448 - 
Table 4.1 June Supplement to the 1980 Current Population Survey: Selected Variables 
Variable 
Description 
date of birth 
day, month, year 
in years 
age 
married civilian spouse present 
current marital status 
married military spouse present 
married spouse absent 
widowed, divorced 
never married 
number of times married 
date of marriage 
month, year of first, second, last 
race 
white, black, other 
ethnicity 
Mexican American, Chicano, Mexican, 
Puerto Rican, Cuban, Central or South 
American, Other Spanish, Other ethnicity 
highest grade completed 
grades 0, 1,..., 20+ 
and other ethnicity. 
I used data on year of birth to create four large cohorts of women born between 1880- 
1909, 1910-1929, 1930-1949, and 1950-1965. Respondents were asked about the highest grade 
attended and if she had completed this grade; note that both items are current for the date 
of interview and so have a somewhat ambiguous interpretation in the context of first marriage 
especially for women who continued their education after marriage. I coded a completed edu¬ 
cation variable by calculating the highest grade completed and then collapsing responses into 
grades 0-9, 10-12, 13, and 14 or more. The separate category of 13 years of education (one year 
of college) requires some explanation. A preliminary examination of frequencies revealed that 
a large number of women reported that their highest educational level completed was one year 
of college. Because I expected that these women might differ from those who completed two or 
more years of college, I coded a separate category for these women. 
4.2 Exploratory Methods for Event History Analysis 
This section presents a brief and self-contained description of the smoothed hazard estimator 
used extensively throughout this chapter. Readers who desire a more detailed and leisurely 
description of the methods used should see Chapter 3. Other readers who are primarily interested 
in the exploratory results and substantive implications of these results may wish to proceed 
directly to Sections 4.3 and 4.4, taking the estimator and confidence intervals on faith. 
Although there are a number of standard theoretical perspectives that inform research on 
marriage, existing theories of marriage offer little or no guidance about the specific shape of 
the rate of first marriage. In such a situation, nonparametric methods have practical attrac¬