Rate of Decline of Arctic Sea Ice and Regression Modeling

By Amanda Walker (Texas State University, San Marcos)


This activity is taught in an introductory statistics college course using statistical software, Rguroo, approximately 40 students in a computer classroom. The activity can be adapted to a high school course using Excel or TI calculator.

Students will explore properties of linear and quadratic regression using Arctic Sea ice measurements collected from NASA satellites. Arctic Sea ice levels have declined significantly since measurements began in 1979. To make sense of this reduction and understand the rate at which sea ice levels are shrinking, students will create scatterplots, construct LSRL, describe the rate of change by interpreting the slope of the LSRL, and discuss the implications of extrapolation to make predictions of sea ice levels in future years.

Students will interpret residual plots and determine if a nonlinear model, quadratic, would be more appropriate. Comparing coefficient of determination and residual plots for each model students explore selecting the “best” model, and implications for the rate of decline in sea ice levels. After comparison, it is clear the quadratic model is the superior model.
Witt, G. (2013). Using Data from Climate Science to Teach Introductory Statistics. Journal of Statistics Education, 21(1), http://jse.amstat.org/v21n1/witt.pdf