Chapitre 12. R´ egression lin´ eaire simple et multiple
517
> legend("bottomright",c("SMOKE=0","SMOKE=1"),
+
col=c("red","blue"),lty=1)
●
●
● ●
●
● ●
●
●
●
● ●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
● ●
●
●
●
● ●
●
●
●
●
● ●
●
●
●
● ● ●
●
●
●
●
●
●
●
● ●
● ●
●
●
●
●
●
● ● ●
●
●
● ● ●
●
●
●
●
●
●
●
●
●
● ●
●
●
● ●
●
●
●
●
● ●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
● ●
●
●
●
●
● ●
●
●
●
● ●
●
●
●
●
●
● ● ● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
15
20
25
30
35
40
45
1000
2000
3000
4000
5000
BWT = β 0 + β 1 AGE + β 2 SMOKE + β 3 AGE x SMOKE + ε
AGE en années
Poids de naissance (g)
●
●
● ●
●
● ●
●
●
●
●
●
●
●
●
●
● ●
● ●
● ●
●
●
●
●
●
●
● ●
●
●
●
●
● ●
● ●
●
●
●
● ●
● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
● ● ●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
SMOKE=0
SMOKE=1
Figure 12.8 – Effet de l’ˆ age sur BWT dans un mod` ele avec interaction.
Nous pouvons tester la significativit´ e du terme d’interaction en analysant
les r´ esultats du mod` ele 6.
> summary(modele6)
Call:
lm(formula = BWT ∼ AGE * SMOKE)
Residuals:
Min
1Q Median
3Q
Max
-2187.8 -456.6
52.8
526.6 1522.2
Coefficients:
Estimate Std. Error t value
Pr(>|t|)
(Intercept) 2408.38
292.24
8.241 0.0000000000000305 ***
AGE
27.60
12.15
2.271
0.0243 *
SMOKE
795.38
484.42
1.642
0.1023
AGE:SMOKE
-46.36
20.45 -2.267
0.0245 *
--Signif. codes: 0 ‘ *** ’ 0.001 ‘ ** ’ 0.01 ‘ * ’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 709.4 on 185 degrees of freedom
Multiple R-squared: 0.0683,
Adjusted R-squared: 0.05319
F-statistic: 4.521 on 3 and 185 DF, p-value: 0.004378