Browsing by Subject "bias"
Now showing 1 - 4 of 4
Results Per Page
Sort Options
Item Bias Reduction and Goodness-of-Fit Tests in Conditional Logistic Regression Models(2011-10-21) Sun, XiuzhenThis dissertation consists of three projects in matched case-control studies. In the first project, we employ a general bias preventive approach developed by Firth (1993) to handle the bias of an estimator of the log-odds ratio parameter in conditional logistic regression by solving a modified score equation. The resultant estimator not only reduces bias but also can prevent producing infinite value. Furthermore, we propose a method to calculate the standard error of the resultant estimator. A closed form expression for the estimator of the log-odds ratio parameter is derived in the case of a dichotomous exposure variable. Finite sample properties of the estimator are investigated via a simulation study. Finally, we apply the method to analyze a matched case-control data from a low-birth-weight study. In the second project of this dissertation, we propose a score typed test for checking adequacy of a functional form of a covariate of interest in matched case-control studies by using penalized regression splines to approximate an unknown function. The asymptotic distribution of the test statistics under the null model is a linear combination of several chi-square random variables. We also derive the asymptotic distribution of the test statistic when the alternative model holds. Through a simulation study we assess and compare the finite sample properties of the proposed test with that of Arbogast and Lin (2004). To illustrate the usefulness of the method, we apply the proposed test to a matched case-control data constructed from the breast cancer data of the SEER study. Usually a logistic model is needed to associate the risk of the disease with the covariates of interests. However, this logistic model may not be appropriate in some instances. In the last project , we adopt idea to matched case-control studies and derive an information matrix based test for testing overall model adequacy and investigate the properties against the cumulative residual based test in Arbogast and Lin (2004) via a simulation study. The proposed method is less time consuming and has comparative power for small parameters. It is suitable to explore the overall model fitting.Item Net pay evaluation: a comparison of methods to estimate net pay and net-to-gross ratio using surrogate variables(2009-06-02) Bouffin, NicolasNet pay (NP) and net-to-gross ratio (NGR) are often crucial quantities to characterize a reservoir and assess the amount of hydrocarbons in place. Numerous methods in the industry have been developed to evaluate NP and NGR, depending on the intended purposes. These methods usually involve the use of cut-off values of one or more surrogate variables to discriminate non-reservoir from reservoir rocks. This study investigates statistical issues related to the selection of such cut-off values by considering the specific case of using porosity () as the surrogate. Four methods are applied to permeability-porosity datasets to estimate porosity cut-off values. All the methods assume that a permeability cut-off value has been previously determined and each method is based on minimizing the prediction error when particular assumptions are satisfied. The results show that delineating NP and evaluating NGR require different porosity cut-off values. In the case where porosity and the logarithm of permeability are joint normally distributed, NP delineation requires the use of the Y-on-X regression line to estimate the optimal porosity cut-off while the reduced major axis (RMA) line provides the optimal porosity cut-off value to evaluate NGR. Alternatives to RMA and regression lines are also investigated, such as discriminant analysis and a data-oriented method using a probabilistic analysis of the porosity-permeability crossplots. Joint normal datasets are generated to test the ability of the methods to predict accurately the optimal porosity cut-off value for sampled sub datasets. These different methods have been compared to one another on the basis of the bias, standard error and robustness of the estimates. A set of field data has been used from the Travis Peak formation to test the performance of the methods. The conclusions of the study have been confirmed when applied to field data: as long as the initial assumptions concerning the distribution of data are verified, it is recommended to use the Y-on-X regression line to delineate NP while either the RMA line or discriminant analysis should be used for evaluating NGR. In the case where the assumptions on data distribution are not verified, the quadrant method should be used.Item Underrepresentation of Hispanic/Latino Students Identified with Emotional Disturbance in IDEIA: What's the Teacher's Role?(2011-10-21) Massa, IdaliaHistorically, Hispanic/Latino (H/L) students have been under-referred, under-identified, and under-served by the U.S. Special Education (SPED) system, particularly under the emotional behavioral disturbance (EBD) category. This finding is alarming given that numerous federal sources report that H/L students continue a disturbing trend of struggling academically as well as being at a higher risk for poor mental health outcomes such as elevated levels of depression, anxiety, and suicidality when compared to their peers. Unfortunately, the existing mental health and education literature on H/L students provides limited guidance in understanding the disproportionate underrepresentation of H/L in the EBD category of the SPED system; an underrepresentation well-documented in the report to congress on the implementation of the Individuals with Disabilities Education Improvement Act (IDEIA). Using survey methods, the purpose of this study was to shed light on the possible mediating role teachers' perceptions have on the SPED referral and identification decisions by looking at teacher ratings of risk for EBD-like behaviors of students across behavioral conditions (i.e., internalizing versus externalizing types of behaviors) and across ethnic/racial groups (i.e., White, African Americans, and H/L students) using a response-to-intervention framework. Using the Qualtrics software, an online survey tool, 114 self-selected pre-service teachers were surveyed; data was collected and analyzed using a One-way Analysis of Variance. Two main effects and two interaction effects were explored: does the students' ethnic/racial background moderate the teachers' at risk score (ARS) regardless of the behavior displayed?; does the type of behavioral expression moderate the ARS regardless of ethnic/race?; is there an interaction effect between H/L students exhibiting internalizing behaviors that systematically results in a lower ARS and AA students exhibiting externalizing behaviors that systematically results in a higher ARS? Results indicated that (a) when compared to White, Hispanic/Latino students are indeed less likely to be perceived by the pre-service teachers as exhibiting EBD-like behaviors regardless of the behavior (externalizing, internalizing, or neutral) displayed, (b) externalizing behaviors was the strongest predictor for perceiving someone as at-risk for having EBD-like behaviors, and (c) no interaction effects were found.Item Universal Screening for Behavior: Considerations in the Use of Behavior Rating Scales(2012-08-20) Mason, Benjamin 1972-Universal screening for behavior is the use of a measure of social, emotional or behavioral function across an entire population with a goal of preventing future difficulties by intervening with students identified by the screening protocol. Multiple screening procedures have been used, with most including behavior rating scales in the selection process. The purpose of the present research was to investigate two central questions related to the use of universal screeners for behavior in school settings: first, can scores on universal screeners be used as an outcome measure investigating program based interventions, and second, what evidence of teacher bias exists when an external criterion of behavior is included. The purpose of study one was to determine if differences in teacher-rated behavior could be detected between a sample of students that attended public preschool and a nonattending peer group matched for ethnicity, gender, and a gross measure of socioeconomic status (total n= 138). Results of Study One indicated no significant differences between preschool-attending and nonattending groups (p=.61) or between Hispanic and Caucasian participants. Limitations related to sampling and measurement were discussed. In study two, a best-evidence synthesis of peer-reviewed articles investigating teacher bias in behavior ratings of students was conducted. Strict inclusion criteria were chosen to allow for inferential judgment of teacher accuracy. Results of Study Two found a final total of 25 studies of teacher bias that suggested mixed evidence for bias due to student ethnicity or gender and stronger evidence for bias due to expectancies (disability label), teacher culture, unrelated behaviors (halo effects), and teacher training and experience. Limitations, implications for practice and directions of future research were discussed.