Reporting and Predicting Grades in IB
- Kieran Burgess
- Mar 10
- 5 min read
Are your reports truly reflecting what you value the most in your school? Do you have a consistent, reliable methodology for predicting final grades? How are you bridging the gap between IB grading and your local or national system? In a recent Ibicus webinar we addressed the challenges in and options for reporting student progress and attainment in academic and lifelong learning areas across the IB continuum, paying special attention to the Diploma Programme (DP). We explored how to systematically predict grades, and acknowledged the inherent challenges in predicting DP grades for university applications.
We discussed different reporting practices: what different schools report, including criteria achievement, approaches to learning (AtLs), learner profile attributes, competency mastery; how you could report clearly and concisely, and when in a way that provides most reporting contexts with the information they need. We also covered key principles of converting IB grades to align with local and national systems.
Predicting Grades
Aggregated data from grade analysis contractors and a consortium of IB grade researchers indicate that just under 50% of grade predictions made at the end of the DP course (those predicted grades submitted to IB just before exams) are accurate. Less than half. Why?
Teachers often fall into a number of traps when it comes to predicting grades (PGs):
Basing PGs on elements that are not explicit assessment or curriculum outcomes (e.g. effort, attitude or linguistic fluency)
Basing PGs on too few data points
Using PGs as a motivator or a punishment
Relying on too narrow a window of assessment methodologies (e.g. past exam paper only)
Conflating marks with grades
Mixing up multiple grade boundaries from multiple years out of multiple contexts.
These last two are very common. There is a logical fallacy that we can only predict grades based off a student completing a like for like assessment: a mock exam or past paper question under timed conditions. But this discounts a few important points:
Students and their families prepare and make space for the highest possible performance during exam time. In many cases, the exam will be their best version of themselves because it's the only thing in their life for that month that they have to do.
IB exams are not one-shot only. Every grade is based on multiple components over multiple days, including papers sat on different days and IAs that take place in different months.
Each IB exam attracts marks, which are only converted to grades based on many factors: the quality of the cohort compared to previous years; the difficulty that the cohort had in answering the question; global events, and so on. The grade descriptors are used to determine which mark hits which grade. The mark only gets applied to a grade level after all the papers have been awarded a definitive mark. Thus, an in-school mock exam will (hopefully) mix past questions from several years prior, sat by very different students in very different contexts. The grade boundaries only applied to them, and may even have varied between years. That mock exam you've just marked can give you a specific mark but, at best, will only offer a range of grades, because of the mess of the mixing of contexts and grade boundaries. In short, you can only apply last year's grade boundaries to last year's exam questions, to last year's cohort.
Instead, teachers should use the IB grade descriptors for their subject (or group) to determine what a student has demonstrated - a series of can-do (or have-done) statements, essentially. They should look at all the evidence they have of a student's understanding first. Every assignment they completed. Every presentation they gave. Every piece of classwork they did. Every time a student said something relevant in a group discussion, or put up their hand to offer an answer or insight. Their IA thinking and product. By always being alert to which grade descriptor a student is demonstrating they can do, and by knowing the standards of their DP course, the teacher will build up a much more accurate picture of the level of comprehension and skill a student has when compared to the IB standards for the subject. A large body of evidence, accumulating over time, built up of many, many tiny data points - the student won't even be aware that many of them are even assessments - and some larger ones. The teacher then looks not at the outliers in the data, but at the trends and consistencies. Then, which grade descriptor does the student's evidence best fit into? That's the grade they're looking at.
A student who demonstrates a grade 3-4 consistently in class work, discussions, 1-1 conversations, homework... who then scores top marks on an end of unit test one day, isn't then suddenly bumped to a 7 for their predicted grade. The anomalous data point there is the full marks on the test. Why is that? Did they fluke it? Did they cheat? Was the test badly designed? Unless they can consistently show that level across multiple methods and topics, I'm going to notice it, but ignore it. Likewise the student who consistently demonstrates 6-7 grade level competency, but who scores a 3-4 in the mock exam. Maybe they had a bad day. Maybe they were ill. Maybe their grandfather died yesterday and their mind wasn't in the zone. Maybe they were thrown by the sudden introduction of an environment they weren't expecting. I will notice the anomalous low grade and talk to them about it, but not hold that against the consistency of the 6-7 work. But if I've pinned all my predictions down to 1 or 2 major assessment events then I don't have the luxury of using my professional judgement to accurately assess their true understanding.
The knock-on effect to reporting
Learning is messy. It goes backwards sometimes, and never goes at the pace or down the paths we intend or expect. If students fear that every mistake will be numerically affecting the predicted grade that will be reported to universities, and thus they worry that every mistake will ruin their chances of their 'dream' university (or their parents' dream), they aren't going to learn effectively, and may develop serious wellbeing issues. We want to encourage students to fail fast and often so that they learn faster and better. We want them to take risks, to learn from mistakes and to experiment. So when we report on progress, we should do 2 things:
Smooth out the bumps: where will the student likely be at the end of the course, based on all evidence to date and the teacher's expert knowledge of their course? Only reporting on currently evidenced skill or competency might not achieve this, especially in subjects that just need time to build skill.
Report what we truly value as a school. If we say that our school cares most about the holistics development of the whole child, but then report primarily or only academic grades, then we: 1) tell the students and parents that we don't really care about the whole child, and 2) tell them that we don't mean what we say, and that that's ok for them too.
In the webinar, we looked at reporting on other areas beyond grades, and how this might look on a report. We touched on some systems out there already that are moving away from old-fashioned reduction of human beings to a list of numbers, and explored how data can be a powerful tool for achieving 3 things:
Representing that we do what we say, and that we are who we say we are.
Reporting 'internally' within the teachers, students and parents to identify how learning can be better.
Reporting 'externally' to universities and IB with integrity, in a way that elevates, not diminishes, learning as a vital human process.
The slides are here!
Comments