Summary: | 碩士 === 國立臺北大學 === 統計學系 === 107 === Illicit drugs use has always been the public eye’s issue of focus. Public security problems associated with drug use, such as homicide, robbery, larceny, sexual crime etc., have significant impact on society. Finding possible factors related to illicit drug use will benefit government policy review on planning and prevention of drug abuse, effectively reducing illicit drug use.
The purpose of this study is to investigate the possible factors related to illicit drugs use by employing logistic regression and Bayesian logistic regression. We focus on a program of NHANES in the U.S. from the years 2013 to 2016. Through the survey we learned that individuals who ever used illicit hard drugs only takes 16% of whole sample. In order to increase the accuracy, we made each case matched with 1 control where control is defined as not ever used illicit hard drugs. Moreover, 500 repeated samples were selected in each imputation sample, and then the models validation is evaluated through the cross-validation.; where 90% of the samples are selected as a training dataset while the rest is treated as a validation dataset.
Two models in our study got consistent conclusions, finding that gender, age, race, BMI, annual family income, education level, occupation, mental health, alcohol use, Tabaco use and marijuana use have significant relationship with illicit drugs use; particularly marijuana use strongly associates with illicit hard drugs use. The U.S. data used in this study may need further confirmation since the sampled cities’ information was unavailable. For instance, some U.S states where marijuana use might be legal thus further confirmation is needed on issues of marijuana use, and authorities concerned could refer to other factors for drugs prevention.
|