Non-Technical Summary:
A great deal of research has focused on the decision to attend college and college outcomes; vocational education has received far less attention. Individual ability and family background are strong predictors of a student’s educational path. Less advantaged and less able youth are substantially more likely to choose vocational education. We use a nine-year panel following two cohorts of students completing compulsory education in Denmark to examine the decision to enroll in and the probability of completing vocational education, taking into account several complicating factors.
One of the most obvious factors an analysis of vocational training (VET) should address is the heterogeneous nature of VET programs. We differentiate between technical programs providing the skills needed in the construction, transportation, and manufacturing sectors; educational and health related programs providing the skills needed for medical and teaching assistants; and business related programs providing the skills needed for purchasing, financial, and general office assistants.
We also recognize the multidimensional nature of skills. Nationally administered and normed math and language exams are employed to capture these two different types of cognitive skills. Given these controls for cognitive skills, data on teacher-assigned grades are used to capture non-cognitive skills. Just as different skills are predictive of success in completing different academic programs of study, we posit that different skills will have different effects on completion of different VET programs. Some evidence from apprenticeship programs suggests that math skills are important for both technical and business programs. Language skills are likely particularly important for business programs.
As is the case with academic programs of study, attrition is a problem. As many individuals enter VET training later in life, we must also acknowledge those who are still enrolled at the conclusion of the sample period. We model VET completion separately by gender and program type, controlling for selection into VET and those still enrolled. As expected, we find that all skills are inversely related to VET enrollment. The association between skills and completion varies considerably by program type. While math scores are positively related to certification for all VET tracks, language skills are more important for those in non-technical tracks, and non-cognitive skills appear important primarily for the business track.
Thus, we provide evidence that cognitive and non-cognitive skills are important not just for success in academic education, but also for VET education, and that different VET programs require different skills. These results demonstrate that employing a one-size-fits-all approach to an analysis of student persistence in VET is inappropriate. Future research should take these factors into account in order to model VET better and to aid decision making by students, parents, educators, and policy makers.