Hi all,
I've done some anovas, found some significance and proceeded to do the
post hoc tests to determine the source of the significance. I used the
multcmp package as suggested by Andy Liaw (thank you).
I am, however, unsure how to interpret the results of the call to
simint. Could someone help me with this?
I've attached the code and the results it produced. There are 3 groups
(ctrl,long,short) and I'm trying to see if the data in the Mean_12
column (in the example below) differs based on group membership.
for (meanCol in meanColumns) {
roi.err.aov<-aov(eval(parse(text=meanCol)) ~ Group, data=roi.errs);
HSD.err<-simint(eval(parse(text=meanCol)) ~ Group, data=roi.errs,
type="Tukey");
cat("################################################################################\n");
cat("### Summary of AOV for ", meanCol, " cluster: ", i,
"\n");
print(summary(roi.err.aov))
print(summary(HSD.err))
i<-i+1;
}
################################################################################
### Summary of AOV for Mean_12 cluster: 12
Df Sum Sq Mean Sq F value Pr(>F)
Group 2 6368494 3184247 8.8536 0.001627 **
Residuals 21 7552730 359654
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05
'.' 0.1 ' ' 1
Simultaneous 95% confidence intervals: Tukey contrasts
Call:
simint.formula(formula = eval(parse(text = meanCol)) ~ Group,
data = roi.stops, type = "Tukey")
Tukey contrasts for factor Group
Contrast matrix:
Groupctrl Grouplong Groupshort
Grouplong-Groupctrl 0 -1 1 0
Groupshort-Groupctrl 0 -1 0 1
Groupshort-Grouplong 0 0 -1 1
Absolute Error Tolerance: 0.001
95 % quantile: 2.519
Coefficients:
Estimate 2.5 % 97.5 % t value Std.Err. p raw p Bonf
Grouplong-Groupctrl -134.246 -930.286 661.794 -0.425 316.076 0.675 1.000
Groupshort-Groupctrl 978.056 182.016 1774.097 3.094 316.076 0.005 0.016
Groupshort-Grouplong 1112.303 400.303 1824.303 3.934 282.707 0.001 0.002
p adj
Grouplong-Groupctrl 0.906
Groupshort-Groupctrl 0.015
Groupshort-Grouplong 0.002
Thanks in advance,
Regards,
Colm.
