Demographic Research: Volume 21, Article 6
subject than before, whether immediately after completing their first degree, or
sometime later in the life course. If a survey only collects the first date respondents
obtained their highest degree, imputations based on this information will misrepresent
time respondents in reality spent enrolled after this date to obtain a second degree at the
same level.
Gaps in respondents’ education histories before they obtain their highest degree
can be problematic for imputation as well. For the imputations, we assume that
respondents were enrolled continuously up until they received their highest degree.
Therefore, mis-imputations occur whenever respondents return to education and obtain
a higher level degree at a later point in time after having interrupted education for some
time. Gaps between education spells during which they held lower level degrees and
were not enrolled will then be misrepresented as continuous enrollment. It may be
easier for childless women than for women who had a child sometime during their
lower degree spell to reenter education. Then, it will occur more frequently for women
who remained childless that time they actually spent holding a lower level degree and
not enrolled in education is imputed as time spent continuously enrolled in education.
This is illustrated in Figures 1 and 2. Figure 1 shows the case of a person who
returned to education at a later point in time. First, this person was enrolled in education
and by the end of her first education spell had acquired a lower level degree. She then
spent some time not enrolled in education, during which she may have been employed,
for example. During this time, she was holding her lower level degree. She then
returned to education and gained a higher level degree. Her last spell is again spent not
enrolled and holding a higher level degree. In the imputed education history,
represented by the dotted line, she is assumed to have been enrolled continuously up
until the time she gained her highest degree. The time during which in reality she was
not enrolled and was holding a lower level degree is highlighted in red in the imputed
history, which misrepresents this spell.
Figure 2 illustrates the case of a person who does not return to education. This
person is also first enrolled in education and by the end of her education spell has
acquired a lower level degree. Then, in her second spell, she is not enrolled and is
holding a lower level degree, just like in the case illustrated in Figure 1. During this
spell, she has a first child. Because combining education and childcare is difficult in her
particular country context, she decides not to reenroll in education. Thus, she continues
to hold a lower level degree up until the time of interview. In this case, the imputed
history matches the original history, since the respondent actually was enrolled
continuously for the entire time before she gained her highest degree.
Given that motherhood reduces the propensity of returning to education,
imputation is more likely to be correct (in this particular respect) for people who had a
child (Figure 2) than for those who did not have a child (Figure 1). For people who did
http://www.demographic-research.org
139