mpjashby/crimemapping

Warn about interpreting correlations visually

Closed this issue · 2 comments

In some of their assessments, students have claimed 'strong relationships' or similar between variables based on a scatter plot, even when there is little or no actual relationship. Add a note to the section on scatter plots to warn students not to attempt to judge relationships visually.

This misunderstanding may be related to the text that teaches students to add a trend line to a scatter plot, since R will add a trend line regardless of the strength of any relationship. Maybe illustrate this by showing a matrix of scatterplots based on random data (with some variables skewed), showing the different trend lines produced.

Will be fixed with #76