That paper defines lethality as homicides divided by the sum of homicides and aggravated assaults. Correcting by that factor would work to compare one year to another if the ratio of attempted murders to other aggravated assaults remains constant. The biggest reason the calculation might be inaccurate is if the definition or reporting of assault changes over time. It sounds kind of like an ugly hack, except that homicide lethality per year seems to match motor vehicle crash lethality per year, giving some confidence to the calculation.
If we try to fix lethality at the 1960 level, then homicide rates are 5.1 per 100,000 in 1960, hit a peak at 25.5 in 1993, and decline to 14.6 today.
One good comparison would be to find data on how many people are shot, each year, and plot the adjusted murder rate against that. I did a quick sanity check to make sure that adjusted murder rate doesn't exceed total shootings (it did not for the year i checked), but didn't compare the two time series.
At some point, I hope to analyze 1960 to the present, and assess the effects of various factors (drugs, lead, abortion, etc) on adjusted murder. But it's a big project that would take a lot of effort to get right.
Anyways, the main thing to note here is simply that the present day might be 3 times more violent than 1960, even though the per capita murder rates look the same.
