Two thoughts:
1). Murder data are the most reliable crime data we have. Hard to conceal, not report, etc. murders. They are widely reported in the media, etc.
2). When one reports data for a large number of elements, e.g., 100 largest cities, one needs some type of summary data otherwise one can cherry pick individual cases to make whatever point one wants. E.g., "For the 100 largest U.S. cities, homicides in 2015 were N, which is a X% increase from the number reported in 2014 for those same cities.' As the article indicates, homicides rose for a number of cities and declined in others. This is what one would expect with this type data.
Although most don't care, one could use rates of homicides per 100,000 U.S. population. The U.S. population has increased much which would lower homicide rates.
Bookmarks