What it measures
ViCRI establishes and maintains a standardized measurement and scoring methodology to rate international cities on personal risk and safety dimensions. The metrics focus primarily on the probability of violent criminal victimization for residents and visitors.
What makes it unique
Several commercial and open sources aim to rate urban risks. These vary significantly in methods and scope, often combining street, property, and violent crimes, or even road accidents, infrastructure aspects, and vulnerability to natural disasters.
The ViCRi index improves legacy risk rating initiatives by addressing many well‑documented methodological challenges and fallacies ranging from data quality, language ambiguity, legal definitions and enforcement inconsistencies, the use of ordinal scales, and more (see methodological highlights below.)
Anyone can look up their home city or travel destination for personal safety, security awareness, and planning.
Public officials, including city managers, law enforcement personnel, and other stakeholders, can leverage and contribute to the Index for reference, research, and planning.
Who is it for
Corporate, non-profit, and non-governmental organizations can consult and refer to the Index for travel and event planning, duty of care, and resource allocation purposes.
Insurers can gain meaningful insights for risk ratings and various forms of coverage relating to travel, Duty of Care, Key Person, and other specialty products.
Risk meaning and definitions ambiguity plagues most international rating efforts and is a key source of errors in popular indices. Criminal codes and law enforcement practices differ across jurisdictions, even in seemingly narrow and specific crime phenomena such as robbery or rape. The effects of ambiguity are further exacerbated by pseudo-quantitative ratings and scales that combine multiple non-additive incident types and classes.
ViCRi reduces this error by (a) narrowing the metrics to ensure accuracy and (b) focusing on the phenomenological (material) characteristics of incidents rather than the legal definitions, matching relevant and corresponding (codified) event data whenever possible.
National or regional crime and incident rates are often dangerously uninformative at the urban or local level, where incident rates are seldom reflective of the mean. Similarly, victimization rates differ significantly between demographic groups. This fallacy is known as the flaw of averages.
Our efforts here center on adopting the most effective geographic, demographic, and temporal granularity level. Assuming that most residents and visitors will visit and engage in activities across multiple neighborhoods over 12 months (the standard analytical time frame for crime data,) annual city-level rates are considered the most effective level of spatio-temporal granularity for an international index.
Official records (recorded/reported crimes) only represent a portion of actual incidents in virtually all cases. Unreported crimes, authorities' failure to capture/retain data, legal and procedural constraints and other factors result in many crimes going unmeasured. The ratio of recorded to actual crimes is a function of several factors, including but not limited to the legal framework, law enforcement quality and resources, cultural characteristics (stigma), type and severity of the crime, victims' demographics, etc.
The index ratings take unreported crimes into account, offering total incident numbers projections based on documented or estimated percentages and ratios of captured vs. uncaptured data.
The common yet controversial practice of combining murder rates with other violent crimes, such as aggravated assault and rape, without consideration for their severity has been debated for years. Given the higher frequency of lower-impact crimes, homicide rates and trend insights can easily be missed in unweighted aggregate statistics, resulting in weak analyses of criminal phenomena.
A proprietary weighting formula designed to mimic and align violent crime victimization with health morbidity and mortality weighting practices, developed considering the range and shape of the respective distributions.
Continues non-linear distributions, such as city-level robbery or murder rates, are a poor fit for ranked/discreet ordinal scales. The range compression resulting from this practice produces ranks and classes that fail to adequately represent the differences between members of some portions of the distribution and distorts the relative risk perception in the information consumer.
A logarithmic conversion to discreet ranking and the adoption of a number of classes derived by real-world interpretation of the data distribution - including an overflow bin.