Colleges and universities are under mounting pressure to meet enrollment goals in the face of declining college attendance. Insight into student-level probability of enrollment, as well as the identification of features relevant in student enrollment decisions, would assist in the allocation of marketing and recruitment resources and the development of future yield programs. A logistic regression model was fit to predict which applicants will ultimately matriculate (enroll) at Concordia College. Demographic, geodemographic and behavioral features were used to build a logistic regression model to assign probability of enrollment to each applicant. Behaviors indicating interest (campus visits, submitting a deposit) and residing in a zip code with high alumni density were found to be strong predictors of matriculation. The model was fit to minimize false negative rate, which was limited to 18.1 percent, compared to 50-60 percent reported by comparable studies. Overall, the model was 80.13 percent accurate.

Binary response data is often modeled using the logistic regression model. Experimental design theory for the logistic model appears to be increasingly important as experimentation becomes more complex and expensive. The optimal design work is extremely valuable in areas such as biomedical and environmental applications. Most design research dealing with the logistic model has been concentrated on the one-variable case. Relative little has been done for the two-variable model. The primary goal of this research is to develop and study efficient and practical experimental design procedures for fitting the logistic model with two independent variables. Optimal designs are developed addressing D optimality, Q optimality, and the estimation of interaction between the design variables. The two-variable models with and without interaction usually have to be handled separately. The equivalence theory concerning D optimal designs is studied. The designs are compared using their relative efficiencies in the presence of interaction. Robustness to parameter misspecification is investigated. Bayesian design procedures are explored to provide relatively more robust experimental plans. / Ph. D.

Predictive models have established a niche in the field of archaeology. Valued as tools in predicting potential archaeological sites, their use has increased with development of faster and more affordable computer technology. Predictive models highlight areas within a landscape where archaeological sites have a high probability of occurrence. Therefore, time and resources normally expended on archaeological exploration can then be more efficiently allocated to specified locations within a study area. In addition to the resulting predictive surface, these models also identify significant variables for site selection by prehistoric or historic groups. Relationships with the environment, whether natural or social, are extremely pertinent to strengthening the resource base. In turn, this information can be utilized to better interpret and protect valuable cultural resources. A predictive probability model was generated to locate Union Civil War fortifications around Petersburg, Virginia. This study illustrated the ease with which such analysis can be accomplished with the integrated use of a Geographic Information System with statistical analysis. Stepwise logistic regression proved effective in selecting significant independent variables to predict probabilities of fortifications within the study area, but faired poorly when applied to areas withheld from the initial building stage of the model. Variation of battle tactics between these two separate areas proved great enough to have a detrimental effect the model's effectiveness. / Master of Science

No description available.

No description available.

Faculty members are influenced by a complex set of factors when making decisions about when to retire. These factors generally include both financial and non-financial characteristics. This study was designed to examine the non-financial factors related to the retirement process for selected faculty groups. Key components of the design included selecting faculty groups for analysis and identifying the non-financial factors related to the retirement process. Two faculty groups were selected: (a) faculty who had previously retired from another position and (b) faculty members with no plans to retire in the next three years. The non-financial factors were identified through a review of the literature and included (a) employment characteristics, (b) demographic characteristics, (c) activity measures, and (d) satisfaction items. The study was based on secondary analysis of NSOPF: 99 data. A combination of descriptive statistics and logistic regression was used to analyze the data. Major findings include (a) previously retired faculty members may be a substantial pool of qualified, productive talent intrinsically motivated to be part of an academic environment on a part-time basis because their financial status is not solely dependent on basic salary from the institution; (b) additional indicators distinguishing age at retirement from a career position versus age at retirement from all paid employment may also be needed to fully describe the issue; (c) employment status, years in current position, program area, age, gender, geographic region, average class size, and satisfaction with other aspects of the job (excluding instructional duties) were distinguishing characteristics of previously retired faculty members; (d) a sizeable portion of older faculty has not yet reached traditional retirement age; (e) the impact of uncapping mandatory retirement ages for tenured faculty may not have yet been fully realized — even eight years after the legislation took effect; (f) evidence does not support some of the objections from the higher education community in opposition to uncapping; and (g) control of institution, program area, years in current position, age, marital status, number of dependents, recent publications, career publications, and satisfaction were distinguishing characteristics of faculty members with no plans to retire in the next three years. / Ph. D.

For this study, I have constructed a logistic regression model, using existing datasets of environmental parameters to predict the probability of As concentrations above 5 parts per billion (ppb) in Virginia groundwater and to evaluate if geologic or other characteristics are linked to elevated As concentrations. Measured As concentrations in groundwater from the Virginia Tech Biological Systems Engineering (BSE) Household Water Quality dataset were used as the dependent variable to train (calibrate) the model. Geologic units, lithology, soil series and texture, land use, and physiographic province were used as regressors in the model. Initial models included all regressors, but during model refinement, attention was focused solely on geologic units. Two geologic units, Triassic-aged sedimentary rocks and Devonian-aged shales/sandstones, were identified as significant in the model; the presence of these units at a spatial location results in a higher probability for As occurrences in groundwater. Measured As concentrations in groundwater from an independent dataset collected by the Virginia Department of Health were used to test (validate) the model. Due to the structure of the As datasets, which included As concentrations mostly (95-99%) = 5 ppb, and thus few (1-5%) data in the range > 5 ppb, the regression model cannot be used reliably to predict As concentrations in other parts of the state. However, our results are useful for identifying areas of Virginia, defined by underlying geology, that are more likely to have elevated As concentrations in groundwater. Results of this work suggest that homeowners with wells installed in these geologic units have their wells tested for As and regulators closely monitor public supply wells in these areas for As. / Master of Science

In today's world the challenge of logistics becomes more and more important for all companies, which consequently contributes to the need of an integrated system of in-house logistics. Goods must flow easily between stations and departments in order to achieve the best utilization of transportation as well as maintaining a good distribution structure to handle these processes in an efficient and effective way to enhance business performance and competitiveness. This report aims to provide research on how to strategically organize, plan and structure the flow of internal logistics within an organization. The research in this study originates from 3 main methods, a theoretical review, a case study and a benchmark. The main idea of the thesis was to contribute with scientific knowledge of improvements that can be achieved within internal logistics targeting companies operating within the service sector. In order to understand the scope of the problem the current state at the case Tropical Islands was analysed and served to identify areas where possible improvements could be achieved. Further on, in combination with the theory and empirical findings the main conclusions of the thesis was the need to strategically organize the flow of internal logistics to improve efficiency and enhance competitiveness. It was made evident that planning is essential in order to successfully implement the milk runs. Other tools could be utilized to further eliminate drawbacks that might occur.

Multicollinearity can be present in the propensity score model when estimating average treatment effects (ATEs). In this thesis, logistic ridge regression (LRR) and principal components logistic regression (PCLR) are evaluated as an alternative to ML estimation of the propensity score model. ATE estimators based on weighting (IPW), matching and stratification are assessed in a Monte Carlo simulation study to evaluate LRR and PCLR. Further, an empirical example of using LRR and PCLR on real data under multicollinearity is provided. Results from the simulation study reveal that under multicollinearity and in small samples, the use of LRR reduces bias in the matching estimator, compared to ML. In large samples PCLR yields lowest bias, and typically was found to have the lowest MSE in all estimators. PCLR matched ML in bias under IPW estimation and in some cases had lower bias. The stratification estimator was heavily biased compared to matching and IPW but both bias and MSE improved as PCLR was applied, and for some cases under LRR. The specification with PCLR in the empirical example was usually most sensitive as a strongly correlated covariate was included in the propensity score model.

The distributional properties of improvement-over-chance, I, effect sizes derived from linear and quadratic predictive discriminant analysis (PDA) and from logistic regression analysis (LRA) for the two-group univariate classification were examined. Data were generated under varying levels of four data conditions: population separation, variance pattern, sample size, and prior probabilities. None of the indices provided acceptable estimates of effect for all the conditions examined. There were only a small number of conditions under which both accuracy and precision were acceptable. The results indicate that the decision of which method to choose is primarily determined by variance pattern and prior probabilities. Under variance homogeneity, any of the methods may be recommended. However, LRA is recommended when priors are equal or extreme and linear PDA is recommended when priors are moderate. Under variance heterogeneity, selecting a recommended method is more complex. In many cases, more than one method could be used appropriately.