The figures on Suç Endeksi are compiled from incidents reported in the press and expressed relative to population. This page explains how to read the rates, the index and the classes, and which questions the data answers and which it does not.
What the data measures
Two indicators are published as rates and indices. Homicide counts people killed intentionally. Armed assault counts attacks with a firearm or a knife or blade that resulted in death or injury. Theft, snatching, fraud and harassment are not part of this layer: for those types, incidents reported in the press are a small share of official records and are unevenly distributed across provinces, so they are not presented as rates.
Most of the incidents that shape how safe a place feels belong to that second group. The rates here are therefore not a measure of general safety, but a measure of violence that ends in death or involves a weapon.
Rates rather than counts
Absolute counts grow with population. The fact that İstanbul records the most homicides in a year does not make it the riskiest province; it makes it the most populous one. Comparison is therefore made per 100,000 people.
The index and the classes
The index divides a place's rate by the national rate and multiplies by 100. The national average is 100, so a value of 140 means the rate per 100,000 people in that province is 40 per cent above the national average. The limits of the five classes on the maps are fixed and remain comparable across years.
Small numbers mislead
A province or district with a small population records a handful of incidents a year. One incident more or fewer can halve or double the rate; that fluctuation is chance and carries no information about the level of crime there. The index, rank and class are therefore put through a small-number adjustment, so that the fewer the incidents, the more the result is drawn towards the national average. Even after the adjustment, the position of a small place in the ranking is not settled by a single year of data.
In practice this means that a district at the top or the bottom of the ranking may fall to the middle the following year. As years accumulate, the ranking becomes stable.
Documented figures and true figures
Incidents that are not reported in the press cannot be counted. The published figure is therefore the documented one and falls below the official figure. The share the press captures varies between provinces and is higher where populations are large. At national level the undocumented part is modelled, and the estimated total is published alongside the documented count. No estimate is published at province level, because annual provincial counts are too small for one.
Differences in coverage feed directly into the ranking of provinces. Where the local press is strong, incidents are more likely to be reported and the documented rate rises. Small differences between two provinces should therefore not be read as differences in the level of crime.
What the data answers
The data shows how many people were killed in a place in a year, where that figure stands relative to the national average once population is taken into account, how incidents are distributed across the months, the part of the offender–victim relationship that is known, and how much of fatal violence takes place between partners or within the family.
The data does not show whether a place is safe. That question is not answered by a single indicator or a single year, and it cannot be answered at all while official figures for common offences such as theft and harassment are not published at province level. The situation of a particular neighbourhood or street is not measured by this data at all.
Comparing two places
Comparison is made with rates, not counts, and between places of similar size; the population group and region figures on the province pages are given for that purpose. The trend over several years is taken as the basis rather than the difference in a single year. Small differences are not treated as meaningful, because they can be explained by differences in coverage.
The details of the calculations and the formulas are on the methodology page.