decision; it is not the only rule.
- Convert weights into feasible marks or counts. Reconcile rounding, question values, internal choice and time constraints.
- Review the selected tasks. Confirm their actual demands, context and scoring guidance, then compare the final paper with the grid.
There is no universal twenty-minute completion time for this work. It may expose a missing outcome, unsuitable questions or an unrealistic paper length that needs further planning.
A checked 50-mark example
The following is an original illustrative Grade 9 Biology planning example, not a particular syllabus requirement. The cells contain marks, not question counts. Its categories and distribution should not be copied as a default for every Biology paper.
| Topic | Periods | Target weight | Remember | Understand | Apply | Analyze | Marks |
|---|---|---|---|---|---|---|---|
| Cell theory and cell types | 3 | 15% | 4 | 3 | 1 | 0 | 8 |
| Cell organelles and functions | 5 | 25% | 5 | 4 | 2 | 1 | 12 |
| Cell membrane and transport | 5 | 25% | 2 | 4 | 4 | 3 | 13 |
| Cell division | 4 | 20% | 3 | 3 | 2 | 2 | 10 |
| Microscopy and measurement | 3 | 15% | 1 | 1 | 3 | 2 | 7 |
| Total | 20 | 100% | 15 | 15 | 12 | 8 | 50 |
Every row and column reconciles to 50 marks. The column shares are 30%, 30%, 24% and 16%; these describe this example only. They are not a recommended universal cognitive distribution.
The target weights are also not the exact shares after rounding. The rounded topic allocations are 8, 12, 13, 10 and 7 marks, equivalent to 16%, 24%, 26%, 20% and 14% of the paper. Label this difference instead of claiming the final marks exactly match the original percentages.
Handle rounding and mixed marks explicitly
A 25% share of 50 marks is 12.5 marks. In this illustration the two such rows receive 12 and 13; the two 7.5-mark targets receive 8 and 7. The total remains 50. Record why you chose each adjustment and check that the resulting tasks still fit the intended emphasis.
A 13-mark allocation could contain thirteen 1-mark items or one 3-mark task and ten 1-mark items. Both total 13 marks, but they differ in count, coverage and response demand. Check the actual questions rather than inferring those properties from the total.
For a task assessing several outcomes, document how its marks are allocated without double-counting. For internal choice, distinguish printed marks from attempted marks and check the coverage of every permitted route.
Classify the thinking required, not just the verb
A question beginning “explain” is not automatically harder than one beginning “identify.” Its demand depends on the content, context, prior teaching and response required. Use Bloom’s question stems as prompts, then inspect the actual task.
Do not force every topic into every column. Some outcomes need a different assessment method, such as observed practical performance. A paper blueprint cannot replace evidence the paper does not elicit.
Use the blueprint to review alternate forms
Two papers can match the same grid while differing in reading load, difficulty, response time and access requirements. Items with matching topic and skill labels are not automatically interchangeable. A blueprint does not establish equivalent scores.
Compare the actual alternatives and each version’s key. For consequential comparisons, additional empirical evidence and specialist review may be needed. The alternate-forms guide explains this distinction; the moderation checklist helps record the review and unresolved actions.
Turn the plan into a paper in Examiar
Examiar organises questions and creates papers and answer keys. Use the blueprint as your planning and review document, then configure the requirements supported by the current app workflow. Do not assume an uploaded Excel grid automatically becomes a fully enforced generation specification.
Check the finished paper against the grid, marking guidance and accessibility checklist. Try Examiar free with a small, non-confidential sample before relying on a larger workflow.
Frequently asked questions
Does a table of specifications prove content validity?
No. It documents intended coverage and supports review. The actual tasks, scoring, administration and intended score use still need suitable evidence.
Should topic weights always follow teaching time?
No. Teaching time is one input. Consider the intended outcomes, their importance, the applicable specification and students’ opportunities to learn.
Should cells contain questions or marks?
Either can be useful if clearly labelled. When question values differ, marks show score weighting while counts show the number of tasks; track both when needed.