Demographic Research: Volume 21, Article 6
Figure 2: No reenrollment: imputed and original education histories
high degree, not enrolled
low degree, not enrolled
enrolled in education
first birth
original histories
.......... imputed histories
The same type of problem may occur if people obtain several different educational
degrees at their highest level in sequence but the survey only asks for the last date they
obtained their highest degree. If people obtain one degree, then work for some period of
time, and then reenroll in education in order to obtain an additional degree at the same
level (though perhaps in a different subject than before), the same selectivity problems
described above may occur. If the information used for imputation refers to the date the
respondent last obtained a degree at their highest level, time they spent outside of
educational enrollment between spells would be misrepresented as a time of educational
enrollment in the imputed histories. If reenrollment is more likely for those who did not
have a child after obtaining their first degree, similar distortions to the ones described
above are likely to occur.
Further problems may arise in contexts where it is common to obtain additional
educational degrees on a part-time basis while employed or pursuing another activity.
If, following our method of imputation, we use the date respondents received their
highest degree to impute their duration of educational enrollment, the problem is that
we are not accounting for differences between part- and full-time students. Educational
enrollment may not have been respondents’ main activity, and fertility rates could differ
strongly between full-time students and students who are simultaneously employed.
For the present context, the question is to what extent these imputation problems
might apply to Germany. Findings from previous research indicate important obstacles
to educational reentry at later ages in Germany. If it is uncommon to resume education
later in the life course, that would make imputation less problematic. Schütze and
http://www.demographic-research.org
141