Fall 2017 Colloquium: Graduate Portfolio in Applied Statistical Modeling






Pushkar Kumar Jain Dec. 12, 2017 1:00 to 1:30 pm GDC 7.402 "Dynamically Adaptive Data-Driven Simulation of Extreme Hydrological Flows"
Sunyoung Park Dec. 12, 2017 1:30 to 2:00 pm GDC 7.402 "Using Total Sample Size Weights in Meta-analysis of Log-Odds Ratios when Missing Cell Frequencies"
Rachel Donnelly Dec. 13, 2017 1:00 to 1:30 pm GDC 7.402 "Contextualizing Diseases of Despair: How Job Insecurity Shapes Health Behaviors"
Joy Wyckoff Dec. 13, 2017 1:30 to 2:00 pm GDC 7.402 "Gossip as an Intrasexual Competition Strategy: Predicting Negative Information Sharing by Competitor-potential Mate Goal Misalignment"
Shreya Gupta Dec. 12, 2017 2:00 to 2:30 pm GDC 7.402 "Ranking Routes in Semi-conductor Wafer Fabs"

Pushkar Kumar Jain

(PhD student in Engineering Mechanics, supervised by Dr. Clint Dawson)

Title: "Dynamically Adaptive Data-Driven Simulation of Extreme Hydrological Flows"

Abstract: Hydrological hazards such as storm surges and tsunamis are physically complex events that are very costly in loss of human life and economic productivity. Such disasters could be mitigated through improved emergency evacuation in real-time and through the development of resilient infrastructure based on knowledge of how systems respond to extreme events. Data-driven computational modeling is a critical technology underpinning these efforts. Our investigation will focus on the novel combination of methodologies in forward simulation and data assimilation. The forward geophysical model will be based on adaptive mesh refinement (AMR), a process by which a computational mesh can adapt in time and space to the current state of a simulation. The forward solution will be combined with ensemble based data assimilation methods, whereby observations from an event can be assimilated to improve the veracity of the solution. The novelty in our approach is the tight two-way coupling of AMR and ensemble filtering techniques. The technology will be tested with twin experiments and using actual data from the event of the Chile tsunami of February 27, 2010.

 Shreya Gupta

(Phd student in Operations Research & Industrial Engineering, supervised by Dr. John Hasenbein)

Title: "Ranking Routes in Semi-conductor Wafer Fabs"

Abstract:  The goal of this project is to analyze the quality of processing routes in the wafer fabrication process. In particular, we focus on two steps in a recipe (say, deposition and annealing) and define a route by the specific tools used in each of the two steps. The purpose of categorizing routes is two-fold. First, identification of the ``best'' and ``worst'' routes is needed for recipe probing. Specifically, when adjustments are made to recipes, it is useful to identify the best and worst routes in the existing process as these are likely to provide good bounds on performance for the adjusted recipe. Second, route categorization is useful for scheduling. For example, once a wafer has completed the first step in a route, the analysis can identify the best tool to be used at the next stage. 

We develop methods for ranking routes based on three different types of metrics. First, we consider score-based metrics, in which a higher  value of the metric always indicates a "better" product. The specific  metric considered for this project is yield, although the methodology applies to any score-based metric. Next, we develop ranking methods for  target-based metrics. In this case, the analysis is more complex, because there is no canonical way to define "better" (for example, a better process should be both accurate and precise, but the balance between these measures could be dependent on the decision maker). An example of a target-based metric is thickness where the objective is to produce wafers with the thickness as close as possible to a target set by engineering. Finally, we consider count-based metrics, in which the metric takes values that are non-negative integers and lower values are better. In particular, 0 is the best possible value of the metric. In this case, we examine defect counts and provide ranking methods. We also compare different ranking procedures and provide some recommendations.

Sunyoung Park

(Phd student in Educational Psychology, supervised by Dr. Tasha Beretvas)

Title: "Using Total Sample Size Weights in Meta-analysis of Log-Odds Ratios when Missing Cell Frequencies"

Abstract: The log-odds ratio (ln[OR]) is commonly used to quantify a treatment’s effect on a dichotomous outcome and then pooled across studies using inverse-variance (1/v) weights. Calculation of the ln[OR]’s variance requires four cell frequencies for two groups crossed with two outcomes. While primary studies report the total sample size () many do not report all four frequencies. Using real data, we demonstrate pooling of ln[OR]s using versus 1/v weights. In a simulation study we compared the two weighting approaches under several conditions. Efficiency and Type I error rates for versus weights used to pool ln[OR] estimates depended on sample size and the percent of studies missing cell frequencies. Results are discussed and guidelines for applied meta-analysts are provided.

Rachel Donnelly

(Phd student in Sociology, supervised by Dr. Mark Hayward)

Title: "Contextualizing Diseases of Despair: How Job Insecurity Shapes Health Behaviors"

Abstract: Mortality rates are stalling for some groups in the United States, and this trend is being attributed to an increase in “deaths of despair,” or deaths due to suicide, drug overdose, and liver disease. Recent narratives suggest that Americans are facing stress and hopelessness due to bleak economic prospects, which leads them to engage in destructive behaviors that drive these deaths of despair. One such source of stress that is becoming increasingly common in the United States is a sense of uncertainty about job security. I analyze 12 waves of nationally representative data from the Health and Retirement Study to examine how job insecurity shapes patterns of health behaviors and depressive symptomology among workers. Moreover, I consider heterogeneity in this association by birth cohort and gender. Preliminary results suggest that experiences of job insecurity contribute to increases in unfavorable health behaviors over time. Understanding the health consequences of job insecurity is particularly important in light of the rise in deaths and diseases of despair, wherein the most commonly hypothesized explanation is work-related stress. Moreover, documenting the heterogeneity of harmful health behaviors in relation to job insecurity is essential for creating and implementing effective solutions.

Joy Wyckoff

(Phd student in Psychology, supervised by Dr. Greg Hixon)

Title:"Gossip as an Intrasexual Competition Strategy: Predicting Negative Information Sharing by Competitor-potential Mate Goal Misalignment"

Abstract: We propose that negative information sharing (i.e. gossip) evolved as an intrasexual competition tactic and posit that individuals’ decisions to share negative information about a competitor is sensitive to the probability of that information diminishing a competitors mate value. According to Sexual Strategies Theory (Buss & Schmitt, 1993), men and women have evolved different psychological mechanisms that underlie short-term and long-term mating strategies and display different preferences for long-term and short-term mate choice. In two experiments, we utilized a 2x2 factorial within-subject design to manipulate the potential mates mating strategy (long-term vs. short-term) and characteristics of a rival that indicated a rival’s mating strategy (long-term vs. short-term). We predicted that participants would be more likely to share the same information about a competitor when that information mismatched the potential mates preferences. Using Linear Mixed Effects Modeling, Study 1 found that men and women rated that they were more likely to share that a competitor is promiscuous when the potential mate was interested in a long-term mate than when the potential mate was interested in a short-term mate. In other words, people rated being more likely to share information about a rival that did not match the potential mates mating strategy. Study 2 demonstrated similar findings for ratings of effectiveness of sharing the information.