Utilizing different data types for grading

As teachers, we know that “content is king”, but it is no longer the entire kingdom. We have to look at both qualitative and quantitative data when assessing and determining outcomes for students on whatever it is that we are teaching. Yes, report cards and transcripts still rely on grades (whatever that means), but more and more there are also learning behaviors, predicted grades and other things that we have to consider. 

Classroom conduct, readiness to learn, cooperation and teamwork, etc are all factors that fall under learning behaviors. Generally, these are qualitative data points that are not quantifiable. You can track each learning behavior every day and come up with quantitative data, which would generally look like a percentage of time that the student is demonstrating the learning behavior. This is great information, but who has time to track all of that data?

Academic scores used to fall under the category of quantitative data (points earned, percentage correct/incorrect, etc), but are moving towards less numeric data to values like Exceeding Expectations, Meets Expectations, etc. While the former is easy to calculate, average**, trend, the latter is not. So how do we track qualitative data when we need to use it as a quantitative value? Here’s some workarounds and techniques I’ve used.

First, as mentioned with the learning behaviors, you could track the frequency that the student is demonstrating and get a percentage of time they demonstrate the trait out of the total. While a single percentage is adequate, I believe that it is better to show the visual as well. Pie charts (or donut charts) are the best for showing percentages of a total.

Simple percentages are fine for your own calculations or analysis.

But a graph is much more impactful if communicating with the student or parent.

When I need to show how a student is doing using a rating scale vs a numeric point or percentage system, I look at 2-3 different statistics to try to get the fullest picture possible. 

Like the learning behaviors, I will convert the rating scale to numeric values. Let’s say we have a rating scale with 5 levels: Exceeding, Meeting, Approaching, Progressing, and Attempting. I convert those to 1–5 with 5 being the top score and 1 being the lowest. Then I will (a) track their progression over time, (b) look at formative and summative data as a whole, (c) look at just the summative data, and (d) if I have a cumulative assessment, I will also consider that. All four of these metrics tell me a story about the student. Here’s a sample student named Chandler who has been studying the volume unit in math.

Finding "averages" from qualitative data may not make sense, but it helps get us into a "ballpark".

We can see his data values for formative and summative assessments as well as “averages”. How do you get a 3.3 or a 3.5? It doesn’t make much sense, other than to know that on average he is falling between Approaching and Meeting. So maybe that is where I start.  I also see that his final summative assessment is a 4, so then it might mean that I consider him to be Meeting. And when I look at his trend over time…

Formative assessments are grey and summative assessments are purple.

Now I can see that while he did have one formative assessment up at a 5, his trend of summative assessments and his work over time (average) really does point him more towards a 4 or Meeting. Note: you could also incorporate the learning behaviors here too, but really they should probably be separated as this is looking primarily at attainment of a certain standard.

This is one of the reasons why I believe that teachers need to be much more data literate. They need to understand how to be flexible with the data, but not be held captive by an average. 

**For my friends who are still exclusively using points or percentages, be weary of your averages. If a student doesn’t hand something in (resulting in a 0) or gets a very low grade (say 30%), know that it can be mathematically impossible to get out of the debt of this one assignment/assessment. Consider bumping to a 50% or some other value that still demonstrates that they did not do what they were supposed to do and does not show understanding, but still lets them dig out of the mathematical debt they just put themselves into.