How to tell if IELTS practice material is any good before you trust it
Wrong answer keys and over-easy questions teach the wrong lessons. A ten-point audit for practice tests, and what to do when a key is plainly wrong.
ComparisonsGeneralUpdated 7 min read
A bad practice test is worse than no practice test, and the damage is not the wasted hour. A wrong answer key teaches you to distrust reasoning that was correct, which is the single hardest habit to unlearn before exam day; an over-easy question set teaches you a pace and an accuracy rate you cannot hold on the real paper. Both faults are common in freely available material and in paid material alike, and neither announces itself. So audit the source before you study from it. Fifteen minutes with the official format specification will tell you most of what you need to know.
The ten-point audit
Run these against any Listening or Reading set before you sit it, and against the Writing and Speaking material before you time yourself on it. Points 1 to 9 are mechanical: they either match the published format or they do not.
| # | Check | Official specification | Fail signal |
|---|---|---|---|
| 1 | Listening length | 40 questions | 30 or 50 items |
| 2 | Listening structure | four parts | three parts, or five |
| 3 | Listening playback | each part heard once only | a replay button, or "listen again" |
| 4 | Reading length | 40 questions across three passages | two passages, or 36 questions |
| 5 | Reading timing | 60 minutes for the whole paper | a 75-minute allowance |
| 6 | Reading transfer | no extra transfer time on the Academic paper | 10 extra minutes offered at the end |
| 7 | Completion word limits | a limit printed on every completion task | tasks with no limit stated |
| 8 | Writing tasks | Task 1 at least 150 words in about 20 minutes; Task 2 at least 250 words in about 40 minutes | "write 120 words", or one long task |
| 9 | Speaking format | 11 to 14 minutes in three parts | a single 20-minute interview |
| 10 | Key transparency | each answer carries a paragraph or line reference | a bare list of letters and words |
Point 3 matters more than it looks. A replay button quietly rewrites the skill being tested: single-hearing Listening is about committing to an answer while the audio moves on, and material that lets you rewind trains the opposite reflex. Point 6 is the most frequently muddled, because the ten minutes of transfer time belongs to the paper-based Listening test, not to Reading. If a practice Reading set hands you extra minutes at the end, your timing data from it is inflated.
Point 10 is the one that separates usable material from decorative material. A key that only says "14. Not Given" cannot be checked. A key that says "14. Not Given, paragraph D, the text gives no figure for the interval" can be checked in twenty seconds, which is the whole point.
The one-sentence test for the answer key itself
Every True/False/Not Given and Yes/No/Not Given item has to be defensible from the text alone. That is not a style preference; it is what the item type means, and the decision rule that separates False from Not Given rests on exactly this. So apply the one-sentence test: write a single sentence justifying the key using only wording from the passage. If you cannot, one of two things is true, and you must decide which.
Suppose the passage says: The 1911 survey recorded 240 nesting pairs; no comparable count was attempted until 1985. The statement is Nesting pair numbers fell between 1911 and 1985. The justification writes itself: no count exists for the interval, so the text supports no direction of change. Not Given, defensibly.
Now the broken version. The passage says the earliest systematic count and the statement is The 1911 survey was the first attempt to count the birds, keyed False. There is nothing in the passage contradicting it; at worst it is an unsupported inference, which makes it Not Given. A key that says False here is teaching you a rule the examiner does not use. One such item is a typo. Five across a set is a broken source, and you should stop using it for accuracy work.
When a key is plainly wrong
Do not litigate it for twenty minutes. Run this procedure and cap it at two minutes.
- Write both answers down, yours and the key's, plus the exact sentence from the passage you used. Quote it, do not paraphrase it. Half the disputes die here.
- Check the mechanics before the meaning. Did you exceed the stated word limit? Misspell it? Pluralise a word the passage left singular? Answers are marked on spelling — the answer-marking rules on capitals, plurals and brackets are stricter than most candidates assume — so most "wrong keys" turn out to be your own transfer slips.
- Re-read the sentence before and the sentence after the one you used. Genuine misses live there far more often than in the sentence itself.
- Check the rule, not an opinion. For word limits, Not Given definitions and timing, go to the published format and scoring information, and write down the line that settles it next to the disputed item.
- If it survives all four, log it as unresolved and move on. One line in a notebook: set, item number, your answer, the key's. No further argument.
- Count unresolved items per 40. Nought or one is normal noise. Two to five means use that source for timing and stamina only, never for accuracy. More than five means retire it.
Step 6 is the payoff. It converts a vague suspicion into a decision you can act on, and it stops you doing the thing candidates do instead, which is keeping a set they half-trust because it was inconvenient to find.