What IQ Do You Need to Be a Scientist?

No research post or doctoral programme asks for an IQ score. Occupational data places research scientists in a band with a median around 120 to 125, and theoretical physics and mathematics higher again at roughly 125 to 130, the highest band recorded for any occupational group.

What makes science unusual is that somebody also went and measured the people at the top of it, which almost nobody has done for any other profession.

The study of eminent scientists

In the early 1950s the psychologist Anne Roe assembled a sample of 64 eminent scientists, selected through membership lists of scholarly organisations and expert nomination. They were well known in their fields without being quite at Nobel level, and they spanned physical scientists, biologists and social scientists.

She then did something unusual: rather than using a standard test that would run out of difficulty, she had a purpose-built high-ceiling instrument administered to each of them.

The scores were extraordinary. Later estimates place the typical subject far above 130, with analyses putting the median well into territory corresponding to something like one in ten thousand of the general population.

Two caveats matter, and most retellings drop both. The conversion to a modern IQ scale is contested, because Roe used a custom instrument rather than a normed one, and the high-end figures come from later re-analysis rather than her own claims. And the sample was eminent scientists, chosen for distinction, which is a different population from scientists generally.

The gap that matters

Put the two figures side by side and the interesting thing appears.

• Research scientists generally: a median around 120 to 125
• Eminent scientists in Roe’s sample: far higher

That gap is much larger than the gap between, say, doctors and engineers, or between lawyers and accountants. Within science, distinction appears to track cognitive ability more steeply than entry to the profession does.

Which fits what the long-running work on high-ability populations suggests: even within an already-selected group, further differences keep predicting output. That evidence is set out in our guide to an IQ of 130. Returns diminish going up the scale. They do not switch off.

It does not follow that ability produces distinction. Roe sampled people who had already become eminent and measured them afterwards, which cannot separate cause from selection. Plenty of people with those scores never produce anything notable.

The famous counter-example

Worth setting against Roe, because the two are usually quoted by opposing camps.

Lewis Terman’s longitudinal study followed around 1,500 Californian children selected for very high scores. None won a Nobel Prize. Two boys screened and rejected for scoring too low, William Shockley and Luis Alvarez, later did.

Usually told as proof that testing misses real scientific talent. The correction is more interesting: modern analysis suggests Terman had only a small probability of catching either of them, simply because Nobel Prizes are astronomically rare and his cut-off was high. Missing them is what you would predict from base rates, not evidence of failure. The detail is in our guide to what counts as a genius IQ.

Different sciences, different shapes

Roe’s sample spanned physical scientists, biologists and social scientists, and one of her more durable findings was that the balance of verbal, spatial and mathematical ability differed between fields rather than simply the overall level.

That matches the pattern running through this whole series. Law loads on verbal, engineering on spatial, software on working memory. Within science the same thing happens one level down: theoretical physics and the social sciences select for different profiles, and a single composite averages exactly that distinction away.

What actually gates a research career

• A doctorate, which takes three to seven years at low pay
• Publication record, which compounds and is heavily influenced by which lab you joined
• Postdoctoral positions, often several, frequently requiring international moves
• Funding, which is the real bottleneck and increasingly competitive
• Tolerance for insecurity, since permanent academic posts are far scarcer than the people qualified for them

The attrition in science happens overwhelmingly after the cognitive filter. People leave research because of funding, geography and job security, not because they could not do the work. That is worth knowing if you are weighing a research career against your score.

Frequently asked questions

The bottom line

No IQ is required to be a scientist. The occupational median sits around 120 to 125, rising to 125 to 130 in theoretical and mathematical fields. What distinguishes science from the other professions here is that we also have data on the people at the top, and Roe’s eminent scientists scored far above the working median, though the exact conversion is contested.

The honest reading is that within science, distinction tracks ability more steeply than entry does, and that this is a correlation measured after the fact rather than a recipe. What actually ends most research careers is funding, geography and job security, long after the cognitive filter has stopped mattering.