What is the trend of gun-related homicide in the US history? Do issues like ethnicity play any role?
What do the places with good records do differently than the places with bad records? This report was an attempt to find answers to such questions.
Statistical techniques that were used in the report were correlation analysis, chi-square test, and logistic regression.
Software used for analysis was R.
Some of the interesting findings are:
- The highest number of homicides take place in July and August compared to other months
- Rifle is more common firearm among Native Americans/Alaska Natives compared to other ethnicities
- Intra-race homicide (e.g. whites murdered by whites) is way more common than inter-race ones (e.g. blacks murdered by whites)
- Though inter-race homicide is more common between whites and Native Americans (close to 40% of the homicides in native Americans were done by white)
- The highest number of murderers are acquaintances not strangers!
- Unlike common notion, places with less homicide are not the places with the strictest gun laws.
- Rather intuitively enough they are better at solving cases (they have a better ratio of solving case compared to the places with bad records).
2. Agenda Background
• Research
Background
• Research Questions
• Overview of Data
1
Study Detail
• Study Findings
• Summary
2
Conclusion
• Implication
• Recommendation
• Future Scope of
Research
3
4. Background of the Study
▶ 88.8 guns per 100 people, or about
270,000,000 guns, which is the highest
total and per capita number in the
world
▶ 22% of Americans own one or more
guns (35% of men and 12% of
women)
▶ Turned in to a political agenda
Source:
CNN. Com || CNN Money.US (2007)
5. Research Question
What are
the driving
factors
behind
gun
homicide?
What is the overall trend of homicide in United States over the years or within a year?
Is there any regional trend in homicides?
Is there any pattern in the types of gun used?
How does ethnicity play role in homicide incidents?
Which factors play predictive roles in an intentional homicide?
Is there any variability in the rate of homicides using guns between states having gun
control and states having no (lesser) gun control?
Is there any intra state variability in the rate of homicides using guns between the pre and
post gun control implementation year?
6. Description of Data
▶ Source
▶ ‘The Murder Accountability Project’ sourced from FBI and Freedom of
Information Act
▶ Content of Dataset
638,454 incidents
23 variables
20141980
▶ Brady Score
▶ A comparative ranking of states based on gun laws
▶ Brady Campaign to Prevent Gun Violence – named in honor of Jim Brady,
President Reagan’s press secretary
▶ 40 features of state gun laws, grouped into five broad categories
Source:
Lanza, S. P. (2014)
7. Description of Data Cont..
Variables
Record ID Crime Type Perpetrator Race
Agency Code Crime Solved Perpetrator Ethnicity
Agency Name Victim Sex Relationship
Agency Type Victim Age Weapon
City Victim Race Victim Count
State Victim Ethnicity Perpetrator Count
Year Perpetrator Sex Record Source
Month Perpetrator Age
21. 1.
Overall Handgun is the
most widely used
weapon.
2.
Above average use of
Rifles in 10 States e.g.
Vermont, Maine, North
Dakota, South Dakota,
Alaska, Hawaii, Idaho,
and Montana
22. 1.
No significant difference
on gun choice between
men and women
2.
Rifle is more common
among Native
American/Alaska
Americans compared to
overall trend
23. Finding: 03
➢ Handgun is the most widely
used weapon
➢ Sex doesn’t seem to impact
weapon choice
➢ States and race seem to
have a little impact on
weapon choice
24. Question: 04 How does ethnicity play role in
homicide incidents?
25. 1.
Intra race homicide is more common
2.
Exceptions: Native American & Asian/Pacific Islanders
26. 1.
Lowest Unsolved case %: Native American/Alaska Natives
2.
Highest unsolved case % %: Black
27. Model 02 is the better model to explain because of lower
P value and better McFadden’s R square
Logistic Regression model 01:
Crime solving vs Victim’s race
DV: Crime solve Status
IV: Victim Race
n/a
McFadden’s R2: 0.01077
Logistic Regression model 02:
Crime solving vs Victim’s race +
Victim Sex + State
DV: Crime solve Status
IV: Victim Race, Victim Sex, State
P-value: 2.2e-16
McFadden’s R2: 0.0714
28. No significant multicollinearity among the independent
variables
GVIF DF GVIF^(1/(2*Df))
Victim-race 1.361592 4 1.039336
Victim-sex 1.097827 2 1.023608
State 1.25054 50 1.002238
30. Explanations of results from logistic regression model
Considering other variables constant, the odds of the case being solved
goes down by 51.85% for male victims compared to female victims
Considering other variables constant, the odds of the case being solved
goes up by 49.86% for white victims compared to black victims
Considering other variables constant, the odds of the case being solved go
down by respectively 88.50%, 78.20%, 67.53% and 67.78% for DC, New York,
California and Illinois respective compared to Alabama.
31. Finding: 04
➢ Overall same race homicide
is more common
➢ There maybe some hate
crimes by whites against
native Americans and
Asians/Pacific islanders
➢ Race, place and sex of victim
have significant relationship
with the case solve status.
32. Question: 05 Which factors play predictive roles
in an intentional homicide?
33. Murderers are more likely to be acquaintance than stranger!!
Figure 12 Most vulnerable relationship
34. Lowest average age: Native American/Alaska Native
Figure 13 Age distribution of perpetrators Figure 14 Age distribution of victims
38. Explanations of results from logistic regression model
Considering other variables constant, the odds of the homicide being
murder goes up by 5.07% with 1 year increment of the perpetrator’s age
Considering other variables constant, the odds of the homicide being
murder goes down by 64.53% if the perpetrator is White rather than Black
Considering other variables constant, the odds of the homicide being
murder goes down by 48.19% if the firearm used is Rifle compared to
Handgun
39. Finding: 05
➢ Perpetrators gets acquainted
with the victims before
homicide
➢ Native American/Alaska
Natives are tend to be
involved as well as victim at
the earliest of age
➢ Perpetrator age, race and
weapon choice have
significant relation with the
intention of homicide
40. Question: 06 Does gun control law play a role
to reduce gun control?
41. Gun laws are strict in the places with higher homicide and vice versa
42. Homicide and Brady score tend to change in the same direction
e.g. both goes up or both goes down at the same time
Sample Est:
Correlation
P-Value
Brady Score-2010 &
Homicide-2011
0.37159 0.008567
Brady Score-2011 &
Homicide-2012
0.36321 0.009677
Brady Score-2013 &
Homicide-2014
0.243046 0.08208
43. There is hardly any change in the laws year on year
Brady Score-
2010
Brady Score-
2011
Brady Score-
2013
Brady Score-2010 1 0.9983 0.957043
Brady Score-2011 0.998 1 0.955259
Brady Score-2013 0.95704 0.95526 1
44. Finding: 06
➢ States having low Brady score
tend to have low number of
homicide
➢ On an average states
doesn’t tend to make big
changes in their existing laws
46. Implications
➢ Help the law enforcement agencies to narrow down their initial suspect
based on the most important predictive variable.
➢ Add another perspective in this existing debate on gun control and help to
clarify some of the areas.
➢ Challenge some of the commonly believed stereotypical associations
between race and homicide or region and homicide.
47. Limitations
➢ Unavailability of unbiased and complete comparative scale on
effectiveness of gun control law
➢ Publicly unavailability of existing comparable scale’s score
48. Future Scope of Research
➢ Unified and unbiased comparative scale for state wise gun law
effectiveness
➢ Study on possibility of inter-race hatred in inter race gun homicide
➢ Study on reasons of early age involvement in homicide in specific races