Crime statistics refer to systematic, quantitative results about crime, as opposed to crime news or anecdotes. Notably, crime statistics can be the result of two rather different processes: scientific research, such as criminological studies, victimisation surveys; official figures, such as published by the police, prosecution, courts, and prisons. However, in their research, criminologists often draw on official figures as well. There are several methods for the measuring of crime. Public surveys are occasionally conducted to estimate the amount of crime that has not been reported to police. Such surveys are usually more reliable for assessing trends. However, they also have their limitations and generally don't procure statistics useful for local crime prevention, often ignore offenses against children and do not count offenders brought before the criminal justice system. Law enforcement agencies in some countries offer compilations of statistics for various types of crime. Two major methods for collecting crime data are law enforcement reports, which only reflect crimes that are reported, recorded, and not subsequently canceled; and victim study (victimization statistical surveys), which rely on individual memory and honesty. For less frequent crimes such as intentional homicide and armed robbery, reported incidences are generally more reliable, but suffer from under-recording; for example, no criming in the United Kingdom sees over one third of reported violent crimes being not recorded by the police. Because laws and practices vary between jurisdictions, comparing crime statistics between and even within countries can be difficult: typically only violent deaths (homicide or manslaughter) can reliably be compared, due to consistent and high reporting and relative clear definition. The U.S. has two major data collection programs, the Uniform Crime Reports from the FBI and the National Crime Victimization Survey from the Bureau of Justice Statistics. However, the U.S.