Walden Education Grade Prediction Project​

We conducted a small-scale experiment on Twitter to investigate how accurately teachers are able to predict grades (https://twitter.com/Waldenkent/status/980385548727660544).  Over 600 teachers attempted to predict the results of a small group of A-level Chemistry students using a short biography and a significant range of data collected over 2 years (including Year 12 exam results, end of topic assessment average percentage scores, and Year 13 mock results).

Headline figures:

- Highest prediction score: 10/14   (71% accurate)

- Average prediction score: 4/14    (29% accurate)    

Although the results were not necessarily shocking, they were telling.  Our ability to accurately and reliably predict the future is highly questionable. These 'predictions' were essentially made at the last possible moment (two months before the actual exam). Interestingly almost a third of the teachers involved in this challenge are required to provide predicted outcomes for pupils in year 8 onwards.


Some data use in schools can produce work for teachers without providing clear returns for pupils in terms of improving their learning. Time spent producing, analysing and responding to wobbly grade predictions can easily fall into this category.  Furthermore, some school data systems are premised on this predictive data being highly accurate and reliable. Based on this little experiment, it doesn't seem to be.


Real Grade Predictions Project


However we are interested to see if teachers are more accurate with their own students so we are conducting a larger experiment. It will provide you with an easy way to check your own prediction accuracy - and us with insight into the accuracy of teacher predictions in 2018. 


If you would like to participate, click on the red button above to enter your own predictions for your classes this year - and then enter the actual results in August. The sheet will automatically calculate the accuracy of your predictions. 


In order to protect anonymity the form does not require pupils names- just the predicted grade now, and the actual grade when you update it in August/September.  We will only use the data in anonymised form schools and teachers will never be referred to by name. 

We'll send you a reminder to update once the real results are out in August. 

The spreadsheet will automatically calculate your percentage accuracy.  It is a Google Sheets document for easy sharing - however if you would like an Excel version, please email:  mail@waldeneducation.org