the student appears to order the remaining cases. Fifty students
represented orthogonal covariation with the variables on opposing axes. In some
cases axes were unlabeled, but units made clear the variable measured on at least
one axis. Thirty represented study time on the horizontal axis and scores on the
vertical, whereas 20 interchanged the axes. Forty students used conventional
ordering of values on the axes, that is, increasing value as one moves up or right;
seven reversed the values on one axis (giving the visual impression of a positive
covariation); and three showed values unordered in bar graphs (giving the visual
impression of no covariation). Thirty-one responses appeared to indicate a perfect
linear fit with values of equal spacing on each variable, and the other 19 showed
some variation of a perfect linear fit. Students differed in the form of graph used: bar
graphs (25), scattergraphs (7), line graphs (5), and line graphs of connected dots
(13). Figure 7d (G9m) shows a line graph of connected dots with conventional axes
and linear fit.