REGANOVA Procedure

@REGANOVA is a post-processor for a linear regression which displays an ANOVA table. It has no parameters, picking all information out of the accessible variables. To use it, run a LINREG first, then do @REGANOVA.

@REGANOVA(no options or parameters)

Example

* Makridakis et al, Forecasting Methods and Applications, 3rd edition

* Example from pp 205-207

open data pcv.dat

data(format=prn,org=columns) 1 19 gdpw_eur pdpeall

linreg pdpeall

# constant gdpw_eur

* Scatter plot with regression line

scatter(footer="Figure 5-12 PCV Sales Regression",line=%beta,$

hlabel="GDP Western Europe",vlabel="PCV Industry Sales")

# gdpw_eur pdpeall

* Residuals

scatter(footer="Figure 5-13 Residual Plot",$

hlabel="GDP Western Europe",vlabel="Residuals")

# gdpw_eur %resids

* Analysis of variance table (page 214)

* (The F-statistic is in the standard regression output)

@reganova

Sample Output

This first shows the output from the LINREG, with the usual F-statistic. The @REGANOVA output is below that and gives the breakdown on the components that go into the F. The ratio of the two mean-squares is the F.

Linear Regression - Estimation by Least Squares

Dependent Variable PDPEALL

Usable Observations 19

Degrees of Freedom 17

Centered R^2 0.9010023

R-Bar^2 0.8951789

Uncentered R^2 0.9999442

Mean of Dependent Variable 8.7249938421

Std Error of Dependent Variable 0.2129504110

Standard Error of Estimate 0.0689450232

Sum of Squared Residuals 0.0808080758

Regression F(1,17) 154.7211

Significance Level of F 0.0000000

Log Likelihood 24.9113

Durbin-Watson Statistic 2.4076

Variable Coeff Std Error T-Stat Signif

************************************************************************************

1. Constant -4.207701265 1.039835249 -4.04651 0.00083836

2. GDPW_EUR 1.598565305 0.128515536 12.43869 0.00000000

Regression ANOVA for Dependent Variable PDPEALL

Source Sum of Squares DF Mean Square

Regression 0.7354537202 1 0.7354537202

Residuals 0.0808080758 17 0.0047534162

Total 0.8162617960 18