Dissertations / Theses on the topic 'Statistics|Public policy'

Although women obtain degrees in the fields of Science, Technology, Engineering, and Mathematics (STEM) at lower rates than their male counterparts, this difference does not alone account for the similarly large disparity of men and women working in STEM jobs. Using data from the American Community Survey for 2015, I run a survey of models, including linear and logistic regressions as well as propensity matching, to investigate the extent to which women are underrepresented, even after accounting for education. The results show that for women, the benefit of a STEM degree on STEM job placement rates is significantly lower than it is for their male counterparts; in some estimates, the effect is halved. The models diverge somewhat on the extent to which this is correlated with a lower baseline of women without STEM degrees working STEM jobs (compared to similar men), but all provide grounds to reject the notion that workforce disparity is attributable solely or even primarily to education.

Opioid Substitution Therapy (OST) attempts to curb opioid addiction by substituting a treatment opioid (i.e. methadone, buprenorphine, naltrexone, etc.) for an addict’s primary drug of abuse (i.e. heroin, oxycodone, etc.). However, insofar as patients continue abusing their preferred drug during treatment, OST programs may be increasing the absolute number of drugs patients are abusing. To the extent that some OST patients “divert” their treatment drugs into illicit markets, OST programs may also be increasing the absolute number of opioids abused by the surrounding population, as well. If corroborated, these trends would indicate a connection between OST treatments and the phenomenon most correlated with drug overdoses— “polydrug use”, or the concurrent use multiple drugs by one person.

To ascertain whether and to what extent OST treatment provisioning may be affecting regional polydrug use, this study models the annual number polydrug treatment episodes reported in a state as a function of that state’s OST patient population. The study relies primarily on two administrative datasets—the National Survey of Substance Abuse Treatment Services (N-SSATS) and the Treatment Episode Data Set-Admissions (TEDS-A)—collected between 2002 and 2006.

Using a two-way fixed-effects model that controlled for both state and annual fixed-effects, as well as for state populations and population densities, this study discovered a statistically significant positive correlation (p<.002) between a state’s OST patient population and the number of polydrug treatment episodes reported in the state. The model predicts that a doubling of a state’s OST patient population will be correlated with a 6.16-percent increase in polydrug episodes. These results suggest that OST treatment may be producing a dangerous side effect. At the very least, they suggest that, when considering potential expansions to OST programs, circumspect policymakers should also consider simultaneous expansions to services that address the predicted increase in polydrug use.

The dissertation analyzes multilevel models to predict mortgage origination and the allocation of subprime credit pre-and-post Great Recession. With representative samples from two full years of mortgage applications filed in the top 100 U.S. metropolitan areas, the dissertation uncovers evidence of persistent disparities by race and neighborhood minority concentration despite controls for socioeconomic, demographic, assimilation and housing variables. Mortgage outcomes varied by applicant race, neighborhood racial composition and neighborhood racial change. Findings suggest evidence of Fair Housing Act violations and disparate impacts towards minority homebuyers and minority neighborhoods. Results lend support for spatial assimilation theories in explaining much of the gap white-Asian gap and non-Hispanic white-Hispanic gap. Notable gaps remain between blacks and whites that are better explained by the place stratification model. In both 2004 and 2010 black, Asian and Hispanic prospective homebuyers were less likely to originate their mortgage relative to whites. In the earlier 2004 mortgage sample, black and Hispanic borrowers were much more likely to take out subprime loans relative to whites and Asians less likely. In 2010, the results held that black homebuyers were more likely than whites to take on subprime mortgages and Asian homebuyers less likely. The applicant coefficient for Hispanic was not a significant predictor of subprime in 2010. Asian homebuyers were indeed at a disadvantage relative to white homebuyers, both before and after the Great Recession and the disparity did not change significantly over time. Women were more likely than men to be denied mortgages and also to obtain subprime mortgages. Black neighborhoods were disadvantaged in 2004 and 2010, and increasing numbers of black neighbors reduced the likelihood of mortgage origination in 2004, but not in 2010. In 2004 and 2010, the presence of Asian and Hispanic neighbors increased the likelihood of mortgage origination but neighborhood racial ethnic change was less well received—with negative impacts for increasing numbers of Asian neighbors in both years and reduced mortgage odds for increasingly Hispanic neighborhoods in 2010. Neighborhoods with concentrated single mother households experienced disinvestment including denied mortgages and higher likelihood of subprime status when lenders infuse mortgage capital, though the effect size was very small. The research overcomes limitations of previous studies by employing advanced methods which account for neighborhood effects and by incorporating dynamic measures such as house-price-to-income ratios and regional credit score averages. The primary novel contribution of the dissertation is that this is the first study to examine the effect of neighborhood racial ethnic change on mortgage outcomes and the allocation of subprime mortgage credit. Explicit Fair Housing Act violations spur recommendations for reducing unequal disbursement of mortgage credit by race and gender, most importantly, by urging American schools to provide compulsory primary school coursework in financial literacy.