Wouldn't put too much stock in this particular StatsCan General Social Survey when it comes to the rates (per 1000) we see in the Census Metropolitan Areas. Besides the fact that the GSS is voluntary, for CMAs as an individual strata, the sampling isn't quite enough to be all that reliable (with the exception of the few largest population centres).
Within the table indicating CMAs in the official report, StatsCan has a notation to exercise caution with most every CMA where a rate is reported (including Regina). It also withheld many other CMAs for lack of reliability (again, likely sampling amounts in relation to the CMA population). There is an additional note that further caution should be exercised when comparing CMAs, as not all (in fact, most) are not statistically significant.
It comes as no real surprise that the article posted at the top of this page does not mention that StatsCan has cautionary notes regarding the reliability of the stats from the Regina (and other) CMAs, as well as the poor reliability when comparing CMAs.
IMO, the much more reliable StatsCan release regarding the state of crime in Canada and the reporting of crime by CMAs, is the Uniform Crime Reporting Survey, released annually. Additional, by virtue, victimization statistics should not be considered as an indicator of overall crime rates. We also must consider that one's perception of victimization does not imply that a criminal code infraction has occurred. There would also likely be substantial degrees of what constitutes victimization by individual perception (ie: an old lady who gets bumped into at the supermarket in the rush to the shelf of canned meat on sale may positively indicate in responding to this survey that she was, indeed, a victim in the past year; another individual with the same experience may brush-off such a triviality and forget about it).
http://www.statcan.gc.ca/pub/85-002-.../11340-eng.pdf
http://www.statcan.gc.ca/cgi-bin/imd...db&adm=8&dis=2