Verbal reasoning: analogies, odd one out and the classic logic trap

"If it rains, the road is wet. The road is wet. What can be concluded?" Many people answer that it is raining. It might be, but it does not follow: a street cleaner could have passed, or a pipe could have burst. Logicians call this mistake affirming the consequent, and it is one of the most common errors in everyday reasoning.

That question comes from our verbal reasoning test. Verbal reasoning is the ability to work out relations and conclusions from information given in words. It overlaps with vocabulary, but it is not the same thing: the words in a good verbal reasoning task are easy, and the difficulty lies in what you do with them.

Three kinds of tasks

The test has 15 questions in three groups of five. Each question has four answers, and only one is correct.

Analogies follow the pattern A is to B as C is to what? "Bird is to nest as spider is to..." You need to name the relation between the first pair (where the animal lives) and apply it to the second (web).

Classification asks you to find the word that does not belong. In "rose, tulip, oak, lily", three are flowers and one is a tree. The harder items have more than one possible rule, and you have to find the one that leaves exactly one word out.

Inference gives you premises and asks what follows from them. "No fish is a mammal. A dolphin is a mammal." The correct answer is that a dolphin is not a fish, not that it is smarter than a fish or swims like one. Some items have "cannot be determined" as the right answer, which people often skip because it feels like giving up.

Where these tasks come from

Analogies and odd-one-out tasks have been part of intelligence testing since its early days. In 1904 Charles Spearman noticed that people who did well on one kind of mental task tended to do well on others, and he called the common factor general intelligence, or g. Analogies turned out to be among the tasks most closely tied to it.

We wrote all 15 questions ourselves in this tradition. The test has not been validated or normed, so your result shows how well you did in each part. It is not an IQ score, and it cannot place you against the population.

How to read your score

Look at the three parts separately. Mistakes in inference usually come from adding knowledge that is not in the premises, as in the wet road example. Mistakes in classification often come from stopping at the first rule you notice. Mistakes in analogies often mean the relation in the first pair was named too loosely: "a spider is related to a web" is less useful than "a spider lives in a web".

On all three, it helps to say the rule out loud in one sentence before choosing. If you cannot, you have probably guessed.

The test is also short, so one careless click moves your score by a noticeable amount. If a result surprises you, check which items you got wrong before drawing conclusions.

If you want a broader picture

Verbal reasoning is one part of general ability. Our IQ test adds numerical, logical and spatial tasks and working memory, and the cognitive ability test covers four areas in 20 questions. For what online IQ tests can and cannot tell you, read what an online IQ test measures and how it works.