How to automate grade checks and create grade statistics

If you’re the person responsible for grades, do you want to manually search through 6,500 grades in the grade catalogs, or would you rather do it with the push of a button? If you’re a principal, do you want to hand out a grade catalog containing only grades for teachers to analyze at the end of the year? Or would you prefer to give teachers and school leadership a color-coded catalog, with average scores, subject-by-subject statistics as well as boys/girls like Skolverket, together with absence and national tests? Also with, like, one button press. If so, I have a script for you!

I’ve been responsible for the administrative part of grades at Årstaskolan for many years. In short, that means making sure teachers can enter grades for their students in Skolplattformen (Stockholm), printing a paper preliminary grade catalog that teachers review and make their changes in.
At the same time I go through all students’ grades in the grade catalog, about 6,500, to see if there are any problems. Then I communicate with the teachers, make corrections and finally print a final grade catalog along with the grade documents for the students. After the city has sent the grades to Gymnasieintagningen, I check that all Year 9 students’ grades are correct and finally answer for anything SCB contacts me about.

Problem 1 - the data we have

The big problem is that the grade catalog and the grades only come in PDF format. So you can’t do much with them in terms of analysis. I don’t know how many people are sitting manually converting grades into their values to get various numbers, for example which students should receive scholarships and similar. Please, all of you who decide on and buy platforms, services and the like. We in the operations need to be able to extract the data we put into the systems!

Problem 2 and 3 - Troubleshooting the grade catalog and Analysis of results

Troubleshooting Trying to find errors among 6,500 grades across a bunch of grade catalogs, to make sure everything is legally secure, is no joke. Above all it’s insanely tedious work.
After all these years it can be things like a missing * asterisk that marks final grades, or there being a * even though it shouldn’t be there, missing grades, grades present in both svenska and svenska som andraspråk, extra checks for students with adapted study paths and so on.

Analysis Just analyzing grades and assuming that grades equal outcomes isn’t great, since grades are no longer a very good indicator of results. Automating at least to produce a catalog with averages, broken down by gender and similar (as Skolverket does) is a start. But the time saved through automation gives us the opportunity to spend extra time adding additional data points, such as national test results and absence in the same catalog. There is of course more data we can add, like Stanine, whether the student has access to additional accommodations, special-teacher support and more.

If you’re a principal planning and allocating resources for the coming school year you naturally want as much data as possible. Similarly, as a teacher of incoming Year 7s it can be useful to have both grades and national test results collected for the current Year 6s you’ll be receiving, to plan your teaching.
A student health/welfare team might be able to find more students at risk of falling through the cracks and also look at correlations between absence and results or grades. You can of course copy and save all grade summaries into an Excel sheet so you can access and follow a class over several years, which many people ask for.

The solution for you and me :)

Based on my experience working with grades, I coded together a script that partly performs a troubleshooting pass and creates an Excel catalog where errors are color-coded. The second part creates a color-coded analysis catalog (Excel) when you’re done with the troubleshooting.

If you work in Stockholms stad, this will work for you. If you work in another municipality but also have the same kind of grade catalog it should work too. Otherwise you can tinker with the code a bit if you want. The script is verified to work on Windows. Since many people use PCs and Windows I want it to work for as many of you as possible.

Here’s how to do it!

If you haven’t seen the video at the top of the post, watch that first. You can find the code and instructions at github.com/mickekring/betyg_statistik_och_kontroll If you want to know how to install Python and run a script on a Windows computer you can find that at mickekring.se/sa-har-kor-du-ett-pythonscript-pa-din-windowsdator

What have I gained from this?

Regarding troubleshooting, Sweco (on assignment from SCB) contacts me with all the errors that are made and must be fixed for the official statistics to be reliable. This spring they didn’t contact me. No errors. The term before it was two errors. Before that an average of around 20. Above all it’s also a damned boring and time-consuming task that I no longer have to do during an otherwise quite stressful time. Regarding analysis I hope it will help the school move forward. Unfortunately I won’t be part of that journey.