In: Math
General Electric recently conducted a study to evaluate filaments in their industrial high intensity bulbs. Investigators recorded the number of weeks each high-intensity bulb would last before failure for three test filaments (Groups 1, 2, and 3) and the standard filament (Group 4). The results are as follows. Using ? = 0.01,
Group 1 2 3 4
15 14 25 28
18 18 19 31
21 20 22 27
16 16 20 32
17 15 18 23
20 16 24 25
18 22 27 30
14 18 27
24 25
26
we have given 4 groups and ask to do a ANOVA test of them, so
H0: means of all groups are equal
vs
H1: not H0
we get R output as below
Analysis of Variance Table
Response: bulbs
Df Sum Sq Mean Sq F value Pr(>F)
group 3 613.48 204.493 24.835 2.836e-08 ***
Residuals 30 247.02 8.234
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
here p-value < 0.05 we can say that the group effect is significantly effective and we are rejecting H0
since we get the group means are different we search for which are different by Tukey HSD
the R output is
Tukey multiple comparisons of means
95% family-wise confidence level
Fit: aov(formula = bulbs ~ group)
$`group`
diff lwr upr p adj
2-1 -0.9821429 -5.02031317 3.056027 0.9107047
3-1 4.0317460 0.09966253 7.963830 0.0428315
4-1 9.5428571 5.69774991 13.387964 0.0000010
3-2 5.0138889 1.22256203 8.805216 0.0059628
4-2 10.5250000 6.82395578 14.226044 0.0000001
4-3 5.5111111 1.92611606 9.096106 0.0012635
here p-value for 1 and 2 difference is >0.05 so we can say that group 1 and 2 are same mean where as all other combinations are significantly effective..
R-code for this problem is
data=read.csv("Book1.csv")
bulbs=data$bulbs
group=as.factor(data$group)
anova(lm(bulbs~group))
TukeyHSD(aov(bulbs~group))