NCT common mistakes
NCT · Understand the traps and how to avoid them.
Reading a question about a specific neuropsychological test and answering from its general reputation (e.g., 'WCST measures executive function') rather than from the exact score pattern or index definition described in the item.
Candidates memorize test labels and broad constructs during study but never rehearse the operational details - which index is compared to which, what a discrepancy means, and how impairment cutoffs are defined. Under time pressure the label surfaces first and feels sufficient.
What to do instead
For each major instrument on your study list, build a one-page operational card: full name, what it actually measures, the specific indices/subscales, the direction of better performance on each, and typical cutoff or discrepancy rules. Before answering, locate the exact index named in the item and confirm the direction of the finding.
Answering a lesion-localization or syndrome question by matching a single keyword (e.g., 'confabulation' or 'apraxia') to a brain region without checking whether the rest of the described profile is consistent with that localization.
Keyword-to-region associations are easy to memorize and are reinforced by flashcard study, so they dominate retrieval. Items are usually written with a full clinical picture, and one keyword is deliberately compatible with more than one option.
What to do instead
Practice reading the entire vignette and listing all described signs before looking at options. For localization items, verify that every sign listed fits the chosen region and that at least one sign would be unexpected there. Build lesion profiles as clusters of signs, not single associations.
Treating base-rate and pre-test probability information as background and answering a diagnostic-accuracy item using only sensitivity and specificity figures.
Sensitivity and specificity are the numbers most heavily drilled, while prevalence and predictive values are treated as secondary. Candidates also confuse positive predictive value with sensitivity because both sound like 'how often the test is right.'
What to do instead
Before answering any accuracy item, write down which of the four quantities is being asked for (sensitivity, specificity, PPV, NPV) and whether prevalence is given. Remember that PPV and NPV move with base rate while sensitivity and specificity do not, and rehearse a 2x2 table construction until it is automatic.
Choosing the most comprehensive-sounding test battery or the most prestigious instrument for a referral question instead of the measure whose psychometric properties and format actually fit the described patient.
Candidates equate thoroughness with correctness and worry that a brief or targeted measure looks like an incomplete answer. They also overlook practical constraints in the vignette such as sensory limits, language, fatigue, or the specific decision the referral requires.
What to do instead
For every test-selection item, first state the referral question in one sentence and the patient's limiting factors (vision, hearing, motor, language, stamina, culture/language of test). Eliminate options that cannot answer the question or cannot be validly administered, then choose among what remains on psychometric grounds.
Applying a statistical significance rule mechanically to a single-subject or small-n clinical question, concluding 'no change' because a score difference is not statistically significant, when the item is asking about clinically meaningful or reliable change.
Coursework emphasizes null hypothesis significance testing, so candidates default to p-values. They conflate reliable change (a psychometric property of the measure) with statistical significance in a group study, and ignore the minimal clinically important difference.
What to do instead
Separate three questions explicitly: is the change reliable (RCI or similar), is it statistically significant in a group sense, and is it clinically meaningful for this person. Practice computing or interpreting a reliable change index and identifying when a test's standard error of measurement makes a small difference uninterpretable.
On research-design items, accepting a study's stated conclusion without checking whether the design supports a causal claim - for example, treating a cross-sectional correlation or a pre-post single-group comparison as evidence of treatment effect.
Vignettes are written to sound authoritative, and candidates read for content rather than for design features. Time pressure encourages matching the conclusion to the topic rather than auditing the comparison group, randomization, and blinding.
What to do instead
Adopt a fixed three-question audit for every design item: what was compared to what, was assignment random, and who was blinded. Name the design (RCT, cohort, case-control, cross-sectional, single-case) before reading the options, and reject any option that claims more causal strength than that design permits.
Ignoring the direction and magnitude of a stated effect or score when selecting an interpretation, so an option describing 'significantly worse than average' is chosen for a score that is actually within normal limits or above average.
Candidates scan for the construct named in the stem and stop reading the numbers. Percentile ranks, scaled scores, and T-scores use different metrics, and mixing them up reverses the meaning of the same underlying performance.
What to do instead
Convert every reported score to a common metric (percentile or T-score) before evaluating options, and note whether higher or lower is better for that specific instrument. Practice a mixed set of conversions until T-scores, scaled scores, and percentiles can be compared at a glance.
Answering an ethics or professional-practice item with the option that seems most helpful or compassionate (e.g., releasing records to a family member who requests them, or continuing testing past informed consent limits to finish the battery) rather than the option that follows the actual standard.
Clinical instincts toward beneficence and avoiding conflict with families pull against procedural rules. Candidates also half-remember standards and pick the closest-sounding phrase rather than the one that matches the situation.
What to do instead
For ethics items, identify the specific obligation in play (informed consent, confidentiality and its limits, competence boundaries, multiple relationships, test security, release of records) and choose the option that satisfies that obligation while using the least intrusive step first. Review the actual wording of the relevant standards rather than summaries.
Rushing the last items or leaving them blank when the exam does not penalize wrong answers, or conversely changing correct first answers on a review pass without a specific reason.
Fatigue and clock anxiety near the end of a long exam, combined with a belief that unanswered items are safer. Answer-changing is driven by anxiety rather than by newly recalled evidence.
What to do instead
Budget time so the final minutes are reserved for ensuring every item has an answer. Change an answer only when you can articulate a concrete reason (a misread word, a recalled fact, a re-derived calculation); otherwise keep the first response, which is more often correct.
Studying by rereading notes and recognizing terms, then discovering on exam items that recognition does not support applying the concept to a novel vignette.
Recognition feels like mastery, so study sessions end without retrieval practice. Exam items require transferring a concept to an unfamiliar clinical scenario, which is a different skill from identifying a definition.
What to do instead
Switch to active retrieval: closed-book recall of test properties, lesion profiles, and design features, followed by writing your own short vignettes and answering them. Space the review and mix topics within a session so you must select the relevant framework rather than knowing which topic is being tested.