This general method is a worthy one

closer to the bottom

of the shaded area

than to the top of the

area. This fact shows

that, though there

are several graduates getting high salaries much above the average

income, nevertheless, if all of the dots were shown, most of the dots

would fall below the average line rather than above the average line.

A few extremely large incomes near the maximum line of the chart could

greatly raise the average line but still most of the dots would be found

beneath the average.

Fig. 166 is a very interesting correlation chart. Here a single dot has been used for each observation and the dots are shown in the chart. Instead of showing each dot in its exact position, the dots have been grouped so that all the dots are arranged uniformly inside of squares formed by co-ordinate lines spaced 10 units apart in the scales for examination marks. Classification by class limits 10 units apart on the scale of marks is sufficiently accurate for all practical purposes, as is proved by the gradual change in shading on the chart as a whole. With few observations it might be desirable to show dots on the chart to actual scale rather than in classes by tens.

Ordinarily a line drawn like the heavy wavy line in Fig. 166 would be so placed that the points on the line would be at the center of gravity for the dots vertically on either side of the line. Here, however, the line is so drawn that there are an equal number of dots on either side of the line, at right angles to it at any point throughout its course.