it] = 0-1 indicator variable that takes a value of one if individual i attrits (namely is not interviewed) in year t.
i(t+1)] = indicator variable signaling individual i attrits in year t + 1;
For example, if the youth who attrit tend to be the low earners in the population, then the earnings distributions based on the NLSY will suggest greater growth over time than actually occurs because the lower tail of the distribution will thin due to the systematic loss of its members.
The second set of rows (percentage of attriters who never return) show that of the persons who attrit in the early years, around 10 to 25 percent are never seen again by 1991.
For the population at large listed in the first column (labeled "All"), individuals aged 18-20 who are in school are 8 percentage points less likely to attrit in the subsequent year than their nonemployed counterparts, and employed persons are 3 percentage points less likely to attrit; the in-school difference is significant at the 5 percent level.
This implies that persons dropping out of high school are no more likely to attrit than those remaining in high school.
School attendance is not linked to attrition for women in high school and college, with the exception of Black women participating in college (ages 18-20) who are less likely to attrit than their nonemployed contemporaries.