# Continuous Data

How do you handle an independent variable that is not yearly data, but is continuous data and is presented in discrete categories, or is ordinal data?

In general, the joinpoint software can handle independent variables that are not yearly data. For example, if the data is reported in months instead of years, the model can be run exactly the same as using “year” as independent variable. The interpretations of annual percentage change (APC) and average annual percentage change (AAPC) should change accordingly. APC now is “monthly percentage change”, and AAPC “average monthly percentage change”. However, if the data the monthly data is coded in years (i.e. 1/12th, 2/12ths,…_ instead of 1,2,…) then the APC would be interpretable as an “annual percent change”.

If the independent variable represents chronological age instead of calendar years, then the APC would represent “percent change per year of age”, and if it represents dollars, then the APC would represent “percent change per dollar”.

If the independent variable represents intervals, e.g. age intervals of 10-19, 20-29, 30–39, etc, then it is recommended using the mid-point of the interval (e.g. 15, 25, 35) or the mean or median of the observations in each interval. For open-ended groups (e.g., age 80+), one should consider eliminating it or one may use the median (or mean) of all the observations that are above 80 years old for this group.

If the independent variable is an ordinal variable such as stage of disease (e.g. Stage I, II, III, IV), while the analysis can still be run, the APC cannot be interpreted because the underlying data is not continuous.

Related content