In 1995, firearm homicides were falling dramatically. But in 1999, the direction reversed, and killings started increasing...until about 2005, when they started falling again. Best-fit curve algorithms show the direction could go either way. Is the overall trend down or up? Are there trends in components of overall firearm fatalities that illuminate the answer? What lessons can we learn from the trends for ourselves?

This article is a sidebar to the thorough analysis, "A Benthamite Solution to Gun Control" (Yofiel, 2015) on this site.

### Examining the Trends

Which direction are firearm fatalities headed? The 30-year information is too ambiguous, but since 2005, the homicides have been decreasing steadily. So to help foresee the possibilities, this article examines the trend since 2005, and projects it forwards for 10 years, linearly. That provides a coherent standard against which the actual future can be gauged. But first, we need to consider another important fact, and that is that the 30-year trend of firearm homicides, shown above, are only part of the overall number of fatalities due to firearms. There are also suicides and accidents, which together comprise three-quarters of all firearm fatalities.

#### Top-Level Decadal Trends

First, some claim overall firearm fatalities appear to be rising, but due to population growth being more than the rate of increase, the fatalities are actually falling as a ratio to the population. But this decadal analysis STILL found the total rate to be rising 0.31%, even after Annual Growth Adjustment (AGA). Also, suicides are still rising at 2.12%, but homicides are falling at twice the rate, at a combined -5% with AGA. In fact, after projecting the actual total cases of murder from known cases, involuntary fatalities actually overtook murder in 2013. The following illustration displays the top-level category values, statistical analysis, trends, and projected values for the next decade (see the later section "Source Data and Methodologies" for details).For the most recent year of known data, suicide and involuntary deaths still total more than three quarters of all fatalities. With current trends continuing as they are, total fatalities will be 37,143 in 2024. A logarithmic plot of firearm fatalities illustrates the relative relationships of decreasing murder and crime, with lower increases in all other categories, resulting in the total increase of annual fatalities.

These two pie charts show how small changes can add up over time.

#### Suicide and Injury Trends

The following table combines some of the above results with suicide and injury data.It is thus estimated that 22,662 firearm suicides will occur in 2016, or 65.8% of all firearm suicides. The raw number of firearm suicides is increasing at 2.9% annually. With USA's steady 0.78% annual increase in population, this is 2.90%/year, average-growth adjusted (AGA). Total firearm fatalities are increasing at 1.1%/year (non-AGA, 0.31% AGA). So firearm suicides are increasing between twice and thrice the rate of total firearm fatalities, resulting in more deaths, and more deaths as a proportion of firearm fatalities, as illustrated below. This makes firearm suicide an important topic for gun control. Compared to the mean number of successful suicides in the USA for the last known decade (19,063), 52.2% chose firearms as the method (AGA). Thus, the Brady campaign's claim that firearms are the method of suicide for more than half of all successful suicide attempts is true by only a very small margin. The following plot shows the current trends for suicide injuries and deaths, with the X axis logarithmically scaled so that the trends are clearly visible.

The logarithmic plot shows there are ~700,000 suicide injuries resulting in hospitalization. Of these, ~600,000 are injuries, but only ~5,000 are injuries if a firearm was used (0.1%). In contrast, there are ~40,000 successful suicide attempts, and deaths, and half the deaths are with firearms (50%). So if a suicide victim does not have access to a gun, but still attempts to commit suicide, the chances of survival are ~500 times greater.

This leads to the natural question if the same ratios are true for violent homicides.

In 2016, Violent crime overall may not be increasing linearly, and variations make it difficult to predict. Looking at averages for the last known decade, violence resulted in ~40% all firearm injuries. In one in seven cases where a firearm was used in violence, death resulted. However, while cases of violent injuries were increasing (1.4% AGA), incidents resulting in death were decreasing (-3.5% AGA). with a total spread of 5% between rising injuries and falling deaths, the difference has already increased by 35% since the middle of the decadal sample range (July 2009), and since 2010, injuries from violent crime appears to be compounding, whereas deaths appear to be continuing to decrease linearly.

#### Crime Homicide and Murder Trends

The following illustration shows the resulting trend groups data for 2005-2014, with the same population growth variance and linear regression, and projection methods as for the top-level data.This breakout reveals that the majority of FBI-reported violent crime is due to arguments and brawls (together, 9.9% of total firearm fatalities). Gangland and juvenile gang killings are, together, 3.3% of total firearm fatalities. Other causes of murder are more frequent than gangland killings, at 4.7%. A breakdown of crime homicides reveals that the majority are due to theft or burglary (~3.0% of total firearm fatalities). Drugs are the next most prevalent cause (~1.8% of total firearm fatalities). Homicides from other crimes, such as rape and arson, are rare (~1.1% of total firearm fatalities).

It is sometimes claimed crime rates due to gang activity are increasing. The following chart shows the decadal trend for the above defined breakout categories.

This illustrates that there has not been an increase in gang activity, nor will there be from a simple linear projection. On the contrary, it appears that all major defined categories of murder and crime homicide are falling drastically in various directions, as shown in the following 20-year trend.

Three times as many firearm fatalities are now due to arguments and brawls than due to gang activity.

#### Mass Shootings

The FBI counts any homicide where four or more die as a mass shooting. "A Study of Active Shooter Incidents in the United States Between 2000 and 2013" (*U.S. Department of Justice*, 2014) shows mass shootings increasing at 12%/year. The following illustration shows the data from this report, as well as linear trendlines until 2024.

The following graph shows the mass-shooting incidents and casualties for 2005-2013, as well as a projection until 2024, based on delta change from the mean of known decadal values. However, due to R^{2} values below 0.3, the projection really cannot indicate more than a general trend upwards, which may be linear, logarithmic, exponential. or continually chaotic. By projecting from only those values in the decadal range of this study, the rate of increase was 3% than the D0J projection, which included the prior 5 years since 2000.

From data in "Guns are More Dangerous to Owners than Criminals" (*Yofiel*, 2016), it is at the very least reasonable to expect that domestic disputes will increasingly result in more mass shootings, partly due to the larger number of guns and handguns in particular owned by the attacker, and partly due to the increasing number of guns owned in fewer households, and partly due to continuing conflict of interest between those promoting gun sales and those trying to prevent gun injuries.

### Self Defense Fatalities Compared to Crime

"Guns are More Dangerous to Owners than Criminals" (*Yofiel*. February 2016) estimated 430.18 accidental fatalities of due to acts of attempted self defense in 2015. This finding was combined with the factor of 236.0 (the estimated number self-defense homicides for 2015) to provide a ratio of mistaken killings to successes in self defense (430:18:236.0), which via linear projection from known decadal span provides a forward projection. This is compared with previously calculated data above for homicides during robbery/theft, and with mass shooting victims, to provide the following data. It. already includes annual population variance from above calculations.

The following chart plots the data and trend lines for mass shootings, attempted self defense (successful, mistaken, and total), together with homicides due to theft and burglary.

The trends show that, as of 2012, the total number of people killed in attempted self defense, including those killed by mistake, overtook the number of those killed by crime from theft and burglary. Then in 2023, the number killed by mass shootings will surpass those killed by criminals during theft and burglary. Public attention has focused on the latter trend, rather than the more significant former fact.

Finally, there are about six times as many killed in accidents from attempted self defense than killed in mass shootings now. The number killed in mass shootings is still small compared to those being killed in attempted self defense. Thus, by linear projection, homicides due to mistakes in self defense will still exceed those from mass shootings in the next decade, even though the rate of increase in mass shootings is much larger.

### Source Data and Methodology

For total firearm fatalities and suicide data, this study retrieved data from the*Center for Disease Control* (CDC) "Wonder" and "WISQUARS" databases. The following table aggregates the retrieved data.*Tap on any picture for a lightbox zoom, and tap anywhere on the picture to close it*.

For homicides, this study collates the "FBI Criminal Justice Information Services" spreadsheet data, as follows:

#### Notes on Methodology

In the tabulated data, the yellow cells are original data; the green cells are projected values; and the blue cells are adjusted values and statistical characteristics, as follows:- Rate:For all except the population variance, the rate of change is calculated each year, for each category, then averaged together. For population variance, the rate column contains the percentage increase for the linear regression in population growth from which the population variance percentages provide annual deviation from the linear growth, which is applied to the adjusted data values.
- Δ:The annual step value for the ''best-fit"' linear extrapolation (''delta'), calculated using
*ordinary least-squares*(OLS) regression. - Mean:The average, across the known decadal values.
- Σ
_{err}:The*standard error from mean estimate*('sigma'), across the known decadal values. The probability of any source data falling within one multiple of Σ_{err}from the regression line is ~68%; within two multiples, 95%; and withing three multiples, 99%. There is more than one way to determine Σ_{err}.For a low number of discrete equidistant values, this was thought most appropriate:*s*[A*i*, A*j*. . .A*n*] / √ #[A*i*, A*j*. . .A*n*]

- R
^{2}:The*coefficient of determination*, across the known decadal values. R^{2}values closer to unity indicate more accuracy in the linear regression. - Population Growth Adjustment The adjusted data includes compensating factors derived fromUS. Census bureau population reports. From this was deduced: (1) a best-fit linear estimate for the decadal range; (2) the annual growth rate, assuming linear growth; (3) the annual deviation from the linear growth rate; and (4), the annual variance from the linear growth rate. For best possible linearity, annual variances were also reduced by 0.27% (365/366) in the leap-year values.
- Adjusted Data: As the variance from a linear regression is randomly distributed and <0.7% for the sampled decade, the annual variance is simply applied to each year in the adjusted data. Rate, Δ, Mean, Σ
_{err}, and R^{2}reflect the resulting improvements. - Rate
_{AGA}and Δ_{AGA}:The reduced rate and Δ values for each decadal prediction that result from compensating for the linearly increasing population growth, in the predicted annual rise or fall of values. - Main Categories: Mutually-exclusive categorical values for causes of firearm fatality were created from the above data sources, as follows:
- Suicide:Total suicide fatalities by firearm, from CDC 'Wonder database' query.
- Murder and Crime: From the main categories in the FBI homicide spreadsheets. Then, to improve the trend R
^{2}certainty factor, unsolved homicides for each year were split proportionally into the crime and murder categories. This was found to improve R^{2}, most likely due to law enforcement having more constantly limited resources to investigate homicides, and with the downwards trend, less cases were fully investigated earlier in the decade. Thus, the higher quality of more recent solved-cases ratios for lower total counts improved the predictability of the trend. - Involuntary:Involuntary manslaughters (including acts by minors 16 and under), determined by subtracting the sum of all suicides and homicides (including police and self-defense actions) from the total number of reported firearm fatalities.
- Police and Self Defense: From the FBI homicide spreadsheets.

- Homicide SubcategoriesFor the FBI data, some of the subcategories are very small, leading to unreliable trends (as indicated by the very low R
^{2}values). Also, there is an 'unsolved' homicide category, and separate unspecified subcategories for felony, suspected felony, and murder. Therefore the FBI subcategories with numbers too small to define trendlines were collated into mutually exclusive, semantic groups as follows:- Burglary and theft: Robbery, burglary, larceny-theft, and motor-vehicle theft
- Narcotics crime: Narcotics drug laws
- Other crime: rape, arson, prostitution and commercialized vice, other sex offenses, gambling, and 'other-not specified under 'felony type'
- Arguments and brawls: Romantic triangles, brawls due to influence of alcohol, brawls due to influence of narcotics, arguments over money or property, and other arguments.
- Gangs:Gangland killings and juvenile gang killings
- Other murder:Child killed by babysitter, institutional killings, sniper attack, and 'other not specified' under 'other than felony type total.'

- On Suicide: Many have objected that suicide is not gun violence, and that guns are therefore not the cause of suicide. See "Suicide and Firearms,"
*Yofiel*, January 2016.

#### References

Direct links are provided were possible.**Total Firearm Fatalities, Suicide, and Injuries**:

"http://wonder.cdc.gov/" (*Center for Disease Control (CDC)*database portal, 2016).

"WISQARS" (*Center for Disease Control (CDC)*database portal, 2016).

"Suicide and Firearms" (*Yofiel*, 2016).**Crime and Murder Homicides**:

2014: "Expanded Homicide Data: Circumstances by Weapon Type, 2014" (*FBI*, 2015).

2013: "Expanded Homicide Data: Circumstances by Weapon Type, 2013" (*FBI*, 2014).

2012: "Expanded Homicide Data: Circumstances by Weapon Type, 2012" (*FBI*, 2013).

2011: "Expanded Homicide Data: Circumstances by Weapon Type, 2011" (*FBI*, 2012).

2010: "Expanded Homicide Data: Circumstances by Weapon Type, 2010" (*FBI*, 2011).

2009: "Expanded Homicide Data: Circumstances by Weapon Type, 2009" (*FBI*, 2010).

2008: "Expanded Homicide Data: Circumstances by Weapon Type, 2008" (*FBI*, 2009).

2007: "Expanded Homicide Data: Circumstances by Weapon Type, 2007" (*FBI*, 2008).

2006: "Expanded Homicide Data: Circumstances by Weapon Type, 2006" (*FBI*, 2007).

2005: "Expanded Homicide Data: Circumstances by Weapon Type, 2005" (*FBI*, 2006).

2004: "Expanded Homicide Data: Circumstances by Weapon Type, 2004" (*FBI*, 2005).

2003: "Expanded Homicide Data: Circumstances by Weapon Type, 2003" (*FBI*, 2004).

2002: "Expanded Homicide Data: Circumstances by Weapon Type, 2002" (*FBI*, 2003).

2001: "Expanded Homicide Data: Circumstances by Weapon Type, 2001" (*FBI*, 2002).

2000: "Expanded Homicide Data: Circumstances by Weapon Type, 2000" (*FBI*, 2001).

1999: "Expanded Homicide Data: Circumstances by Weapon Type, 1999" (*FBI*, 2000).

1998: "Expanded Homicide Data: Circumstances by Weapon Type, 1998" (*FBI*, 1999).

1997: "Expanded Homicide Data: Circumstances by Weapon Type, 1997" (*FBI*, 1998).

1996: "Expanded Homicide Data: Circumstances by Weapon Type, 1996" (*FBI*, 1997).

1995: "Expanded Homicide Data: Circumstances by Weapon Type, 1995" (*FBI*, 1996).**Justified Homicides**:

Self Defense, 2010~2014: "Expanded Homicide Data: Justified (Civilian) 2010~2014" (*FBI*, 2015).

Self Defense, 2005~2009: "Expanded Homicide Data: Justified (Civilian) 2005~2009" (*FBI*, 2010).

Police, 2010~2014: "Expanded Homicide Data: Justified (Law Enforcement Officials), 2010~2014" (*FBI*, 2015).

Police, 2005~2009: "Expanded Homicide Data: Justified (Law Enforcement Officials), 2005~2009" (*FBI*, 2010).

Combined report, 2000-2004: "Expanded Homicide Data: Circumstances by Weapon Type, 2004" (*FBI*, 2005).

Combined report, 1995-1999: "Expanded Homicide Data: Circumstances by Weapon Type, 1999" (*FBI*, 2000).**Accidental Deaths:**"Guns are More Dangerous to Owners than Criminals" (*Yofiel*, 2016).**Mass Shootings:**"A Study of Active Shooter Incidents in the United States" (*FBI*, 16 September 2013).**Population:**: "Population Estimates" (*United States Census Bureau*, 2015).