Study the GRT by item family, not by generic aptitude drilling. Name the relationship in verbal items, set up the full arithmetic chain before computing in numerical items, and identify the transformation rule in figure series. Then practice mixed timed sets so moving between families becomes automatic, and measure readiness with a rubric based on process, not just score.
How the GRT's Reasoning Item Families Differ
The GRT is used in federal public service graduate recruitment, and graduate recruitment reasoning assessments are commonly organized by item family: verbal, numerical, and figural or spatial. Treat this as a practice framework; the PSC publishes the official description of the assessment's content.
Verbal items test relationships between words and ideas: analogies, word meaning in context, and logical deductions from written statements. The operation is identifying the structure that links two terms and mapping that same structure onto a second pair. Vocabulary alone is not enough; an item can use everyday words and still hinge entirely on whether you read the relationship correctly rather than the surface topic.
Numerical and figural items test different operations again. Numerical items require setting up and composing small calculations, often percent change, ratios, or series rules, where the setup matters more than arithmetic speed. Figural items require detecting one transformation rule, such as rotation or element count, that generates a sequence of shapes. Because the operations differ, diagnosing the family before answering is the first trainable skill. Check the issuer's official material for the exact composition of your sitting, and use the families here as the organizing frame for practice.
Verbal Analogies: Naming the Relationship Before Choosing
For analogy items, state the relationship as a short sentence before looking at the options. Then test each option in that same sentence. Choosing on surface association instead of relationship structure is the trap this section is built to expose.
Scenario: 'Chapter is to book as scene is to ___' with options including 'theatre', 'act', and 'script'. A surface-association reading links 'scene' with 'theatre' because scenes happen in theatres. But the relationship in the stem is part-to-whole: a chapter is a component of a book. Testing each option in the sentence 'a scene is a part of a ___' eliminates 'theatre' and lands on 'act'. The better decision is to name the relationship first and verify it in both directions.
This matters because analogy stems recycle a small set of relationship structures: part-to-whole, cause-to-effect, tool-to-function, degree of intensity, category-to-example, and object-to-characteristic user. Learning to label these structures converts an open-ended puzzle into a matching task. In practice, write the relationship sentence above the stem in your margin style, even mentally, before reading the options. If two options both fit, the stem's structure is probably more specific than your first reading, and you should tighten the sentence rather than guess between them.
Percent Change and Ratio Items: Composing the Operations
Numerical reasoning items usually require composing two or more operations in the correct order. Convert every percentage into a multiplier and every ratio into a count model before you calculate, and check whether the final quantity is a total, a share, or a change.
Scenario: 'A salary rises 20 percent, then falls 20 percent. Is the final salary higher, lower, or equal to the original?' A plausible mistake is cancelling the changes and answering 'equal'. The better decision is to compose multipliers: 1.20 multiplied by 0.80 equals 0.96, so the final salary is 4 percent lower. The order of the changes does not matter here; what matters is that percentages act on different bases, so they multiply rather than cancel. The same multiplier habit handles successive discounts, population changes, and fee markups.
For ratio items, the useful move is converting the ratio into a total-parts model. If roles are split 3 to 5 and you know one group's headcount, divide by the number of parts for that group to get the value of one part, then scale. A common slip is dividing by the total number of parts when the question gives a single group's quantity, which produces a consistent but wrong fraction. Practise writing, in one line, what 'one part equals' before answering; that single habit exposes most setup errors while the arithmetic itself stays simple.
Figure Series: Rotation, Count, and Shading Rules
Figure series items are governed by one or two explicit transformations. Check rotation and reflection, then element count, then shading or fill changes, then position movement. Two rules can combine, so verify each candidate rule across every frame, not just the last two.
Rotation and reflection are the first distinction to master because they look similar under time pressure. Rotate an arrow shape mentally through fixed steps, such as quarter turns per frame; if the sequence alternates between a shape and its mirror image, the rule is reflection, not rotation, and the direction of the flip matters. A practical check is to trace one distinctive feature, like a filled corner, and follow only that feature across the frames rather than trying to rotate the whole figure at once.
Count and position rules often run in parallel with orientation rules: the number of dots may increase by one while the whole figure turns. This combination is where a plausible mistake occurs: you confirm the count rule on the last two frames and answer, while the orientation rule predicts a different option. The better decision is to state both rules verbally, for example 'count rises by one and the figure rotates a quarter turn clockwise', and confirm each rule on at least three frames. If no single rule fits all frames, look for alternating rules applied to odd and even positions.
Choosing What to Drill: A Family-by-Family Comparison
Each reasoning item family rewards a specific first move, and each has a characteristic trap that comes from borrowing another family's habits. Use this table to decide which family to drill next, based on which first move you currently skip.
Read the table by column when diagnosing a practice set: the 'first move' column is the habit to build, and the 'characteristic trap' column tells you what a wrong answer probably came from. If your errors cluster in one family, drill that family's first move in isolation before returning to mixed sets. If your errors scatter across families, the switching itself, not any single content type, is the gap to train.
Keep the table beside you during untimed review, but put it away during timed practice. The goal is that the first move becomes automatic enough that you no longer need the reminder. When you review errors, classify each one against this table: a trap-driven error means the first move was skipped, while a careless-arithmetic error means the setup was right and only the execution slipped. These call for different fixes.
| Item family | Core operation | First move | Characteristic trap |
|---|---|---|---|
| Verbal analogy | Identify and map a relationship structure | State the relationship as a sentence before reading options | Matching surface association instead of structure |
| Written logic / deduction | Derive only what the statements entail | Translate each statement into symbols, then test conclusions | Accepting a conclusion that is plausible but not entailed |
| Numerical word problem | Compose ordered operations on correct bases | Convert percentages to multipliers; write 'one part equals' for ratios | Cancelling successive percent changes; wrong ratio base |
| Number and letter series | Find the generating rule | Check differences, then ratios, then alternating patterns | Fitting a rule to only the last two terms |
| Figure series / spatial | Detect one or two transformation rules | Trace one distinctive feature across all frames | Verifying a rule on two frames while a second rule contradicts it |
A Timed Mixed-Set Exercise with a Self-Check Rubric
Build a twelve-item mixed set: four verbal, four numerical, four figure series, and attempt it in ten minutes. Score with the rubric below. Process observations, not the raw count, tell you what to change before the next set.
For the numerical items, use a scenario like this: a team of 40 is split in a 3-to-5 ratio, and the larger group then grows by 25 percent. Setup: five parts equals 25 people, so one part equals 5, the smaller group is 15, and 15 multiplied by 1.25 is 18.75, or 19 when a whole-person answer is expected. The plausible mistake is computing 25 increased by 25 percent, which answers a different question. The exercise trains reading what quantity is actually asked for before any arithmetic.
Self-check rubric: give yourself one point per item for a correct answer, and one process point each time you (a) wrote the relationship sentence on a verbal item, (b) wrote the multiplier or one-part value before calculating, and (c) named the figure-series rule aloud before choosing. Observations to record: which family consumed the most time, whether skipped items were verbal or figural, and whether process points correlated with correct answers. These are learning milestones for tracking progress across sets, not a prediction of your official result. Aim to hold process points steady as the timer tightens.
A Six-Week Preparation Sequence and Readiness Checks
Sequence practice in three phases: two weeks diagnosing families untimed, two weeks drilling the weakest family's first move, and two weeks of mixed timed sets. Finish by running the readiness checks below; the Public Service Commission of Canada's website holds administrative details and self-assessment resources for candidates.
Weeks one and two: take one short untimed set per family and classify every error using the comparison table. Weeks three and four: drill only the weakest family, focusing on its first move, in ten-minute blocks, and re-test it untimed at the end of each block to isolate speed from accuracy. Weeks five and six: run two mixed timed sets per week using the exercise format above. In parallel, use the self-assessment tests the Public Service Commission publishes for candidates, since practising on issuer-provided material keeps your expectations calibrated. One short note: question counts, timing, and booking logistics belong to the issuer, so confirm them on the PSC website rather than from secondary sources.
Readiness checks before test day: you can state a relationship sentence for an analogy within a few seconds of reading the stem; you write multipliers and one-part values before calculating on every numerical item; you can name a figure-series rule covering all frames or consciously flag 'alternating rules' instead of guessing; and across your last two mixed sets your process points stayed high even where answers were wrong, meaning errors came from reasoning, not skipped habits. If any check fails, return to that family's isolated drill rather than adding volume. Keep practice sessions honest and self-contained, and never rely on recalled live test content, which undermines the merit-based assessment the Commission safeguards.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.