Using JMP Student Edition, Third Edition

138 6 Fit Y by X Platform Introduction

Figure 6.10 Shaded Confidence Curves

The equation of the line, as well as several computed statistics, are found in the Linear Fit report. Val-

ues of the slope and intercept are also printed in the

Parameter Estimates section of the report.

Aside from the graphical confidence-curve method detailed above, there are numerical measures of the

significance. One is the p-value associated with the slope of the line, also found in the Parameter Esti-

mates report. In this case, the p-value is 0.0045, significant by almost any standard, reinforcing the

graphical results above.

Correlation Coefficient

Another measure of fit is the correlation coefficient, frequently denoted by r. Its value does not appear on

any of the reports so far, although the square of its value (r

2

) is listed beside RSquare in the Summary

of Fit

text report. To compute the value of r itself, request a density ellipse.

 Select

Density Ellipse > .95 from the platform menu on the plot’s title bar.

A new report named

Bivariate Normal Ellipse appears at the bottom of the text reports. It is initially

closed, but can be opened by clicking on the disclosure icon (Figure 6.11).

Figure 6.11 Correlation

The correlation coefficient is listed under the word Correlation. It is interesting to note that its signifi-

cance (p=0.0045) is the same as that listed for the slope coefficient in the

Parameter Estimates table,

and the same as the

Prob>F value in the Analysis of Variance table.

Two-Way Contingency Tables

In the next example, both X and Y are categorical variables. The question is whether the method of

washing denim has an effect on thread count. The analysis uses contingency tables—orderly ways of

arranging count data. To generate a contingency table for this problem,

 Select

Analyze > Fit Y by X from the menu bar.