It’s easy to make a list of the major events each generation experienced. Boomers, for example, were young during the Vietnam War, and Gen Z grew up after 9/11. But listing major events doesn’t tell you much about how the generations differ in their personality traits, attitudes, and behaviors. That’s the approach I take, analyzing nationally representative survey data on 39 million people.

For example, my book Generations relies on 24 nationally representative surveys. Most include data across several decades. That’s powerful, because they allow us to compare the generations at the same age. That means we can know what is really different across the generations apart from being a certain age.

Here are examples of some of the datasets I have worked with:

Monitoring the Future, 1976-present
General Social Survey, 1972-present
Programme for International Student Assessment (PISA), 2000-present
Behavioral Risk Factor Surveillance Survey, 1984-present

Most of these datasets don’t yield their secrets easily. Finding the generational trends involves downloading the datasets, merging the files, recoding variables, and running analyses. I’m often asked why the dataset webpages don’t show the trends I document. Most often, it’s because it takes so much time to merge the datasets and do the analyses. That’s what I have spent my career doing. (And making graphs. Generations has 282 graphs!)