# [R] Stepwise Regression

Martina Erdbrügge erdbruegge at statistik.uni-dortmund.de
Wed Dec 6 13:16:19 CET 2000

```Dear all,

I would like to carry out a stepwise regression using the function step.
If I use either ~ (A + B + C + D)^4 or explicitly all main effects and
interactions for the scope argument, the procedure only considers the
four main effects for addition or elimination in each iteration step.
What did I do wrong?

I'm using R version 1.1.1 on Windows NT.
(I'm sorry if this is a stupid question but I didn't find the answer in
the help file or the FAQs.)
Thanks in advance for any help.

Martina Erdbrügge

Here is my code with results:
> lm44 <- lm(betamat2[, 2] ~ 1, data = fndaten, weights = fndaten\$varin)
> step(lm44, ~ (A + B + C + D)^4)
Start:  AIC= -20.5
betamat2[, 2] ~ 1

Df Sum of Sq      RSS      AIC
+ C     1    1.3500   2.5706 -25.2555
+ B     1    1.2323   2.6882 -24.5393
<none>                3.9206 -20.5016
+ A     1    0.0132   3.9074 -18.5554
+ D     1    0.0072   3.9134 -18.5309

Step:  AIC= -25.26
betamat2[, 2] ~ C

Df Sum of Sq     RSS     AIC
+ B     1     1.430   1.141 -36.255
<none>                2.571 -25.255
+ A     1     0.096   2.475 -23.862
+ D     1     0.002   2.569 -23.265
- C     1     1.350   3.921 -20.502

Step:  AIC= -36.26
betamat2[, 2] ~ C + B

Df Sum of Sq     RSS     AIC
<none>                1.141 -36.255
+ A     1     0.079   1.062 -35.398
+ D     1     0.053   1.088 -35.018
- B     1     1.430   2.571 -25.255
- C     1     1.548   2.688 -24.539

Call:
lm(formula = betamat2[, 2] ~ C + B, data = fndaten, weights =
fndaten\$varin)

Coefficients:
(Intercept)            C            B
1.2245      -0.4140       0.4284
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

```