Comparing the same indiviual in datatset

General
#1

Dear all,

I'm a bit of in newby with R, so please be gentle (: Currently I am visualizing my data using boxplots (ggplot2). In my data set I have individuals (df$Fish_ID) who all are tested twice (df$Exposure) which is eighter "1" or "2". I've made multiple boxplot for different treatments and different treatment, all using all the data to compare. Is there is possiblity where I can compare the difference in mean( difference between Exposure1 vs Exposure 2) and make this visual in a boxplot? So for example; instead of comparing all the data of Exposure 1 vs Exposure 2, only compare this for every individual apart. Is this possible in ggplot2 or do I have to perform statistics first?

I hope my question is clear, thank you in advance for your help!

#2

You can certainly easily compare the boxplots in ggplot (just add group = Exposure in your aesthetic), but to calculate statistics (e.g. means, t-test p value of difference) on the differences I would do it outside ggplot first and put it in a separate dataframe before plotting. The datanovia website has a lot of excellent example of running and plotting stats in R.

#3

Thank you for your reply! This would compare the mean of Exposure 1 to the mean of Exposure 2, if I am correct. This is not the same as comparing both the exposures for each individual apart, which I want to know. For Fish$ID "12" compare Exposure 1 to Exposure 2, for Fish$IS "13" compare Exposure 1 to Exposure 2 etc. Or is this the same as ANOVA?

#4

ANOVA is for comparing the differences in mean between groups. There are different ways to define the "groups" and this is why there are different kinds of ANOVA.

#5

Data is not normally distibuted, therefor I can't preform an ANOVA. What to do know?

#6

If each Fish is tested twice, and you have a single non-Normal response at each test, then wouldn't a Paired Samples Wilcoxon Test be most suitable and simpler? This is a non-parametric version of a paired t-test.

Would you want to visualise something like this:

image
image672×576 18.9 KB

The image is taken from the linked article here: