Students' Understanding of Linear Regression

Abstract

This paper analyzes students¿ beliefs and difficulties in understanding linear regression methods. Two groups of students are considered. The first one is involved in an introductory statistics course in the computer science engineering degree. The other one is formed by students in the fourth year of the environmental science degree. In this second group the statistic course contains more advanced topics. Learning and assessment activities of both courses are presented. Results show that reasoning about association when using graphical and numerical methods, implies misunderstandings such as evaluating the degree of correlation, and interpreting that it implies causation. Teaching of linear regression models is designed to understand that these models are useful approximations to reality, but they should never be considered the final word.