A inclusão do escore de propensão como uma covariável em modelos de árvores de regressão Bayesianas para inferência causal pode reduzir o viés existente nas estimações de efeitos de tratamento, o qual ocorre devido ao fenômeno de confudimento induzido por regularização. Este estudo defende o uso do escore de propensão por meio de um panorama de seleção de variáveis totalmente Bayesiano, e através do uso de Gráficos de Expectativa Individual Condicional, que se trata de um elemento que pode aprimorar a análise de efeitos de tratamento. Tal ferramental pode ser utilizado como meio de identificar grupos que possuem diferentes respostas ao tratamento aplicado e para analisar o impacto de cada variável no efeito de tratamento estimado. / The inclusion of the propensity score as a covariate in Bayesian regression trees for causal inference can reduce the bias in treatment effect estimations, which occurs due to the regularization-induced confounding phenomenon. This study advocates for the use of the propensity score by evaluating it under a full-Bayesian variable selection setting, and the use of Individual Conditional Expectation Plots as a graphical tool to improve treatment effect analysis. These tools can be used to form groups with different responses to the applied treatment, and to analyze the impact of each variable in the estimated treatment effect.

The purpose of this study was to explore the relationship of dual enrollment course participation by comparing first-time, full-time traditional community college students who participated in dual enrollment (N=246) to peers (N=986) that did not participate. Dual enrollment participation was defined as taking one or more dual enrollment courses. The population for this study (N=1,232) included first-time, full-time students who graduated from public high schools in the service area of Northeast State Community College over a five year span from 2008 through 2012. Propensity score matching eliminated self-selection bias by controlling for confounding covariates such as parental education, high school GPA, and ACT scores. The major findings of the study included the following: dual enrollment participants (a) were nearly four times less likely to take remediation than non-participants, (b) earned approximately 1 extra credit hour in the first semesters of college, (c) earned higher first semester GPAs, (d) were 2.5 times more likely to graduate in 2 years (100% of degree time) and, (e) were 1.68 times more likely to graduate in 3 years (150% of degree time). The study concluded that dual enrollment benefits community college students in Tennessee, both at the beginning and completion of college. This is a significant justification for the current investment in dual enrollment by the State of Tennessee and for further increasing access to dual enrollment for all students, especially for students that live in rural areas, experience poverty, or are underrepresented in higher education.

This this three-article dissertation sought to explore the potential causal link of students’ collegiate residence with three broad categories of student outcomes. Using data from the Wabash National Study of Liberal Arts Education, each article employed propensity score matching in an effort to reduce selection bias associated with a student’s decision to live on campus. The first manuscript examined academic achievement, retention, four-year graduation, and satisfaction with the college experience and found that living on campus had no direct effect on any of these outcomes. The second manuscript explored the effect of living on campus on students’ overall health, alcohol consumption and binge drinking, smoking behaviors, exercise frequency, and psychological well-being. Findings suggest that living on campus has a positive effect on students’ first-year alcohol consumption, frequency of binge drinking, and exercising behaviors. These findings do not persist beyond the first year. Some conditional effects were uncovered, with a significant interaction between race and campus residence on some outcomes. The final study considered the effect living on campus has on student engagement. Living on campus was found to have a direct effect on positive peer interactions, frequency of interactions with student affairs staff, and co-curricular involvement. Like the second study, conditional analyses were conducted and revealed significant interactions mostly among race and campus residence.

Objective: To assess the ability of propensity score methods to maintain covariate balance and minimize bias in the estimation of treatment effect in a time-to-event setting. Data Sources: Generated simulation model Study Design: Simulation study Data Collection: 6 scenarios with varying covariate relationships to treatment and outcome with 2 different censoring prevalences Principal Findings: As time lapses, balance achieved at baseline through propensity score methods between treated and untreated groups trends toward imbalance, particularly in settings with high rates of censoring. Furthermore, there is a high degree of variability in the performance of different propensity score models with respect to effect estimation. Conclusions: Caution should be used when incorporating propensity score analysis methods in survival analyses. In these settings, if model over-parameterization is a concern, Cox regression stratified on propensity score matched pairs often provides more accurate conditional treatment effect estimates than those of unstratified matched or IPT weighted Cox regression models.

During the last decade higher education organizations and educational policy makers have substantially increased efforts to incentivize study abroad participation. These efforts are grounded in the longstanding belief that study abroad participation improves intercultural competence - an educational outcome critical in a globalized 21st century economy. Yet decades of evidence that appear to support this claim are repeatedly limited by a series of methodological weaknesses including small homogenous samples, an absence of longitudinal study design, no accounting for potential selection bias, and the lack of controls for potentially confounding demographic and college experience variables. Thus, a major competing explanation for differences found between students who do and do not study abroad continues to be the possibility that these differences existed prior to participation. The current study sought to determine the effect of study abroad on intercultural competence among 1,593 participants of the 2006 cohort of the Wabash National Study on Liberal Arts Education. The Wabash National Study is a longitudinal study of undergraduates that gathered pre- and post-test measures on numerous educational outcomes, an array of institutional and self-reported pre-college characteristics, and a host of college experiences. The current study employed both propensity score matching and covariate adjustment methods to account for pre-college characteristics, college experiences, the selection effect, and the clustered nature of the data to both cross-validate findings and provide guidance for future research. Under such rigorous analytic conditions, this study found that study abroad generated a statistically significant positive effect on intercultural competence; an effect that appears to be general rather than conditional. Moreover, both covariate adjustment and propensity score matching methods generated similar results. In examining the effect of study abroad across the three constituent subscales of the overall measure of intercultural competence, this study found that study abroad influences students' diversity of contact but has no statistically significant effect on relativistic appreciation of cultural differences or comfort with diversity. Finally, the results of this study suggest that the relationship between study abroad and intercultural competence is one of selection and accentuation, holding important implications for postsecondary policy makers, higher education institutions, and college impact scholars.

Observational studies are an indispensable complement to randomized clinical trials (RCT) for comparison of treatment effectiveness. Often RCTs cannot be carried out due to the costs of the trial, ethical questions and rarity of the outcome. When noncompliance and missing data are prevalent, RCTs become more like observational studies. The main problem is to adjust for the selection bias in the observational study. One increasingly used method is propensity-score matching. Compared to traditional multi-covariate matching methods, matching on the propensity score alleviates the curse of dimensionality. It allows investigators to balance multiple covariate distributions between treatment groups by matching on a single score. This thesis focuses on the large sample properties of the matching estimators of the treatment effect. The first part of this thesis deals with problems of the analytic supports of the logit propensity score and various matching methods. The second part of this thesis focuses on the matching estimators of additive and multiplicative treatment effects. We derive the asymptotic order of the biases and asymptotic distributions of the matching estimators. We also derive the large sample variance estimators for the treatment effect estimators. The methods and theoretical results are applied and checked in a series of simulation studies. The third part of this thesis is devoted to a comparison between propensity-score matching and multiple linear regression using simulation.

<p>In this paper I study how the repeal of the Swedish wealth tax (1 of January 2007) has affected people´s labour supply behaviour. This particular issue is relevant because it may help us understand some of the effects of the earnings tax changes that have taken place in Sweden. Accoring to standard economic theory a repealed wealth tax is similar to an income effect for the persons who previously paid the tax. That means that they theoretically will want to consume more leisure, that is decrease their labour supply. The method I am using to test this hypothesis is a difference-in-difference approach where the treatment group consists of persons who previously paid the tax and the control group of comparable persons who did not pay the tax. The data I am using is taken from a Swedish database called LINDA, compiled by the Swedish Central Agency for Statistics (SCB). My main result in this paper is that the repealed wealth tax does not seem to have had any influnece on the labour supply behavior of the persons who previously paid the tax.</p>

<p>Aktivitetsgarantin är ett arbetsmarknadspolitiskt program som startades i Sverige den första augusti år 2000. I den här uppsatsen skattas aktivitetsgarantins effekt på sannolikheten att erhålla arbete. För att utvärdera den genomsnittliga effekten används en matchningsestimator eftersom programmet inte hade en experimentell utformning. De empiriska resultaten är något varierande, men ett entydigt stöd för att deltagande i aktivitetsgarantin medför ökad sannolikhet till arbete är svårt att finna. Däremot visar det sig att om arbetslösheten fortgår under åtminstone 6 månader efter samplingsperioden hittas positiva effekter av deltagande i aktivitetsgarantin.</p>

Aktivitetsgarantin är ett arbetsmarknadspolitiskt program som startades i Sverige den första augusti år 2000. I den här uppsatsen skattas aktivitetsgarantins effekt på sannolikheten att erhålla arbete. För att utvärdera den genomsnittliga effekten används en matchningsestimator eftersom programmet inte hade en experimentell utformning. De empiriska resultaten är något varierande, men ett entydigt stöd för att deltagande i aktivitetsgarantin medför ökad sannolikhet till arbete är svårt att finna. Däremot visar det sig att om arbetslösheten fortgår under åtminstone 6 månader efter samplingsperioden hittas positiva effekter av deltagande i aktivitetsgarantin.

This thesis examines determinants of the adoption of total mixed ration (TMR), and the effects of the adoption of TMR on the farm level productivity and profitability of Ontario dairy farms. A sample of 320 farm level data from 2004-2008 is taken from the Ontario Dairy Farm Accounting Project (ODFAP). A probit model is estimated to examine the factors affecting the adoption of TMR; and the propensity score matching analysis is used to explore the influence of the use of TMR on sample farm’s productivity and profitability. Results from the probit model show that farmer’s age, herd size, region, breed type and feeding system have significant effect on the adoption of TMR. In turn, the adoption of TMR feeding practice has positive influence on both farm productivity and profitability. Under the propensity score matching method, the use of the TMR feeding practice has an economically significant effect on farm profits (i.e., for average farm with approximately 73 cows, the use of TMR feeding practice increases farm profits by CAD$37,091.30/year approximately) and a statistically significant increase in milk production by 1075.41 hl/cow per year.