Is there any way to roughly approximate someone’s IQ simply based off of observing their behavior, mannerisms, speech, social interactions, etc? In other words, are their certain patterns of behavior that people exhibit as a function of their intelligence, allowing for the approximation of their IQ by observing them display these patterns (and if so, what are some examples throughout the IQ spectrum)?
There's a body of research that shows that thin slices allow for reasonable discernment of intelligence and personality. Just two recent articles show the following:
Negative affect, extraversion, conscientiousness, and intelligence were judged moderately well after 5-s exposures.
Inferences of intelligence from thin slices of behavior strongly predicted intelligence test scores. A particularly strong single predictor was how persons read short sentences.
Note that this is alongside a large body of research done to show how thin slices of intelligence corresponds to real IQ. Both studies have further citations supporting this; they were just the first two I found off of Google.
Dana R. Carney, C. Randall Colvin ¤, Judith A. Hall. (2007). A thin slice perspective on the accuracy of Wrst impressions. Journal of research in personality. http://faculty.haas.berkeley.edu/dana_carney/slices.in.press.page.numbers.oct.2007.pdf
Borkenau, P., Mauer, N., Riemann, R., Spinath, F. M., & Angleitner, A. (2004). Thin slices of behavior as cues of personality and intelligence. Journal of personality and social psychology, 86(4), 599. doi: 10.1037/0022-3518.104.22.1689 https://psycnet.apa.org/record/2004-12052-006
After seeing the answer that cites the Borkenau study, I wondered how the reading of short sentences factors in, so I pulled up the full article [PDF]
It didn't entirely have the concrete details I was hoping for (I imagined something about intonation, body language, or the size of "chunks" read at once). The following seems to mean that people are reported to be more intelligent when they appear to quickly read and understand text, perhaps without any sign of hesitation or struggling, and that this does correlate strongly with measured intelligence.
Pg. 11: "Inferences of intelligence from Task 13 (reading 14 newspaper headlines and subtitles aloud) were more strongly related to the LPS–PC in both participant subsamples, and to the APM score in one subsample, than were inferences of intelligence from any of the 14 other tasks."
Pg. 14: "Performance on intelligence tests is most strongly related to inferences from observing targets who read short phrases aloud, as in the headlines task. This suggests that the way people read standard text provides particularly diagnostic cues of their psychometric intelligence. This resembles findings by Borkenau and Liebler (1993), who found that inferences of intelligence based on observations of how strangers read a standard text allow for quite accurate judgments of their intelligence. This finding is interesting from a theoretical as well as from an applied perspective. Concerning theory, the finding is consistent with the assumption that intelligent persons perform simple tasks more efficiently (Deary, 2000). According to this view, measured intelligence should not only be related to how persons solve complex problems but also to how efficiently they perform on everyday tasks like understanding and reading a short text. In this context, the strength of the correlation of judgments of intelligence from the headlines task with the performance measures of intelligence becomes important. The finding that the LPS is related to the headline task nearly as strongly as it is related to another test of intelligence shows that the headline task is a quite precise measure of intelligence."