The analysis is based on youths ages 16 to 25 in the Decennial Censuses and 2007 American Community Survey (ACS) public use micro samples. The Integrated Public Use Micro Sample (IPUMS) versions of these data, compiled by the University of Minnesota Population Center, were used. The 2007 ACS is a 1% sample of the U.S. population. In 1970, school enrollment information is available only in the Form 2 samples. The three 1970 Form 2 samples were combined to form a 3% sample of 1970 youths. For 1980, 1990 and 2000, the 5% IPUMS files were used. For Hispanic and non-Hispanic black youths, all the observations were used. But, for ease of tabulation, random 1-in-10 subsamples of non-Hispanic white youths were utilized. So, for 1980, 1990, and 2000, the non-Hispanic white results are based on a 0.5% sample of non-Hispanic white youths.
All the results published are based on the appropriate sample weights. The unweighted sample sizes of youths are as follows:
![](https://assets.pewresearch.org/wp-content/uploads/sites/7/2009/10/2009-pathways-20.png)
Commencing with the 2000 Decennial Census, the Census altered the racial identification question by permitting respondents to declare multiple racial identifications. Non-Hispanic black and non-Hispanic white youths before 2000 and after 2000 are made comparable by using the IPUMS RACESING variable for the racial designation. See the IPUMS documentation for further details on how RACESING bridges the old and new racial classifications: http://usa.ipums.org/usa-action/variableDescription.do?mnemonic=RACESING
The 1970 Form 2 samples do not inquire about citizenship status. They do, however, have information on place of birth. For this reason, nativity in this report is based on the youth’s place of birth rather than citizenship status. Youths who were born in Puerto Rico or other outlying areas were included in the foreign-born population.
The American Community Survey does not reveal the type of institution where institutionalized youths live. Hence, in the ACS it is not possible to identify youths incarcerated in correctional facilities from other institutionalized youths and only the total count of institutionalized youths is available. For comparability, the Decennial Census counts on incarcerated youths shown herein also refer to the number of institutionalized youths in all institutions, not just correctional facilities. As Appendix B shows, most (more than 90%) institutionalized young males are in correctional facilities.
Results involving the employment or labor force status of youths for 2000 were not published. Census Bureau evaluation studies indicate that the 2000 Decennial Census may not accurately measure a youth’s employment status (Palumbo and Siegel, 2004). Census 2000 underestimates employment and overestimates the number of people not in the labor force (relative to the Current Population Survey for the Census 2000 time period). The classification problems appear to be particularly acute for young age categories and lesser-educated individuals.
The Decennial Census/American Community Survey data have a number of advantages for the examination of the activities of youths. Unlike some other data sets, the universe for the Census is the resident population. As such, youths residing in institutions and those employed in the armed forces are enumerated. Both the institutionalized and those in the military are particularly important for understanding the activities of young males. The census has extremely large samples of youths, providing precise estimates for narrowly defined age and nativity groups. The Current Population Survey (CPS) is often used to examine the school and labor market activities of youths. The CPS did not routinely inquire about citizenship status until 1995 and hence cannot be used to document trends for immigrant and native-born youths. Finally, a number of well-known studies of the activities of black men and immigrants have used the Decennial Census (Welch, 1990; Betts and Lofstrom, 2000).