Entry into the literature
As a starting point, Simonton (2009) provides an excellent introduction to the field of historiometric assessment of intelligence. To quote the abstract:
Running parallel to mainstream research on the psychometric assessment
of intelligence is another tradition of research on the historiometric
assessment of intelligence and closely affiliated variables.
Historiometric assessment is based on four data sources: (a)
personality sketches (e.g., Intellectual Brilliance), (b) developmental histories (e.g., IQ), (c) content analyses (e.g., integrative
complexity), and (d) expert surveys (e.g., Openness to Experience).
The first two represent major lines of intelligence research that
involved key figures in the development of corresponding psychometric
methods (e.g., Galton, Terman, and Thorndike), whereas the last two
constitute independent research paradigms that later intersected with
the first two. The literature on U.S. presidents then provides an
integrated illustration of the four historiometric approaches and how
they converge on the same broad conclusions. Significantly,
historiometric investigations on the relation between broadly defined
intelligence and adulthood achievement obtain about the same effect
size as that found in psychometric research (i.e., $r$ or $\beta$ .25 $\pm$
.10). Because historiometric and psychometric studies have rather
distinctive methodological advantages and disadvantages, this
consistent outcome provides corrobora- tive support for both sets of
The review summarises various studies aiming to assess intelligence of royalty, U.S. politicians, and eminent figures.
For current and recent public figures substantial additional information would be available. In particular, many powerful predictors of IQ would often be available including university grades, scores on standardised assessments, SAT scores and so on. Furthermore, the existing literature on IQ correlates is huge and provides substantial information that could inform predictive models of IQ prediction. In contrast where such models are applied to the past, the predictive power of such models may be wrong.
I'd also be extremely wary of estimates of intelligence that yield estimates like 200 (i.e., around 7+ standard deviations above the mean). E.g.,
After comparing the chronological age in which .. achievements
appeared with the mental age that they represented, Terman decided
that Galton’s IQ must have been close to 200. (Simonton, 2009, p. 319)
The expertise literature has highlighted the importance of not confusing eminence with IQ. Eminence is substantially influenced by years of practice, training, focus, opportunity, and luck. IQ is useful, but the correlation between IQ and eminence is insufficient to yield the kind of certainty in prediction to result in IQ predictions above something like 3 or perhaps 4 standard deviations above the mean (i.e., 145 to 160).
- Simonton, D.K. (2009) The “other IQ”: Historiometric assessments of intelligence and related constructs. Review of General Psychology, 13, 315. PDF