OLSAT common mistakes
OLSAT · Understand the traps and how to avoid them.
In antonym items, choosing a word that is merely unrelated to the prompt word (e.g., pairing 'generous' with 'money') instead of a true opposite, because the related word 'feels' connected.
Candidates scan for any word that co-occurs with the prompt in real life, and 'money' appears in contexts of generosity. Association masquerades as opposition, so the first familiar link gets selected.
What to do instead
Before looking at the options, silently generate your own opposite and its part of speech. Then test each option by inserting it into the frame 'X is the opposite of Y' and asking whether the two could both describe the same thing at the same time. If they could (a generous person can have money), it is not the antonym. Eliminate association-only options first, then choose among the remainder.
In sentence completion, picking an option that fits the blank grammatically and 'sounds right' without checking it against the logical cue words (because, although, therefore, instead) elsewhere in the sentence.
The blank is the most visually salient gap, so attention locks on it. Cue words are small function words that are easy to skim past, and a fluent-sounding option produces a false sense of correctness.
What to do instead
Read the sentence twice: first pass, cover the options and mark the cue word that sets the logical relationship (contrast, cause, result, example). Second pass, predict the blank's meaning from that cue before reading options. Only then evaluate options, and reject any that reverses the cue's logic even if it reads smoothly.
In sentence arrangement, ordering sentences by what 'sounds like a good opening' rather than by tracking pronoun and transition chains, so a sentence beginning with 'This process...' or 'However...' gets placed first.
Candidates judge flow by topic familiarity; the sentence that names the topic feels like the natural start. They do not systematically scan for anaphoric openers (this, these, it, such) that require a prior referent.
What to do instead
Use a two-step mechanical method: (1) find the sentence with no backward-pointing words and no contrast/result transition - that is the opener; (2) chain forward by matching each sentence's opening pronoun or transition to a noun or idea in the previous sentence. Any candidate order that puts an anaphoric opener first is eliminated immediately, regardless of how it reads.
In verbal analogies, matching on a single shared attribute (both are animals, both are large) and declaring the pair complete, instead of naming the precise relationship between the stem pair.
Surface category membership is the fastest thing to notice, and under time pressure the first available match is accepted. The bridge between stem words is never stated explicitly, so it cannot be tested against options.
What to do instead
Write a short bridge sentence for the stem in the form 'A is a ___ that ___ B' using a specific relationship (part-to-whole, tool-to-user, cause-to-effect, degree, function). Then apply that exact sentence to every option. If an option fits only a looser version of the bridge, reject it. Practise stating bridges aloud until it becomes automatic.
In classification (odd-one-out) items, choosing the option that is 'most different' in personal experience - for example, picking the only item the candidate has never seen - rather than the item that breaks a rule shared by all the others.
The task is framed as 'which does not belong', which invites subjective judgement. Candidates do not first identify the common property of the majority, so unfamiliarity or personal relevance drives the choice.
What to do instead
Always work in the positive direction first: state the rule that unites four of the five options (all are tools, all are mammals, all can be worn). The answer is the one that violates that stated rule. If two possible groupings exist, prefer the more specific, testable rule over a vague one, and say the rule to yourself before selecting.
In following-directions items, acting on each instruction as it is read (ticking boxes, drawing arrows immediately) instead of holding the full set of conditions and checking them together at the end.
Instructions arrive sequentially and each one seems actionable, so candidates execute in reading order. Later constraints then contradict earlier actions, and the candidate does not go back to re-verify.
What to do instead
Read all instructions once without writing anything, then encode the goal in one sentence ('place exactly two items, none adjacent'). Execute step by step, and after the final step run an explicit verification pass against every stated condition, including 'none', 'at least', and 'exactly' constraints. Budget the last few seconds of the item for that check.
In picture analogies and picture classification, choosing by overall visual similarity or colour matching instead of by the relational or categorical rule the pictures encode.
Pictures are processed holistically and fast; the visual system flags 'looks alike' before deliberate reasoning engages. Colour and shape are the most salient dimensions, so they dominate over function or category.
What to do instead
Convert each picture to a word or short phrase before comparing ('this is a hammer, used for driving nails'). Then apply the same bridge-sentence method as verbal analogies: state the relationship in the stem pair, then test options against that statement. Explicitly ignore colour and orientation unless the stem pair shows that colour or orientation is the rule being varied.
In figural series, extrapolating only the most obvious changing feature (size grows, shape rotates) while missing a second, independent rule such as the number of sides or the position of an internal element.
Series items typically combine two or more simultaneous rules, but attention fixates on whichever change is most noticeable. Once one pattern is found, the search stops - a classic satisficing error under time pressure.
What to do instead
For every series item, run a fixed checklist on each figure: shape type, size, rotation, number of elements, position of elements, shading/fill, and line count. Note what changes and what stays constant for each dimension across the sequence. Confirm the predicted next figure satisfies every changing dimension, not just the first one you spotted.
In figural analogies and figural classification, applying a transformation in the wrong direction or to the wrong element - e.g., rotating the answer figure when the stem rotated the first figure to produce the second.
The A-to-B relationship has a direction, and candidates often recall 'there was a rotation' without recording which figure was rotated and by how much. Mirror-image and rotation errors are especially common because both produce plausible-looking options.
What to do instead
Annotate the stem pair explicitly: write 'A changed by X to become B'. Then apply X to C and check the result against options. For rotation, note the direction and the amount (quarter turn clockwise) rather than just 'rotated'. Distinguish reflection from rotation by checking whether a distinctive asymmetric feature (a notch, a dot) ends up on the same side or the opposite side.
In number series and number analogies, searching for a single arithmetic difference and abandoning the item when it fails, rather than testing the standard family of rules in a fixed order (differences, ratios, alternating/interleaved sequences, squares/cubes, two-step operations).
The first rule taught is constant difference, so it becomes the default and its failure is read as 'I don't know this one'. Many series are interleaved (two alternating sequences) or use increasing differences, which a single-difference check cannot detect.
What to do instead
Apply a fixed hierarchy every time: (1) first differences; (2) if not constant, second differences or a repeating difference pattern; (3) ratios; (4) split into odd and even positions and test each as its own sequence; (5) compare to squares, cubes, and primes; (6) test a two-step rule (multiply then subtract). For number analogies, state the operation as a formula ('B = 3A − 1') and verify it on the stem pair before applying it to the target.