Average IQ by College Major (1970s)

Razib Khan and Emil Kirkegaard have estimated the average IQ by college major using GRE and SAT scores, though I am fortunate enough to possess the Project Talent dataset, which has both IQ scores (measured at ages 12-18, for most) and college majors. They also test verbal/visual/mathematical ability, so I took advantage of that and plotted ability tilts by major, using the method Razib Khan used.

This dataset also includes measures of parental SES, extraversion, self-reported verbal ability, and maturity which I have included in the plots. Parental SES and IQ correlated at .45 in this dataset, identical to the value calculated in me and Emil’s study of race and honesty, suggesting that the data is probably accurate. The correlation between self-reported and measured verbal ability was .4, above what you tend to see in the literature, but not too out of the ordinary.

Technical information regarding how the estimates were made is available in the appendix.

Here is the average IQ by major:

Here are the plots:

correlation: .5, p < .01

correlation: .74, p < .001

correlation: .23, p = .19

correlation: .58, p < .001

correlation: .54, p < .001

correlation: -.54, p < .001

correlation: .84, p < .001

Appendix/methodology:

Non-cognitive traits were standardized at a mean of 0 and standard deviation of 1, while cognitive traits were standardized at a mean of 100 and standard deviation of 15. This was done within the whole dataset, so the averages of the majors should differ from the mean.

Cognitive testing:

• IQ was measured using the first general factor of general knowledge, memory, processing speed, verbal ability, mathematical, and visual ability

• General knowledge was measured using a composite of technical and cultural knowledge

• Technical knowledge was measured using the following variables: phsk, aerok, eleck, mechk, sportk, huntk, fishk

• Cultural knowledge was measured using the following variables: screening, litk, musk, sstudiesk, mathk, biok, sciattk, farmk, homek, artk, lawk, healthk, engk, archk, jourk, fork, militk, salesk, prack, clerick, biblek, colorsk, etiqk, outdk, photok, gamesk, theatk, foodk, misc

• Memory was measured using the following variables: sentmemory, wordmemory

• Verbal ability was measured using the following variables: spell, capital, punct, english, express, words, reading, create, disgwords

• Mathematical ability was measured using the following variables: arith, bmath, amath

• Visual ability was measured using the following variables: mech, twod, threed, abst

• Processing speed was measured using the following variables: arithfast, table, cler, obj

• Subtest scores were adjusted for age

• Missing values were removed.

• Scorers with missing values were not scored.

Non-cognitive data:

• Parental SES was measured using their ‘socio-economic index, measured in the first wave.

• Maturity was measured using their ‘mature personality’ variable

• Extraversion was measured using a composite of the ‘sociability’ and ‘social sensitivity’ variables

• Overestimated ability was calculated by combining people’s self-reported reading/writing ability and controlling it for measured verbal ability.

• As always, missing values were removed

• Self-reported verbal ability was calculated by combining self-reported reading and writing ability

College major:

• Statistics and math were combined

• Biochemsitry and chemistry were combined

• Data from the 11 year followup was used. If this was not available, data from the 5 year followup was used.

• Individuals who had not completed their Bacherlors were excluded

Correlations between relevant variables (within the entire dataset):

Parental SES and IQ:

Parental SES and Extraversion:

IQ and Extraversion (note: these do not correlate in most datasets):

Correlation between Verbal (0.23) and Mathematical (0.1) ability with Extraversion:

Correlation between self-reported verbal ability and measured verbal ability:

dataset variable conversions for cognitive ability:

nd$arithfast <- nd$BY_F410

nd$table <- nd$BY_F420

nd$cler <- nd$BY_F430

nd$obj <- nd$BY_F440

nd$spell = nd$BY_R231

nd$capital = nd$BY_R232

nd$punct = nd$BY_R233

nd$english = nd$BY_R234

nd$express = nd$BY_R235

nd$sentmemory = nd$BY_R211

nd$wordmemory = nd$BY_R212

nd$disgwords = nd$BY_R220

nd$words = nd$BY_R240

nd$reading = nd$BY_R250

nd$create = nd$BY_R260

nd$mech = nd$BY_R270

nd$twod = nd$BY_R281

nd$threed = nd$BY_R282

nd$abst = nd$BY_R290

nd$arith = nd$BY_R311

nd$bmath = nd$BY_R312

nd$amath = nd$BY_R333

nd$screening = nd$BY_R101

nd$vock = nd$BY_R102 + nd$BY_R162

nd$litk = nd$BY_R103

nd$musk = nd$BY_R104

nd$sstudiesk = nd$BY_R105

nd$mathk = nd$BY_R106

nd$phsk = nd$BY_R107

nd$biok = nd$BY_R108

nd$sciattk = nd$BY_R109

nd$aerok = nd$BY_R110

nd$eleck = nd$BY_R111

nd$mechk = nd$BY_R112

nd$farmk = nd$BY_R113

nd$homek = nd$BY_R114

nd$sportk = nd$BY_R115

nd$artk = nd$BY_R131

nd$lawk = nd$BY_R132

nd$healthk = nd$BY_R133

nd$engk = nd$BY_R134

nd$archk = nd$BY_R135

nd$jourk = nd$BY_R136

nd$fork = nd$BY_R137

nd$militk = nd$BY_R138

nd$salesk = nd$BY_R139

nd$prack = nd$BY_R140

nd$clerick = nd$BY_R141

nd$biblek = nd$BY_R142

nd$colorsk = nd$BY_R143

nd$etiqk = nd$BY_R144

nd$huntk = nd$BY_R145

nd$fishk = nd$BY_R146

nd$outdk = nd$BY_R147

nd$photok = nd$BY_R148

nd$gamesk = nd$BY_R149

nd$theatk = nd$BY_R150

nd$foodk = nd$BY_R151

nd$misc = nd$BY_R152

Other variables used:

Y11_E305 - year 11 followup, college completion

Y11_E321 - year 11 followup, college major

Y5_E321 - year 5 followup, college major

Y5_E305 - year 5 followup, college completion

BY_P828 - self-reported writing skills

BY_P829 - self-reported reading skills