Web-Based Lectures



Title: Unequal Probability, High Entropy, and Balanced Sampling Designs: An Overview
Presenter: David Haziza, Université de Montréal
Date and Time: Thursday, January 26, 2017, 1:00 p.m. – 3:00 p.m. Eastern time
Sponsor: Survey Research Methods Section

Registration Deadline: Tuesday, January 24, at 12:00 p.m. Eastern time

Description:
Unequal probability sampling designs are widely used in surveys for a number of reasons. In this talk, we present an overview of some important sampling designs, including probability proportional-to-size sampling designs. We discuss the concept of entropy of a sampling design and discuss some well-known approximations of the second-order inclusion probabilities in the context of high entropy sampling designs. These approximations may prove useful at the variance estimation stage. Finally, balanced sampling will be discussed. Implementation through the Cube algorithm of Deville and Tillé (2004) and the rejective procedure of Fuller (2009) will be presented.

Registration Fees:
SRMS Members: $60
AAPOR Members: $60
ASA Members: $75
Nonmembers: $95

Each registration is allowed one web connection and one audio connection. Multiple persons are encouraged to view each registered connection (for example, by projecting the webinar in a conference room).

Register

Access Information
Registered persons will be sent an email the afternoon of Tuesday, January 24, with the access information to join the webinar and the link to download and print a copy of the presentation slides.



Title: Differential Privacy: Protecting Individuals from Re-Identification
Presenters: Daniel Kifer, Pennsylvania State University, Vishesh Karwa, Duke University
Moderator: Aleksandra Slavkovic, Pennsylvania State University
Date and Time: Friday, February 10, 2017, 1:00 p.m. – 2:00 p.m. Eastern time
Sponsor: ASA Committee on Privacy and Confidentiality

Registration Deadline: Wednesday, February 8, at 12:00 p.m. Eastern time

Description:
Differential privacy is a mathematical framework for protecting privacy interests in statistical databases by focusing on the disclosure risk to an individual being included in a data set. In this webinar, two research experts explain this methodology and how they apply differential privacy methodology as a data protection method for protecting data files from the risk of re-identification. A key benefit of the Differential Privacy methodology is that in many cases, appropriate privacy protection can be achieved if random noise is properly added to the actual results. For example, rather than simply reporting the sum, the data provider can inject noise based on a distribution. The calculation of “how much” noise to inject can be made based only on knowledge of the function to be computed. This webinar will cover the basic principles of differential privacy, how it works, and how it can successfully be applied to current statistical databases.

Registration Fees:
This webinar is free to anyone who would like to attend. However, registration is limited so you must register to receive the access information. The access information will be emailed to everyone who has registered the afternoon of Wednesday, February 8.

Each registration is allowed one web connection and one audio connection. Multiple persons are encouraged to view each registered connection (for example, by projecting the webinar in a conference room).

Register



Title: UC Berkeley’s Data Science Course for Undergraduates: Computational and Inferential Thinking for the 21st Century
Presenters: Ani Adhikari, Department of Statistics, UC Berkeley, David Culler, Department of Electrical Engineering and Computer Sciences (EECS) , UC Berkeley, John DeNero, EECS, UC Berkeley
Date and Time: Thursday, February 16, 2017, 2:30 p.m. – 3:30 p.m. Eastern time
Sponsor: ASA-MAA Joint Committee, Section on Statistical Education, Section on Statistical Learning and Data Science

Registration Deadline: Tuesday, February 14, at 12:00 p.m. Eastern time

Description:
UC Berkeley's data science education program focuses on all undergraduates—not just those in computer science, statistics, or another STEM field. Building on the premise that data literacy is part of what it means to be an educated person in the 21st century, Berkeley faculty have developed a very popular data science course for freshmen, with no prerequisite background in mathematics, statistics, or programming. The course, which has been attracting national and international attention, is unusual also in that it has "connector" courses in numerous other disciplines, typically taken concurrently. This webinar will describe the course – how it works, whether it works, and if so why it works – and why the faculty who are teaching it can't imagine themselves ever teaching in the old way again.

Registration Fees:
This webinar is free to anyone who would like to attend. However, registration is limited so you must register to receive the access information. The access information will be emailed to everyone who has registered the afternoon of Tuesday, February 14.

Each registration is allowed one web connection and one audio connection. Multiple persons are encouraged to view each registered connection (for example, by projecting the webinar in a conference room).

Register



Title: An Introduction to Bayesian Nonparametric Methods for Causal Inference in Pharmacoepidemiology
Presenter: Jason Roy, University of Pennsylvania
Date and Time: Thursday, February 23, 2017, 12:00 p.m. – 2:00 p.m. Eastern time
Sponsor: Biopharmaceutical Section

Registration Deadline: Tuesday, February 21, at 12:00 p.m. Eastern time

Description:
In this webinar we provide an overview of Bayesian nonparametric (BNP) approaches to causal inference from observational data. One of the concerns about using fully Bayesian methods in these kinds of studies has been possible mis-specification of models for high dimensional conditional or joint distributions (such as the conditional distribution of outcome given confounders). Recent advances in Bayesian nonparametric methods, however, opens the door to using fully Bayesian methods that make minimal assumptions about the observed data. The combination of the observed data model and causal assumptions allows for identification of any type of causal effect - differences, ratios, or quantile effects, either marginally or for subpopulations of interest. In the first half of the webinar we will review BNP methods. In the second half, we will focus on causal inference problems and illustrate with examples in pharmacoepidemiology. Software and implementation will also be discussed.

Registration Fees:
Biopharmaceutical Section Members: $0
ASA Members: $59
Nonmembers: $74

Each registration is allowed one web connection and one audio connection. Multiple persons are encouraged to view each registered connection (for example, by projecting the webinar in a conference room).

Register

Access Information
Registered persons will be sent an email the afternoon of Tuesday, February 21, with the access information to join the webinar and the link to download and print a copy of the presentation slides.



Title: Introduction to Functional Neuroimaging
Presenter: Martin Lindquist
Date and Time: Thursday, April 13, 2017, 11:00 a.m. – 1:00 p.m. Eastern time
Sponsor: Mental Health Statistics Section

Registration Deadline: Tuesday, April 11, at 12:00 p.m. Eastern time

Description:
Understanding the brain is arguably among the most complex, important and challenging issues in science today. Neuroimaging is an umbrella term for an ever-increasing number of minimally invasive techniques designed to study the brain. These include a variety of rapidly evolving technologies for measuring brain properties, such as structure, function and disease pathophysiology. The analysis of neuroimaging data is an example of a modern ‘big data’ problem, and the data is not only large but also has a complex correlation structure in both space and time. Statistics plays a crucial role in understanding the nature of the data and obtaining relevant results that can be used and interpreted by neuroscientists. In this talk we will focus on methods for performing functional neuroimaging (e.g., functional MRI) and discuss how these techniques can be used to detect areas of the brain activated by a task, determine how different brain regions are connected and communicate with one another, and how brain measurements can be used for prediction and classification purposes.

Registration Fees:
Member of the Mental Health Statistics Section: $60
ASA Member: $90
Nonmember: $110

Each registration is allowed one web connection and one audio connection. Multiple persons are encouraged to view each registered connection (for example, by projecting the webinar in a conference room).

Register

Access Information
Registered persons will be sent an email the afternoon of Tuesday, April 11, with the access information to join the webinar and the link to download and print a copy of the presentation slides.





Title: Intensive Longitudinal Data Analysis Using Mplus
Presenters: Bengt Muthen, University of California, Los Angeles
Date and Time: Thursday, April 20, 2017, 12:00 p.m. – 2:00 p.m. Eastern time
Sponsor: Mental Health Statistics Section

Registration Deadline: Tuesday, April 18, at 12:00 p.m. Eastern time

Description:
This talk discusses new methods for analyzing intensive longitudinal data, such as obtained with ecological momentary assessments, experience method sampling, ambulatory assessments, and daily diaries. Typically, such data have a large number of time points, T = 20-150. Single-level (N=1) as well as multilevel (N > 1) time series models with random effects varying across subjects are handled using a dynamic structural equation model (DSEM) and Bayesian estimation implemented in the Mplus Version 8 software. DSEM for N=1 time series analysis can be used to model the dynamics within a particular individual over time. Additionally, N > 1 multilevel DSEM includes extensions of time series models, such that at level 1 a time series model is used to model the within-person dynamics of a process over time, while at level 2 individual differences in the parameters that capture these dynamics are modeled. DSEM can handle multivariate outcomes as well as latent variables, and random effects can be both predicted from but also predictors of level 2 variables. DSEM is available with auto-regressive and moving-average components both for observed-variable models such as regression and cross-lagged analysis and for latent variable models such as factor analysis, IRT, structural equation modeling, and mixture modeling. DSEM also handles time-varying effect modeling (TVEM) where parameters change not only across individuals but also across time, making it suitable for assessing intervention effects. Several examples are discussed from application areas such as:

  • multilevel AR(1) model with random mean, random AR, and random variance
  • multilevel AR(1) model with measurement error
  • multilevel ARMA(1,1) model
  • multilevel cross-lagged modeling
  • multilevel AR modeling with a trend
  • latent multilevel AR(1) model with multiple indicators
  • latent multilevel VAR(1) model and dynamical networks
  • dynamic SEM
  • dynamic latent class analysis using hidden Markov and Markov-switching AR models

Registration Fees:
Member of the Mental Health Statistics Section: $60
ASA Member: $90
Nonmember: $110

Each registration is allowed one web connection and one audio connection. Multiple persons are encouraged to view each registered connection (for example, by projecting the webinar in a conference room).

Register

Access Information
Registered persons will be sent an email the afternoon of Tuesday, April 18, with the access information to join the webinar and the link to download and print a copy of the presentation slides.