The article discusses how to analyze data using a logistic regression model. Instead of looking at simple correlations, the researchers used cumulative logistic models with different types of response variables. They found that when collapsing the categories of the response variable into binary options, the results of the analysis changed. This means that the way we interpret and understand data can vary depending on the type of model we use.