Overview

Before we move further on the journey of learning R, analyzing our data, and making figures, it’s helpful to stop and think of the best practices for data analysis.

Balance

As much as possible, we should aim to have relatively even numbers of individuals in whatever treatments we may have. Unbalanced experimental designs lead to uneven sample sizes, which can cause problems for our analyses.

Randomization

It’d be best if we had individuals randomly allocated into whatever treatment groups we have. We must always make sure that we set up our experiment without any knowledge of the possible preexisting nature of our individuals. For example, if we’re doing a growth experiment, we shouldn’t pick all the most prominent individuals in one particular treatment since that’ll bias our results. We must make sure to allocate individuals blindly. This can be accomplished by numbering all of our individuals and then randomizing the numbers into our treatments.

Replication

The more data we have, the better off we are. However, replication doesn’t just mean having more individuals. Replication refers to repeatedly setting up the experiment so that we have ...