Health researchers often have a list of addresses that they would like to map. Refer to the ArcGIS documentation for procedures for using the software for geocoding. For US data, researchers often geocode the data using the TIGER roads from the US Census. If you have a small number of addresses (e.g. < 500) it might be best to geocode them yourself. This way you will know exactly what assumptions were made and be able to control the entire process. You should use the "offset" feature with an appropriate distance (e.g. 20 m) to locate addresses to the right or left of the road centerlines. This is especially important if you plan to assign addresses to census tracts or blocks, since their boundaries will correspond with roads. To assign addresses (points) to census tracts (polygons) do a spatial join and specify that the result should contain all attributes of the polygon the point falls in.

There are many vendors who provide geocoding services at reasonable cost. Vendors will have more accurate and complete road data and software to check addresses for mistakes. If you do not have too many addresses, it is helpful to read through them before geocoding or sending them out to vendors to catch obvious errors. I always remove extraneous information from address fields, such as apartment numbers, building names, etc. before geocoding. If you have post office boxes you will need to decide whether to assign them to the centroid of the ZIP code or remove them from your dataset.

Typically most addresses can be geocoded automatically pretty easily. It would be reasonable to get about 80% of a datasets' addreses geocoded without much difficulty. Geocoding the remaining 20% of the addresses can be a very labor intensive and painstaking process. If time and resources are available geocoding difficult addresses might include research using the internet, using electronic or paper records available at state and local government agencies or libraries, and even going into the field to locate addresses with a GPS.

For your convenience I am including here a link to a company I have used for geocoding, but there are many others available.

Mapping Analytics - one of several companies that provide geocoding services

In July 2003 the journal Epidemiology published a commentary "Place, Space, and Health: GIS and Epidemiology" by Nancy Krieger of Harvard School of Public Health, followed by three articles on geocoding:

Hurley SE, Saunders TM, Nivas R, Hertz A, Reynolds. Post Office Box Addresses: A Challenge for Geographic Information System-Based Studies. Epidemiology 2003;14(4):386-391.

McElroy JA, Remington PL, Trentham-Dietz A, Robert SA, Newcomb PA. Geocoding Addresses from a Large Population-based Study: Lessons Learned. Epidemiology 2003; 14(4):399-407.

Bonner MR, Han D, Nie J, Rogerson P, Vena JE, Freudenheim JL. "Positional Accuracy of Geocoded Addresses in Epidemiologic Research. Epidemiology 2003; 14(4):408-412.