The article discusses different ways data can be missing in scientific studies and how this affects the analysis. The researchers show that the method used to handle missing data depends on how the missing values are related to the other data. They use an example from a study on dental retainers to explain three types of missing data: completely random, at random, and not at random. The chance of missing data depends on variables like age and other data values. This relationship can be shown visually using directed acyclic graphs.