# Average Harvard IQ

Harvard is one of the World’s premier higher education establishments. Harvard University was established in 1636 and today admits has 7,000 undergraduates and 14,000 graduate students.

As most US undergraduate degrees last four years, it is easy to work out that Harvard should therefore be admitting 1750 students per year to keep the undergraduate student population stable at 7,000 (assuming no attrition). However, 2,175 were admitted in 2013 which suggests that either the drop / non completion rate is high, or that Harvard is trying to grow the size of its undergraduate student body.

But Harvard is an intellectual elite school. 27,500 people applied for entry but only 8% were admitted in 2013.  On this basis, it comes as no surprise that many people are intrigued as to what the average Harvard IQ might be?

One way to estimate what the average Harvard IQ might be through SAT scores. SATs have been described by Harvard educator Howard Gardner as thinly disguised IQ tests.

It has been reported that most Ivy league admissions officers focus on the Math and Critical Reading sections of the SAT test. To estimate the average Harvard IQ, one should therefore start with understanding the average admission criteria.

The bottom 25% of the class of 2013 had the following scores on the different sections of the SAT 1.

• Math: 710
• Writing: 710

The above scores sum up to a total composite score of 2120. Although this score represents the bottom 25% of newly admitted Harvard students, such a score in fact corresponds to the 97th percentile of scores for the population of SAT test takers as a whole. In other words, the top 3%.

If SAT scores were exactly analogous to fluid IQ scores, the top 3% would correspond to an IQ score of 129 (assuming a standard deviation of 15) or an IQ score of 131 (assuming a standard deviation of 16).

Coming back to the class of 2013 admissions data, students at the 75% percentile (i.e. the top 25% of new Harvard students) had the following SAT scores

• Math: 790
• Writing: 800

For a total composite score of 2390. This translates into a  combined score that is well inside of the top 1% of the population. In IQ terms, the top 1% of the population would translate into a score of 134 (15 Standard deviation) and 137 (16 SD).

On the basis of the SAT figures above, and knowing the admissions values for the 25th and 75th percentiles, we can guess roughly that the average SAT score at the 50th percentile (i.e. the average) may be somewhere around the 2200 mark, which translates roughly into the 98% percentile of the top 2% of  SAT test takers.

The corresponding IQ in the top 2% would be 130 (15 SD) or 132 (16 SD), which happens to be the threshold for admission into Mensa.

## Estimating average Harvard IQ – are SATs scores the answer?

Prior to 1994, Mensa used to accept SAT scores as prior evidence for qualification to the society. SAT scores are no longer accepted by American Mensa as it is no longer believed that modern-day scores correlate with IQ.

I have seen estimates online of average Harvard IQs being as high as 140 and as low as 127. So what is the answer for the true average Harvard IQ? No one knows for sure.

One valid argument is that is it not entirely correct to equate the distribution of SAT scores with that of a fluid intelligence test. This is so because IQ distributions should be based on the general / overall population, whereas SAT scores distributions leave out the lower left hand tail of the general population. That is, people who struggle academically and with low IQs are less likely to sit an SAT test. So based on this argument, one would conclude that achieving a high percentile on an SAT distribution (relatively to filtered distribution which excludes the lowest IQ scores) corresponds to an increased percentile relative to the general population. In plain English, a SAT score around the 2,200 (98th percentile relative to other SAT test takers) for instance, might equate to a significantly higher IQ score relative to the general population.

However, this distribution bias effect might well be negated by the fact that you can prepare for the SAT test, whereas you should not have any indication of the contents of a properly administered fluid IQ test. In other words, the test prep element could well negate any distribution-related biases resulting from the comparison of fluid IQ and SAT score distributions.

On the basis that you can prepare and study for an SAT test, but that you cannot effectively do the same for a fluid IQ test, I would contend that the average SAT percentiles are likely to be over-inflated relative to fluid IQ scores. The only real challenge to this hypothesis is that there is little supporting evidence that SAT test prep reliably produces tangible results. Nevertheless, it is irrefutable that you can study for the SAT, which this is likely to lead to score gains relative to a fluid IQ test.

Coming back to SAT scores, for which the average eventual Harvard class of 2017 has a score in the top 2% of the population, of test takers this is probably an upper bound for the corresponding average level of fluid intelligence. I propose that an IQ score of 127  (95.4% percentile with 16 SD) as the average Harvard IQ sounds plausible. So the average Harvard university student would fall short of the IQ level required for Mensa (132 assuming 16 SD).

This is not to say that Harvard would not have plenty of students with Mensa (132+) and genius-level IQs (>140), but rather that the percentile achieved on the SAT is likely to be high compared to that cohort’s corresponding average level of IQ.