Test Deconstruction

Jason Buell

I used to pre-test my students, for two main purposes:

1. I wanted to use it as a diagnostic to see what they already knew.
2. I wanted to focus them on what was coming.

As a diagnostic, the test was pretty useless. All my students come in with essentially zero content knowledge of what we're going to learn. A few might be able to shout out a half-remembered vocabulary word, but I haven't had any students who can go beyond that. I get the information I really need from whatever I use to launch a topic; for example, with the ball and hoop demo for when we learn about atoms or by just having them predict what will sink or float when we start on density and buoyancy.

In terms of focus—well, pre-tests didn't work so great for that either. Students would fail their way through the pre-test, and because they had zero pre-exposure, none of what they saw on the test would stick. They didn't have anything to anchor it with.

Now, instead of pre-tests, I use test deconstruction. In my teaching cycle, this happens after students have run their own experiment and teased out the big ideas. I give students a copy of a test I'm planning to give them. The questions aren't identical, but the format is the same and the questions test the same standards. In their notebooks, they draw four columns.

The first column is just the letter of the standard, which they can find listed next to the test questions.

The second column is, "What do I need to do?" They should write what the question actually requires them to do. Do they need to label a diagram? Explain something? Draw a picture? Fill in the blank? They have a copy of Costa's questions in their notebooks to help them along (see Figure 1).

Figure 1. Costa's Levels of Questions

The third column is, "What do I need to know?"

The fourth column is key vocabulary. Their chart should look something like this:

Obviously, identifying the content knowledge and vocabulary required is an important reason for creating this table, but there are two other things I'm trying to accomplish.

1. I want them to understand how fundamentally different the question in the figure below is from (A) "Draw and label" or (B) "Select from the word bank and label the diagram" or (C) "List the subatomic particles of an atom."
   
   

   
   
   In the first example, they'll need to know the parts of an atom, where they are located, and what those little symbols might mean. They're not asked to do the actual drawing from memory, as in example A (this comes later on, when we start with the periodic table).

   In B, they're only asked to recognize, not memorize, the names and know the locations. In example C, they're only asked to remember the names, but not the locations, or to know what those symbols are. However, they are required to know what "subatomic" means.

   These are different questions that will require different levels of knowledge and different skills. More important, they need to prepare for these differently. I can't just tell failing kids to study. They don't know how to study. It's something that needs to be taught.

2. I want to create student-friendly learning goals. One of my rules for teaching is "don't do the thinking for your students." If your learning goals are in teacher language (e.g., "Identify and label the subatomic particles of an atom") and you translate that for students, you're doing the thinking for them.I want my kids to know what "identify," "label," and "subatomic" mean. Why in the world would I translate that for them? The filled-in columns of our deconstructed test questions translate easily into a learning goal for each standard. Students write them up in their science notebook next to that topic's table of contents. Whenever we add something to the table of contents, we can refer to the learning goals at the same time.

Now when I say, "Today we're going to work on identifying and labeling the subatomic particles of an atom," they've already got something written up to decode what that means.

Jason Buell is a middle school science teacher in San Jose, Cal. He blogs at Always Formative. Reach him on Twitter @jybuell.

ASCD Express, Vol. 7, No. 9. Copyright 2012 by ASCD. All rights reserved. Visit www.ascd.org/ascdexpress.