The article discusses how analyzing binary and ordinal data in research can be challenging due to non-normal distributions and recoding of variables. The researchers found that categorizing continuous variables into ordered categories can lead to a loss of statistical power. They also discovered that using Pearson correlations instead of maximum likelihood estimates can bias the results. Additionally, odds ratios can be influenced by the prevalence of affected individuals in the population. Treating binary data as continuous can underestimate genetic and environmental variance components. Properly modeling ordinal data may be more time-consuming but is crucial to avoid biased results.