HOMICIDES IN CALIFORNIA, 1988 - 2008
Precipitating Event = rape
Year	Total Homicides	Precipitating event reported as rape	Facts provided do not permit determination of circumstances
1988	2,947	39	406
1989	3,159	39	322
1990	3,562	38	447
1991	3,876	41	632
1992	3,920	31	581
1993	4,095	21	643
1994	3,699	19	527
1995	3,530	14	595
1996	2,910	11	389
1997	2,579	12	424
1998	2,170	9	336
1999	2,006	16	304
2000	2,074	8	314
2001	2,201	12	317
2002	2,392	8	370
2003	2,402	7	410
2004	2,394	6	320
2005	2,503	7	526
2006	2,483	6	568
2007	2,258	3	825
2008	2,143		751
Source: Department of Justice, Criminal Justice Statistics Center's
Homicide files for 1988 through 2008.

The sharp increase in "no determination" cases since 2004 is worrisome to anyone thinking of doing research with these numbers, but let's focus on 1991 to 2004. Crime generally peaked nationwide and declined through the 1990s. We also see California murders dropping like a rock, with the 1999 number less than half the 1993 number. If we had per capita numbers, the drop would be even sharper. The total number of murders then rises somewhat in the early 2000s. Through 2004, the number of "no determination" homicides is roughly proportional to the total number of homicides, but the number of rape-murder cases is dropping much faster. There are 41 in 1991 but only 6 in 2004.  The number of homicides only dropped by about 38% in the interval 1991-2004, but the number of rape-murders plunged by 85%.

I'm skeptical about these circumstance numbers. Even in the period where the no-determination cases are a reasonably steady proportion of the total, their portion is large. As much as a sixth of your total sample coming in with the variable of interest unknown is a lot of missing data, and we can't assume the unknown cases are proportional to the known cases in their circumstances.

On the other hand, one would expect that rape cases would be less likely than others to qualify for the label, "facts provided do not permit determination of circumstances."  Sexual assault is a violent act separate from the homicide, and it very often causes damage discernible in an autopsy. So while rape-murders may not have dropped to zero, it is hard to escape the conclusion that they have indeed dropped further than the homicide rate generally.

When I am asked about the drop in the number of death sentences in recent years, I generally note that the largest single factor in that drop is simply the drop in the number of homicides. But that only accounts for part of it.

I am told, anecdotally, by some prosecutors that they have seen the number of homicides so heinous as to cry out for the death penalty to be dropping even faster than the homicide rate generally. Rape-murder is certainly high on the list of cases for seeking the death penalty. If the rate of rape-murder is indeed dropping faster than the rate of homicide generally, as California's figures indicate, that would be a partial empirical validation of the anecdotal evidence.

I will get some information on this from other states when I have a chance.