Visualizing Value-Added
In our previous post, we examined how value-added scores are calculated. Now let's visualize them.
To calculate a value-added score, we have:
value-added score = actual growth - expected growth
expressed in grade levels per year. Generally, expected growth is 1 grade level per year, plus or minus a bit for economic hardship and other factors. We're going to start by showing the expected growth at the center of our new graph, marked by a bold vertical line. Let's say it's 0.85 grade levels / year:
Having expected growth at the center emphasizes that value-added is a difference from this expected growth. Next, we add the actual growth. Let's say it was 0.92 grade levels / year:
Finally, we draw a bar between the two, and add a text box with the score to the left of the graph so viewers will see it first. The separate text box emphasizes that the value-added score is the most important number here, being the result of the calculation. The actual and expected growth are of secondary
importance, being inputs to the calculation.
I'm going to recommend a fourth item. We should show the error in the calculation, since it can be quite large. We'll add a triangle pointer and a bar like this. We show the error amount under the score to emphasize that the utility of the score depends on the error.
The key differences? Many visualizations, such as in the current state report card, report the value-added score only and highlight whether the score is positive or negative using green and red. The result, while accurate, can look like a Christmas colored bingo card. The viewer flounders around unfocused. Error information, in a separate graph, is rarely looked at - thought it should be, because the error might be large.
Showing and centering the expected growth anchors the viewer and gives us an alternative way to describe positive and negative, using a right-stretching or left-stretching bar. No holiday colors needed. Adding error bars directly to the graph immediately tells the viewer how much confidence to place in the estimate. (A shout-out to the state of Wisconsin for this idea!)
​
Going even further, viewers may ask, why is the expected growth that value? Some methods for calculating value-added give us insight into this (we'll discuss methods in our next post). If our method allows, we can show the details of the expected growth calculation. This could be an expandable box or hover-over to avoid cluttering the main graph. The factors are flexible. I picked a few as an example.
For multiple grades and subjects, we can show one graph per grade / subject pair. We can label the first graph thoroughly, then omit some text from subsequent graphs to save space. Note the graphs don't have to be large to adequately convey the meaning. With separate graphs, we can also sort them, such as from highest to lowest value-added score.
​
OK, so that was 5 items total for those of you counting at home. Still not bad! Now we have a clear, accurate visualization. Next we'll look at the heart of computing value-added: how to calculate the expected growth.
One, two, three. There you have it!
​
If there's a category rating, such as a letter grade or "Does not meet"/"Meets"/"Exceeds", we can add that information using a gradient background and labels.
