To calculate the SDS values, information about ``normal'' people is required. In DPV , data from a European and an American exploration have been entered, but no matter which one should be used the statistic, always uses the European one.
Due to the fact that for each patient and each year, only the date of the latest schooling should be used, several of them were entered for one patient in the same year. This made the following bug obvious: not only the last one is used to calculate the statistic, but also the other entries about a schooling in the table schooling (schulung).
Next, the program's behavior was tested if no information was given for the year before or after the educational step. In this case, the whole schooling should be ignored, because the statistic would give no information about the success of the treatment. All tests -- including the three cases: only data before a schooling, only after it, and no data at all -- did not detect any faults.
Thereafter a boundary value test was done. This time a bug could be found. If all patients schooled in one year are in their first year of diabetes , no output should be produced. This is done correctly, but the information about those patients is still being collected, and the variables used for this are not reset at the end of the appropriate year. So the next year data is available for will also include the information about the foregoing year.
Finally, the computation of the median used in this part
was tested analog to the tests described in chapter 
on
page , but no further faults could be
found.