Wednesday, November 27, 2019

You can't change what you don't understand


Charles Ungerleider, Professor Emeritus of Education, The University of British Columbia

[permission granted to reproduce if authorship acknowledged]


Confronted by what seems to be a problem, educators often jump to a solution without knowing exactly where the problem lies. It’s not surprising, but it is a distressingly common and costly situation in education that occurs at the classroom, school district, and at the provincial level.

Imagine eight schools in which only 75% of students meet or exceed the “provincial standard” using a levels approach to educational measurement. What I mean by a levels approach is that the score a student earns on some assessment indicates that: the student has not begun learning or was excused from the assessment (level 0); has begun learning but hasn’t made much progress (level 1); s/he is progressing but is not quite at grade level (level 2); is firmly performing at grade level (level 3); or is exceeding the expected performance for students at the grade level (level 4). Although the meaning of the symbols may differ, the levels approach is the same as assigning grades (A, B, C, D, F). 

On paper, the 75% figure for level 3 and 4 students may look good. But merely adding together the level 3 and 4 students doesn’t inform decision-making at the school, district or provincial levels about students who are not at or above provincial standard.
Furthermore, in a school where 25% of the students have not met the provincial standard there may be a relatively large proportion of students who have not even achieved level 1 or who have been excused from the assessment. Acknowledging the percentage of students at each level provides a more complete understanding of student performance.
Providing Additional Information


Weighted averages provide more information than can be conveyed in a simple percentage. A weighted average – instead of treating all scores equally – considers the proportional relevance of each score.
I have computed weighted averages for the performance of students for a set of fictional schools. I did that by multiplying the number of students at each level by the value of the level achieved (level 0, level 1, level 2, level 3, or level 4) and adding the products together. I divided the sum of those products by the sum of all the students assessed including the students below level one and the students excused from the assessment to find a weighted average for each school. I think this approach provides a more nuanced look at differences among schools.
Here’s how it looks: an illustration

The table below contains the data from the fictional schools. 75% of the students in each of the schools are at or above the provincial standard (levels 3 and level 4), but there are some important differences among them that affect the weighted average of their scores. For example, the proportion of students at levels 3 and 4 at all the schools are the same, but at Bay View 50% are at level 3 and 25% at level 4, while at River View those proportions are reversed. The students at River View (weighted average 3.15) are outperforming the students at Bay View (weighted average 2.90). At Coleman the proportion of students at levels 3 and 4 are the same as Bay View, but the proportion of students at levels 1 and 2 at Coleman are exactly the opposite of those at Bay view (15 % level 1 and 10% level 2 versus 10% level 1 and 15% level 2). That small percentage change is reflected in the difference in the weighted average for Coleman (2.85).

By comparing the schools where the colours are the same (Bay View with Westbrook, River View with Sea View, Coleman with Oceanside, and Queen Anne with St. Lawrence), you can see the weighted averages decline when students who have been excused and students who are below level 1 are included in the calculation. This provides a more accurate picture of the results obtained. It also has a positive impact on morale when teachers efforts across the entire range of levels are recognized and acknowledged, as opposed to reporting only the percentage that met or exceeded the provincial standard.
Schools in which the percentage of students meeting the provincial standard were identical can be differentiated from one another. Examined more closely, the differences paint a more subtle and complete portrait of student performance. The numbers are just a starting point. Once you have them its time to dig deeper to diagnose what’s going on. You cannot change something without understanding it.