Compared to students in other countries, the dropout rate in Norway is relatively high (at least if we use OECD data as a starting point), and Norwegian students also tend to spend long time to complete a degree, implying that estimated time to degree commonly is exceeded. This phenomenon is partly linked to the patterns of HE attendance among students in Norway; their HE attendance is rarely linear. Compared to other countries it is common for students to transfer from one program to another during their studies, particularly for students in general university undergraduate programs. The same type of pattern is not found among students attending more professionally oriented programs at university colleges, such as nursing, social work and engineering at the bachelor's level, even though the admission requirements for both types of programs are quite similar. The presentation will address the structure of dropout in higher education in Norway, and also talk about how institutions work to reduce or prevent dropout.

There has been a growing emphasis on the use of measurement and data to underpin and direct educational decisions and practices. These extend right from teaching the individual child, and as envisaged in the data intensive AI systems, {Luckin, 2016 #3605}{Luckin, 2016 #3605}to the bases of educational policies at national or even global levels. Given the enormous thrust in this direction, I note that there are several challenges to address. I will ask if there is an unwarranted faith in the guiding value of data implicit in a multitude of advisory documents and national and international policy efforts extending to all levels of the educational edifice. I will argue that data, however good, gives far less guidance for action than is often assumed.