, , , , , , , , , , , , , , , , ,

Although I came of age at the peak of Oprah’s cultural influence, I was not particularly interested in her until the day I overheard her talking on TV about the fact that her hats have to be custom made because her head is so large. It quickly occurred to me that at some time in her her life, Oprah may have been not only the most powerful and prosperous woman in the world, but also, the biggest brained woman in the world. Historically, Social Darwinism was used to diminish women, minorities, and the poor, so to see a black woman from such humble origins be so impressive on perhaps the two most Darwinian correlates of intelligence (money/power and brain size) was incredibly inspirational.

In 2010, Time.com asked Oprah and 11 other of America’s top female business leaders (as selected by Fortune magazine) to answer five questions. The questions were:

1) What is the best and worst decision you’ve ever made?
2) What was your dream job as a kid and why?
3) What do you think is the most significant barrier to female leadership?
4) What woman inspires you and why?
5) What will be the biggest challenge for the generation of women behind you?

I emailed the author of the Time piece telling him how scientifically valuable it was to see so many successful women all agreeing to answer the same set of questions. I did not tell him the purpose of my research but he informed me that he had published the exact unedited words of the women themselves. Most of the women (including Oprah) had emailed their answers to him directly, though a few responded by phone.

Because Oprah had answered the same questions, in mostly the same way (email, complete sentences, a few paragraphs per question), for the same publication, for the same purpose, as 11 other women with similar occupational status, this was a rare opportunity to measure Oprah’s cognitive ability on a highly standardized task against a very comparable control group. But how does one measure intelligence from a sample of writing? A Promethean once, somewhat facetiously, estimated my IQ by doing a Flesch-Kincaid readability calculation on an article I had published. I’ve explored this methodology and while it seems to give somewhat valid results in children, in adults it seems to reward a long-winded ostentatious writing style, which doesn’t make for efficient communication. Clearly a better method was needed…

The Nun Study

While searching for ways to estimate intelligence from writing, I came across the most fascinating study I have ever read. In the Nun Study, Dr. David Snowdon, a leading expert on Alzheimer’s disease, convinced a group of elderly nuns to take cognitive tests and allow their brains to be examined postmortem. This allowed Snowdon and his colleagues to identify dementia and Alzheimer’s disease in the older nuns, but Snowdon wondered if dementia in the elderly could be predicted from cognitive ability in young adulthood. Unfortunately, there were no available mental test scores from when the nuns were young.

Then Snowdon and his team discovered a goldmine. A stack of handwritten one page autobiographies that the nuns had written roughly six decades earlier when they first joined the convent as young women. Because the nuns had written autobiographies of similar length, for the same purpose, following identical instructions about what topics to include, these writing samples were highly comparable and provided an excellent window into the nuns’ early life cognitive ability. But Snowdon asked the same question I would ask years later. How do you measure cognitive ability from a language sample? Snowdon tried various methods like looking at the level of vocabulary each autobiography contained, but craved a more effective technique.

Idea Density

According to his book about the Nun Study (Aging with Grace), Snowdon decided to contact a language expert to help him analyze the biographies. He phoned the brilliant psycholinguist Dr. Susan Kemper who suggested that the most powerful way to measure mental ability from language samples is to quantify both idea density and, separately, grammatical complexity. Grammatical complexity is related to working memory and ranges from simple one-clause sentences to complex sentences with multiple forms of embedding and subordination. By contrast, idea density measures how succinctly you express yourself.

Intelligence can be defined as the cognitive ability to adapt (i.e. problem solve, turn situations to your advantage). If an advantage is a benefit, and a disadvantage is a cost, one could define intelligence as the mental capacity for low cost/benefit behavior, which is why making a lot of money for doing very little work is considered smart, all else being equal. In a trivial sense, every word we say or write has a cost (in the time it takes to write it, and the time it takes others to read it), but every idea we express is a benefit because we’ve communicated something of substance. So generally speaking, the more ideas we can express with as few words as possible, the lower the cost/benefit ratio of our behavior and the more intelligently we’ve behaved.

Amazingly, the idea density from just the last 10 sentences of the nun’s autobiographies correlated a potent 0.6 with their scores on the Mini-Mental State Examination (a brief measure of global cognitive ability) administered roughly six decades later. This is higher than the correlation of two different actual IQ tests administered in youth and old age, for example an outstanding study by scholar Ian Deary and his colleagues, published in the prestigious journal Intelligence found that the Moray House Test scores at age 11 correlated 0.48 with Raven scores at age 77, though the correlation was higher in women (0.55). Thus, the 0.6 correlation roughly six decades later with the Mini-Mental State Examination would seem to validate idea density as a measure of intelligence; indeed idea density is said to reflect overall neurocognitive development, rather than just a specific talent like verbal ability.

Snowdon and his colleagues were not surprised that idea density (and to a lesser extent grammatical complexity) predicted cognitive functioning in later life. The theory of cognitive reserve predicts that people with extra capacity (i.e. large brains, well developed minds) are able to delay the cognitive symptoms of Alzheimer’s longer because when their brains are afflicted by disease, they have enough extra brain mass and enough extra brain power to compensate for the damage. However the surprising finding of the Nun Study was that idea density also strongly predicted getting Alzheimer’s in the first place. Snowdon claimed in his book that simply by measuring the idea density in the last ten sentences of autobiographies written in the early 20s, they were able to predict with 80% accuracy which nuns would have the Alzheimer’s level of brain tangles at autopsy six decades later. This suggests that by age 20, many people already have very incipient Alzheimer’s that shows up linguistically in extremely subtle ways, but doesn’t progress into dementia until six decades later.

The results of the Nun Study were so astonishing that it made the cover of Time magazine and Snowdon was interviewed by Oprah’s old talk show rival, Phil Donahue.

The Nun Study makes the cover of Time magazine

Measuring the idea density of Oprah and other female elite business leaders

There was something symbolic about using a study of nuns (spiritual women) as a model to estimate Oprah’s intelligence since Oprah was preaching at her church at the age of three and decades later, would emerge as perhaps the most influential spiritual leader, ushering in a culture of secular new-age thinking. And just as the highly educated nuns studied by Snowdon and Kemper were exceptionally accomplished women for the standards for women in their generation, Oprah and the 11 female business elites I compare her idea density to, are exceptionally accomplished for the standards for women today.

I measured the idea density of Oprah and the 11 other elite female business leaders by pasting all the answers they gave to Time.com’s questions into a computer program called CPIDR 3.2.2785.24603. To illustrate how CPIDR works, consider the following two sentences:

1) I live in a house that is big.
2) I live in a big house.

Both sentences say the same thing, but because sentence 2, says it with fewer words, it has a higher idea density (also known as proposition density). CPIDR scores sentence 1 at 0.375 while sentence 2 gets a score of 0.5. Before scoring language samples on CPIDR, Dr. Susan Kemper advises that all multi-word names of people, places or dates be replaced by placeholders because long names add more words without adding more information and thus spuriously lower the ratio of propositions to words. Thus, the sentence “My name is John Smith and I was born in Dallas, Texas on October 31, 1877” should be entered into CPIDR as “My name is NAME and I was born in PLACE on DATE.”

Once such changes were made, Oprah’s answers to Time.com’s questions clocked in at 0.542. By comparison, the answers of the other 11 elite business leaders (the reference group) averaged 0.532 (Standard Deviation (SD) = 0.031)(Range = 0.489 to 0.610). Thus Oprah scored 0.32 SD above the reference group. It should be noted that these idea densitities are generally lower than those reported for nuns in the nun study, and that’s because CPIDR gives an indirect measure of idea density which produces lower scores than the directly measured idea density of the nun study. But the two methods are extremely high correlated, despite this systematic difference.

It should be noted however that Oprah was born in 1954 and the average birth year of the other 11 women was about 1960. Because Kemper’s research shows that idea density declines (within the same person) precipitously with age in a very linear way, beginning perhaps in the 20s, age adjustments are essential. Kemper found that a group of older adults (mean age 76.4) had a mean proposition density that was 1.56 SD lower than younger adults (mean age 22.8). This implies that idea density declines at a rate of 0.029 SD a year. Since Oprah is six years older than the average woman in the reference group, I added 6(0.029 SD) to her score, which increased it to 0.494 SD above the reference group.

Converting idea density to IQ

If one assumes that idea density correlates about as well with IQ as two different IQ tests correlate with one another, then one can convert idea density into IQ equivalent scores using a psychometric technique known as equipercentile equating or score pairing (see section 8.4.1 of this Prometheus document). Score pairing assumes that if a group of people get scores on two quite g loaded mental tests, x and y, then the distribution of x will mirror the distribution of y. That is to say if 50% of a sample is above the 98 percentile on X, then 50% of the sample should be above the 98 percentile on Y. That doesn’t mean the same individuals in the sample will be above the 98 percentile of both X and Y, it just means that being above the sample mean on X is as statistically rare as being above the sample mean on Y, and X and Y are equivalent and interchangeable measures of intelligence.

Thus, in order to convert Oprah’s idea density into an IQ equivalent, one must know the IQ distribution of the reference group she’s being compared to. The reference group were all women Fortune magazine ranks among the 50 most powerful in business. In a previous post, I noted that the average IQ of all Fortune 500 CEOs is likely about 124, but the tiny subset of women who crash through the glass ceiling to reach the top of American business likely average a brilliant IQ of 131. Assuming they have a similar IQ variability as the general population (SD 15), and assuming idea density and IQ scores are statistically interchangeable, then Oprah being 0.494 SD more idea dense than this outstanding group implies an extremely high IQ of 0.494(15) + 131 = 138. An IQ of roughly 140 is consistent with a previous analysis where I used multiple regression to predict Oprah’s IQ from Darwinian correlates of intelligence (income and brain size). It is also consistent with my historiometric estimate of Oprah’s childhood IQ. When disparate methodologies converge on one conclusion, probable truth is implied.

In the following video, Oprah displays high idea density by asking a complex question succinctly:

It may seem surprising that a mere daytime talk show host from the backwoods of Mississippi is more intelligent than most elite female business leaders, many of whom have attended the best colleges in America and are CEOs of some of the biggest high tech companies. But one must remember that Oprah’s brain size, wealth and impact on the culture is immensely greater than that of the other elite female business leaders, implying greater ability, especially since she overcame poverty and adversity to dominate the ultra competitive, improvisational and creative field of TV talk shows. High intelligence was likely a factor in Oprah immediately overtaking the very bright former talk show king Phil Donahue in the television ratings. Back in the 1980s, Newsday‘s Les Payne observed:

Oprah Winfrey is sharper than Donahue, wittier, more genuine, and far better attuned to her audience, if not the world.

And British actress and Cambridge graduate Thandie Newton believes incredible intelligence made a Oprah a good film actress too, stating:

I’ve worked with lots of good actors, and I know Oprah hasn’t made many films. I was stunned. She’s a very strong technical actress, and it’s because she’s so smart. She’s acute. She’s got a mind like a razor blade.

A few caveats

Of course this entire analysis assumes idea density is a good measure of intelligence. While the Nun Study showed that idea density in youth does an impressive job predicting intelligence and Alzheimer’s in old age, one can’t necessarily assume that idea density is a good measure of pre-elderly intelligence. Although it’s normally true that if a test predicts intelligence decades later, it does an even better job predicting intelligence contemporaneously, one of the ironies of the Nun Study is that idea density measured in youth failed to correlate with the nuns’ academic grades in youth. This may suggest that idea density is a very sensitive measure of intelligence that can be impaired by the smallest tangles in the brain many decades before the tangles increase to the point of diminishing intelligence on the whole. If so, idea density in the relatively young may be a poor measure of contemporaneous intelligence, but an excellent measure of how well one will maintain their intelligence with age.

The strongest evidence of Oprah having extremely high intelligence all comes from when she was extremely young (she could read, write, recite, and do arithmetic by age 3). And idea density suggests she will have extremely high intelligence when she is extremely old (relative to the elderly). And yet during her peak intellectual years (late teens), she studied only easy subjects at a very modest college. Perhaps extremely precocious toddlers regress to the mean in young adulthood, only to come home to their original brilliance in old age. But of course, drawing such sweeping conclusions from largely anecdotal evidence involving only one person would be extremely foolish.