Both are done. Typically we try to give researchers access to the populations directly if that's the type of research they're doing. Some of the database research is done using existing Stats Canada databases and stuff. But no, as much as possible, we try to link the researchers with the relevant populations.
For example, one of the top PTSD researchers in the country, Dr. Jitender Sareen from the University of Manitoba, has assembled a national team together to look at conducting the Canadian community health survey again for veterans. Going back to Mr. Harris' question, it was done in 2007, and there are some old data, but they want to redo it and get new data, again with a particular focus on military and veterans. He's going across the country doing town halls with veterans.