By Hillary Rettig
If ever a revolutionary masked himself in pedant's tweeds, Steve Ruggles has. Balding, with a round face and bushy beard, Ruggles gives himself away through his dark, intense eyes. They belie the University of Minnesota history professor's otherwise placid appearance. Ruggles could play the mad genius in a B movie without much of a stretch, or - better yet - Santa's dark twin. He seems calm, but is really multitasking - making lists and checking them twice.
Ruggles's mission, like that of all revolutionaries, is to change the way we see society and ourselves. But the "bombs" he lobs are not Molotov cocktails - they are data. His goal is to organize the data from 65 million US Census returns - a statistically significant sample of every census since 1850 - in a vast database called IPUMS, for Integrated Public Use Microdata Series. It will be the largest ever released for the public study of a human population.
A database that size makes all kinds of things possible. Ruggles - who proudly defines himself as "a quant (quantitative) kind of guy" - uses it to perform what he calls "multivariate societal analyses where time is a variable." This means he can tell us, using hard data, how our ancestors lived over the past century and a half, and how in general their lives were shaped by the forces of race, immigration, industrialization, and crime, as well as the rise of public education and the political and economic emancipation of women.
This information is not, in the 1990s, without a certain relevance. Take race, for instance: in his 1994 article "The Origins of African American Family Structure" (published in the American Sociological Review), Ruggles debunked the common notion that welfare programs have caused a breakdown in family structure among African Americans. This idea, popularized in a famous 1965 report by Senator Daniel Patrick Moynihan, has recently been used by conservative social critic Charles Murray for his argument on abolishing welfare.
Ruggles's analysis demonstrated that African Americans have, since 1850, had a higher-than-average percentage of single-parent families. He also found that the rate of single parenthood among blacks over the past few decades has increased no more than that among whites. "The real question is what produced the initially higher numbers of single parenthood," says Ruggles, who believes that it was most likely either slavery or African culture. He adds: "Anyone who thinks Great Society-era welfare programs increased the rate of single parenthood among African Americans is either ignorant or willfully misrepresenting the data."
Zealots of all political stripes, in fact, have reason to fear historical demography. As the African-American family structure fracas demonstrated, it has the potential to explode some cherished myths about our society. Multiculturalist dogma notwithstanding, historical demography demonstrates that all immigrant groups have rapidly assimilated throughout American history. Other historical demography research recently debunked the 1991 Harvard and Yale study - vastly hyped in the media - about how professional women, by delaying their marriages, doomed themselves to spinsterhood.
According to Ruggles, who is studying the correlation between women's economic status and their divorce and separation rates, historical demography confirms what may have seemed obvious all along to fans of 18th- and 19th-century novels: there has always been a significant pool of single women in America. Ruggles believes the proportion of these women in 1980 and 1990 society is not exceptional, and faults the Harvard and Yale group for using baseline data from the late 1950s and early 1960s - "the peak years for marriage in our country."
There's no question that historical demography studies can have enormous political and societal ramifications. Says Ruggles: "Society has gone through wrenching changes over the past couple of centuries, and continues to do so. If the IPUMS database can help us set current social and economic changes in the context of the past couple of centuries, it might help us to figure out what is going on now."
According to Professor Charles Wetherell, director of the Laboratory for Historical Research at the University of California at Riverside, Ruggles's research "provides historians and social scientists of all types with the biggest and best data set in American history. And I don't have to tell you how much social policy is made on the basis of demographic analyses."
All of that changed in the 1960s, when universities began acquiring computers, and historians and sociologists began using them to study population databases. During the 1970s, many academic departments grudgingly tolerated such research as a novelty. "Many historians don't like the quant stuff," says Matt Sobek, a researcher working with Ruggles. "It's pretty dry and boring compared to narrative work - not to mention that a lot of the people who do it write terribly."
By the 1980s, however, historical demography had achieved enough prominence to invoke an anti-quant backlash among the traditionalists. It was not a happy time to be a quant, according to Ruggles, who says that a lot of academic departments and professional journals were uninterested in the research. But that attitude has now mostly passed. These days, historical demography is conducted in history and sociology departments throughout the US, Canada, England, France, and other countries.
During the bleak 1980s, Ruggles suffered along with the rest of the quants. (When he was looking for a job in 1983, he was rejected 87 times before landing his first interview.) But he has rebounded - with a vengeance. He was not only granted tenure at an unusually early age, but has won several major professional awards, and commands - by history standards - lavish funding. He is considered the bellwether of the modern historical demography movement, the "young Turk" who is defining the path along which historical demography will progress. Says UC Riverside's Wetherell: "Steve is a superstar - one of the best and the brightest young historians practicing today. I'm surprised he hasn't received a MacArthur [genius] award. "
Still, debate exists even within the historical demography community about the limits of the approach. Most scholars agree that it can't be used to accurately describe "values" or "culture"; however, not all of them draw the line at the same place. Tony Wrigley, a professor of economics at the University of Cambridge, England, and a historical demography researcher, has been one of Ruggles's sharpest critics. While acknowledging the "logic" of the historical demography approach, he feels there are constraints that limit the usefulness of most, if not all, historical demography databases. "If you're interested in measuring behavior in the past, the available materials pose inevitable problems of interpretation because of factors such as migration and death that change the nature of the sample. The group that remains intact throughout the course of the study may not have the same characteristics as the group that left." Wrigley calls these factors "truncation effects," and says they are most serious in the type of long-term studies Ruggles specializes in. Ruggles, in turn, says that while people may drop out of other data sets, they "don't fall out of observation from the national census."
Daily life in the IPUMS labs is chaotic but pleasant, and Ruggles, by all accounts, runs a pretty relaxed sort of operation. Says Sobek: "Steve is about as laid-back as they come. He doesn't care when you do the work as long as it gets done."
Within his department, Ruggles is respected as the founder of the Friday afternoon "brouhaha," during which large numbers of historians go out on the town for beers, then bowling at Stardust Lanes, which Sobek describes as Minneapolis's "most serious" bowling alley. Technology is not a big issue in Ruggles's lab because the technology used is ordinary, not cutting-edge. Most of the 20th-century data is in machine-readable format on tape, although the lab is imposing a common format on it; the lab is entering most of the 19th-century data from microfiche copies of the original census books. The IPUMS data sets are stored as Unix-compressed ASCII files on a Sun Microsystems SPARCStation 20 with 64-Mbyte RAM and 29-Gbyte storage. The IPUMS team's biggest challenge is coming up with a consistent data-field structure, given that the raw data was collected by thousands of people over 150 years. (Self-pitying programmers take note!) "In 1910, enumerators hand wrote answers, so you wind up with something like 200 family relationship categories," explains Ruggles. "But by 1960, answers were coded, and nonstandard entries were listed under 'other,' so we have only seven categories."
Birthplace is also tricky: "There are a whole bunch of countries that don't even exist anymore, like Austria-Hungary." Ruggles and his colleagues have made their way through about one-third of their project in the last five years, which is expected to continue - funding allowing - until 2007. (Part of the reason the project will take so long is that census data is withheld for 70 years from the public for privacy reasons. The 1920 census became available only in 1992, and 1930's will become available in 2002.) He considers it something of a miracle that so much of the original census data survives: only the 1890 census has been lost - to fire. Not only has his group codified the data from 1850 (the first census to enumerate individuals rather than households), but also from 1880 ("a terrific census"), 1900, and 1910; they are about one-third of the way through 1920, and are awaiting funding to start 1860 and 1870. These data sets can be used in studies alongside census data from 1960 onward, which the US Census Bureau digitized. The databases - still in beta, but due for official release by mid-1995 - are available via the Internet: anonymous ftp legohead.hist.umn.edu. ("Legohead" comes from a graduate student who said, when the first project SPARCStation was about to be moved into an already crowded office: "We'll have to put Legos on our heads and sit on top of each other.")
Ruggles and his colleagues represent the first generation of historical demography scholars to benefit from almost unlimited access to cheap computing power and storage. As the tools and data sets improve, so do the types of studies historical demography can take on. Says Ruggles: "Before, we were all studying population size and the things that affect it - births, deaths, and migrations. Now, it's changes in population characteristics, such as education, ethnicity, occupation, and family structure."
Historical demography may also be the harbinger of a new era in which inexpensive information technology changes the way social sciences, and even the "softer" liberal arts, are conducted. Many people will no doubt find that transition jarring, but Ruggles seems equally at home in both worlds. Part of the reason may be quant genes: both his parents were Yale economists, and granddad Clyde Orville Ruggles was an economist at Harvard. Ruggles quants not just in his professional life, but in his personal as well: for years, he has been tracking the age, time of arrival, group composition, and costume choices among the trick or treaters who visit his house each Halloween. And some predict that his new baby, christened Abigail Kristin by Ruggles and his wife (and fellow historian), Lisa Norling, will become "the most quantified kid in history."
Hillary Rettig (hrettig@bix.com) is president of RCS Inc., a Manhattan-based technology consulting firm that advises Fortune 500 companies and others on technological trends. She writes frequently on technology and society.
