NNAT3 Level F common mistakes
NNAT3 Level F · Understand the traps and how to avoid them.
On Pattern Completion items, committing to the first option that matches the visible pattern in the exposed portion of the matrix and never checking whether it stays consistent across the whole figure.
Working memory holds only the most recent feature inspected, so once one attribute (e.g., stripe direction) matches, the brain treats the search as finished; Level F matrices add layered attributes (number, orientation, shading, position) that make partial matches easy to find.
What to do instead
Adopt a fixed attribute checklist (count, orientation, fill, position, symmetry) and verify every option against all five before choosing; explicitly name the attribute that eliminates each rejected option.
On Serial Reasoning, extrapolating a sequence by continuing the visible step between the first two or three figures instead of testing whether the step itself changes (constant, increasing, or alternating).
The first visible difference is the most salient cue and pattern-seeking intuition treats it as the rule; Level F sequences frequently use a step that doubles, alternates direction, or resets every third frame.
What to do instead
Before predicting the next frame, compute the step between frames 1-2 and 2-3 separately and compare them; if they differ, classify the progression as arithmetic, geometric, or alternating before choosing.
On Analogies, treating the A:B relationship as a property match (both are triangles) rather than a transformation (rotation, reflection, size change, element addition or removal), then picking C:D that merely looks similar to C.
Perceptual similarity is processed faster than relational reasoning, so surface resemblance wins under time pressure; the distractor set is built to reward exactly this shortcut.
What to do instead
State the A-to-B transformation aloud as a verb phrase (e.g., 'rotate 90 degrees clockwise and remove one dot') and apply that phrase mechanically to C before scanning options.
On Spatial Visualization items, mentally rotating the whole figure as a rigid block when the task requires folding, unfolding, or reflecting part of it, producing an answer that is a valid rotation but an invalid fold.
Whole-object rotation is the practiced default from earlier levels; folding tasks demand tracking which edges meet and which faces become hidden, a distinct operation that rotation intuition does not supply.
What to do instead
For fold items, mark one reference edge and trace where it lands after each fold; for reflection items, check chirality (handedness) explicitly, since reflection reverses it and rotation does not.
Spending equal time on every item, so a single hard matrix consumes two or three minutes and the last items are answered in a rushed final minute.
Level F items are not ordered by difficulty within a section, and the desire to 'not leave a blank' makes candidates grind; the exam is strictly timed and unanswered items score the same as wrong ones.
What to do instead
Set a per-item ceiling (roughly 45-60 seconds), flag and move on when it is hit, and return only after all items have a first-pass answer; practice with a visible timer to calibrate the ceiling.
Changing a first answer on review without a concrete, articulable reason, especially on Pattern Completion and Analogies items.
Anxiety near the end of a timed section produces doubt-driven switching; first answers on nonverbal items are often produced by accurate implicit pattern recognition that verbal second-guessing overrides.
What to do instead
Only revise when a specific attribute violation can be named (e.g., 'option C has four sides, the rule needs three'); otherwise keep the original response and log revisions during practice to see whether they help or hurt.
Ignoring the answer-option set and deriving the answer purely from the stem, then choosing the closest available option even when it violates a rule.
Candidates who learned to 'solve first, then look' carry that habit over, but on nonverbal reasoning the options are part of the problem: elimination by attribute is often faster and more reliable than full construction.
What to do instead
After forming a hypothesis, immediately test each option against it and against the checklist; if no option fits, the hypothesis is wrong and must be revised rather than approximated.
Answering items in the order presented and skipping the opportunity to bank quicker Serial Reasoning items before slower Spatial Visualization ones.
Sections mix item types, and candidates treat the section as a single homogeneous block; without a triage plan, hard spatial items early in the section eat the time budget for easier items later.
What to do instead
In the first 10 seconds of each item, classify it as fast (analogy, simple series) or slow (fold, complex matrix) and do a rapid first pass on fast items before returning to slow ones, subject to the platform's navigation rules.
Failing to transfer a rule across item types, e.g., not recognizing that the 'number of sides increases by one' rule seen in a Serial Reasoning item also governs an Analogies or Pattern Completion item in the same section.
Rules are encoded with the surface context in which they were learned, so candidates store 'the triangle series rule' rather than the abstract progression, and do not retrieve it when the surface changes.
What to do instead
After each practice item, write the rule in abstract form ('sides +1', 'fill alternates', 'element moves clockwise one position') in a running rule log and review it before the next practice set.
Practicing only full-length timed tests in the final week and never reviewing wrong items to identify which specific attribute (rotation direction, count, shading) was misread.
Timed practice feels productive and reduces test anxiety, but without error analysis the same misread recurs; candidates remember the score, not the failure pattern.
What to do instead
Split practice time roughly 60/40 between timed sets and untimed error review; for every miss, record the item type, the attribute misjudged, and the correct rule, then re-attempt similar items a day later.