a rigorous test of the Straus and Howe generational hypothesis

William Straus and Neil Howe proposed that history followed a generational cycle, with an alternating sequence of Prophet, Nomad, Hero, and Artist generations, each representing people born within 20-year periods. They started their story in 1433 and carried it through at least 2014, when the last of Generation Z will be born. Their evidence was mostly a series of examples of people who exemplified the ostensible spirit of their generation. They also proposed a logic for why the cycles would recur.

The danger is selection bias. For 2,000 years, physicians convinced themselves that leeches cured disease by sharing stories of patients who were bled and then improved. Pierre Charles Alexandre Louis put that practice to scientific test, using what he called the “numerical method,” and found that it did not work. The essence of the “numerical method” is to create an appropriate sample and then look for the predicted patterns.

The Straus/Howe hypothesis says that all else being equal, a person is more likely to have the attributes of a “Prophet” (per their definition) if that person was born within a time span when prophets predominated–e.g., between 1945 and 1964. One’s generation should be a statistically significant predictor of one’s public character, even controlling for gender, race, education, class, region, religion, parents’ opinions, etc. Moreover, one’s generation should be a better predictor than one’s birth year: in other words, people should come in 20-year blocs divided by what Straus and Howe called “turnings.” Someone born in 1966 should more closely resemble an individual born 19 years later (since they are both Xers, who are Nomads) than someone born one year earlier.

Here is a proposal for how to test the hypothesis. The American National Biography is a reference book that describes 17,400 notable Americans. I would take a random sample of those entries and conceal their names, gender and race, and birth year. I would ask several people to read their brief bios and code them as Prophet, Nomad, Hero, or Artist. I would check the consistency of coding and retrain coders until they could reach the same conclusion at least 80% of the time. I would then add a dummy variable for generation and see whether there was a statistically significant relationship between that variable and the coding. Finally, I would control for at least gender, race/ethnicity, and birth year to see if any relationship survives (or emerges).

The Straus/Howe theory has enough plausibility–and a vast enough following–that it seems to deserve a test. I don’t want to do it, so the idea is available for anyone ….

This entry was posted in Uncategorized on by .

About Peter

Associate Dean for Research and the Lincoln Filene Professor of Citizenship and Public Affairs at Tufts University's Tisch College of Civic Life. Concerned about civic education, civic engagement, and democratic reform in the United States and elsewhere.