Career quizzes, and how to use one properly

Every one of these hands you something that reads like an answer: clean type, a list of occupations, a paragraph describing you warmly and accurately. That confidence is a design decision, not a measure of what the result knows about your life. It is a decent starting point, mis-sold as a finish line.

What a questionnaire can and cannot see

Two jobs, not equally possible. What you are drawn to can be measured tolerably well, is fairly stable across years, and relates to the work people stay in. How well you would do it is not predicted well by any questionnaire. Nor can one see what decides whether a job is liveable: pay, commute, hours, whether your manager is reasonable, whether anyone is hiring near you.

Three worth knowing, and what each is worth

The O*NET Interest Profiler. Free, no sign-up, produced by the US Department of Labor. Take it at mynextmove.org/explore/ip. A serious instrument built on the interest model below, and the best free option we know. Its honest limit: the occupation lists come from US labour data, so read them as illustrations, not a shortlist.

Holland's RIASEC model. The framework underneath most careers questionnaires you will meet, including the one above. It ranks you across six interest areas — realistic, investigative, artistic, social, enterprising and conventional. Its standing for interests is decent; for performance it is weak, and your top three letters say very little about how good you would be at anything. Everyone has some of all six, so this is a ranking, not six boxes with you in one.

The Myers-Briggs family. This covers the official instrument and the free four-letter lookalikes online, which are not the same thing and are usually presented as though they were. It is by far the most popular set of tools here, and it hands people vocabulary they lacked. Its evidence for predicting job performance is weak — a plain statement, and not a controversial one among people who study selection. People also come out as a different type on retest more often than a stable measure should allow. Use it for language, never for decisions.

In the UK, the National Careers Service skills assessment is the free local equivalent.

The method: turn a result into a Tuesday

Take the most surprising line, not the most flattering one. The flattering lines are why it feels accurate; the surprising one carries the information.

Write three questions about a working day that would be true if that line were true. They need observable answers — not would I enjoy this, but questions answerable by describing what happened. If the result points towards work with a strong social component: how many hours of a Tuesday go on talking to people who are not colleagues; when something goes wrong at four o'clock, who deals with it; how much of the week could be done without speaking.

Ask a real person doing the job now. Not the careers page, not someone who did it in 2011. Ask for twenty minutes and an average Tuesday hour by hour, then the worst week of the year — everybody describes the good version by default.

Write down which of the three questions the conversation changed. That is your finding. The quiz result never was.

Where this pays and where it costs

It pays when you have no vocabulary and no starting point. A questionnaire generates candidates you would not have written down yourself, cheaply and in private, and a description that fits gives permission to take an option seriously that you had dismissed. The vocabulary alone makes the next conversation with an adviser far more useful.

It costs by closing options as efficiently as it opens them. A four-letter result narrows what you read next, because you search for the label rather than the work. The category hardens into identity and people decline things on the grounds of what they are. It is also an effective way of confirming a decision already made.

What the setting can do

Careers services buy questionnaires because questionnaires scale and conversations do not. Ask for the conversation instead, held to a format: twenty minutes with someone doing the work, an average Tuesday and the worst week of the year. Ask that the write-ups be published, so the next person need not arrange the call.

One rule is worth pushing for wherever you study or work: no type result used in group allocation, placement matching, promotion or hiring. This happens often, and there is no basis for it.

Small things to try this week

1. Take the O*NET Interest Profiler in one sitting. Close it before reading the job list, and write down what surprised you.
2. Turn that into three Tuesday questions with observable answers.
3. Ask one person for twenty minutes. Most say yes, and most are more honest than a careers page.

If none of this shifts anything after a few weeks, that is useful too — it usually means the question you are stuck on is about money, geography or timing, which no questionnaire has ever helped with.