Wednesday, November 5, 2014

Skyward: Elementary Trend Calculations

PPS Elementary Teaching Staff:

We have received a number of questions the last few days regarding how the Skyward gradebook calculates the M, P, and L proficiency levels. This, admittedly long, yet important read, addresses these questions.

The Skyward gradebook uses trend-based calculations when determining the M, P, or L for a student's level on any given skill in the gradebook. The theory is, we want to report out where a student is on a skill today, and not a simple average of their scores over time. However, there are cases where a trend calculation can produce a proficiency level that, while correct from a mathematical perspective, does not seem to make sense. ​As with any automatic grade calculation, we recommend you spot check gradebook.

Overriding a Trend Calculation

This is a simple process - if you click on a skill's title in the gradebook, you will see a view which shows all of the scores incorporated into the skill, the trend score the gradebook is reporting, along with the mean and median scores for your reference.

As you review these scores, you can quickly change any calculated value to one that makes more sense if needed. In the example below, I have highlighted the three scores and overwritten one that didn't seem correct. (The student has a calculated P due to the last assessment having an L, but perhaps an M makes more sense?)

What Causes Trend Calculations to Not Work Well

I have looked at a number of gradebooks where teachers have reported odd results. In most cases, the odd marks are caused by one of three things:
  1. Low Max Point Values  Trend-based calculations do not work well when assessments are scored with a point total of 10 or less. The lion share of the gradebooks I have reviewed where the trend calculation is not working have assessments scored on 4 points or less. A student earning a 1/4 on an assessment will have that score converted to a 25% on the back end and will have a hard time, mathematically, overcoming it. (This happens in traditional gradebooks as well.) Solution: Enter low point assessments into the gradebook using M, P, and L marks directly instead of using points. 
  2. Overuse of One Date  Trend calculations incorporate the date between assessments into the calculation. If you enter in several assessment scores, all on the same date, the gradebook cannot differentiate for the purposes of a trend. Solution: During assessment entry, use the actual date of the assessment on the date fields, and not the date you are entering in the score. (It is not a problem to wait to enter data, just make sure you set the event dates to match when you gave the assessment.​ 
  3. An Assessment with Scores Vastly Different than Others  If you have a situation where a lot of students receive an L on an assessment whereas the bulk of of previous assessments on the same skill were at the M or P level, you can see the scores drop significantly. Solution: Consider not counting assessments that are outliers from other scores.

Looking Forward

For what it is worth, our last gradebook also used trend calculations when calculating a P, D, or B. However, it used a different algorithm. For example, with three or less scores it used the mean score and not the trend. It is our goal to work closely with Skyward to see if changes can be made to the way the grades are calculated in order to produce a grade that more closely and consistently aligns with the student's performance over the quarter.