This post continues the conversation about The End of Average by Todd Rose.
There are plenty of times where considering the average of a group makes sense. It’s a way to improve predictions or estimations about large sets. Weather forecasts, experimental data, political polling, insurance pricing, and medical predictions are all improved through measures of central tendency. Averaging data makes a great deal of sense intuitively and adds value to many processes.
The problems with averaging, especially in education, arrive when we make what Peter Molenaar calls “the ergodic switch” – replacing information about an individual with information taken from an average. Knowing an average about a group might improve a prediction or estimation, but it doesn’t tell you much with certainty about an individual.
According to ergodic theory, you are allowed to use a group average to make predictions about individuals if two conditions are true: (1) every member of the group is identical, and (2) every member of the group will remain the same in the future. (Rose, 2016, p. 63)
Of course, many data sets where we use averages are not actually ergodic. The fact that the month of July averages only 0.71 inches of rainfall is small consolation if that rainfall occurs during your wedding or fireworks display! Knowing the average skills of students entering kindergarten doesn’t tell us much about the performance of the specific students in our classes. The fact that people are not ergodic is a key reason for all forms of assessment, especially pre-assessment. We need to know our students to design effective learning experiences for them.
Applying averages to students is further complicated because there are so many ways in which students are potentially different from one another and over time. Students vary in their interest in certain topics, performance during different times of the day, working memory, and social skills to pick just a few dimensions that would be relevant to any teacher. Furthermore, any one of those dimensions contains multiple components: social skills contains distinct skills like small group communication, large group communication, listening, leadership, etc. Over time, students will grow in these areas (and more) at different rates, highlighting the need for personal learning profiles (PLPs) that are continuously updated by both educators and students. Summative assessment data won’t be nearly enough to make valuable PLPs – non-cognitive skills, preferences, and student voice will all need to be included.
The next post in the series will look at some specific ideas and recommendations from Rose and how our shift toward personalized learning will benefit learners throughout the district.