Demographic Research: Volume 21, Article 6
interview and the date this level was reached. Using only this rudimentary information,
we conduct four different imputations based on alternative assumptions about how the
education question is phrased in surveys that do not include complete histories. For
additional comparison, we also impute histories using only the highest degree at
interview, without any information on the date this degree was attained. We then re-
analyze first birth risks using only these imputed education histories and compare the
results with those produced using complete education histories. Our aim is to see to
what extent the results are comparable.
The following section discusses possible problems that imputation may provide for
fertility analysis. The third section describes the data set, our data preparation, and the
procedure used for imputing education histories. The fourth section then presents the
results of our model for the impact of education on rates of transition to first birth using
complete as compared to imputed histories. We provide a summary in the fifth, and
final, section.
2. Potential sources of bias when using imputed histories
The basic method of imputation used in this study is to consider respondents to be
continuously enrolled in education up until the time they attained their highest degree.
Education histories could be imputed perfectly in this manner if everyone followed very
regular educational pathways. If everyone completed school2 at a normative age,
transferred directly from school to vocational or university education, and without
interrupting their vocational or university education completed exactly one degree,
imputed histories would correspond precisely to actual education histories.
However, educational trajectories will always be irregular to some extent. The
diversity of education histories, and thus also whether education histories can be
approximated sufficiently closely using only the date of attainment of the highest
degree, or even only the highest degree itself, depends strongly on the country context.
In some countries it is common to exit and reenter the educational system and to
acquire higher levels of education even after a considerable period of employment. This
appears to be the case especially in the United States, where young labor market
entrants often return to education on a full-time basis, as well as in Great Britain, where
vocational credentials are frequently acquired on a part-time basis at the same time as
2 The term ‘school’ used here refers to secondary education, not to post-secondary education. This
corresponds to the terminology most often used in the German context. In the remainder of this paper,
secondary education will continue to be referred to as ‘school,’ since this is shorter. Vocational and university
education on the other hand will be referred to as ‘post-secondary education.’
http://www.demographic-research.org
137