Zabel: Imputed education histories and fertility analysis in the western German context
histories during which respondents held a vocational degree was correctly allocated to
the category ‘vocational degree’ in the imputed histories (Table A1a). However,
additional exposure time is also allocated to this category in the imputed histories as can
be seen in Table A1b. Only 93% of the exposure time allocated to the category
‘vocational degree’ in the imputed histories actually stems from this category in the
original histories, while 6% originates from the category ‘in vocational training’ in the
original histories. In this imputation, we are assuming that we know only the first date
the respondent obtained the highest degree they held by the time of interview. As
described in the methods section, respondents are assumed to be enrolled in education
up to this date and then to be holding this degree without being enrolled in education up
until the date of interview. Thus, if respondents reenter education after the first time
they received a degree at their highest level, this reenrollment is not registered in the
imputed histories. This appears to be reflected in the 6% of the imputed exposure time
for ‘vocational degree’ during which respondents actually were enrolled in vocational
training in the original histories.
The other way around, for the imputation using the last date the respondents
received their highest degree (Tables A2a - A2c), not all of the exposure time from the
category ‘vocational degree’ in the original histories is correctly allocated to that
category in the imputed histories. Of the exposure time that respondents actually spent
holding a vocational degree and not enrolled in education, 10% is instead allocated to
educational enrollment in the imputed histories (Table A2a). In this imputation, we are
assuming that we only know the last date the respondents received their highest
degrees, and we are imputing them to be enrolled in education up to that date.
Therefore, whenever there are gaps between education spells, original time spent
holding a vocational degree in these gaps is mis-imputed as time spent enrolled in
education in the imputed histories. By contrast, for this imputation, 99% of the imputed
time spent holding a vocational degree actually originates from that category in the
original histories (Table A2b).
Thus, the extent of misallocation of exposure time into (first imputation) or away
from (second imputation) the category ‘vocational degree’ is comparable and not very
large. In the first imputation, we only have the first date the respondents received their
highest degree. As we have seen from Table A1b in the appendix, time that is actually
spent enrolled in education after educational reentry is imputed as time spent holding a
vocational degree without being enrolled in education. This will tend to lead to an
underestimation of risks of first birth for the category ‘vocational degree,’ because time
spent enrolled in education, during which first birth risks are typically very low, is
added to the ‘vocational degree’ category in the imputed histories. This is reflected in
the disproportionately low number of events (Table A1c) added to the category
‘vocational degree’ as compared to the amount of additional exposure time (Table
150
http://www.demographic-research.org