The PRIISM seminar series consists of research seminars of interest to an applied statistics audience, from innovative applications of applied statistics to novel statistical theory and methodology. An archive of past events (2008 - 2015) and (2015 - 2018) are available, as well as a list of related seminars.

Our seminar series generally consists of three categories of talks: Statistical Methodology, Didactic, and Data and Social Impact seminars. The statistical methodology talks cover new advances in methods for analyzing data, and may be more technical. Didactic seminars provide an opportunity to learn a new area of statistics or way of analyzing data. Data and social impact lectures take a critical look at how we can use data to learn about and provide recommendations for solutions to social problems through quantitative research. These talks vary in technical level.

Upcoming and Recent Events: Fall 2018 and Spring 2019

Date, Time, LocationTalk CategorySpeaker Name, AffiliationTopic (click for more info)
10/24/2018 (Weds)
11:00 am - 12:00 pm,
Kimball 3rd Fl Conf Rm 
Statistical Methodology/Didactic  Leslie McClure (Drexel)
Abstract: Planning for randomized clinical trials relies on assumptions that are often incorrect, leading to inefficient designs that could spend resources unnecessarily. Recently, trialists have been advocating for implementation of adaptive designs, which allow researchers to modify some aspect of their trial part-way through the study based on accumulating data. In this talk, I will introduce the concept of adaptive designs and describe several different adaptations that can be made in clinical trials. I will then describe a real-life example of a sample size re-estimation from the Secondary Prevention of Small Subcortical Strokes (SPS3) study, describe the statistical impact of implementing this design change, and describe the effect of the adaptation on the practical aspects of the study.
10/31/2018 (Weds)
10:30 am - 12:00 pm,
Kimball 3rd Fl Conf Rm 
Didactic  Chuck Huber (Stata Corp)
Abstract: This talk introduces the concepts and jargon of structural equation modeling (SEM) including path diagrams, latent variables, endogenous and exogenous variables, and goodness of fit. I demonstrate how to fit many familiar models such as linear regression, multivariate regression, logistic regression, confirmatory factor analysis, and multilevel models using -sem-. I wrap up by demonstrating how to fit structural equation models that contain both structural and measurement components.

11/14/2018 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Data and Social Impact  Ravi Shroff (NYU)
Abstract: Policymakers often seek to gauge discrimination against groups defined by race, gender, and other protected attributes. A common strategy is to estimate disparities after controlling for observed covariates in a regression model. However, not all relevant factors may be available to researchers, leading to omitted variable bias. Conversely, controlling for all available factors may also skew results, leading to so-called "included variable bias". We introduce a simple strategy, which we call risk-adjusted regression, that addresses both concerns in settings where decision makers have clear and measurable policy objectives. First, we use all available covariates to estimate the expected utility of possible decisions. Second, we measure disparities after controlling for these utility estimates alone, omitting other factors. Finally, we examine the sensitivity of results to unmeasured confounding. We demonstrate this method on a detailed dataset of 2.2 million police stops of pedestrians in New York City.

12/5/2018 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Statistical Methodology   Russell Steele (McGill)
Abstract: Every statistical analysis requires at least some subjective or untestable assumptions. For example, in Bayesian modelling, the analysis requires specification of hyperparameters for prior distributions which are either intended to reflect subjective beliefs about the model or to reflect relative ignorance about the model under a certain notion of ignorance. Similarly, causal models require assumptions about parameters related to unmeasured confounding. Violations of these untestable or subjective assumptions can invalidate the conclusions of analyses or lead to conclusions that only hold for a narrow range of choices for those assumptions. Currently, researchers compute several estimates based on either multiple “reasonable” values or a wide range of “possible” values for these inestimable parameters. Even when the dimension of the inestimable parameter space is relatively small, the sensitivity analyses generally are not systematically conducted and may either waste valuable computational time on choices that lead to roughly the same inference or will miss examining values of those parameters that would change the conclusions of the analysis.

In this talk, I will propose the use of Bayesian optimization approaches for decision-driven sensitivity analyses. We assume that a decision will be made as a function of the model estimates or predictions from particular model which relies on inestimable parameters. We use a Bayesian optimization approach to identify partitions of the space of inestimable parameter values where the decision based on the observed data and assumed parameter values change, rather to rely on non-systematically chosen values for the sensitivity analysis. We will illustrate our proposed approach on a hierarchical Bayesian meta-analysis example from the literature.

The work that will be presented was done in collaboration with Louis Arsenault-Mahjoubi, an undergraduate mathematics and statistics student at McGill University.

2/6/2019 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Data and Social Impact   Melody Goodman (NYU)
Abstract: The utility of community-engaged health research has been well established. However, measurement and evaluation of community engagement in research activities (patient/stakeholder perceptions of the benefit of collaborations that indicate how engaged the patient/stakeholder feels) has been limited. The level of community engagement across studies can vary greatly from minimal engagement to fully collaborative partnerships. Methods for measuring the level of community engagement in research are still emerging in the field due to the methodological gap in the assessment of stakeholder engagement, likely due to the lack of existing measures. There is a need to rigorously evaluate the impact of community/stakeholder engagement on the development, implementation and outcomes of research studies, which requires the development, validation, and implementation of tools that can be used to assess stakeholder engagement.

We use community-engaged research approaches and mixed-methods (qualitative/quantitative) study design to validate a measure to assess the level of community engagement in research studies from the stakeholder perspective. As part of the measurement validation process, we are conducting a series of web-based surveys of community members/community health stakeholders who have participated in previous community-engaged research studies. The surveys examine construct validity and internal consistency of the measure. We examined content validity through a five round modified Delphi process to reach consensus among experts and construct validity is assessed through participant surveys.

Research that develops standardized, reliable, and accurate measures to assess community engagement is essential to understanding the impact of community engagement on the scientific process and scientific discovery. Implementation of gold standard quantitative measures to assess community engagement in research would make a major contribution to community-engaged science. These measures are necessary to assess associations between community engagement and research outcomes.

2/27/2019 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Data and Social Impact   Eric Loken (UConn)
Abstract: Science is responding well to the so-called reproducibility crisis with positive improvements in methodology and transparency. Another area for improvement is awareness of statistical issues impacting inference. We explore how some problematic intuitions about measurement, statistical power, multiple analyses, and levels of analysis can affect the interpretation of research results, perhaps leading to mistaken claims.

Speaker: Eric Loken is in the Neag School of Education at The University of Connecticut. He studies advanced statistical models including hierarchical models, measurement models, factor and mixture models, and their applications in health and education research. He works extensively in educational measurement with applications to large scale testing. Recent work has addressed issues surrounding statistical inference, and the relationship to failures to replicate research results.

4/3/2019 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Data and Social Impact   Gregory Ridgeway (UPenn)
Abstract: The police are chronically a topic of heated debate. However, most statistical analyses brought to bear on questions of police fairness rarely provide clarity on or solutions to the problems. This talk will cover statistical methods for estimating racial bias in traffic stops, identifying problematic cops, and determining which officers are most at risk for police shootings. All of these methods have been part of investigations of police departments in Oakland, Cincinnati, and New York and show that statistics has an important role in prominent crime and justice policy questions.

4/10/2019 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Statistical Methodology   Lawrence Wu (NYU)
Abstract: We examine difference-in-differences procedures for estimating the causal effect of treatment when the outcome is a single-decrement demographic process. We use the classic case of two groups and two periods to contrast a standard and widely-used linear probability difference-in-differences estimator with an analogous proportional hazard difference-in-differences estimator. Formal derivations and illustrative examples show that the linear probability estimator is inconsistent, yielding estimates that, for example, evolve with time since treatment. We conclude that knowledge of how the data are generated is a necessary component for causal inference.

4/17/2019 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Statistical Methodology   Irini Moustaki (LSE)
Abstract: In this talk we will discuss some primary results from the modelling of dyadic data that provide information on intergenerational exchanges in the UK. We will use longitudinal data from three waves of the UK Household Longitudinal Survey, to study and explain associations between exchanges of support from the respondent to their parents and to their children. The data resemble the structure of dyadic data, they are collected across time and they are also multivariate because constructs of interest are measured by multiple indicators. Support is measured by a set of binary indicators of different kinds of help.
We propose two different joint models of bidirectional exchanges with support given and support received treated as a multivariate response, and covariances between responses measuring the extent of reciprocation between generations. Moreover, joint modelling of longitudinal data allows for the possibility that reciprocation may occur contemporaneously or may be postponed until the donor is in need of help or the recipient is in a position to reciprocate.

5/1/2019 (Weds)
11:00 am - 12:00 pm
Kimball 3rd Fl Conf Rm 

Data and Social Impact   Joseph Cimpian (NYU)
Abstract: From the time students enter kindergarten, teachers overestimate the abilities of boys in math, relative to behaviorally and academically matched girls, contributing to a gender gap favoring boys in both math achievement and confidence. Using data from numerous nationally representative studies spanning kindergarten through university level, as well as experimental evidence, I demonstrate how girls and young women face discrimination and bias throughout their academic careers and suggest that a substantial portion of the growth in the male–female math achievement gap is socially constructed. Each of the studies leads to a broader set of considerations about why females are viewed as less intellectually capable than their male peers. The studies also demonstrate that biases can be exhibited and perpetuated by members of negatively stereotyped groups (e.g., female teachers demonstrate greater bias against girls than do male teachers), and raise questions about the root causes of their biases and the long-term effects of being negatively stereotyped oneself. This research also suggests that comparing boys and girls on metrics such as standardized tests and grades may contribute to a false belief that education systems promote the success of females. Together, the studies suggest several implications for research, teacher professional development, and policy.

Sign up for our mailing list to receive upcoming event reminders. 

Links to Related Seminars: