Blood on the Cascadia: A Data-Driven Investigation of Pacific Northwest Homicides
Introduction
The Pacific Northwest, known for its gloomy weather and lonely highways, also harbors a dark history of violent crime. This paper delves into homicide and serial killings within this unique region, exploring patterns and trends. This research aims to illuminate this dark topic and contribute to the development of effective strategies for preventing future tragedies.
The paper was created using data from the Murder Accountability Project and Dr. Michael Aamodt of Radford University. Special thanks to Dr. Kirsten Gore of Willamette University for advising on this project.
Section I: Data Quality
The source data from the Murder Accountability Project was derived from the Federal Bureau of Investigation’s Supplemental Homicide Report (SHR). This data goes back to the 1970s and is known to be missing data; not all police departments report their homicides to the FBI and certain categories are known to be biased or use now outdated terminology. For example, some relationships include “Common Law Wife” or “Homosexual Relationship”. While validating the data quality, it was noticed that for multiple victim homicides the relationship of one member was assigned to all. This resulted in eight year boys assigned to the “Wife” relationship. A manual cleansing process corrected these errors.
Section 2: Homicide Trends
Below is an unadjusted annual homicide trend chart. The Pacific Northwest followed national trends with violent crime decreasing during the 1990s. Due to changes in policing, prosecution, and other trends the final data set was windowed to 1990 through a partial 2021.
There does not appear to be a strong seasonal component to homicides rates in the Pacific Northwest.
The early 20s appear to be the ages with the greatest incidence of committing homicide. Juvenile homicides, under age 14, are frequently accidents where children are playing with guns or other arguments, but there are a few cases of intential homicide committed by young children such as babysitters or tween/teenaged gang members.
Section 3: Dangerous Relationships
One of the difficulties with solving homicides is that many of them do not have apparent connections between the victim and offender.
The two most dangerous relationships appear to be an acquaintance of a killer or their female romantic partner. Circumstances are wide ranging, but the most common identified reason is an argument regarding anything but money.
Section 4: Geography
Most the counties in Washington and Oregon are rural. It is unsurprising that most homicides occur in the most populous counties which are King and Pierce Counties in Washington, which contain the cities of Seattle and Tacoma. For Oregon, Multnomah County is mostly the city of Portland. Counties in grey do not have data available.
Applying a log10 transformations allows us to better compare the counties with fewer homicides. Wahkiakum County in western Washington and Wheeler County in eastern Oregon now stand out for having very few homicides during this time period.
Are Seattle, Tacoma and Portland that dangerous? What if we convert the homicide counts to rates of homicide per 100,000 residents? On a per 100,000 rate, the focus shifts form the western population centers to the rural east side.
Sherman County in 2020 stands out as the murder capital of the region due to a single murder in 2020 and a population of 1,800 residents.
Section 5: Sex and Weapon and Victim Choice
Firearms, whether handguns or rifles, are far and away the most popular killing instrument.
Looking at the combination of offender sex and the age of the victim, we see that women kill far fewer people than men, but when women kill they are more likely to target a child.
Combining weapon and offender sex, we see that women choose different weapons and that the “Unknown” category more strongly resembles the male offender pattern.
Adding the victim age category back, and we see that adults are targeted in fairly similar ways. Female killers of adults do favor narcotics, drugs, and knives more than male killers. Female killers of children greatly prefer beatings and asphyxiation as compared to the male counterparts.
Section 6: Circumstances
The circumstances provided by the FBI have numerous categories. For this analysis, several categories were combined such as all “gang” categories. A few new categories of interest were added though. Family annihilators, those who kill multiple family members in a single instance, had a surprisingly large number of victims in this data set.
The 135 victims of family annihilators and older family annihilators were identified through relationships in the FBI data and most were confirmed through newspaper articles (notes are available upon request). Most family annihilators targeted an adult romantic partner and at least one child. There are also those who target their parents and adult siblings. Both of those groups are included in the following table:
| Offender Sex | Victim Count | Youngest Offender | Oldest Offender | Average Offender Age | Youngest Victim | Oldest Victim | Average Victim Age |
|---|---|---|---|---|---|---|---|
| Female | 10 | 21 | 43 | 30.200 | 1 | 61 | 27.800 |
| Male | 125 | 16 | 63 | 35.768 | 0 | 55 | 19.104 |
The ten victims of female annihilators are spread across three cases. Six of those victims were the 2007 Carnation, Washington murders committed by a husband and wife team against her parents, brother, and his children. Given the much higher case count for male perpetrators, it is unsurprising that they have a much larger age range for when they commit this crime.
Given the whiteness of the Pacific Northwest, it is unsurprising that most of these offenders are White followed by Asian. These crimes were mostly committed with firearms with knives coming a distant second.
(https://www.census.gov/quickfacts/fact/table/WA/PST045222)
Section 7: Race
In a reverse of the trends in family annihilators, there are more Black victims of homicide than Asian victims.
Plotting out the races of the victim and offender, we see that most offenders favor victims of their own race.
Section 8: Statistical Analysis
But those are just the raw numbers. Can we find a correlation between victim and offender race? What other categories are correlated? As it turns out, the choice of weapon and victim sex, the weapon and circumstances, and the weapon and relationship type are all correlated to a statistically significant degree. P-values for all displayed categories combinations were at or near zero.
| OffRace | Asian | Black | Native American | Pacific Islander | Unknown | White |
|---|---|---|---|---|---|---|
| Asian | 216 | 40 | 2 | 0 | 6 | 117 |
| Black | 55 | 956 | 23 | 2 | 12 | 529 |
| Native American | 3 | 11 | 91 | 0 | 1 | 126 |
| Unknown | 133 | 562 | 44 | 5 | 82 | 1337 |
| White | 114 | 286 | 89 | 3 | 102 | 6221 |
| Pacific Islander | 0 | 1 | 0 | 6 | 0 | 1 |
Section 9: Serial Killer Analysis
Last, we look at the most famous of killers. The Pacific Northwest has been home to several infamous serial killers such as Theodore Bundy and Gary Ridgway. Whether or not we have more than our fair share is outside the scope of this analysis.
Of the 75 counties in Washington and Oregon, only 24 have any victims of serial homicide. Following the general trend of homicides, these victims are most numerous in the major population centers of Spokane, Tacoma, Portland, and to a lesser extent Spokane.
When converting to a per 100,000 rate, the central Oregon county of Wasco stands out. This is a very rural area with a population of less than 30,000 residents and four victims of a single killer in 1991: Douglas Franklin Wright.
Can we identify patterns in the homicides committed serial killers? How are they different from other homicides?
A chi square test between serial killer victim status and several fields in the data show some interesting correlations. The state and victim age have low chi square and high p-values. However the county and victim sex are significantly correlated with serial killer victim status. The date of killing has the highest statistical significance. This is most likely because there are more identified serial killer victims earlier in the data set; these types of killings peaked in 1997 and 1992 with 14 victims in each year.
Who is most likely to be targeted by such a killer? The age range of their targets largely follows homicide trends in general.
Narrowing our focus on just serial killer victims, we see that the age ranges of male and female victims are fairly similar.
The victim race trends match the general homicide trends with White victims being the most common followed by Black victims. The gender distribution within in each race tends toward female with the exception of Black victims.
Delving a little deeper, serial killers prefer firearms, but there are certain types of weapons use predominantly or exclusively used on female victims.
While circumstances are largely unknown, theft and rape are the two most common identified circumstances. The gender distribution of each circumstance is unsurprising. Rape and other sex offenses impact women while brawls and gang activity impact men.
Conclusion
Each one of these cases is a tragedy to the individual, their family, their friends, and their community. By examining trends in homicide, we strive understand and, eventually, dismantle the conditions that allow such acts to occur. The first step in preventing tragedy is to understand the motivations of those who cause them. It is my hope that readers now understand some rarely discussed aspects of homicide.