 |
|
February 9, 2010
Courses Course Schedule
  |
 |
Data and Methods Click on a researcher's name to display only his or her recent research projects.
In work using formal demography, Thomas Espenshade, Simon Levin (Ecology and Evolutionary Biology), and OPR graduate student Analia Olgiati have initiated a study on population momentum. The purpose of the project is to decompose total population momentum into two constituent and multiplicative parts called "weak" momentum and "strong" momentum. Weak momentum depends on deviations between a population's observed age distribution and its implied stable age distribution. Strong momentum is a function of deviations between a population's implied stable and stationary age distributions. In general, the factorization of total momentum into the product of weak and strong momentum is a very good approximation. The factorization is exact, however, if the observed age distribution is stable or if initial fertility is already at replacement. The authors provide numerical illustrations by calculating weak, strong, and total momentum for 176 countries, the world, and its major regions. The paper brings together disparate strands of the population momentum literature and shows how the various kinds of momentum considered by researchers fit together into a single unifying framework. A paper on this project was presented in a session on formal demography at the annual meetings of the Population Association of America.
Multistate life tables provide us with estimates of the length of remaining life that individuals can expect to live in different states, like healthy versus unhealthy, married versus unmarried, etc. (called state expectancies). The traditional approach to producing these tables does not produce interval estimates, but instead, produces only a point estimate that fails to reflect the uncertainty with which state expectancies are estimated. Additionally, the traditional approach does not allow us to answer important questions about heterogeneity in state expectancies across the population. Over the past several years, Scott Lynch has developed a method that addresses these two limitations. More recently, he has been extending this method to handle cross-sectional data. Most life table methods require panel data so that transition probabilities between states across time can be observed and modeled. These transition probabilities are then used as input for life table estimation. However, panel data are substantially less common than cross-sectional data. As a consequence, many researchers use "Sullivan's method" to produce multi-state-like estimates of state expectancies. Yet the same limitations to the traditional approach to multi-state life table estimation also apply to Sullivan's method. Lynch's new method overcomes these limitations.
Scott Lynch published a book entitled Introduction to Bayesian Statistics and Estimation for Social Scientists (Springer). This book shows what Bayesian statistics is about and how Bayesian analysis is performed. The book is highly applied and includes a number of R programs that can be used to estimate parameters from common social science models.
Matthew Salganik's research has addressed a number of questions at the intersection of social networks and statistics, much of which involves efforts to collect better data about the populations at greatest risk for HIV/AIDS. Salganik has worked to develop respondent-driven sampling, a network-based sampling method that has been used for disease surveillance among high-risk groups in more than 20 countries. Salganik is also currently working to develop network-based methods to estimate the sizes of high-risk groups. Salganik's other main area of research involves using the Internet and other new technology for social research. For instance, he is working with others to develop software that can be installed on mobile phones allowing researchers to study segregation in space and time. He is also developing methods to allow groups to solicit and then collectively evaluate new ideas. The method elicits suggestions from members of the group, but rather than relying on any central authority, it divides the process of evaluating and sorting these suggestions into human-size chunks that are then distributed to group members. In additional to democratizing the evaluation process, this distributed procedure allows groups to process literally thousands of suggestions, a task beyond the scope of a single individual. A pilot study of this approach was recently completed with the Princeton Undergraduate Student Government.
Samuel Schulhofer-Wohl's recent methodological research focuses on age-period-cohort analysis. The failure of identification in age-period-cohort models due to the perfect linear relationship between birth year, age, and current year is one of the longest-standing methodological problems in the social sciences. In a recent paper, Schulhofer-Wohl and sociologist/demographer Yang Yang (University of Chicago) develop a novel model of continuously evolving age and cohort effects. The conventional linear age-period-cohort model assumes that the influence of age is the same in all time periods, that the influence of present conditions is the same for people of all ages, and that cohorts do not change as they age. The new model relaxes these assumptions and should be useful for studying a wide variety of social scientific topics, such as changes in the pattern of mortality or the pattern of consumption inequality over the life course. Source: OPR Annual Reports.
| |||||
| top | |
||||||
|
|||||||
| |||||||
Search
Back
Next