The Dawn of Canada’s Century: Hidden Histories ed. by Gordon Darroch (review)

This book, written by many of Canada’s leading historical scholars, sheds new light on a number of topics, including language identity, aboriginal populations, family and household arrangements, immigration, politics, and social class and mobility at the beginning of the twentieth century. Its most important feature is the analysis of historical census microdata. The four introductory chapters deal with census data and methods, the new 1911 census microdata sample, and parallel projects for samples of the 1851/52, 1871, 1881, 1891, and 1901 censuses. In addition to the 1911 census sample, comparable samples are also underway for the 1921, 1931, 1941, and 1951 censuses. The other twelve chapters present an analysis of historical census microdata. All of them analyze 1911 data, but several use other historical data sets. A chapter about Quebec City explores the linkage of individuals over several censuses for the 1871 to 1911 period. About one-half of the chapters examine Canada’s national population. Other chapters deal with subnational populations, including Trois-Rivières, Newfoundland, Quebec City, and Hamilton.

The development and availability of census microdata samples has been the basis for improvements in research during the past five decades. The methodological foundation for these new data emerged in the 1930s and 1940s when statisticians provided the theory and methods for survey sampling. By the 1950s, researchers were able to conduct national surveys of 1,000 respondents, with valid inferences about entire populations. Because of new sampling methods, better statistical software, and big advances in computing technology, census microdata has progressed in four stages.

First, survey sampling techniques were employed for the first time after the 1960 U.S. census to make data tapes with public-use microdata samples (pums) available. These pums files were either 1-in-1,000 or 1-in-100 (1 percent) samples of individual records—omitting identifiers that could reveal individual identities—with information about age, sex, ethnic origin, nativity, marital status, family relationships, occupation, income, and other data collected in the census. These first pums files proved to be a treasure chest; researchers could, for the first time, prepare their own tabulations rather than depend on tables published by the U.S. Census Bureau. Moreover, researchers could use modern multivariate statistical techniques for the analysis of census microdata. By the end of the 1960s, academic journals routinely included articles with regression and other multivariate analysis of census data.

The second major advance occurred when census microdata for several countries became available for comparative analysis. By the 1970s, the census microdata samples for many countries could be used for the study of such topics as the factors associated with international variations of female employment.

When researchers realized the value of pums files for several censuses, they began to develop pums files for earlier censuses—the third stage. In the United States, researchers initially took samples from the 1940 and 1950 censuses in order to make longer-term comparisons with existing 1960, 1970, and 1980 censuses. Currently, 1 percent pums files exist for U.S. decennial censuses from 1850 to the present, comprising one of the most valuable quantitative data sets for historical research.

We are currently in the midst of a fourth stage—the development of large census-data collections that are both historical and comparative, as evidenced by the pioneering work of ipums-International (https://international.ipums.org/international/), which now includes pums files for 258 censuses from 79 countries. For a comparative study of, say, southern Latin America, ipums-International currently includes sixteen census pums files for Argentina, Chile, and Uruguay that could be used to analyze trends from the 1960s to the present. The next frontier for census microdata analysis is the comparative study of change over time in, for instance, the determinants of fertility variations, correlates of family structure, and factors affecting the living arrangements of elderly adults.

U.S. census microdata files—including individual data with detailed codes for age, country of birth, ethnic origin...