TY - JOUR AU - McCaffrey,Daniel F. AU - Pacula,Rosalie Liccardo AU - Han,Bing AU - Ellickson,Phyllis TI - Marijuana Use and High School Dropout: The Influence of Unobservables JF - National Bureau of Economic Research Working Paper Series VL - No. 14102 PY - 2008 Y2 - June 2008 UR - http://www.nber.org/papers/w14102 L1 - http://www.nber.org/papers/w14102.pdf N1 - Author contact info: Daniel F.. McCaffrey RAND Corporation 4570 Fifth Ave, Ste 600 Pittsburgh, PA 15213 E-Mail: danielm@rand.org Rosalie Liccardo Pacula RAND Corporation 1776 Main Street P.O. Box 2138 Santa Monica, CA 90407-2138 Tel: 310/393-0411 ext. 6494 Fax: 858/350-8798 E-Mail: pacula@rand.org Bing Han RAND Corporation 1776 Main St., P.O. Box 2138 Santa Monica, CA 90407-2138 E-Mail: Bing_Han@rand.org Phyllis Ellickson RAND Corporation 1776 Main St., P.O. Box 2138 Santa Monica, CA 90407-2138 E-Mail: Phyllis_Ellickson@rand.org AB - In this study we reconsider the relationship between heavy and persistent marijuana use and high school dropout status using a unique prospective panel study of over 4500 7th grade students from South Dakota who are followed up through high school. Propensity score weighting is used to adjust for baseline differences that are found to exist before marijuana initiation occurs (7th grade). Weighted logistic regression incorporating these propensity score weights is then used to examine the extent to which time-varying factors, including substance use, also influence the likelihood of dropping out of school. We find a positive association between marijuana use and dropping out (OR=5.68), over half of which can be explained by prior differences in observational characteristics and behaviors. The remaining association (OR=2.31) is made statistically insignificant when measures of cigarette smoking are included in the analysis. Because no physiological justification can be provided for why cigarette smoking would reduce the cognitive effects of marijuana on schooling, we interpret this as evidence that the association is due to other factors. We then use the rich data to explore which constructs are driving this result, determining that it is time-varying parental and peer influences. ER -