Professional Documents
Culture Documents
Brandon Starcheus
Response 5
Although it is common to hear of homicide in the US, it is less common to hear of child
murderers. David Hemenway and Sara J. Solnick summarize their study of 146 incidents of
murders involving children aged 0-14 as the suspects in “The epidemiology of homicide
perpetration by children”. Their research spanned all incidents reported to a database which
includes 16 states of the US from 2005 to 2012. They use their best judgment to create 5
categories to describe the cases, and then provide statistics for each. Although their study does
provide as large a sample size as they could obtain with the database and they do address many
limitations of the study, their research still holds several other unidentified weaknesses.
Specifically, as Ioannidis discusses in “Why Most Published Research Findings Are False”, the
research still holds several biases, and research itself is more likely to be conclusive than
inconclusive.
The study’s main form of bias comes from the subject matter itself, as well as the
arbitrary categorization of the murders. Throughout the study they read reports of murders and
determined what fit into the category of a murder, what fit into their definition of “child” (0-14
years), and then categorized the murders themselves from there. All of these identifications are
open for debate, such as the “accident” of a child accidentally shooting a friend. Deciding that
the study was only interested in children 0-14 rides on the assumption that by the age of 15
people have the ability to intelligently choose between life and death, and that their case no
longer applies to the study. Finally, they choose five categories to place murders in, and have
another one for all cases that don’t fit into any of the five. The selection of these categories is
open to interpretation and could be different for different analysts, which limits the validity of
the research. They address these concerns in the article, which is good, but they do not provide
Intermediate Composition
Brandon Starcheus
Response 5
any solutions. Even though the five categories are able to encapsulate most of the cases, 43 of the
146 are in the “other” category. This is almost 30% of the cases, which is a significant amount to
simply put off to the side. Perhaps the study should have selected better categories to include
more data.
Perhaps most significantly of all, the study made use of data from the database, which
only has suspects and does not necessarily include the definite murderer. In this way, the entire
study is based on hypothetical data. The children could have been wrongly accused, or there
could be thousands of cases of murder out there not reported because the child was able to cover
up their involvement. In either case, there is a subset of the population of child murders that has
not been accessed for the study. Although they address the small sample size in the article, the
inaccuracy of the data they do have could be harmful as well. Instead of putting trust in the
thoroughness of the US legal system, they should have increased their sample size. However
with the tools the study had available, their research is extremely thorough.
Finally, the study used statistics directly from their data instead of doing statistical
calculations to incorporate margins of error due to their small sample size. This also contributes
to the validity of the numbers. For larger samples, farther reaching assumptions can be made, but
this study made very broad suggestions without any inclusion of error. Even though the study did
use all relevant cases in their database and their categorizations are quite evident, there are too
many arbitrary selections and too little regard for error that makes the research very limited in
scope. Until a more thorough approach surfaces, though, this is the most complex and
comprehensive study on child murderers and their motives. The numbers may not be completely
accurate, but the study’s suggestions for increased safety in the US have much validity.