Workshops

Thursday, April 10, 8:00 AM – 5:00 PM

Over the past decade, we’ve seen an explosion in tools designed to use large quantities of data to make very accurate predictions on new observations that are very similar to the old. These tools, falling under the broad label of machine learning and, more recently, artificial intelligence, were largely designed with very different problems in mind compared to those population researchers usually face. Nonetheless, the combination of more easily accessible AI/ML tools and increasing challenges with data collection (e.g. rising costs and decreasing response rates in traditional surveys) facilitates a growing wave of population research that’s highly dependent on these tools.  

This workshop provides a hands-on introduction to commonly used AI/ML tools and, critically, the implications of these tools for population research.  The workshop will be interactive using a series of exercises and activities in R. The course has two parts. In part 1, participants will learn about commonly used AI/ML models (e.g. lasso, random forests, neural networks).  They will also practice fitting these models and evaluating their predictive performance. Part 2 addresses the *implications* of using AI/ML tools in the context of population research. We examine three specific implications: inference, bias, and privacy. We explore bias through an interactive exercise that illustrates the connection between data used to train algorithms and bias in prediction. For statistical inference, we illustrate how using predicted outcomes gives incorrect regression coefficients (and, equivalently, basic descriptive summaries like breaking an outcome down by respondent demographics) and invalid statistical inference, then work through solutions using the ipd package for R. Finally, we will discuss the potential issues for respondent privacy that arise in the context of AI/ML models.

Thursday, April 10, 8:00 AM – 12:00 PM

The purpose of the workshop is to introduce attendees to two extraordinary new data resources for studying the long-term effects of education and other early-life factors on later-life morbidity, mortality, and cognition. Both long-running cohort studies began as large, nationally representative, and diverse longitudinal studies of education. Both have been repurposed as studies of aging, health, mortality, and cognition by reviving them at midlife and following sample members forward. The Education Studies for Healthy Aging Research (EdSHARe) project is conducting these two cohort studies. Attendees will be positioned to efficiently and effectively develop innovative new population research projects using EdSHARe data.

Thursday, April 10, 1:00 PM – 5:00 PM

This half day workshop is designed for researchers and implementors working in family planning programming and are interested to layer and/or refocus FP programming using a women’s empowerment lens. The workshop will provide an overview of the EMERGE framework (Raj et al, 2024), which was developed to conceptualize and operationalize gender empowerment measures for use across disciplines. Participants will grapple with the concepts of choice, agency, backlash, and norms in the context of FP and SRHR, and learn to operationalize these concepts for measurement using the EMERGE Framework and emerge.ucsd.edu. This participatory workshop will ask participants to share their experiences and learnings on choices and actions women and girls are taking to protect their bodily integrity and reproductive health, and we will discuss how these map onto choice and agency measures in family planning.  Attendees will increase knowledge and capacity on measuring rights- and choice- based approaches to FP programming and evaluation, centering women and girls choice and agency in family planning.

Thursday, April 10    8:00 AM – 12:00 PM

The workshop objective is to equip participants with the technical skills necessary to access, manipulate, and integrate satellite derived and remotely sensed climate data from repositories like the Famine Early Warning Systems Network (FEWS NET) with surveys, such as the Demographic and Health Survey (DHS) or their own personal work. The goals will be to introduce and together walk through a step-by-step R program to learn the above skills. The expected outcomes will be that participants are more knowledgeable about the types of datasets they can use and have introductory knowledge of the R packages and code structures that can be used to analyze them.

Sunday, April 13    1:00 PM – 5:00 PM

This workshop will enable participants to acquire innovative insights, practical skills, and expertise at the intersection of statistics and artificial intelligence. Attendees will: (1) Understand the fundamental principles of artificial intelligence (AI) and their application in the analysis of complex statistical data, (2) Explore machine learning as a complementary tool to traditional statistical methods, (3) Learn how to use AI models to solve problems related to big data, missing variables or non-linear relationships, (4) Acquire skills in data visualization using tools integrating AI technologies to better interpret analysis results, (5) Promote an interdisciplinary approach that combines AI and statistics to address contemporary challenges in research and scientific communication.

(e.g., ethnicities, language, religion, and subdistricts)

Thursday, April 10    1:00 PM – 5:00 PM

This workshop will introduce participants to key challenges in integrating data across international datasets by complex categories, provide hands-on practice using SocioMap to overcome challenges across the entire workflow of merging new comparative datasets, and demonstrate how registered users can add, cross-check, and share new categories, crosswalks, and dataset metadata.

Participants will be able to use SocioMap’s functions to efficiently and transparently overcome challenges across the entire workflow of merging new comparative datasets, including translating, merging, and sharing merging decisions for reuse and replication.

Thursday, April 10    1:00 PM – 5:00 PM

Participants will learn how to use Stata to analyze time-series data, fit models, interpret their results, and run diagnostics to check their modeling assumptions.

Thursday, April 10    1:00 PM – 5:00 PM

Attendees will gain an understanding of the design and content of NHATS and NSOC, learn about complex methodological considerations of NHATS and NSOC, including the use of weights, linkage of data files, and conducting cross-sectional and longitudinal analyses - and learn about resources available, including other learning tools and educational opportunities. 

Our goal in presenting the workshop is to introduce demographers to the study design, content, and analysis of NHATS/NSOC data to further study of population aging, disability and care. Attendees should expect to leave with an understanding of how NHATS/NSOC can further their research agenda on questions related to population aging, disability and care, a preliminary understanding of how to work with NHATS and NSOC data files, and know how to find more information and resources after the workshop

Thursday, April 10    8:00 AM – 12:00 PM

This workshop will familiarize participants with the content, structure, and scope of a new publicly available datasource to advance maternal health equity among underserved, underrepresented, and underreported populations. This course demonstrates practical applications of the dataset in answering maternal health equity research questions while helping participants understand the uses and limitations of the dataset. Participants are guided through a hands-on session with the dataset in commonly-used statistical analysis software packages.

In the end, attendees will have an increased familiarity of the sources of data, data dictionary, and user guide. Participants will understand the structure, content, and potential application of the new maternal health equity dataset. Attendees will identify potential maternal health equity questions that can be answered with the dataset.

Thursday, April 10    1:00 PM – 5:00 PM

This workshop will provide an overview of the genealogical, longitudinal design and content of the Panel Study of Income Dynamics (PSID), describe the design and content of the PSID Child Development Supplement (CDS) and Transition into Adulthood Supplement (TAS), introduce available genomic data from CDS children and their caregivers, interactively demonstrate how to access data, documentation, and tutorials through the publicly available online Data Center. Also, the course will interactively demonstrate data tools that allow users to connect individuals in PSID to their kin to support intergenerational and family-level analysis. Programs and data files will be available to workshop participants who register as PSID users.

Thursday, April 10    8:00 AM – 5:00 PM

This course aims to explore the impact of LGBTQ population from a queer perspective on the field of population studies. Throughout the workshop, we will introduce concepts and definitions related to queer studies, delving into challenges in advancing the field. Additionally, we will explore widely used measurements in representative surveys and censuses and their limitations. We expect that after the workshop students that are not familiar with the topic, feel more encouraged to work with LGBTQ data and those that have some familiarity can improve their understanding in the topic in the field of population studies.

Thursday, April 10    1:00 PM – 5:00 PM

This workshop highlights a freely-available, high quality data source for the study of population-level trends and consequences in mortality. It also provides an orientation for researchers who may not be familiar with applying common statistical approaches for the study of mortality, such as Cox Proportional Hazards models, to complex survey-NDI linked data. It is an excellent source for the study of health inequities by characteristics of interest to many PAA 2025 attendees, such as age, sex, race, education, nativity, and sexual orientation.

Navigate the IPUMS NHIS data access systems to identify covariates of interest for examining mortality outcomes, use online documentation to learn about these variables, and create customized data files to answer research questions. Describe the strengths and limitations of the LMF data for examining mortality outcomes. Apply NHIS LMF data to calculations of mortality statistics.

Thursday, April 10    8:00 AM – 5:00 PM

Population projections have traditionally been done deterministically using the cohort component method, yielding a single value for each projected future population quantity of interest. Over the past decade, the United Nation Population Division has adopted a probabilistic approach to project fertility, mortality, migration and population for all countries. In this approach, the total fertility rate, female and male life expectancy at birth, and net migration rates are projected using Bayesian hierarchical models estimated via Markov Chain Monte Carlo. These are then combined with a cohort component model which yields probabilistic projection for any quantity of interest. The methodology is implemented in a suite of R packages which has been used by the UN analysts producing the most recent revision of the World Population Prospects. More recently the methods have been extended to be applied to subnational population projections for states and counties.

This course will teach the theory and practice behind Bayesian subnational population projections. Ideas of the Bayesian hierarchical modeling will be explained. In hands-on exercises, students will become familiar with the functionality of the R packages. By the end of the course, they will have a basic understanding of the methods, be able to generate projections using their own data, and visualize probabilistic projections for many quantities of interest using various output formats, such as graphs, tables, maps, and pyramids.

Thursday, April 10    8:00 AM – 12:00 PM

The Future of Families and Child Wellbeing Study (FFCWS, formerly the Fragile Families and Child Wellbeing Study) is the longest running birth cohort study in the US based on a national probability sample. It is based on a stratified, multistage sample of 4898 children born in large U.S. cities (population over 200,000) between 1998 and 2000, where births to unmarried mothers were oversampled by a ratio of 3 to 1. This sampling strategy resulted in the inclusion of a large number of Black, Hispanic, and low-income families and provides sufficient data to examine racial inequalities. Mothers were interviewed shortly after birth and fathers were interviewed at the hospital or by phone. Follow-up interviews were conducted when children were approximately ages 1, 3, 5, 9, 15, and 22. When weighted, the data are representative of births in large US cities. With the latest data release for the Year 22 study, which focuses on the focal children’s transition to the young adulthood, researchers are able to explore a variety of topics on the health and wellbeing of the young adults, including social economic status, family formation, health and behavior, relationships, systems involvement, identity, and substance use. In addition to those topics, the study also collected data on the impacts of the pandemic on the young adults’ and their parents’ lives.

This workshop is designed for both researchers who are interested in learning about using The Future of Families and Child Wellbeing Study data for the first time and current FFCWS data users who would like a deeper dive into navigating our data. This workshop will introduce researchers to 1) a brief overview of the history and data collection of the FFCWS, including the variety of survey and activity components included in our available data; 2) an overview of our Year 22 data and the latest contextual data files; 3) a guide to the file structure, contents, and data conventions used in FFCWS including tips and advice from experts in FFCWS data support; 4) an interactive tutorial on the contents and use of our documentation and metadata web interface; and 5) an exploration of how the FFCWS data can be used for multi-disciplinary research topics and for teaching. Our goals for this workshop are to provide a solid background for working with FFCWS data and to support participants as they begin variable exploration and selection relevant to their own research questions. 

The FFCWS provides free access to its public use data, which covers a wide range of topics, including education, employment, housing, relationships, parenting, health, income, religion, and more. It also releases restricted use contract data to the research community via a Contract Data Agreement. Those restricted use contract data include residential contextual files, school contextual files, as well as biological and health files. The longitudinal information on individual, family, school, and neighborhood characteristics makes the FFCWS a rich data source to support research on a wide range of public policy topics, including economic wellbeing, safety net, healthcare access, substance use, racial inequity, criminal justice, child support, gun violence, policing, determinants of health, and more.

Thursday, April 10    1:00 PM – 5:00 PM

The main goal of this workshop is to introduce researchers to the Annual Survey of Refugees (ASR), which is an annual survey that collects information on refugees during the first five years after arrival in the US. The ASR has household and individual information about demographics, experiences before arrival, education and skills, economic self-sufficiency, health, social connection, well-being and receiving community experience, and government program receipt, among other topics. 

Through this workshop, researchers will learn about the range of information collected in the ASR; other users’ experience with the ASR and examples of prior research that analyzed the ASR; key methodological features; how to download the datafiles from the Inter-University Consortium for Political and Social Research (ICPSR); how to export the dataset to R, Stata, SPSS, or SAS; and how to use the survey weights to produce subgroup estimates. The programming language for the workshop will be determined through a survey shared with registered participants. Researchers will also learn about the importance of using weights and tips for determining whether to use person-level or household-level weights. The workshop will also discuss how to merge multiple years of data and considerations for doing analysis with a multi-year dataset.