Clerical checking: study guide

Worked examples and pitfalls

Word-by-word or letter-by-letter: two correct answers, one rule: File these four names: Van Dyke, Vandenberg, Van Horn, Vance. Under word-by-word filing, each space-separated unit is compared in turn and 'nothing files before something', so the first unit 'Van' sorts ahead of both 'Vance' and 'Vandenberg'. The order is Van Dyke, Van Horn, Vance, Vandenberg. Under letter-by-letter filing, spaces are ignored entirely, so the keys are VANCE, VANDENBERG, VANDYKE, VANHORN, and the order is Vance, Vandenberg, Van Dyke, Van Horn. Both orders are correct filing; only one is correct for the system you are working in. Test items state the convention in the instructions, and the candidates who lose marks are almost always the ones applying whichever convention their previous employer used. The same fork appears with prefixes and punctuation: whether Mc and Mac interfile, whether St is treated as Saint, whether a hyphen counts as a space. Read the rule, restate it to yourself in one sentence, then apply it mechanically and do not let a name that 'obviously' belongs somewhere override it.

Alphanumeric codes: A-102 at the front or the back of the drawer: Sort the references A-7, A-12, A-70, A-102, B-3. Under natural numeric ordering, the way a person reads them, the answer is A-7, A-12, A-70, A-102, B-3. Under plain string ordering, the way most software sorts by default, each character is compared in turn, so '1' comes before '7' and the answer is A-102, A-12, A-7, A-70, B-3, with A-102 first rather than last. Both are defensible; a filing item is testing which one the stated system uses. This is also why serious record systems zero-pad their references: rewrite the set as A-007, A-012, A-070, A-102 and the string sort and the numeric sort agree, permanently. If you are ever asked to design or clean a reference scheme, pad to a fixed width and the whole class of problem disappears. In a test, the giveaway is a set that deliberately mixes one-, two- and three-digit suffixes; that mixture exists only to separate candidates who apply the stated rule from candidates who apply the intuitive one.

Coding to a key: every mark is at a boundary: A claims routing key: under 500 pounds with no injury reported goes to CS1; under 500 with injury goes to CI2; 500 to 4,999 with no injury goes to CS3; 500 to 4,999 with injury goes to CI4; 5,000 and over goes to CE5 regardless of injury. Now code five records. R-118 at 499.99, no injury: CS1, since 499.99 is under 500. R-119 at exactly 500.00, no injury: CS3, because 'under 500' excludes 500 itself and the second band starts there. R-120 at 4,999.00 with injury: CI4, since the band is inclusive at its top. R-121 at exactly 5,000.00 with no injury: CE5, because the top band is 'and over' and its 'regardless of injury' clause overrides the injury split that governs the lower bands. R-122 at 86.40 with injury: CI2. Four of those five decisions turn on a boundary, and that is not an accident, coding items are written so that the interior cases are trivial and every discriminating mark sits on an edge. Before coding anything, underline the boundary words: under, up to, and over, between, inclusive. Then decide, once, what each one does to the endpoint, and apply that decision to every record in the batch.

Duplicates that are not textually identical: Three rows arrive in a merge. Row 1: SMITH, JANE | 07/04/1988 | AB123456C. Row 2: Smith, Jane | 1988-04-07 | AB 123456 C. Row 3: SMITH, JANE | 04/07/1988 | AB123456C. Rows 1 and 2 are the same person: normalise case, strip the spaces from the reference, and convert the ISO date and they match exactly. Row 3 is the genuinely hard one. If the file is in day/month order it is a different date of birth and possibly a different person; if that row came from a system using month/day order it is the same record again. You cannot tell from the row itself, so the correct action is to send it to the exception queue with the ambiguity noted, not to merge it and not to discard it. This is the judgement that separates competent records work from the appearance of it: the goal is not to make every row disappear, it is to make every decision defensible. In test form the item usually asks 'how many distinct individuals are represented', and the answer is often given as a range or accompanied by a 'cannot determine' option for exactly this reason.

Where accuracy actually decays on a long batch: Run a self-measurement rather than trusting a number from anyone: take 200 record pairs, split them into four blocks of 50, and log errors per block. Most people find block 1 slightly worse than block 2, a warm-up cost, and then a rise across blocks 3 and 4 as vigilance falls. The reason to measure it is that the arithmetic of small percentages is brutal at volume. Checking 200 pairs at 98 percent accuracy passes 4 bad records; at 99.5 percent it passes 1. Scale that to a realistic month of 20,000 records and the same two accuracy rates mean 400 defects against 100: a four-fold difference in downstream rework from a 1.5 point difference that would look like noise on a single test. Once you know where your own curve turns, the intervention is cheap: a deliberate twenty-second break at that point, or splitting the batch so the hardest records fall in your strongest block. Practice sessions on this site are recorded on your own device, so building a block-by-block picture across several sessions costs nothing but the logging.

Setting an attempt rate from the scoring rule: A 120-item checking test with a 10-minute limit, scored as correct minus incorrect. Suppose practice has told you that you hold 92 percent at ten items a minute and 96 percent at seven and a half. Fast: 10 x 10 = 100 attempted, 92 correct and 8 wrong, score 84. Careful: 10 x 7.5 = 75 attempted, 72 correct and 3 wrong, score 69. Fast wins by 15. Now change the rule to correct minus three times incorrect: fast scores 92 - 24 = 68, careful scores 72 - 9 = 63, and the 15-point gap has shrunk to 5. Now suppose your real accuracy at ten a minute is 80 percent rather than 92. A gap most people do not discover until they measure it. Fast now gets 80 right and 20 wrong: under simple correct-minus-incorrect that is 60, already behind the careful strategy's 69, and under the triple penalty it is 80 - 60 = 20 against 63. Same test, same person, opposite advice, and the two deciding variables are the penalty multiplier, which the instructions hand you, and your own accuracy-at-speed, which only measurement gives you. Do not pick a pace from temperament. Measure two rates in practice, write both accuracy figures down, and do this arithmetic before the test rather than during it.

How to practise this skill

• Before the first record of any batch, write the filing or coding rule at the top of the page in your own words. The single largest source of clerical error is applying a remembered rule from a previous system.
• Underline the boundary words in a coding key (under, up to, and over, inclusive), and resolve each endpoint once. Interior cases carry almost no marks; the edges carry nearly all of them.
• Normalise before you compare: mentally strip case, spaces and punctuation from references, and restate dates in one fixed order. Half of apparent duplicates are formatting, and half of apparent non-duplicates are too.
• When a record is genuinely ambiguous, mark it and move on. Time spent resolving one unresolvable row is taken from thirty rows you could have done correctly, and a guessed merge is worse than a flagged one.
• Measure your accuracy at two different speeds in practice and write both numbers down. You cannot choose an attempt rate rationally without them, and the fast rate is almost never as accurate as it feels.
• Practise on batches long enough to hit your own fatigue point, not on ten-item samples. The construct is specifically about sustained accuracy, and a ten-item drill measures the part of the curve that was never in doubt.

Glossary

Word-by-word filing

An alphabetical convention that compares space-separated units in turn, with a shorter first unit filing before a longer one. Under it, Van Horn files before Vance.

Letter-by-letter filing

An alphabetical convention that ignores spaces and punctuation entirely, comparing the run of letters. Under it, Vance files before Van Dyke.

Natural sort

Ordering that reads embedded digit runs as numbers, so A-7 precedes A-12. Contrasted with string ordering, which compares characters one at a time and puts A-102 first.

Zero padding

Writing references to a fixed width by adding leading zeros (A-007 rather than A-7) so that string ordering and numeric ordering produce the same result. The standard structural fix for alphanumeric filing errors.

Primary and secondary sort key

The field sorted on first and the field used to break ties within it. A batch sorted by site then by surname will look wrong if the two keys are applied in the opposite order.

Canonicalisation

Converting records to a single standard form (one case, no stray whitespace, one date order, punctuation stripped from references) before any comparison, so that formatting differences stop masquerading as data differences.

Exception queue

The destination for records that cannot be resolved from the information available. Routing an ambiguous record here is a correct outcome; guessing it into a merge is not.

Boundary condition

The endpoint of a coded band, where wording such as under, up to or and over decides which side a value falls. Coding tests concentrate their discriminating items here.

Study this next

• Data checking and clerical accuracy skill hub
• Clerical accuracy practice in the Cognitive Skills Lab
• Error detection lesson
• Administrative and clerical entry suite
• Clerical checking suite
• Data checking and clerical accuracy lesson on Novus Learn

Where this material comes from

• All filing sets, reference codes, routing keys and records above were invented for this lesson. The two filing orders, the two sort orders and every score calculation in the attempt-rate example were worked through and checked by recomputation.
• Filing and sorting conventions cross-checked against standard public descriptions such as the Wikipedia articles 'Alphabetical order', 'Natural sort order' and 'Collation'. Terminology only; the examples are original.
• Novus Learn aptitude construct registry (catalog seed) for construct scope and suite mapping.
• Public educational framing only: not affiliated with any official exam board, publisher or employer, and no copyrighted test item is reproduced.

Worked examples and pitfalls

The four transcription error families: Source reference: INV-2024-08871. Four corrupted copies, one of each family. INV-2024-08571 is a substitution: a single character changed, 8 to 5. INV-2024-0871 is an omission: one character dropped, and the string is now a character short. INV-2024-088711 is a duplication: a character repeated, one character long. INV-2024-08817 is a transposition: the final 7 and 1 have swapped places, and crucially the string is the same length and contains exactly the same characters. That last property is why transposition is the hardest of the four to see and the most dangerous in practice. Anything that checks length catches omission and duplication. Anything that compares character sets or sums the digits catches substitution but not transposition, because 0+8+8+7+1 and 0+8+8+1+7 both come to 24. Reading the reference aloud catches substitution reliably and transposition poorly, because the ear is comparing sounds and both versions sound similar at speed. The only defences against transposition are a positional comparison and a weighted check digit.

A fixed scan order finds the one field that moved: Source record: Name Priyanka Ramaswamy / Account 4471-9026-3358 / Date of birth 14/03/1991 / Postcode SW1A 2AA / Phone 0161 496 0872. Copy: Name Priyanka Ramaswamy / Account 4471-9026-3358 / Date of birth 14/03/1991 / Postcode SW1A 2AA / Phone 0161 469 0872. The difference is in the phone field, where 496 has become 469: a transposition inside the middle block. Most candidates find it eventually; the ones who find it in eight seconds are the ones running a procedure. Always compare in the same field order, top to bottom, never skipping a field because it 'looks fine'. Compare long strings in the printed blocks rather than as a single run (4471, then 9026, then 3358) because short-term memory holds about four items and a twelve-digit run does not fit. Say the block silently, look, compare, move on. On a same-or-different item you may stop at the first mismatch; on a 'how many fields differ' item you must complete every field, and the instruction wording is what tells you which regime you are in. Getting this wrong in either direction costs marks: stopping early on a count item, or exhaustively checking a same-or-different item you had already resolved.

Modulus 11: why a weighted check digit catches a transposition: The ISBN-10 scheme multiplies the ten digits by descending weights 10, 9, 8 down to 1 and requires the total to be divisible by 11. Take 0-306-40615-2. Working left to right: 10x0 = 0, 9x3 = 27, 8x0 = 0, 7x6 = 42, 6x4 = 24, 5x0 = 0, 4x6 = 24, 3x1 = 3, 2x5 = 10, and the check digit 1x2 = 2. The sum is 0 + 27 + 0 + 42 + 24 + 0 + 24 + 3 + 10 + 2 = 132, and 132 = 12 x 11 exactly, so the number is valid. Now transpose the 1 and the 5 to give 0-306-40651-2. The digits are identical, so a plain digit sum is unchanged at 27 either way and would report no problem. The weighted sum, however, becomes 0 + 27 + 0 + 42 + 24 + 0 + 24 + 15 + 2 + 2 = 136, and 136 - 132 = 4, so it is not a multiple of 11 and the record is rejected. That is the entire reason check digits are weighted rather than plain: position has to matter, or the commonest human error passes straight through.

Luhn: the number that fails, and the one case it misses: The Luhn check, used on payment and many membership numbers, doubles every second digit counting from the right, subtracts 9 from any doubled result above 9, sums everything, and requires a total ending in zero. Take 4539 1488 0343 6467. The doubled positions contribute 3, 3, 8, 0, 7, 2, 6 and 8, totalling 37; the undoubled positions contribute 7, 4, 3, 3, 8, 4, 9 and 5, totalling 43. The grand total is 80, which ends in zero, so the number passes. Now transpose the last two digits to 4539 1488 0343 6476. The doubled contributions become 5, 3, 8, 0, 7, 2, 6, 8 = 39 and the undoubled become 6, 4, 3, 3, 8, 4, 9, 5 = 42, giving 81. Not a multiple of ten, so the mistyped number is rejected at the point of entry. Luhn catches every single-digit substitution and almost every adjacent transposition: with exactly one blind spot. Swapping an adjacent 0 and 9 leaves the total unchanged, because doubling 0 gives 0 and doubling 9 gives 18 which reduces to 9, so both digits contribute the same amount whether doubled or not. A form that validates with Luhn will happily accept 90 where you typed 09. Knowing the blind spot is the point: a check digit narrows the space of undetected errors, it never closes it.

Negative marking: 44 right can beat 46 right: Many clerical checking sections score correct minus incorrect, with omissions neutral. Under that rule, guessing between two remaining options has an expected value of 0.5 x (+1) + 0.5 x (-1) = 0, exactly neutral, and guessing blindly among four options has an expected value of 0.25 - 0.75 = -0.5, clearly negative. Concretely: on a 60-item section you attempt 48, get 44 right and 4 wrong, and score 40. A colleague attempts all 60, gets 46 right and 14 wrong, and scores 32. More correct answers, a lower score. Reverse the scoring rule to plain raw-correct with no penalty and the ordering flips: 46 beats 44 and leaving a blank becomes strictly irrational. Neither strategy is universally right, which is why the instruction screen is not optional reading. Find the sentence that says whether wrong answers are penalised, and if it is absent, assume raw scoring and answer everything.

Reconcile the batch: consistent is not the same as correct: An invoice lists three lines: 12 at 14.25 = 171.00; 7 at 33.60 = 235.20; 3 at 128.00 = 384.00. The stated subtotal is 709.20, VAT at 20 percent is stated as 141.84, and the total is stated as 851.04. Every downstream figure checks out against the one above it: 709.20 x 0.20 = 141.84 and 709.20 + 141.84 = 851.04. Nothing in the tax or total column is inconsistent. But recompute the lines: 171.00 + 235.20 + 384.00 = 790.20, not 709.20. The subtotal is a transposition of the correct figure, and because every later number was calculated from the wrong subtotal, the document is perfectly self-consistent and perfectly wrong. The true VAT should be 158.04 and the true total 948.24, a shortfall of 97.20. This is the single most valuable habit in the construct: never verify a total against the number printed above it, always recompute it from the underlying items. Internal consistency proves that one person did the arithmetic carefully; it proves nothing about whether they started from the right number.

How to practise this skill

• Drill transposition specifically. Generate pairs where the only difference is two adjacent characters swapped, because that is the family your eye is worst at and the family a naive check will not catch.
• Fix a scan order and never vary it: field by field, top to bottom, and within a long value block by block. Free-roaming comparison feels faster and reliably misses one field per record.
• Learn one check-digit scheme by hand, modulus 11 is the easiest, until you can run it in about twenty seconds. It converts a whole class of 'does this record look right' items from judgement into arithmetic.
• Read the scoring rule before the first item and decide your guessing policy then, not at minute seven when you are behind. Under penalty scoring, an omission is a legitimate answer; under raw scoring it is a wasted mark.
• When you are checking financial or tabular documents, recompute the lowest-level figures first and work upward. Errors introduced at a subtotal propagate perfectly and are invisible from below.
• Log the errors you miss, not the ones you find. A miss log after four sessions usually shows a personal signature (always the middle of long numeric strings, or always the second occurrence of a repeated field), and that signature is what to drill.

Glossary

Transposition error

Two adjacent characters exchanged, as in 496 typed for 469. Length and character content are unchanged, so length checks and plain digit sums both pass it; only positional comparison or a weighted check digit detects it.

Substitution error

One character replaced by another, as in 08571 for 08871. It is the family most reliably caught by reading a value aloud against the source, and by any check digit scheme.

Check digit

An extra digit computed from the others so that a corrupted record fails an arithmetic test. It detects errors; it never corrects them and never proves the record refers to the right person or item.

Modulus 11

A weighted check scheme in which digits are multiplied by descending position weights and the total must divide by 11. Used in ISBN-10 and several banking and national identifier formats.

Luhn algorithm

A modulus 10 check that doubles alternate digits from the right and requires the sum to end in zero. It catches all single-digit errors and most adjacent transpositions except an adjacent 0 and 9.

Miss and false alarm

A miss is a genuine discrepancy passed as correct; a false alarm is a difference flagged where none exists, usually a formatting variant such as a trailing space or a differently written date. Misses are the costlier of the two in records work, which is why procedures normalise formatting first and then bias toward escalating doubt.

Correction for guessing

A scoring rule that subtracts a fraction or multiple of the wrong answers from the correct ones, making blind guessing negative in expectation and turning omission into a rational choice.

Reconciliation

Recomputing an aggregate from its components rather than accepting the printed figure. It is the only check that catches an error introduced at summary level, where everything below and above it still agrees.

Study this next

• Error detection skill hub
• Error detection practice in the Cognitive Skills Lab
• Data checking and clerical accuracy lesson
• Clerical checking suite
• Copyediting and proofreading suite
• Error detection lesson on Novus Learn

Where this material comes from

• The ISBN-10 modulus 11 worked example (0-306-40615-2, weighted sum 132) and the Luhn worked example (4539 1488 0343 6467, sum 80) were computed by hand for this lesson and each was re-verified digit by digit, including the corrupted variants.
• Algorithm definitions cross-checked against the public specifications described in the Wikipedia articles 'International Standard Book Number' and 'Luhn algorithm', including the documented 09/90 transposition blind spot. Records, invoices and reference numbers above are invented.
• Novus Learn aptitude construct registry (catalog seed) for construct scope and suite mapping.
• Public educational framing only: not affiliated with any official exam board, publisher or employer, and no copyrighted test item is reproduced.

Worked examples and pitfalls

A cancellation count you can actually check: Count every 7 in this row: 4 7 1 7 7 3 9 7 2 8 7 5 7 6 0 7. Working left to right and tapping once per hit: hits at the second, fourth, fifth, eighth, eleventh, thirteenth and sixteenth positions. That is seven sevens. Two error modes produce nearly all the wrong answers. The first is the adjacent pair - the 7 7 at positions four and five gets counted once, because the eye takes a repeated character as one perceptual object. The second is losing the place after the 9, where the visually similar 9 and 7 force a moment of re-checking and the scan restarts a character early, producing an over-count of eight. The defence for both is a fixed scan path with a physical anchor: a fingertip or cursor moving one character at a time, never jumping back. Re-scanning to check is the thing that creates the double-count, so if you must verify, verify by counting a second time from the RIGHT-hand end and comparing totals, never by re-reading part of the row.

Hits, misses and false alarms: the eager candidate loses: A checking batch contains 300 records, 40 of which are genuinely faulty. Candidate A flags 46 records and 36 of them are truly faulty. Hits 36, misses 40 minus 36 equals 4, false alarms 46 minus 36 equals 10. Candidate B flags 38 and 34 are truly faulty: hits 34, misses 6, false alarms 4. On hit rate alone A wins, 36 out of 40 which is 90 percent against B's 34 out of 40 which is 85 percent. Now apply the scoring rule that most checking tasks actually use, where a false alarm cancels a hit: A scores 36 minus 10 equals 26, B scores 34 minus 4 equals 30. B wins by four despite catching two fewer faults. This is why 'flag anything that looks odd' is bad advice on a scored checking task, and why the first thing to read on a checking item is whether wrong flags are penalised. Your response criterion - how much evidence you demand before flagging - is adjustable, and it should be set from the scoring rule, not from your temperament.

The transposition that passes every gist check: Compare these two lines and decide whether they match. Invoice 4820-7391-06. Invoice 4820-7931-06. They do not: the middle group reads 7391 in the first and 7931 in the second, with the 3 and the 9 swapped. Transpositions are the most-missed error class in record checking for a structural reason - the character SET is identical, the length is identical, the first and last characters of the group are identical, so every fast check the visual system runs comes back clean. Substitutions and omissions change the character inventory and get caught; transpositions do not. Two habits raise the catch rate. Read digits in fixed groups of two rather than as a whole number, so 73-91 against 79-31 becomes a mismatch at the first group instead of a subtle difference somewhere in a four-digit blur. And check groups in a deliberately non-natural order - last group, first group, middle group - because reading left to right lets the confirmation you built at the start carry you through the middle, which is exactly where the swap is usually planted.

Conflict: why reading fights you: The classic demonstration is Stroop's, published in 1935: the word RED printed in blue ink, with the instruction to name the ink colour. Naming takes measurably longer, and errors go up, because reading a familiar word is automatic and cannot be switched off, so the automatic response has to be suppressed before the controlled one can be produced. The same conflict has a numeric version you can test on yourself in a second: how many characters are in the string 4 4 4? The answer is three, and the digit 4 pulls at you the entire way. In an assessment this appears wherever the salient feature and the asked-for feature come apart - a chart where the tallest bar is not the answer to the question, a form where the highlighted field is not the one being verified, a row where the bold total is not what the stem requested. The practical move is to name the target feature out loud before you look - 'ink colour', 'character count', 'the value for March' - because pre-loading the target biases selection before the automatic reading response gets a chance to win.

Where the errors actually appear in a 45-minute block: Errors in a long checking block are not spread evenly. Mackworth's 1948 clock-watching study established the pattern that gives the effect its name: detection declines over a prolonged watch, with the sharpest deterioration early rather than at the very end. Practically, on a 45-minute self-timed checking block, expect your per-minute error rate in minutes 20 to 45 to run visibly above minutes 1 to 20 even though nothing about the material changed, and expect the subjective sense of effort to lag the actual decline, so it will not feel like you are getting worse. Two things work against it. Break the block into three fifteen-minute segments with a ten-second reset between them - look away, unfocus, breathe out - which costs thirty seconds of a 45-minute block, about one percent of the time, and buys back more than that in caught errors. And score your practice by segment rather than as one number, because a single overall accuracy figure hides exactly the information you need: whether your problem is skill, which shows as flat error rate, or endurance, which shows as a rising one.

Two streams: alternate on a cadence, do not try to merge: A dispatch-style monitoring task: keep a running total of the numbers announced on channel one while watching channel two for the code word AMBER. Genuine simultaneity is not available - the two tasks compete for the same control resource - so the choice is not whether to alternate but whether to alternate deliberately or accidentally. Deliberate looks like this: fix the arithmetic to a rhythm, updating the total only at each announcement and holding a single number between updates, which frees the gaps for channel two. If the total is 34 and the next announcement is 7, you spend under a second reaching 41 and then you are free again. Accidental looks like re-deriving the running total from the beginning because you did not trust it, which locks up the whole window and is when AMBER goes past unheard. The measurable failure of divided attention is almost never a failure to hear the target; it is a failure to have any spare capacity at the moment it arrived. The related phenomenon worth knowing is inattentional blindness, illustrated by Simons and Chabris in 1999: an unexpected and perfectly visible event is missed entirely when attention is committed to a counting task.

How to practise this skill

• Fix your scan path before you start, and never re-scan backwards to double-check. Backward re-scanning is what produces both double-counts and lost places; if you need to verify a count, run it again from the other end and compare.
• Read the scoring rule before the first item. Whether false alarms are penalised changes the correct strategy completely, and a criterion set for the wrong rule costs more marks than slow reading does.
• Check numeric strings in fixed groups of two or three and in a deliberately shuffled group order. Transpositions are the errors that survive whole-string comparison, and they are the ones planted on purpose.
• Time your practice in segments and log accuracy per segment, not per session. A flat error rate means you need speed; a rising one means you need endurance and breaks, and the two problems have opposite fixes.
• Name the target feature out loud before each conflict item. Pre-loading the relevant dimension is the cheapest available defence against the salient-but-wrong answer.
• Practise with an actual distractor present rather than in silence. A test hall has movement, coughing and page-turning, and attention practice in perfect quiet trains a condition you will not meet.

Glossary

Selective attention

Prioritising one source or feature while suppressing competing ones. It is what fails when the visually salient element of a display captures the response instead of the requested one.

Sustained attention (vigilance)

Maintaining detection performance over a long, low-event period. It is the capacity that checking, monitoring and inspection tasks are really sampling.

Vigilance decrement

The decline in detection performance over a prolonged watch, typically steepest early in the block. Mackworth's 1948 clock test is the standard public reference.

Divided attention

Handling two streams that compete for control. Performance is best modelled as fast alternation with a switch cost, not as genuine parallelism.

Hit, miss, false alarm, correct rejection

The four outcomes of any detection decision: flagging a real fault, missing one, flagging a clean record, and correctly leaving a clean record alone. Any honest accuracy claim needs at least the first three.

Response criterion

How much evidence you require before flagging. Shifting it trades misses against false alarms without changing your underlying sensitivity, which is why it must be set from the scoring rule.

Stroop interference

The slowing that occurs when an automatic response, such as reading a word, conflicts with the requested response, such as naming its ink colour. Named for Stroop's 1935 experiments.

Inattentional blindness

Failing to notice a fully visible, unexpected event because attention is committed elsewhere - the effect Simons and Chabris demonstrated in 1999 with a counting task.

Study this next

• Attention and concentration skill hub
• Cognitive Skills Lab: attention practice
• Emergency dispatch and 911 communications suite
• Error detection lesson
• Perceptual speed lesson
• Attention and concentration lesson on Novus Learn

Where this material comes from

• Character strings, invoice comparisons, batch counts and monitoring scenarios above are written for Novus Learn. All figures are original and no published test item is reproduced.
• Terminology and directions of effect follow widely published work: Mackworth (1948) on the vigilance decrement, Stroop (1935) on response conflict, Simons and Chabris (1999) on inattentional blindness, and the standard hit/miss/false-alarm framing from signal detection theory. Concepts only; no result figures are attributed to those studies.
• Novus Learn aptitude construct registry (catalog seed) for construct scope and suite mapping.
• Public educational framing only - not affiliated with any official exam board, publisher or employer, and not a clinical, diagnostic or attention-disorder screening measure.

Worked examples and pitfalls

Chunk the string, stop at the first mismatch: Compare 4820 6193 5507 with 4820 6913 5507. Read them as three blocks, not as twelve digits. Block 1 matches, 4820 against 4820. Block 2 does not: 6193 against 6913, where the 1 and the 9 have swapped. You are done. On a same-or-different item there is no reason to check block 3, and checking it anyway costs you about a second on every item, which over eighty items is more than a minute of pure waste. The reason chunking works is capacity: a twelve-digit run exceeds what most people can hold for a comparison, so reading it whole forces you back to the source repeatedly, while a four-digit block fits in one go. The instruction wording decides whether early stopping is allowed. 'Are these identical?' permits it. 'How many of the five fields differ?' does not, and stopping early there converts a correct three into an incorrect one. Read the instruction once, classify the section, then run the matching procedure without re-deciding it per item.

Homoglyphs: 0 and O are where the time goes: Three pairs to judge. Nguyen, Thi-Hoa against Nguyen, Thi-Hoa: identical. Muller, Jorg against Mueller, Jorg: different, and the difference is in the surname spelling rather than anywhere in the given name, which is where most candidates look first. RN0O1I5 against RNO01I5: different, because positions three and four hold zero-then-letter-O in the first string and letter-O-then-zero in the second. That last pair is the killer, and it is deliberate. In most screen and print faces, a zero and a capital O differ by a few pixels, and the digit one, capital I and lower-case L can be visually identical. Production systems avoid these characters in generated references for exactly this reason; test writers include them for exactly the opposite one. The practical rule: whenever a string mixes letters and digits, drop your rate for that item, and use context to disambiguate. If a field is described as numeric, an O in it is a zero, and if it is a postcode you can often tell from the format which positions are letters. Guessing on a homoglyph is not a speed decision, it is a coin flip you can avoid by spending half a second.

Memorise the key, then never look at it again: A routing section supplies two small tables. Region: NW = 12, NE = 14, SW = 21, SE = 23, CE = 31. Priority: A = 1, B = 4, C = 7. Each item names a region and a priority and asks for the combined code, so 'SE, priority B' is 23-4 and 'NE, priority C' is 14-7. There are only eight mappings. If the section holds forty items in three minutes, you have 180 / 40 = 4.5 seconds per item, and a candidate who glances back at the key twice per item (once for the region, once for the priority) spends around two seconds of that on lookup alone, which is roughly eighty seconds of the section, or nearly half of it. Twenty seconds spent memorising eight pairs before the first item pays for itself about ten items in and then keeps paying. The general principle for speeded sections: any fixed information that appears in every item should be moved into memory as early as the format allows, because per-item costs multiply by the item count and one-off costs do not.

Cancellation drills and mirror-image distractors: Build your own cancellation drill rather than reusing a published one. Write twenty rows of forty characters drawn only from b, d, p and q, each carrying zero, one or two tick marks above or below. Define the target as 'b with exactly two marks' and cross out every occurrence, timed. What the drill teaches is feature overlap: a p with two marks is a vertical mirror of the target and a d with two marks is a horizontal mirror, and both take measurably longer to reject than an obviously different character would, because they share every feature except orientation. The same effect is why 6 and 9, and M and W, are expensive in real checking work. Two things improve with practice on this drill and neither is eyesight. The first is that you stop re-deriving the target on each row and start rejecting non-targets pre-attentively. The second is that you learn your own decay point, the row at which misses start appearing, which is the information you actually need in order to pace a real section.

Raster order on a scanning grid: A ten-by-ten grid of two-digit numbers, and the task is to mark every value ending in 7. Scan strictly left to right, top to bottom, one row at a time, keeping a finger or the cursor on the current row. The reason is not neatness. Free-roaming the eye over a hundred cells means some cells get visited twice and others not at all, and there is no way to know which is which, so you finish with no idea whether you are done. If you re-visit even fifteen percent of the grid, that is fifteen wasted fixations, and at roughly a quarter of a second each that is close to four seconds on one grid: often the difference between completing the section and leaving items blank. On a two-column layout, treat each column as its own raster and finish it before moving; jumping between columns doubles the re-visit rate. The same discipline transfers directly to real work: scanning a picking list, a passenger manifest or a stock count row by row is slower per glance and faster per document.

Correct per minute, not percent correct: Two candidates, four minutes each. A attempts 62, gets 60 right: 15.0 correct per minute at 96.8 percent accuracy. B attempts 96, gets 84 right: 21.0 correct per minute at 87.5 percent accuracy. Under raw scoring, count the correct answers, B wins, 84 to 60. Under correct minus incorrect, A scores 60 - 2 = 58 and B scores 84 - 12 = 72, so B still wins. Under correct minus three times incorrect, A scores 60 - 6 = 54 and B scores 84 - 36 = 48, and A wins. The same two performances rank three different ways depending on a single line in the instructions. This is why percent-correct is close to useless as a self-assessment on a speeded test: it ignores the rate, and the rate is the whole construct. Log both numbers after every practice session (attempts and correct, from which rate and accuracy both fall out), and compare sessions on rate. A session where accuracy improved but the rate fell is not necessarily progress.

How to practise this skill

• Practise in short, intense blocks of two to four minutes rather than long sessions. The construct is a rate under pressure, and a relaxed twenty-minute session trains a different behaviour from the one being measured.
• Never sub-vocalise a string you are comparing. Saying '4820' in your head takes about as long as saying it aloud and caps your rate; comparing the shapes of the blocks does not.
• Deliberately vary the drill format between name checking, number checking, symbol cancellation and table lookup. Speed built on one format transfers less than candidates expect, and real sections mix them.
• Slow down on purpose for mixed alphanumeric strings and speed back up on pure numeric ones. A uniform pace is the wrong strategy when the items have unequal difficulty.
• Fix your scan path before starting (chunk size, direction, whether you finish a column before moving), and do not change it mid-section. Changing procedure under time pressure costs more than the procedure was ever going to save.
• Record attempts and correct answers separately for every session so you can plot rate over time. Scores stay on this device unless you export them, so the log is yours and is worth keeping across weeks rather than days.

Glossary

Speeded test

A test whose items are easy enough that nearly everyone could answer them given unlimited time, so the score reflects how many were completed. Contrasted with a power test, where difficulty rather than the clock is the limiting factor.

Cancellation task

A format in which a target character or symbol must be marked wherever it appears in a dense field of similar non-targets, scored on hits, misses and false marks within a fixed time.

Distractor (perceptual)

A non-target that shares visual features with the target. The more features shared, the longer the rejection takes, which is why mirror-image pairs such as b and d are the most expensive distractors of all.

Homoglyph

A character that looks nearly identical to another in a given typeface: 0 and O, 1 and I and l, 5 and S. The commonest single cause of unavoidable errors in alphanumeric checking.

Chunking

Grouping a long string into units of three or four characters so each comparison fits within working memory. It is the difference between one pass over a reference number and several.

Raster scan

Covering a field in a fixed order, row by row in one direction, so that every cell is visited exactly once. Its value is not tidiness but the guarantee that the search is complete.

Rate measure

Performance expressed per unit of time, such as correct answers per minute. It is the only fair basis for comparing two speeded sessions of different lengths.

Ceiling effect

What happens when a test is easy enough that many candidates reach the maximum score, destroying its ability to discriminate. Speeded sections are deliberately long enough that almost nobody finishes, which is why an unfinished section is normal rather than a failure.

Study this next

• Perceptual speed skill hub
• Perceptual speed practice in the Cognitive Skills Lab
• Processing speed lesson
• Clerical checking suite
• Numeric keypad suite
• Perceptual speed lesson on Novus Learn

Where this material comes from

• All strings, code tables, grids and drills above were written for this lesson. The rate and score comparisons were computed and checked, including the three different orderings produced by the three scoring rules.
• Timing figures given for fixations, lookups and per-item costs are illustrative arithmetic showing the shape of the trade-off, not measured research findings; use your own logged times in their place.
• Terminology (speeded versus power tests, cancellation tasks, chunking, homoglyphs) cross-checked against standard public references such as the Wikipedia articles 'Speed test (psychometrics)', 'Chunking (psychology)' and 'Homoglyph'.
• Novus Learn aptitude construct registry (catalog seed) for construct scope and suite mapping.
• Public educational framing only: not affiliated with any official exam board, publisher or employer, and no copyrighted test item is reproduced.

Worked examples and pitfalls

Main idea: scope, not truth: Passage: 'When the Kirkwall depot adopted route-optimisation software in 2021, its drivers covered 9 percent fewer kilometres in the first quarter, and fuel spending fell by a similar margin. Delivery times improved on urban routes only; rural routes were unchanged, and two drivers reported longer split shifts. The depot manager has kept the software but now overrides its rural schedules by hand.' Which option states the main idea? (a) 'Route-optimisation software reduces fuel costs.' True, and it is exactly one third of the passage - it drops the mixed results and the override, so it is too narrow. (b) 'The software failed at Kirkwall.' Contradicted: mileage and fuel both improved, and the depot kept it. (c) 'Rural deliveries are harder to optimise than urban ones.' A generalisation the passage never makes; it reports one depot, one year, without explaining why rural routes were unchanged. Too broad. (d) 'At one depot the software produced clear mileage and fuel savings but uneven results elsewhere, so it is now used with manual override.' Correct - it covers all three sentences and no more. The lesson generalises: main-idea distractors are usually wrong for scope, and the two commonest failures are a true detail promoted to the main point, and a reasonable-sounding claim the passage never actually makes.

Reading a clause exactly: and, at least, more than, unless: Rule: 'An employee may claim the remote-working allowance if they work from home on at least three scheduled days per week AND their home is more than 40 km from their assigned office, unless they already receive the travel-card subsidy.' Four cases. Priya works from home four days, lives 52 km away, and receives the travel-card subsidy: not eligible - she satisfies both qualifying conditions, and the 'unless' clause overrides both. Tom works from home three days, lives 38 km away, no subsidy: not eligible, because the connective is AND and 38 is not more than 40. Dan works from home two days, lives 60 km away, no subsidy: not eligible, failing the days condition for the same structural reason. Mira works three days, lives 41 km away, no subsidy: eligible - 'at least three' includes exactly three, and 41 is more than 40. Three separate traps live in one sentence: reading AND as OR under time pressure, treating 'more than 40' as including 40, and forgetting that an exception clause outranks the conditions it follows. The reliable method is to rewrite the rule as a checklist - condition 1, connective, condition 2, then exception - before you look at any case.

Vocabulary in context: the evidence is in the next clause: Sentence: 'The auditor's remarks were characteristically dry, and the board took a full minute to realise she had been criticising them.' Which meaning does 'dry' carry here? (a) arid or lacking moisture, (b) dull and lacking interest, (c) understated and quietly ironic, (d) free of alcohol. Option (b) is the distractor that catches most people, because it is the commonest figurative sense and it half-fits an auditor. But the clause after 'and' is doing the work: a remark that takes a minute to land as criticism is understated, not boring - dullness would not delay recognition, it would only reduce attention. The answer is (c). The general method for vocabulary-in-context items is to ignore the word for a moment, read the rest of the sentence as evidence, and ask what property the sentence requires the word to have. Test items are built so that the most frequent sense of the word is available as an option and is wrong; if the most obvious meaning were correct, the item would measure nothing.

Referents: the former, the latter, and the noun three sentences back: Passage: 'Two remedies were proposed: a fixed surcharge on late filings, and a sliding penalty tied to the amount owed. The committee rejected the former on fairness grounds, noting that it would fall hardest on the smallest filers.' Which remedy was rejected? 'The former' is the first item mentioned, so the fixed surcharge. Candidates who skim choose the sliding penalty because it sits nearer to the pronoun, which is exactly why the item is written this way. Notice also the free consistency check built into the sentence: a fixed amount is the one that lands hardest on small filers, because the same sum is a larger share of a smaller liability, while a penalty tied to the amount owed scales with size by construction. When a passage gives you the reason alongside the reference, use it to confirm the reference. The harder version of this item type replaces 'the former' with a bare 'it' after a sentence containing three noun phrases, and the technique is the same: write the candidate referents in the margin and substitute each one into the sentence to see which produces a sentence the passage could have meant.

Whose view is this? Stance versus report: Passage: 'Supporters of the levy argue that it will fund three new depots within a decade. That projection rests on traffic volumes holding at 2019 levels, which no forecast in the sector now expects.' Question: what is the author's position? (a) The author supports the levy. (b) The author opposes the levy. (c) The author doubts the depot projection. (d) The author has no view. The correct answer is (c). The stance markers are 'rests on', which frames the projection as dependent on a condition, and 'which no forecast now expects', which tells you the condition is not met. Option (b) is the trap, and it is a scope error dressed as a tone question: undermining one argument made by supporters is not opposition to the policy, and the passage says nothing about the levy's other merits or costs. Reading for stance means reading the author's verbs and qualifiers rather than the content of the views being described - a passage can spend four sentences laying out a position it goes on to dismantle in the fifth.

Procedures: when two steps both fire: Instructions: '1. Record the meter reading. 2. If the reading is lower than last month's, do not submit; raise a query instead. 3. Otherwise, submit the reading and file the photo. 4. If the reading is more than double last month's, submit the reading and flag it for review.' Case A: last month 4,180, this month 9,020. Step 2 does not apply, since 9,020 is higher. Step 3 applies: submit and file the photo. Does step 4 also apply? Double 4,180 is 8,360, and 9,020 is more than that, so yes - submit, file the photo, and flag for review. Steps 3 and 4 are not alternatives; nothing in the list makes them exclusive, and both instruct you to submit, which is consistent. Case B: last month 4,180, this month 4,090. Step 2 fires: do not submit, raise a query. Steps 3 and 4 never come into play, because step 3 begins 'otherwise' and step 4's condition is not met either. The trap in case A is treating a numbered list as a decision tree where exactly one branch executes. Read each step's condition independently unless the procedure explicitly says to stop.

How to practise this skill

• Map the passage before you answer: read the first and last sentence of each paragraph and write a three-word label for each. On a 400-word passage this takes about twenty seconds and makes every detail question a lookup instead of a search.
• For every detail answer, put a finger on the sentence that supports it. If you cannot point at one, you are answering an inference question by feel, and that is where the marks go.
• On main-idea items, test scope before truth. Ask of each option: does it cover the whole passage, and does it cover nothing outside it? Two options will usually be true and only one will be the right size.
• Rewrite any policy or eligibility clause as a numbered checklist with the connective and the exception written out separately. Assessors build these items around AND read as OR, and around the boundary values on 'at least' and 'more than'.
• Do not pre-read the options on inference and stance items. Answer in your own words first, then find the option that matches; reading four polished options first makes three of them sound reasonable.
• Log wrong answers by cause - scope, connective, referent, stance, boundary value - rather than by passage topic. The topic never repeats; the cause always does.

Glossary

Main idea

The claim that covers the whole passage and nothing beyond it. It is not the first sentence, not the most interesting fact, and not the most general statement available.

Scope

How much an answer option claims relative to the passage. Most wrong main-idea and inference options are true statements that are simply too narrow or too broad, which is why checking truth alone does not separate them.

Referent

The noun a pronoun or phrase such as it, this, the former, or the team points back to. Ambiguous referents are deliberate in test passages and are resolved by substituting each candidate into the sentence.

Connective

A logical joining word in a rule - and, or, unless, provided that, except where. AND requires every condition; OR requires one; unless introduces an override that outranks what precedes it.

Boundary value

The exact number at the edge of a condition. 'At least three' includes three; 'more than 40' excludes 40; 'up to five' usually includes five. Items are written on these edges because that is where careless reading shows.

Stance

The author's own position, signalled by qualifiers, framing verbs and concessions rather than by direct statement. Distinct from any view the author reports.

Skimming versus scanning

Skimming is a fast first pass for structure and gist; scanning is a targeted hunt for a specific word or figure. Comprehension sections reward doing the first once and the second repeatedly.

Topic sentence

The sentence carrying a paragraph's central claim, most often first or last. Locating them across paragraphs gives you the passage's argument in about a fifth of the reading time.

Study this next

• Reading comprehension skill hub
• Practise the reading comprehension bank
• Verbal reasoning lesson
• Policy and program analysis suite
• Immigration and asylum casework suite
• Reading comprehension lesson on Novus Learn

Where this material comes from

• Worked items written for Novus Learn. The passages, eligibility rule, instruction list and figures above are original and invented for this lesson; no published or copyrighted test item is reproduced.
• Terminology follows standard, widely published usage in reading instruction and assessment writing - main idea, scope, referent, topic sentence.
• Novus Learn aptitude construct registry (catalog seed) for the construct scope and the suite mapping shown in the related links.
• Public educational framing only - not affiliated with any official exam board, publisher or employer.