Here's a question that looks easy until your child gets it wrong. Five words sit in a row. Find the two that don't belong with the other three. Sounds simple. Then you watch a clever ten-year-old read it, jab the first word that catches their eye, and move on in four seconds flat. Wrong answer, full confidence.
That's the odd-one-out question, and it's one of the most common verbal reasoning types on GL Assessment papers. It rewards a particular kind of careful thinking and it quietly punishes the opposite. Children who are quick, confident and used to being right are often the ones who drop marks here. So let's look at what these questions actually are, why they trip kids up, and what you can do about it at home.
What an odd-one-out question looks like
The standard GL version gives you five words. Three of them are connected in some way. Two of them aren't. Your job is to find the two that don't fit the group of three.
Take this example: minute, vast, tiny, miniscule, huge. Three of those mean small. Two of them mean large. So the answer is vast and huge. The group of three that belong together are minute, tiny and miniscule. The odd ones out are the two big words.
Notice what makes this harder than it first appears. You're not looking for one strange word in a tidy set. You're splitting five words into a three and a two, and you have to be sure the three left behind really do share something. A child who only finds one odd word has missed half the question.
Here's another: Wales, Yorkshire, Cornwall, Nottingham, Devon. Three of those are regions or counties. Two are cities. Wales, Yorkshire and Cornwall hang together as areas. Nottingham and Devon are the odd ones out, because Nottingham is a city. Devon is the trap. It feels like it belongs with Cornwall, so a rushing child grabs it as a match. The categories don't always line up the way your eye wants them to.
Some papers also use a simpler one-odd-word version, especially in early practice material and CEM-style mixed sections. But the two-words-from-five format is the one that catches most children out, and it's the one worth drilling into until it feels natural.
Why bright children get these wrong
The whole question is built around a first impression that's usually a decoy. When your child reads five words, their brain leaps to the most obvious grouping before they've finished reading. That instinct is exactly what the question is testing against.
Think about a set like rose, tulip, oak, daisy, pine. The obvious grouping is flowers. A fast reader sees three flowers, decides oak and pine are the odd ones out, and they happen to be right. But change one word and the same instinct fails. The examiner knows children grab the first category they see, so the harder questions hide a second, sharper relationship underneath the obvious one.
There's also the problem of stopping too soon. Find one odd word and the relief kicks in. The child thinks they've cracked it and forgets they need two. I've watched my own son do this. He'd circle the clearly wrong word, feel pleased, and never go looking for its partner. Half marks, every time, on questions he fully understood.
Then there's vocabulary, the quiet killer under all of this. If your child doesn't know that "miniscule" means tiny or that "morose" means gloomy, no amount of clever grouping will save them. Odd-one-out questions lean on the same word knowledge that synonym questions do. A child who reads widely has met these words in the wild. A child who hasn't is guessing, however bright they are.
How to teach the slow-down without nagging
You can't just tell a child to slow down. Every parent has tried that and it works for roughly nine seconds. What works better is giving them a routine they can actually follow under pressure.
The routine I'd teach is simple. Read all five words first, before deciding anything. Then ask: what do most of these have in common? Find the group of three that clearly belong together. The two left over are your answer, but check them. Do they share something with each other too? In the small-versus-large example, vast and huge aren't just leftovers, they're a pair that means the opposite of the other three. When the two odd ones out also connect to each other, you can be confident you've split the set correctly.
Get your child to say the grouping out loud when you practise at home. "These three are all small. These two are big." Forcing them to name the relationship stops them guessing. If they can't explain why three words belong together, they haven't found the real pattern yet, and they should look again rather than commit.
The other habit worth building is checking for a hidden second meaning on anything that looks too easy. If your child solves an odd-one-out in two seconds and the answer feels obvious, that's the moment to pause, not relax. The easy ones early in a section are real. The ones designed to catch children dress up as easy and aren't.
Building the vocabulary underneath it
No technique fixes a missing word. So the long game here is the same as it is for every verbal reasoning question type: reading. A child who reads a chapter book most nights meets thousands of words in context, and context is how vocabulary actually sticks. Far better than a list of definitions your child forgets by Tuesday.
When you mark practice papers together, treat every wrong odd-one-out as a free vocabulary lesson. Did they miss it because they didn't know a word, or because they rushed? Those are two different problems with two different fixes. If it was a word they didn't know, write it down. A running list of words your child has actually met on a paper beats any pre-printed vocabulary book. These are the words their specific exam board likes to use.
Want to make it stick? Play with categories away from the desk. In the car, name three things that go together and one that doesn't, and let your child catch you out back. "Apple, banana, carrot, pear." It sounds like a game because it is one, and it's building exactly the grouping instinct the exam rewards. Ten-year-olds will do this for ages without realising they're revising.
How do you know if any of this is landing? You watch the pattern of mistakes over a few papers, not the score on one. If the same child who scored badly in March is splitting five-word sets cleanly and explaining why by June, the technique has gone in. That's the thing you're actually building, not a number on a single test.
Frequently asked questions
Do all 11+ exam boards use odd-one-out questions?
They're most associated with GL Assessment verbal reasoning, which uses a fixed set of recognised question types including this one. CEM-style and mixed papers can include similar grouping questions, but they tend to vary the format more. Don't assume your region's paper looks identical to a GL practice book. Check which board your target schools use before you buy materials.
Is it always two odd ones out, or just one?
The classic GL version asks for two words that don't go with the other three. Some practice material and early worksheets use a one-odd-word version to ease children in. Find out which format your child's actual test uses and practise that one most, because the two-word version needs a different checking habit.
My child knows the words but still gets these wrong. Why?
Almost always it's speed. They grab the first grouping they see and don't check for a second relationship. Or they find one odd word and forget to look for the second. This is a habit problem, not a knowledge problem, and it fixes faster than vocabulary gaps. Get them naming the grouping out loud and the rushing usually slows itself.
How much should we practise this specific question type?
Enough that the routine feels automatic, not so much that your child burns out on one format. A handful of focused questions across a few sessions, mixed in with other verbal reasoning, beats a hundred in one sitting. You're building a habit, and habits form through little and often rather than one long grind.
Want to know whether your child's verbal reasoning is on track, including the question types they're quietly losing marks on? ReadyFor11 gives you an honest, free benchmark with no paywall waiting at the end. Have a look at readyfor11.co.uk and see where your child actually stands.