Operation Manual

Chapter 13: Inferential Statistics and Distributions 241
LinRegTTest
LinRegTTest (linear regression t test; item F) computes a linear regression on the given data and a t test
on the value of slope b and the correlation coefficient r for the equation
y=a+bx. It tests the null
hypothesis H
0
: b=0 (equivalently, r=0) against one of the alternatives below.
•H
a
: bƒ0 and rƒ0 (b & r:ă0)
•H
a
: b<0 and r<0 (b & r:<0)
•H
a
: b>0 and r>0 (b & r:>0)
The regression equation is automatically stored to
RegEQ (VARS Statistics EQ secondary menu). If you
enter a Y= variable name at the RegEQ: prompt, the calculated regression equation is automatically
stored to the specified Y= equation. In the example below, the regression equation is stored to Y1,
which is then selected (turned on).
In the example:
L3={38, 56, 59, 64, 74}
L4={41, 63, 70, 72, 84}
When LinRegTTest is executed, the list of residuals is created and stored to the list name RESID
automatically. RESID is placed on the LIST NAMES menu.
Note: For the regression equation, you can use the fix-decimal mode setting to control the number of
digits stored after the decimal point (Chapter 1). However, limiting the number of digits to a small
number could affect the accuracy of the fit.
LinRegTInt
LinRegTInt computes a linear regression T confidence interval for the slope coefficient b. If the
confidence interval contains 0, this is insufficient evidence to indicate that the data exhibits a linear
relationship.
Input:
Calculated
results: