This image is even more worthless than it seems. The post on Wikimedia is an original work. Its description states: "As the percentage of graduates increases the minimum IQ to include at least that percentage of graduates inherently decreases. Since 2000 the intelligence required to be a college graduate has been less than the intelligence required to graduate from high school in 1940, based on a standard distribution."
It seems the author took the the percentage of the population that graduated high school/college each year and then found the corresponding percentile on an IQ bell curve and used those as the y-values. This methodology only makes sense if you assume that high school/college graduates are exactly the highest IQ population and that everyone who does not graduate isn't intelligent enough to do so. This chart also almost certainly doesn't normalize IQ over time, even though IQ is constantly redefined so that 100 is average while raw intelligence scores have increased over time [1].
What this chart actually shows is the highest possible IQ of the graduate with the lowest IQ in a given year, a statistic that seems to have dubious value.
You’re right it is worse. It looks like they assumed college-going students would approximate the standard distribution of scores for the population as a whole, which… no. An awful assumption for an activity where academic ability is a primary gatekeeper at the same time that they’re attempting to apply a metric of IQ essentially as a proxy for academic ability/knowledge/whatever. (“Whatever” because the entire concept of intelligence is filled with varying definitions)
It seems the author took the the percentage of the population that graduated high school/college each year and then found the corresponding percentile on an IQ bell curve and used those as the y-values. This methodology only makes sense if you assume that high school/college graduates are exactly the highest IQ population and that everyone who does not graduate isn't intelligent enough to do so. This chart also almost certainly doesn't normalize IQ over time, even though IQ is constantly redefined so that 100 is average while raw intelligence scores have increased over time [1].
What this chart actually shows is the highest possible IQ of the graduate with the lowest IQ in a given year, a statistic that seems to have dubious value.
[1] https://en.wikipedia.org/wiki/Flynn_effect