ESTIMATING A LOGIT MODEL WITH RANDOMIZED DATA: THE CASE OF COCAINE USE
Version of Record online: 26 FEB 2008
Australian Journal of Statistics
Volume 36, Issue 1, pages 9–20, March 1994
How to Cite
KERKVLIET, J. (1994), ESTIMATING A LOGIT MODEL WITH RANDOMIZED DATA: THE CASE OF COCAINE USE. Australian Journal of Statistics, 36: 9–20. doi: 10.1111/j.1467-842X.1994.tb00634.x
- Issue online: 26 FEB 2008
- Version of Record online: 26 FEB 2008
- Received July 1990; revised December 1993; accepted January 1994.
- Randomized response survey;
- illegal drug use;
In his influential book, Maddala (1983) suggests combining randomized response survey data with other personal information to estimate logit models predicting immoral, unpopular, or unlawful behaviour. This study is one of the first to implement this technique using real data. Models of college students' recent cocaine use are estimated with academic performance and socio-economic characteristics as determinants. Parameter estimates obtained from randomized response surveys are compared to those obtained using conventional, direct question surveys. The results indicate that randomized response estimates provide useful information on the degree to which inferences regarding the determinants of cocaine use are sensitive to survey type.