R statistics help

nearly_there

Ok so I need to analyse data for my undergrad project. I have experience of working with R but whats really confusing me is the format my data should be in before I convert it to a tab delimited file to import onto R.

I have two genotypes which I am comparing %infection at 6 different treatments. I think I should be using a two way interaction ANOVA for this. I have books from the library but they all deal with R directly not with the data in excel.

If anyone could help me I would be so grateful. Its just the basics, i know, but im really stuck!

Thanks

2Scoops

nearly_there wrote: »

I have two genotypes which I am comparing %infection at 6 different treatments. I think I should be using a two way interaction ANOVA for this. I have books from the library but they all deal with R directly not with the data in excel.

It's pretty standard that you would set up 3 columns. One is a grouping variable for genotype. So, if you have 10 values for genotype#1 and 10 for genotype#2, label 10 cells each in the column with g1 or g2. Column 2 will be a grouping variable for treatment, so put in a label for each treatment and match it up to the number of times it occurs with each genotype. Column 3 will have your actual %infection values.

nearly_there

Thaks 2Scoops,

you see thats what I have exactly but when i import it into R, I cant find normality of each treatment for each genotyope. Should i just check for this in excel first or is it possible to use R?

Im trying to use the tapply function
iV entered this

> tapply ( infection[treat = "1"], type, mean)

where 1 is the first treatment, and type means genotype and iv gotten

Error in tapply(infection[treat = "1"], type, mean) :
arguments must have same length

is there something glaringly obvious im doing wrong?

2Scoops

nearly_there wrote: »

you see thats what I have exactly but when i import it into R, I cant find normality of each treatment for each genotyope. Should i just check for this in excel first or is it possible to use R?

Not sure, tbh. I tend to use GUIs with R so not 100% up to speed with the code. If you have a low N, it may not calculate it. As this is a side-measure, I don't see the problem with doing it quickly in Excel and leave the heavy lifting to R. :pac:

nearly_there

I dont suppose theres any chance anyone could translate this for me? i think il have to start doing them out longhand....:(



Df Sum Sq Mean Sq F value Pr(>F)
type 1 73 73 0.1785 0.6731
treat 1 118360 118360 291.0505 <2e-16 ***
Residuals 197 80113 407
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Tukey multiple comparisons of means
95% family-wise confidence level

Fit: aov(formula = nec ~ type + treat)

$type
diff lwr upr p adj
M-G 1.205 -4.419154 6.829154 0.6731014

Warning messages:
1: In replications(paste("~", xx), data = mf) : non-factors ignored: treat
2: In TukeyHSD.aov(model) :
'which' specified some non-factors which will be dropped

2Scoops

nearly_there wrote: »

Df Sum Sq Mean Sq F value Pr(>F)
type 1 73 73 0.1785 0.6731
treat 1 118360 118360 291.0505 <2e-16 ***
Residuals 197 80113 407

Reading across, it shows the degrees of freedom, sum of squares, mean square, F-value and p-value. Looks like treatment is significant.

nearly_there wrote: »

Tukey multiple comparisons of means
95% family-wise confidence level

Fit: aov(formula = nec ~ type + treat)

$type
diff lwr upr p adj
M-G 1.205 -4.419154 6.829154 0.6731014

These are the results from a Tukey post-hoc test. It shows the mean difference, the 95% CI and the p-value.

nearly_there

Thanks!!