Residuals on the TI-82
by Tom Linton, http://www.central.edu/homepages/lintont
Here is one way to calculate (rather quickly, once you're used to it) residuals
from a linear regression on the TI-82.
-
Using the statistical editor, enter the X values of the data in L1 and
the Y values of your data in L2.
-
Open the [STAT] [CALC] menu (press [STAT][right-arrow]), press [9] and
then [ENTER] to run a linear regression. The intercept a, and the slope
b, should appear on your screen. The least squares regression equation
is y = a + b*x. Store the regression equation asY1 by pressing [Y=] (clear
any old formula from Y1 if one is present) and then press [VARS][5:Statistics]
followed by two right-arrows to select the [EQ] (equation) sub-menu. Press
[7:RegEQ] and the equation of the least-squares regression line should
appear as Y1.
-
You can quickly store the residuals (true Y value minus predicted or regression
line value) in L3 by starting the statistical editor (press [STAT][ENTER]
), moving over to the L3 column, moving up to the top row (which says L3),
pressing [CLEAR] and then [ENTER] to erase L3, and then typing
L2 - Y1(L1) [ENTER]
(be sure to get Y1 from [2nd][Y-VARS] ). The residuals should
appear in the L3 column of the statistical editor.
The residuals are now in L3 and, for example, you can see a residual plot
by (turning off all Y= functions and other stat plots) setting up one of
the stat plots to show a scatterplot of X = L1 and Y = L3. If you use [ZOOM][9]
(zoom-stat), the calculator will select a good window for the residual
plot.