I myself conduct considerable research in which the unit of analysis is the metropolitan area.  This analysis seems to be conducted with the unit of analysis being the State (N = 50).

Causal inference with such data is very hard, and the limitations are fierce.  This article based on the abstract/presentation seems to ignore all of that.

No lagging of the variables is hinted at.  That means that the data are probably consistent with the idea that deaths of children in households lead to more guns in households.  

Also, the data seem consistent with the idea that a culture of violence in a state leads to both.

And of course, this analysis seems on its face to ignore all the difficulties of the ecological fallacy.

I would not want to criticize the authors of the abstract, but the article about it is pretty thin and unconvincing.





-----Original Message-----
From: herb fox <[log in to unmask]>
To: SCIENCE-FOR-THE-PEOPLE <[log in to unmask]>
Sent: Sat, Nov 2, 2013 9:45 pm
Subject: Kids & guns in the house

A                statistical analysis that gives Medical professionals                the basis and obligation to ". . . talk to patients,                particularly parents, about guns in the home . . . "
A Deadly Mix: Kids and Guns in the        House
Published: Nov 1,        2013
By Charles Bankhead,            Staff Writer, MedPage Today
Reviewed by Zalman              S. Agus, MD; Emeritus Professor, Perelman School of            Medicine at the University of Pennsylvania
Note that this study was published as an                        abstract and presented at a conference. These                        data and conclusions should be considered to be                        preliminary until published in a peer-reviewed                        journal.
ORLANDO -- The combination of children and guns in the              same household contributed to a steady rise in gunshot              wounds and deaths in children, according to a review of              national data over a 12-year period.
The number of children hospitalized for gunshot wounds              increased from 4,270 in 1997 to 7,730 in 2009. In-hospital              deaths among pediatric gunshot patients increased from 317              to 503 over the same period.
Among gunshot wounds with known incident locations, 40%              of the wounds occurred in the child's home. The data              showed a significant association between gunshot wounds to              children and the presence of one or more guns in the              household. The association grew stronger when unsafe gun              storage was included in the analysis (P=0.02 to P<0.01),              although some of the numbers were small.
"Handguns are responsible for most of the childhood              gunshot wounds," Arin L. Madenci, MD, of the University of              Michigan in Ann Arbor, said here at the American Academy of Pediatrics              meeting. "Increasing household gun ownership correlated              with increasing childhood gunshot wounds. There was a              stronger likelihood of home gunshot wounds with unsafe              storage practices."
Approximately 2,700 children die of gunshot wounds each              year. Precise data on the epidemiology of childhood              gunshot wounds are difficult to find, in part, because of              limitations in the CDC budget, which includes about              $100,000 (of a $6 billion total) for the study of              gun-related issues, he said.
According to the Children's Defense Fund, the rate of              gun-related deaths among children and teenagers in the              United States is 3.24 per 100,000. That rate is 17 times              greater than the rates of 25 other developed nations              combined, said Madenci.
The U.S. has the highest gun ownership per capita in the              world (nine guns for every 10 people). Gun-related              homicides involving children more than tripled from 1960              to 2010, he continued.
Given the trends in gun ownership and childhood deaths              attributable to gunshot wounds, Madenci and Christopher B. Weldon, MD, PhD, of              Boston Children's Hospital and Harvard, hypothesized that              household gun ownership would correlate with childhood              gunshot wounds.
To study the relationship, they relied on two sources of              data: the CDC's Behavioral Risk Factor Surveillance                System for estimates of household gun ownership and              the Kids' Inpatient Database for              childhood gunshot wounds. Hospitalizations involving              childhood gunshot wounds and in-hospital deaths were              tracked in 3-year intervals.
Over the course of the study period, 31,156 children were              hospitalized for gunshot wounds and 2,159 children died in              hospital of gunshot wounds.
For the proportion of hospitalizations with a known type              of gun, handguns accounted for 77%, shotguns for 19%, and              hunting or military rifles for 4%.
Focusing on the 40% of gunshot wounds that occurred in              homes, Madenci and Weldon examined the relationship              between the presence of a gun in the home and at-home              gunshot wounds involving children. They also identified              the proportion of households with loaded guns and the              proportion with loaded guns in unlocked storage.
The investigators performed regression analysis to              determine the relationship between the percentage of              households with guns and the percentage of childhood              gunshot wounds. Specifically, they calculated the percent              change (up or down) in childhood gunshot wounds that would              be expected with a 1% change in households with guns. The              resulting beta regression coefficient reflected the              change.
Across the 50 states, the proportion of households with              children and at least one gun ranged from 10% to 62%. The              proportion with loaded guns ranged from 1% to 14%, and the              proportion with loaded guns in unlocked storage from              <1% to 7%. Comparing the household gun-related data and              childhood gunshot wounds showed significant associations              for all comparisons: 
Any gun - beta=0.7, 95% CI 0.3-1.1, P<0.01
Loaded gun - beta=1.7, 95% CI 0.3-3.2, P=0.02
Loaded gun in unlocked storage - beta=2.9, 95% CI                0.6-5.2, P=0.02
An analysis of the relationship between the percentage of              gunshot wounds occurring in the home and the proportion of              households with a gun and children produced beta values of              0.7 for any gun (P<0.01), 2.1 for loaded              firearms (P=0.04), and 4.8 for loaded firearms in              unlocked storage (P=0.01).
The study provided more detailed information about trends              in firearm injuries involving children than has been              available in the past, said David Hemenway, PhD, of the Harvard              School of Public Health.
"The results are consistent with a number of other              studies which have found that where there are more guns,              particularly guns in the home, it's more likely that there              will be danger for children," Hemenway told MedPage                Today.
"Also, how you store the gun makes a difference. Storing              the gun unsafely increases the danger for both accidents              and suicide to children 1 to 17 years of age."
The data suggest actions that healthcare providers can              take to help reduce the number of children injured and              killed by guns each year.
"The study suggests, consistent with other studies, that              physicians should talk to patients, particularly parents,              about guns in the home, so they can weigh the risks and              benefits of having a gun in the home and also talk to              patients about how to store a gun," said Hemenway.