Scatter plot for Ph.D. student progress

I experimented with the statistics we have on Ph.D. students in COINS to find out whether there could be meaningful indicators for progress.

A coarse approach to measure progress (or efficiency) is to start with equating a Ph.D. training with 5,400 hours. Most students take more time, but normalised time is 180 ECTS = 5,400 hours of student work. 30 ECTS (900 hours) are dedicated to the course work. During the Ph.D., candidates will write and publish ca. 6 scientific papers that form a cumulative dissertation. Of course, you can obtain a Ph.D. without publishing 6 papers, and you are not at all guaranteed to pass just because you (co-)authored 6 or more papers. Let us assume that it takes 300 hours to write an introduction to the cumulative dissertation. Then 6 papers account for 4,200 hours, each paper contributes 700 hours. (Time to attend COINS events, pursue side projects etc. is not included in this coarse measurement.)

We have most of the data readily available for all COINS students. We know how long they have been working on their Ph.D., we know which publications have been registered with them as co-authors, thanks to the national Cristin database. What we currently lack is data on the course work completed. That might be retrievable from FS, I guess.

I made a scatter plot based on the data we had at hand for some of the students. Here is how it looks like:

The x-axis shows the year of the Ph.D. The y-axis shows the „score“ of the student, calculated as min(#publications,6)*700+min(#ects,30)*30), yielding values in the range 0-5100. Students with no course work completed are not shown for technical reasons.

There are some clusters of students:

  • Typical 1st year students: some course work done, first publication out
  • Typical 2nd year students: more course work done, more publications out, large variation
  • Typical 3rd year students: most course work done, publications in pipeline
  • Students after 1st year with no course work and no publications (not shown in graph) require individual follow-up
  • Students taking more than normalised time for individual reasons require follow-up to not drop out
  • Students with lots of publications and comparatively little course work completed show high research productivity, but risk completing beyond normalised time (and run out of funding)

The year of the student is based on the calendar year of student registration and generation of the graph, i.e. it might deviate +/- 1 year from when calculating with exact dates.

Future work: Automatically include data on completed courses, show all students in a scatter plot, solve the challenge of students having the same score in the same year and still being visible in the graph, put names or pictures in the graph instead of circles.

About Author: Hanno Langweg

Comments are closed.

Search

Postings