The ten-minute typing test
The typing test for this post is scored by the software as follows: gross speed = characters typed ÷ 5 ÷ minutes; net speed = (characters typed ÷ 5 − uncorrected errors) ÷ minutes. A word left blank counts as one error, and an error that has been corrected is not counted at all. The standard notified for this recruitment is a net speed of 95 words per minute with at least 95 per cent accuracy, where accuracy = (words typed − errors) ÷ words typed.
You type for 10 minutes. At the end the software reports 5,180 characters and 18 uncorrected errors. You also remember deliberately skipping 4 words, because the passage contained a technical term you were unsure of and you preferred to move on rather than stall. No corrections were made during the test.
The evaluator also remarks that your gross figure "means you type over a hundred real words a minute", which sounds implausible to you for a passage full of long official terminology.
1. Compute gross speed, net speed and accuracy, and decide whether the standard is met.
Read the worked answer
First convert characters to word-units, using the 5-character convention: 5,180 ÷ 5 = 1,036 word-units. These are not dictionary words; they are keystroke-equivalents.
- Gross speed = 1,036 ÷ 10 = 103.6 wpm.
- Errors chargeable = 18 uncorrected + 4 blanks = 22. A blank counts as an error under the notified rule, and the 4 words are not "unattempted" in any sense the rule recognises.
- Net speed = (1,036 − 22) ÷ 10 = 1,014 ÷ 10 = 101.4 wpm.
- Accuracy = (1,036 − 22) ÷ 1,036 = 1,014 ÷ 1,036 = 97.9 per cent.
Against the notified standard of 95 net wpm and 95 per cent accuracy, both limbs are satisfied with margin - roughly 6 wpm and nearly 3 percentage points to spare. Verdict: qualified, and the four skipped words did not put the result at risk.
2. At about the six-minute mark you notice a mistyped word three lines back. Correcting it would take about four seconds. Using the scoring formula, should you stop to correct it?
Read the worked answer
Treat everything in word-units. Your running rate is 518 characters a minute, i.e. 103.6 word-units a minute, so four seconds of not typing costs 103.6 × 4/60 ≈ 6.9 word-units.
Score effect of the correction: removing one error raises the numerator by 1 word-unit, which over a 10-minute test is 1 ÷ 10 = 0.10 wpm on net speed. The pause costs 6.9 ÷ 10 = 0.69 wpm. Net effect ≈ −0.59 wpm. So the correction is a losing trade on speed, and on accuracy it is worth 1 ÷ 1,036 ≈ 0.1 percentage point against a 5-point cushion.
The general rule: correct only when the accuracy threshold is genuinely at risk, or when the error is in a figure, date or name whose wrongness survives the test even though the software scores it like any other error. At 97.9 per cent accuracy with a 95 per cent floor, neither condition applies - keep typing.
3. Is the evaluator's remark about "over a hundred real words a minute" right? Explain what the 103.6 figure actually represents.
Read the worked answer
It is a unit confusion, not a compliment. The formula divides characters by 5 because the conventional "word" in typing tests is five keystrokes - a standard that approximates average English word length including the space. So 103.6 is a rate of about 518 keystrokes a minute, not 103.6 dictionary words.
In a passage of long official terminology - "Employees' State Insurance Corporation", "contributory", "beneficiary" - the true word count would be materially lower than 103.6, because those words average well over five characters. Conversely, a passage of short words would push the word-unit rate above the real word rate. That is precisely why the convention exists: it makes scores comparable across passages, which counting actual words would not.
The practical implication for a candidate is about practice, not scoring. Drill on long, unfamiliar words and on numbers, since these consume keystrokes without inflating the count; a good gross figure built on short common words can collapse on a technical passage. The remark can be answered simply: the figure is word-units, and the software is right to score it that way.