What is it about?
The research note highlights the potential of flawed analyses when World Health Organization's (WHO) homicide estimates are employed. The WHO homicide estimates are obtained by adjustments and regression modeling. Both procedures generate estimates that should be used cautiously, especially when inferential statistics are the goal. The reason behind this are the covariates used in the estimation procedure. Among others, the infant mortality rate and alcohol consumption are used by WHO researchers to estimate homicide rates. Then, researchers use the homicide estimates (dependent variable) in a regression analysis with the infant mortality (and/or alcohol consumption, etc.) as a predictor and likely cause an artifact of the measurement of homicides.
Featured Image
Why is it important?
First, it is important to highlight the effort of the World Health Organization (WHO) to provide quality homicide estimates. However, application is not always advisable. Further tests are necessary to find out whether the use of estimated homicide rates has an impact or not. Maybe the difference is negligible. The answer to this question is subject of future research. However, it is important to inform the scientific community about this potential problem because researchers already draw upon the WHO homicide estimates.
Perspectives
Among other things, the research note was motivated by the fact that I had to do some analyses using homicide rates. It wasn't until a colleague's hint that I employed homicide rates without even thinking of how those might have been generated. Data availability and the wish for large samples often interfere a cautious use of the data. Additionally, it was quite a challenge to gather precise information of how the WHO procedure works. Yet it is unknown to what extent results between analyses with and without homicide estimates are different. This question immediately raises other questions, especially concerning data availability to enable researchers to conduct meaningful analyses.
Stefan Kanis
Institute for Interdisciplinary Research on Conflict and Violence (IKG)
Read the Original
This page is a summary of: A Cautionary Note about the Use of Estimated Homicide Data for Cross-National Research, Homicide Studies, June 2017, SAGE Publications,
DOI: 10.1177/1088767917715670.
You can read the full text:
Contributors
The following have contributed to this page
