a research used the percentile method to find the prevalence in severe group as below:
The outcome variable was categorized using percentiles to Mild, Moderate, and Severe in the Symptom Severity Scale (SSS) and to Mild-moderate and Severe in the Functional Status Scale (FSS). The three categories (mild, moderate, and severe) of susceptibility to having CTS symptoms were based on the SSS. Individuals suspected of having mild CTS symptoms are within the 50th percentile. Individuals suspected of having moderate CTS symptoms are within the 50th-75th percentile. Individuals suspected of having severe CTS symptoms are above the 75th percentile. The two categories (mild-moderate and severe) of susceptibility to having CTS symptoms will be based on the FSS. Individuals suspected of having Mild-moderate CTS symptoms will be within the 75thpercentile. Individuals suspected of having severe will be within the 4th quartile (above 75th percentile).
In spss the percentile method that is described in the research above is performed using visual binning?
Thanks
Best Answer
This is an absolutely dreadful way to analyze data and another example of how much harm SPSS has done. This violates several statistical principles. It is not appropriate to bin ordinal or continuous variables. Instead analyze continuous trends, e.g., use an ordinal regression model such as the proportional odds model to predict symptom severity as a function of continuous and categorical predictors. I go into this in detail in RMS course notes but start with this. Some of the problems with categorizing variables are (1) we don't know how the category boundaries should be selected, (2) there is too much outcome heterogeneity within bins, and (3) you'll find that some subjects at the edge of one bin are more similar to those at the edge of another bin than they are to other subjects in their own bin.