R-ticulate

11Analysis of Covariance (ANCOVA)

When we have a continuous response variable and a mix of continuous and categorical explanatory variables (at least one of each), we use a so‐called analysis of covariance model (ANCOVA). In R, this type of model can be implemented using the familiar lm function. Let us explore this type of analysis with a dataset consisting of the growth rate (growth, μm day⁻¹) of three different pathogenic fungal species as a function of fungicide concentration (conc, mg l⁻¹). The data are stored in the provided dataset called pathfung. As a ‘sanity check’, we first look at the structure (str) and the summary (summary) of the data.

## Read in data
 > pathfung <- read.csv("pathogenic_fungi.csv")
 
 
## Check structure and summary
 > str(pathfung)
 
 
'data.frame':       54 obs. of  3 variables:
  $ conc   : int  10 8 6 4 2 0 10 8 6 4 ...
  $ growth : num  37.8 85.3 111.4 147.3 236.4 ...
  $ species: chr  "A" "A" "A" "A" ...
 

The str output shows fungicide concentration as an integer and fungal growth as a numeric (continuous) variable, which is fine. However, species is currently coded as character but since it is a categorical variable (factor), it should be coded as a factor (see Box 1.4 in Chapter 1). We will do this right away.

## Recode species as factor
 > pathfung$species <- as.factor(pathfung$species)
 

It is a good idea to rerun the str(pathfung) command at this stage to ensure that species is now a factor.

The summary of this dataset flags no issues. The concentration ranges ...