The original goal for the examination sample was to gather a sample of men who were rejected from service to test whether they were different from those accepted into the Union Army in order to add credibility to the UA sample. The question has arisen whether soldiers who fought in the Union Army were healthier than the men who stayed home. The original sampling procedure for the UA relied on Richard Steckel's findings that rejectees did not look any different than those accepted into service. Unfortunately, we were unable to locate Steckel's original work and Steckel himself could not recreate it. A sample size of 10,000 rejectees was decided upon.

The first step was to try to locate usable records at the National Archives. The best records located were results of medical examinations of enrolled men, which included recruits, rejectees, etc. These records were created in response to the draft call and so only exist for September 1864 through April 1865. (Steckel's work seemed to indicate that he had medical examinaitons earlier than that which were more detailed, but we were unable to locate those records.)

In order to get a sample that reflected the UA sample demographically, researchers determined from which congressional districts soldiers in the UA sample enlisted. The examination sample consists of an equal proportion of men enrolled from each congressional district represented in the UA sample.

After looking at these records, it became clear that fewer men were rejected for service than originally thought, and the sampling procedure had to be reworked. The decision was made to collect the information on everybody, not just rejectees. Since this was too costly, a 20 percent sample of all men enrolled and examined for service was collected. Proceeding state by state, researchers input every 5th line, regardless of examination result. This resulted in a final sample size of 70,800 men.

From these 70,800 men, a subsample of 10,000 rejectees was created. These 10,000 men were linked to the 1860 and 1870 censuses (the census records being the only records available for linking). Due to cost considerations, only 1,000 of these men (with uncommon names which could be easily linked) were linked to every available census and death record where possible.