A Criminologist Questions Sandy Hook

Who in their right mind would even think of questioning accounts of politically momentous events (such as Sandy Hook) proffered by various MSM outlets?

Well, at this point, many will agree that the MSM has dropped even the merest pretense of objectivity; MSM journalism has been fundamentally transformed into sheer advocacy.

Here, we are going to do something the MSM won't: provide context for the Sandy Hook event by conducting a statistical analysis of hard data on 30 years' worth of mass public shootings in the United States. 

Quickly, though, I want to show you an example of an MSM outlet's (CNN) utter failure to do its job on Sandy Hook/Newtown.

A Sandy Hook timeline at CNN contains the following claims:

1.     Police and other first responders arrived on scene about 20 minutes after the first calls.

2.     The gunman took his own life, police said. He took out a handgun and shot himself in a classroom as law enforcement officers approached, officials said.

A March 28 post at CNN asserts the following:

3.     Lanza didn't make it home alive. Nor did the 26 people -- 20 of them schoolchildren ages 6 and 7 -- he shot dead in less than five minutes, firing one bullet roughly every two seconds he was at Sandy Hook Elementary School.

So, the people Lanza killed were dead in less than 5 minutes, Lanza shot himself as "law enforcement officers approached", but it also took "police and other first responders" 20 minutes to arrive on the scene?

Doesn't a reasonable construction of the above at least raise the question: so what was Lanza doing in the 15 minutes or so after the killings and before he shot himself f-- particularly when Lanza is described as trying to kill as many people as possible, and when Lanza is described as having been found dead alongside a multitude of unspent rounds?

Please notice that it doesn't matter for purposes of this issue whether the claims are accurate; what matters is that CNN asserts that they are -- so why aren't they asking the obvious follow up questions?

The MSM Sandy Hook Narrative  

Readers will agree that the MSM narrative regarding the Sandy Hook mass murder event runs something like this, in terms of its basics: a lone gunman (Adam Lanza, age 20), who was perhaps psychotic, acted in a purely private capacity and murdered 26 people (20 children and 6 adults) while wounding only two.

I say "wounding only two."

Why?

In part, for empirical reasons -- and in part for common sense as well as theoretical reasons. 

When you look at the Mother Jones data set of mass public shootings over the last 30 years, you will see that mass public shootings in which there are on the order of 20 or more people killed and as few as 2 people wounded have never, in the last 30 years, happened. 

If you're looking for the closest comparison in the Mother Jones dataset, you will settle on the 2009 Binghamton New York case in which Jiverly Wong killed 13 people (14 including himself) and wounded 4.

It's true that in 1987 David Burke used a gun to commit a mass murder that left 43 people dead (including Burke) and nobody wounded.

Francisco Gonzales behaved similarly in 1964 and left 44 people dead (including Gonzales) and zero people wounded.

But guess what?  These cases involved the use of a gun during a hijack that culminated in plane crashes

Sandy Hook was not a plane crash case, so can we agree to leave the plane crash cases aside?

Mass public shootings on land in the United States never have on the order of 26 people killed and as few as 2 wounded.

There are two types of good reasons for this.

The first type of reason can be easily conceptualized by considering the analogy of coin tosses.

There, the outcome is either a head or a tail, which, by analogy, corresponds with "killed" and "wounded." 

The more times you flip a coin (the more victims shot at), the more tails (wounded) you will expect to observe simply as a matter of chance

Furthermore, this holds no matter the bias of the coin (which, by analogy, corresponds to the accuracy of the shooter.)

That is, unless, of course, the coin (shooter) is perfectly biased (in that the shooter is certain to kill) -- but nobody can make that claim about Lanza since he is said to have wounded 2.

The second type of reason is that the higher the victim total, (killed + wounded) the more "opportunity" there is for chance to come into play in ways that have nothing to do with the shooter's accuracy (where accuracy is considered to be a relatively time stable characteristic of the shooter).

When the victim total is large, weapons may be more likely to require reloading (giving victims more time to flee, thereby creating more distance between themselves and the shooter); more risk may be presented to the shooter; there may be more confusion; more time might be required to generate more casualties (meaning that there is more risk that the shooter will stumble but the weapon discharges anyway; that the sun glints off a window into the eye of the shooter, etc.) and so on.

Obviously, such factors will have different weight depending on the case, but the general idea is clear and uncontroversial.

The preceding discussion addresses a very important empirical fact regarding mass public shooting events in the United States (that land-based public shootings never -- until Sandy Hook --  have had as many as 27 dead and as few as 2 wounded) and supplied reasons that explain that empirical fact.

But, we can do more.

We can, using the hard Fatalities and Injuries data in the Mother Jones data set (which canvasses 62 mass public shootings in the United States since 1982), assess the probability that -- whatever the causal processes are that have functioned to produce mass public shootings over the past 30 years -- produced Sandy Hook Elementary's 27 killed (including Lanza) and 2 wounded, figures.

The analysis drops the Crandon, Wisconsin case because, as Mother Jones notes, the shooting took place in a private dwelling.  In case you are wondering, this minor change does not affect the results.

So, simply put: what does 30 years' worth of data on mass public shootings say about the probability that Lanza, acting alone, purely privately, and possibly psychotic, killed 27 people and wounded only 2?

At this point, some people are saying: "so what if this kind of thing has never happened before."

The Jacksonville Jaguars (2-14 last year) have never won the Super Bowl.  Do you think they'll win next year? 

Statistical Analysis

This statistical analysis deploys a model known as "exact logistic regression."

Exact logistic regression (click here for a gentle introduction) is superior to ordinary logistic regression when sample sizes are relatively small, as is the case with the Mother Jones data set (61 events).

It's an excellent model to use here for substantive purposes as well, since it allows us to model the probability that mass murder events yield -- as a function of the total number of victims produced by the incident -- three or more, in contrast to two or less -- wounded victims. 

So, an exact logistic regression offers one good way to test the reasoning in the previous section.

We are particularly interested in doing so since Newtown generated a scant 2 wounded victims in comparison to 27 dead at the scene. 

What is the probability of such an occurrence?

If we use exact logistic regression to model the general question of the probability of three or more wounded in contrast to two or less as a function of the total number of victims, we can use the output to derive the probability that a specific mass murder event with 29 total victims (Newtown) had two or fewer victims. 

Thus, the outcome variable is the 3 or more wounded/2 or less wounded dichotomy, and the predictor variable is the total number of victims.

The regression was run in STATA 12.  Here is the output:


The "Pr >= Score" value of .0001 indicates that the model is highly statistically significant.

"WoundedSplit" is the dichotomous outcome measure described above.

"Totalvic" is the label for the Total Victims predictor variable.  The "2*Pr(Suff.) of .0000 shows that this variable is highly statistically significant, but what is its effect?

In a word: powerful.  The "Odds Ratio" associated with the "total victims" predictor variable of 1.43 indicates that each time the victim total climbs by one (working from a minimum total of 4, since a minimum of 4 had to be murdered in order to qualify for the Mother Jones dataset), we expect a 43% increase in the odds that 3 or more victims are wounded and not killed.   

With the odds ratio (and its underlying coefficient of .3580897 in hand), we can compute the probability of observing, with a total victim count of 29, a wounded count of 3 or more.  Here is the output:

Now, if, with 29 total victims, the probability of observing a wounded count of 3 or more is .9995, the probability of observing the alternative of 2 wounded victims or less is 1-.9995, or .0005.

.0005 is 5 in 10,000, or 1 in 2,000.

Since the Mother Jones data set contains 30 years of data on mass public shootings, we can conclude:

Whatever causal processes accounted for observed kill/wound figures in the last 30 years of mass public shootings, there is only a 1 in 2,000 chance that they account for what has been reported in this regard about Newtown.

In an era where words like "trillions" are tossed around, this figure might not appear very startling.

However, here is some perspective.

Current Las Vegas odds on the Jacksonville Jaguars (2-14 last year) winning the 2014 Super Bowl are, by comparison, a mere 150-1.

Issues Raised by the Analysis 

The 1 in 2,000 figure provides compelling prima facie evidence that factors accounting for observed kill/wound totals in the last 30 years of mass public shootings do not explain the Newtown figures. 

Newtown seems to have been, from a causal standpoint, different

The question is obviously why.

Many have argued that since the victims were children, they may well have frozen in place (making them easy kill, but presumably not wound, targets.)

There may be something to this, but a few things should be considered by way of response. 

First, there are reports that six children in Victoria Soto's classroom tried to flee, but were shot by Lanza -- so, if this is true, at least six of the child victims were not "frozen."

This report says of six children taken in by Gene Rosen that:

Their teacher, Victoria Soto, 27, had reportedly hid her students from the approaching gunman. The six youngsters who escaped then had to apparently run past her body to safety.

Returning to the Hartford Courant report, we read that:

Lanza next arrived at teacher Victoria Soto's classroom. Soto is believed to have hidden her 6- and 7-year old students in a classroom closet. When Lanza demanded to know where the children were, Soto tried to divert him to the other end of the school by saying that her students were in the auditorium.

But six of Soto's students tried to flee. Lanza shot them, Soto and another teacher who was in the room. Later, in their search for survivors, police found the remaining seven of Soto's students still hiding in the closet. They told the police what had happened.

It seems there is only one way both of these accounts can be true: if there were at least 19 children in Soto's classroom (six who were shot dead, six found by Rosen, and seven who were hidden in the closet and discovered by law enforcement; it seems very unlikely that the seven children found by law enforcement numbered among those found by Rosen, and not only because of the numeric disparity -- why would police have dispatched the children to who knows where after having discovered them?).

No matter what, though, the accounts suggest that at least six children were moving.

And, don't forget that six adults numbered among the dead -- why weren't more adults wounded?

Of the two wounded at Newtown, to the best of my knowledge we have information about only one -- Natalie Hammond. 

The linked article makes clear that Hammond is supposed to have come "face-to-face" with Lanza.

Evidently, Hammond was shot in the foot, leg, and hand

Why was Lanza even aiming at the lower extremities, and, if he wasn't but hit Hammond there anyway, what does that say about his accuracy -- especially when you consider that Hammond is said to have been an adult target in very close proximity to Lanza?

In any event, the damage Lanza dealt to Hammond couldn't have been extraordinarily grave, since 67 days later, sans crutches, Hammond strode across the rink at the Boston Bruins game and dropped the puck. 

What does that say about Lanza's kill accuracy?

And what about the walls in the school -- might rounds have penetrated the walls, and, if so, why weren't more people wounded?

Finally, there is talk that Lanza might have been psychotic.

In their study of a sample of mass murderers, Hempel et al. found that "the average kill to wound ratio of the psychotic subjects was 1:1.4" (p.218).

That ratio does not exactly fit Sandy Hook, does it?

A statistical analysis of 30 years' worth of mass public shooting hard data shows that the odds are 2,000 to 1 against the processes underlying that data accounting for the kill/wound figures of the Sandy Hook shooting.

Therefore, very meaningful and very serious questions arise as to precisely what transpired in Newtown -- questions that are very unlikely to receive a hearing in the MSM.

Why?

Dr. Jason Kissner is associate professor of criminology at California State University, Fresno.  You can reach him at crimprof2010@hotmail.com.

Who in their right mind would even think of questioning accounts of politically momentous events (such as Sandy Hook) proffered by various MSM outlets?

Well, at this point, many will agree that the MSM has dropped even the merest pretense of objectivity; MSM journalism has been fundamentally transformed into sheer advocacy.

Here, we are going to do something the MSM won't: provide context for the Sandy Hook event by conducting a statistical analysis of hard data on 30 years' worth of mass public shootings in the United States. 

Quickly, though, I want to show you an example of an MSM outlet's (CNN) utter failure to do its job on Sandy Hook/Newtown.

A Sandy Hook timeline at CNN contains the following claims:

1.     Police and other first responders arrived on scene about 20 minutes after the first calls.

2.     The gunman took his own life, police said. He took out a handgun and shot himself in a classroom as law enforcement officers approached, officials said.

A March 28 post at CNN asserts the following:

3.     Lanza didn't make it home alive. Nor did the 26 people -- 20 of them schoolchildren ages 6 and 7 -- he shot dead in less than five minutes, firing one bullet roughly every two seconds he was at Sandy Hook Elementary School.

So, the people Lanza killed were dead in less than 5 minutes, Lanza shot himself as "law enforcement officers approached", but it also took "police and other first responders" 20 minutes to arrive on the scene?

Doesn't a reasonable construction of the above at least raise the question: so what was Lanza doing in the 15 minutes or so after the killings and before he shot himself f-- particularly when Lanza is described as trying to kill as many people as possible, and when Lanza is described as having been found dead alongside a multitude of unspent rounds?

Please notice that it doesn't matter for purposes of this issue whether the claims are accurate; what matters is that CNN asserts that they are -- so why aren't they asking the obvious follow up questions?

The MSM Sandy Hook Narrative  

Readers will agree that the MSM narrative regarding the Sandy Hook mass murder event runs something like this, in terms of its basics: a lone gunman (Adam Lanza, age 20), who was perhaps psychotic, acted in a purely private capacity and murdered 26 people (20 children and 6 adults) while wounding only two.

I say "wounding only two."

Why?

In part, for empirical reasons -- and in part for common sense as well as theoretical reasons. 

When you look at the Mother Jones data set of mass public shootings over the last 30 years, you will see that mass public shootings in which there are on the order of 20 or more people killed and as few as 2 people wounded have never, in the last 30 years, happened. 

If you're looking for the closest comparison in the Mother Jones dataset, you will settle on the 2009 Binghamton New York case in which Jiverly Wong killed 13 people (14 including himself) and wounded 4.

It's true that in 1987 David Burke used a gun to commit a mass murder that left 43 people dead (including Burke) and nobody wounded.

Francisco Gonzales behaved similarly in 1964 and left 44 people dead (including Gonzales) and zero people wounded.

But guess what?  These cases involved the use of a gun during a hijack that culminated in plane crashes

Sandy Hook was not a plane crash case, so can we agree to leave the plane crash cases aside?

Mass public shootings on land in the United States never have on the order of 26 people killed and as few as 2 wounded.

There are two types of good reasons for this.

The first type of reason can be easily conceptualized by considering the analogy of coin tosses.

There, the outcome is either a head or a tail, which, by analogy, corresponds with "killed" and "wounded." 

The more times you flip a coin (the more victims shot at), the more tails (wounded) you will expect to observe simply as a matter of chance

Furthermore, this holds no matter the bias of the coin (which, by analogy, corresponds to the accuracy of the shooter.)

That is, unless, of course, the coin (shooter) is perfectly biased (in that the shooter is certain to kill) -- but nobody can make that claim about Lanza since he is said to have wounded 2.

The second type of reason is that the higher the victim total, (killed + wounded) the more "opportunity" there is for chance to come into play in ways that have nothing to do with the shooter's accuracy (where accuracy is considered to be a relatively time stable characteristic of the shooter).

When the victim total is large, weapons may be more likely to require reloading (giving victims more time to flee, thereby creating more distance between themselves and the shooter); more risk may be presented to the shooter; there may be more confusion; more time might be required to generate more casualties (meaning that there is more risk that the shooter will stumble but the weapon discharges anyway; that the sun glints off a window into the eye of the shooter, etc.) and so on.

Obviously, such factors will have different weight depending on the case, but the general idea is clear and uncontroversial.

The preceding discussion addresses a very important empirical fact regarding mass public shooting events in the United States (that land-based public shootings never -- until Sandy Hook --  have had as many as 27 dead and as few as 2 wounded) and supplied reasons that explain that empirical fact.

But, we can do more.

We can, using the hard Fatalities and Injuries data in the Mother Jones data set (which canvasses 62 mass public shootings in the United States since 1982), assess the probability that -- whatever the causal processes are that have functioned to produce mass public shootings over the past 30 years -- produced Sandy Hook Elementary's 27 killed (including Lanza) and 2 wounded, figures.

The analysis drops the Crandon, Wisconsin case because, as Mother Jones notes, the shooting took place in a private dwelling.  In case you are wondering, this minor change does not affect the results.

So, simply put: what does 30 years' worth of data on mass public shootings say about the probability that Lanza, acting alone, purely privately, and possibly psychotic, killed 27 people and wounded only 2?

At this point, some people are saying: "so what if this kind of thing has never happened before."

The Jacksonville Jaguars (2-14 last year) have never won the Super Bowl.  Do you think they'll win next year? 

Statistical Analysis

This statistical analysis deploys a model known as "exact logistic regression."

Exact logistic regression (click here for a gentle introduction) is superior to ordinary logistic regression when sample sizes are relatively small, as is the case with the Mother Jones data set (61 events).

It's an excellent model to use here for substantive purposes as well, since it allows us to model the probability that mass murder events yield -- as a function of the total number of victims produced by the incident -- three or more, in contrast to two or less -- wounded victims. 

So, an exact logistic regression offers one good way to test the reasoning in the previous section.

We are particularly interested in doing so since Newtown generated a scant 2 wounded victims in comparison to 27 dead at the scene. 

What is the probability of such an occurrence?

If we use exact logistic regression to model the general question of the probability of three or more wounded in contrast to two or less as a function of the total number of victims, we can use the output to derive the probability that a specific mass murder event with 29 total victims (Newtown) had two or fewer victims. 

Thus, the outcome variable is the 3 or more wounded/2 or less wounded dichotomy, and the predictor variable is the total number of victims.

The regression was run in STATA 12.  Here is the output:


The "Pr >= Score" value of .0001 indicates that the model is highly statistically significant.

"WoundedSplit" is the dichotomous outcome measure described above.

"Totalvic" is the label for the Total Victims predictor variable.  The "2*Pr(Suff.) of .0000 shows that this variable is highly statistically significant, but what is its effect?

In a word: powerful.  The "Odds Ratio" associated with the "total victims" predictor variable of 1.43 indicates that each time the victim total climbs by one (working from a minimum total of 4, since a minimum of 4 had to be murdered in order to qualify for the Mother Jones dataset), we expect a 43% increase in the odds that 3 or more victims are wounded and not killed.   

With the odds ratio (and its underlying coefficient of .3580897 in hand), we can compute the probability of observing, with a total victim count of 29, a wounded count of 3 or more.  Here is the output:

Now, if, with 29 total victims, the probability of observing a wounded count of 3 or more is .9995, the probability of observing the alternative of 2 wounded victims or less is 1-.9995, or .0005.

.0005 is 5 in 10,000, or 1 in 2,000.

Since the Mother Jones data set contains 30 years of data on mass public shootings, we can conclude:

Whatever causal processes accounted for observed kill/wound figures in the last 30 years of mass public shootings, there is only a 1 in 2,000 chance that they account for what has been reported in this regard about Newtown.

In an era where words like "trillions" are tossed around, this figure might not appear very startling.

However, here is some perspective.

Current Las Vegas odds on the Jacksonville Jaguars (2-14 last year) winning the 2014 Super Bowl are, by comparison, a mere 150-1.

Issues Raised by the Analysis 

The 1 in 2,000 figure provides compelling prima facie evidence that factors accounting for observed kill/wound totals in the last 30 years of mass public shootings do not explain the Newtown figures. 

Newtown seems to have been, from a causal standpoint, different

The question is obviously why.

Many have argued that since the victims were children, they may well have frozen in place (making them easy kill, but presumably not wound, targets.)

There may be something to this, but a few things should be considered by way of response. 

First, there are reports that six children in Victoria Soto's classroom tried to flee, but were shot by Lanza -- so, if this is true, at least six of the child victims were not "frozen."

This report says of six children taken in by Gene Rosen that:

Their teacher, Victoria Soto, 27, had reportedly hid her students from the approaching gunman. The six youngsters who escaped then had to apparently run past her body to safety.

Returning to the Hartford Courant report, we read that:

Lanza next arrived at teacher Victoria Soto's classroom. Soto is believed to have hidden her 6- and 7-year old students in a classroom closet. When Lanza demanded to know where the children were, Soto tried to divert him to the other end of the school by saying that her students were in the auditorium.

But six of Soto's students tried to flee. Lanza shot them, Soto and another teacher who was in the room. Later, in their search for survivors, police found the remaining seven of Soto's students still hiding in the closet. They told the police what had happened.

It seems there is only one way both of these accounts can be true: if there were at least 19 children in Soto's classroom (six who were shot dead, six found by Rosen, and seven who were hidden in the closet and discovered by law enforcement; it seems very unlikely that the seven children found by law enforcement numbered among those found by Rosen, and not only because of the numeric disparity -- why would police have dispatched the children to who knows where after having discovered them?).

No matter what, though, the accounts suggest that at least six children were moving.

And, don't forget that six adults numbered among the dead -- why weren't more adults wounded?

Of the two wounded at Newtown, to the best of my knowledge we have information about only one -- Natalie Hammond. 

The linked article makes clear that Hammond is supposed to have come "face-to-face" with Lanza.

Evidently, Hammond was shot in the foot, leg, and hand

Why was Lanza even aiming at the lower extremities, and, if he wasn't but hit Hammond there anyway, what does that say about his accuracy -- especially when you consider that Hammond is said to have been an adult target in very close proximity to Lanza?

In any event, the damage Lanza dealt to Hammond couldn't have been extraordinarily grave, since 67 days later, sans crutches, Hammond strode across the rink at the Boston Bruins game and dropped the puck. 

What does that say about Lanza's kill accuracy?

And what about the walls in the school -- might rounds have penetrated the walls, and, if so, why weren't more people wounded?

Finally, there is talk that Lanza might have been psychotic.

In their study of a sample of mass murderers, Hempel et al. found that "the average kill to wound ratio of the psychotic subjects was 1:1.4" (p.218).

That ratio does not exactly fit Sandy Hook, does it?

A statistical analysis of 30 years' worth of mass public shooting hard data shows that the odds are 2,000 to 1 against the processes underlying that data accounting for the kill/wound figures of the Sandy Hook shooting.

Therefore, very meaningful and very serious questions arise as to precisely what transpired in Newtown -- questions that are very unlikely to receive a hearing in the MSM.

Why?

Dr. Jason Kissner is associate professor of criminology at California State University, Fresno.  You can reach him at crimprof2010@hotmail.com.