National Crime Surveys Longitudinal File, 1988-1989: (Selected Variables)

This longitudinal file for the National Crime Surveys (NCS) contains selected variables related to whether a crime was reported to the police for households that responded to the NCS on three consecutive interviews between July 1988 and December 1989 and had experienced at least one criminal victimi...

Descripción completa

Guardado en:  
Detalles Bibliográficos
Autor principal: Lohr, Sharon L. (Autor)
Otros Autores: Conaway, Mark R. (Contribuidor)
Tipo de documento: Electrónico Research Data
Lenguaje:Inglés
Publicado: [Erscheinungsort nicht ermittelbar] [Verlag nicht ermittelbar] 1993
En:Año: 1993
Acceso en línea: Volltext (kostenfrei)
Verificar disponibilidad: HBZ Gateway
Palabras clave:
Descripción
Sumario:This longitudinal file for the National Crime Surveys (NCS) contains selected variables related to whether a crime was reported to the police for households that responded to the NCS on three consecutive interviews between July 1988 and December 1989 and had experienced at least one criminal victimization during that time period. Variable names, for the most part, are identical to those used in the hierarchical files currently available for the National Crime Surveys (see NATIONAL CRIME SURVEYS: NATIONAL SAMPLE, 1986-1991 [NEAR-TERM DATA] [ICPSR 8864]). Three new variables were created, and one existing variable was altered. The TIME variable describes whether the interview was the first, second, or third for the household in the period between July 1988 and December 1989. V4410 was recoded to give the most important reason the crime was not reported to the police for all households that responded to questions V4390-V4410. RELNOFF was created from variables V4209-V4267 to reflect the closest relation any offender had to the victim, and INJURE was created from variables V4100-V4107 to indicate minor injury, serious injury, or none at all. The file is sorted by households.
DOI:10.3886/ICPSR06063.v1