When I look at an ANOVA for my dataset, I get a P-value of 0.018, which is significant at alpha=0.05. However, when I do a Tukey's HSD test on that ANOVA, everything is categorized as falling into group A. Can anyone explain what's going on here?
Here's my data:
dataHodag <- read.table(text="
Plot Rate Rep HollowHeartPercent
1 2155 1 1 0.0
2 2252 1 2 0.2
3 2343 1 3 0.2
4 2433 1 4 0.2
5 2211 2 2 0.0
6 2326 2 3 0.0
7 2412 2 4 0.0
8 2136 2 1 0.2
9 2124 3 1 0.0
10 2334 3 3 0.0
11 2421 3 4 0.0
12 2236 3 2 0.2
13 2246 4 2 0.0
14 2354 4 3 0.0
15 2446 4 4 0.0
16 2116 4 1 0.1
17 2145 5 1 0.0
18 2225 5 2 0.0
19 2314 5 3 0.0
20 2451 5 4 0.0", header=TRUE)
I am using the agricolae package for my analysis This is the code that gives me my ANOVA:
TukeyAOV <- aov(HollowHeartPercent ~ Rep + Rate, data=dataHodag)
summary(TukeyAOV)
# Df Sum Sq Mean Sq F value Pr(>F)
# Rep 1 0.00250 0.00250 0.406 0.533
# Rate 1 0.04225 0.04225 6.857 0.018 *
# Residuals 17 0.10475 0.00616
# ---
# Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
And here is the code for my Tukey's HSD test:
HSD <- agricolae::HSD.test(TukeyAOV, "Rate", alpha = 0.05, group=TRUE)
print(HSD$groups)
# HollowHeartPercent groups
# 1 0.150 a
# 2 0.050 a
# 3 0.050 a
# 4 0.025 a
# 5 0.000 a
Any help would be much appreciated. Thanks!!
Did you notice the discrepancy in degrees of freedom? You read Rate and Rep as numeric variables, not as factors. Compare to:
> dataHodag$Plot <- as.factor(dataHodag$Plot)
> dataHodag$Rate <- as.factor(dataHodag$Rate)
> TukeyAOV <- aov(HollowHeartPercent ~ Rep + Rate, data=dataHodag)
> summary(TukeyAOV)
Df Sum Sq Mean Sq F value Pr(>F)
Rep 1 0.0025 0.002500 0.368 0.554
Rate 4 0.0520 0.013000 1.916 0.164
Residuals 14 0.0950 0.006786
I hope this helps.