How can I run a normality test on count data in the following table?-CodePudding

My data is formatted in R as follows:

        Group Light Dark
1  Dermaptera    29   29
2   Oniscidea    72   54
3     Diptera    54   39
4 Lepidoptera    17    7
5  Formicidae    14    6
6   Hemiptera     3    9
7   Diplopoda     8   17

I am certain that this data is not normally distributed as it is count data, and histograms show that it is clearly non-normal, for example using hist (dataframe$Light). When I try to input shapiro.test (dataframe), I get the error code is.numeric(x) is not TRUE, and when I try to instead input shapiro.test (dataframe$Light) or shapiro.test (dataframe$Light dataframe$Dark, the p-value shows the result to be of normal distribution.

How should I instead format this data when putting it into R so that I can test for normality and subsequently test for statistically significant relationships?

CodePudding user response：

library(dplyr)
library(stats)

Provided your data sits here:

dd <- data.frame(Group = LETTERS[1:10], Light = sample(1:10), Dark = sample(1:10))

Shapiro test p-value for each specified column:

pvalues <- dd %>% 
  summarise_at(vars(Light, Dark), funs(shapiro.test(.)$p.value))

You can even find Shapiro test p-value groupwise:

pvalues_by_group <- dd %>% 
  group_by(Group) %>% 
  summarise_at(vars(Light, Dark), funs(shapiro.test(.)$p.value))