Mark D. Hayward

Description:

This course provides an in-depth introduction to survival methods for the analysis of change in time-dependent discrete dependent variables. The course draws on methodological and empirical research from the social sciences. Special attention is directed at the relationship between theories of social change, life course attainment and dynamic models. Moreover, we will examine alternative data collection strategies to obtain longitudinal data. Examples and homework assignments will draw principally on three data sets: the German Life History Study (GLHS), the National Longitudinal Survey of Older Men, and the National Health Interview Surveys.

The course begins by focusing on simple stochastic processes for qualitative variables. We do sothrough the lens of non-parametric, discrete-time life table techniques. The topics include basiclife table relations, single- and multiple-decrement life table models, and increment-decrementlife table models. The purpose is to learn how to describe increasingly complex dynamic social processes. We then turn to continuous-time, discrete-state models. Parametric and partially parametric models will be introduced which allow for the dependency of rates on both explanatory factors and time. We will briefly examine some special topics including competing risk models, local hazard models, unobserved sources of heterogeneity in hazard models, and continuous state-space hazard models. The purpose of this aspect of the course is to communicate the variety and power of multivariate hazard models for hypothesis testing of dynamic processes in the social sciences.

PREREQUISITES:

Students should have completed the graduate statistics core sequence in sociology (or theequivalent) and have a thorough knowledge of the general linear model. A background in loglinear analysis and logistic regression is important, because hazard models are part of the class of log-linear models. Students must have a solid knowledge of a major spreadsheet package (e.g., EXCEL) for computing and graphing.

REQUIREMENTS:

In addition to reading required materials, attending class sessions, and contributing in class, students must complete three sets of written assignments. First, homework occasionally will be assigned. Homework assignments will involve both numerical problem solving and critical writing. The second written requirement is a final exam. The exam will be a take-home exam involving longitudinal data analysis. The basic aim of the exam is to educate you about how one goes about formulating, estimating, and interpreting event history models. Grades will be based on the final exam (75%) and the homework (25%). I also will evaluate the homework assignments to identify concepts or techniques that are problematic. This year, I am also offering the option of writing an empirical paper using event history techniques in lieu of taking the final exam. If you choose this option, you will need to inform me by October 3rd. I will provide more detailed information about the types of analyses that are required if the paper option is chosen.

READINGS:

The technical literature on survival analysis is voluminous, spanning a variety of fields. For example, partial likelihood hazard models were developed primarily within biostatistics, while the much of the development of parametric hazard models took place within sociology and economics. Starting in the late 1980s, a literature emerged which places these different traditions within a more general survival analysis framework. A major work in this vein is the book, Event History Analysis (Blossfeld, Hamerle, and Mayer, 1989). This text also shows how to estimate hazard models using a variety of software packages (e.g., SAS). More recently Techniques of Event History Modeling was published (Blossfeld and Rohwer, 1995, 2002); this is a comprehensive introduction to continuous-time event history techniques. The 2002 2nd edition is the core text for this course.

The textbook has been ordered by the UT bookstore. The full title is: Blossfeld, Hans-Peter, and Götz Rohwer. 2002. Techniques of Event History Modeling. Second

Edition. Mahway, New Jersey: Lawrence Erlbaum Associates.

If you are a Stata user and plan to use event history models extensively in your work, you might want to consider ordering a version of this text that uses Stata to generate the examples shown in the text.

The text is:

Blossfeld, Hans-Peter, Katrin Golsch, and Götz Rohwer. 2007. Event History Analysis with Stata. Mahway, New Jersey: Lawrence Erlbaum Associates.

I also have provided examples of substantive applications for most of the types of survival processes we will be discussing. I may add or delete some of the recommended substantive readings if I think there are better illustrations. Important information about course materials:

1. Course notes for each module are located on the Canvas site for the class. Please print the modules and bring your notes to class.

2. Some material that is difficult to obtain has been scanned and is available in .pdf

format on the class’ Canvas site.

3. Assigned journal articles are available electronically and students are responsible

for obtaining their personal copies.

IMPORTANT DATES:

September 19th – No class

December 5th – Last class day

December 12th- Final exams/papers due

Please cconsult with the academic advisor for more information.

COURSE CONTENT:

This course is intended to provide an in-depth introduction to the major conceptual frameworks and empirical research examining distal and proximate social factors influencing adult health. The underlying question is – “Does the body forget insults and exposures from adverse social conditions over the life course?” In grappling with this question, we will pay close attention to the intersection between the etiology of adult health conditions and the major life-course trajectories of socioeconomic position and family. Of major importance will be the assessment of key social and biological pathways linking social conditions at particular points in the life course with adult health outcomes. 

REQUIREMENTS:

The course is organized as a seminar. Although I will provide overviews and/or supplemental material in lecture, much instruction takes place in the context of guided discussion and exchanges focused on the readings. Students are expected to attend each class and to have read all of the assigned material thoroughly and critically before the class meeting. Reading critically means not only being able to describe the content of an assigned piece, but also to evaluate the scientific issues motivating the research question, the conceptual/theoretical strengths of the research, the adequacy of the study’s methods, the scientific implications of the study’s findings, and the study’s relation to other course material). Students are strongly encouraged to contribute to the discussion both their insights and/or questions from the readings and their own experiences. There are two major tasks to be accomplished in this graduate seminar:

1) The first task involves preparation of the reading for class presentation and discussion. All students are responsible for all of the assigned readings. Class discussions will be oriented around informal student presentations based on the assigned readings. For each class meeting, a student will be assigned an article, and the student will informally present the scientific factors motivating a study, the key gaps in knowledge being addressed, the major findings, and the conceptual/methodological strengths and weaknesses of the study. An outline of the presentation’s key points (1 page) should be distributed to the class on the Tuesday prior to class. The informal presentations should not be longer than 15 minutes – and 10 minutes is desirable!

(20% of grade)

2) The second task is an empirical study on a topic that is relevant to the course’s overall aim. Two assignments are required in accomplishing this task: a poster presentation and a term paper. The empirical study may overlap with research being done either as part of a student’s thesis/dissertation research, in conjunction with work being done in another course (with the permission). The study should be original research, with the aims of making a scientific contribution to the literature and eventual publication in a scientific outlet. Given the time constraints imposed by the semester, I recommend that students rely on publicly accessible datasets (e.g., the Health and Retirement Survey, the National Health Interview Surveys, the National Longitudinal Surveys, Americans’ Changing Lives, and Aging, Status, Sense of Control). Other rich datasets are available from NACDA and ICPSR, two major electronic data archives. Students are required to discuss their topic with me before moving forward. Students should submit 1-page abstracts. . I will use the abstract to gauge the paper’s topic  relevance to the course aims, and as a basis to provide critical feedback to refine conceptual ideas and focus the analysis.

1) Poster presentation:  These sessions will give students a chance to present their work to their colleagues, to field comments,and to refine their ideas and analysis prior to submitting the term paper. Posters should followthe format that one would use if presenting the results at a major scientific meetings such as the Population Association of America. However, you do not actually need to construct an actualposter. Poster “pages” should be constructed using Powerpoint for presentation to the class.

(30% of grade)

2) Term paper  should aim to be a publishable scientific work suitable for submission to a scientific journal such as Demography. Students should review articles in Demography or other major journals for guidance on presentation styles and organizational format. (50% of grade)

DESCRIPTION: 

This course provides an in-depth introduction to survival methods for the analysis of change in time-dependent discrete dependent variables. The course draws on methodological and empirical research from the social sciences. Special attention is directed at the relationship between theories of social change, life course attainment and dynamic models. Moreover, we will examine alternative data collection strategies to obtain longitudinal data. Examples and homework assignments will draw principally on three data sets: the German Life History Study (GLHS), the National Longitudinal Survey of Older Men, and the National Health Interview Surveys. 

The course begins by focusing on simple stochastic processes for qualitative variables. We do so through the lens of non-parametric, discrete-time life table techniques. The topics include basic life table relations, single- and multiple-decrement life table models, and increment-decrement life table models. The purpose is to learn how to describe increasingly complex dynamic social processes. We then turn to continuous-time, discrete-state models. Parametric and partially parametric models will be introduced which allow for the dependency of rates on both explanatory factors and time. We will briefly examine some special topics including competing risk models, local hazard models, unobserved sources of heterogeneity in hazard models, and continuous state-space hazard models. The purpose of this aspect of the course is to communicate the variety and power of multivariate hazard models for hypothesis testing of social science processes. 

PREREQUISITES: 

Students should have completed the graduate statistics core sequence in sociology (or the equivalent) and have a thorough knowledge of the general linear model. A background in log-linear analysis and logistic regression is important, because hazard models are part of the class of log-linear models. Students must have a solid knowledge of a major spreadsheet package (e.g., EXCEL) for computing and graphing. 

REQUIREMENTS: 

In addition to reading required materials, attending class sessions, and contributing in class, students must complete three sets of written assignments. First, homework occasionally will be assigned. Homework assignments will involve both numerical problem solving and critical writing. The second and third written requirements are the mid-term and final exams. Both exams will be take-home exams involving longitudinal data analysis. The basic aim of the exams is to educate you about how one goes about formulating, estimating, and interpreting event history models. Grades will be based on the two exams (45% each) and the homework 

(10%). I also will evaluate the homework assignments to identify concepts or techniques that are problematic. 

I am also offering the option of writing an empirical paper using event history techniques in lieu of taking the final exam. If you choose this option, you will need to inform me immediately after the mid-term exam. 

READINGS: 

The technical literature on survival analysis is voluminous, spanning a variety of fields. For example, partial likelihood hazard models were developed primarily within biostatistics, while the much of the development of parametric hazard models took place within sociology and economics. Starting in the late 1980s, a literature emerged which places these different traditions within a more general survival analysis framework. A major work in this vein is the book, Event History Analysis (Blossfeld, Hamerle, and Mayer, 1989). This text also shows how to estimate hazard models using a variety of software packages (e.g., SAS). More recently Techniques of Event History Modeling was published (Blossfeld and Rohwer, 1995, 2002); this is a comprehensive introduction to continuous-time event history techniques. The 2002 2nd edition is the core text for this course. 

The textbook has been ordered by the UT bookstore. The full title is: 

Blossfeld, Hans-Peter, and Götz Rohwer. 2002. Techniques of Event History Modeling. Second Edition. Mahway, New Jersey: Lawrence Erlbaum Associates. 

If you are a Stata user and plan to use event history models extensively in your work, you might want to consider ordering a version of this text that uses Stata to generate the examples shown in the text. The text is: 

Blossfeld, Hans-Peter, Katrin Golsch, and Götz Rohwer. 2007. Event History Analysis with Stata. Mahway, New Yersey: Lawrence Erlbaum Associates. 

COURSE CONTENT:

This course is intended to provide an in-depth introduction to the major conceptual frameworks and empirical research examining distal and proximate social factors influencing adult health. The underlying question is – “Does the body forget insults and exposures from adverse social conditions over the life course?” In grappling with this question, we will pay close attention to the intersection between the etiology of adult health conditions and the major life-course trajectories of socioeconomic position and family. Of major importance will be the assessment of key social and biological pathways linking social conditions at particular points in the life course with adult health outcomes. 

REQUIREMENTS:

The course is organized as a seminar. Although I will provide overviews and/or supplemental material in lecture, much instruction takes place in the context of guided discussion and exchanges focused on the readings. Students are expected to attend each class and to have read all of the assigned material thoroughly and critically before the class meeting. Reading critically means not only being able to describe the content of an assigned piece, but also to evaluate the scientific issues motivating the research question, the conceptual/theoretical strengths of the research, the adequacy of the study’s methods, the scientific implications of the study’s findings, and the study’s relation to other course material). Students are strongly encouraged to contribute to the discussion both their insights and/or questions from the readings and their own experiences. There are two major tasks to be accomplished in this graduate seminar:

1) The first task involves preparation of the reading for class presentation and discussion. All students are responsible for all of the assigned readings. Class discussions will be oriented around informal student presentations based on the assigned readings. For each class meeting, a student will be assigned an article, and the student will informally present the scientific factors motivating a study, the key gaps in knowledge being addressed, the major findings, and the conceptual/methodological strengths and weaknesses of the study. An outline of the presentation’s key points (1 page) should be distributed to the class on the Tuesday prior to class. The informal presentations should not be longer than 15 minutes – and 10 minutes is desirable!

(20% of grade)

2) The second task is an empirical study on a topic that is relevant to the course’s overall aim. Two assignments are required in accomplishing this task: a poster presentation and a term paper. The empirical study may overlap with research being done either as part of a student’s thesis/dissertation research, in conjunction with work being done in another course (with the permission). The study should be original research, with the aims of making a scientific contribution to the literature and eventual publication in a scientific outlet. Given the time constraints imposed by the semester, I recommend that students rely on publicly accessible datasets (e.g., the Health and Retirement Survey, the National Health Interview Surveys, the National Longitudinal Surveys, Americans’ Changing Lives, and Aging, Status, Sense of Control). Other rich datasets are available from NACDA and ICPSR, two major electronic data archives. Students are required to discuss their topic with me before moving forward. Students should submit 1-page abstracts. . I will use the abstract to gauge the paper’s topic  relevance to the course aims, and as a basis to provide critical feedback to refine conceptual ideas and focus the analysis.

1) Poster presentation:  These sessions will give students a chance to present their work to their colleagues, to field comments,and to refine their ideas and analysis prior to submitting the term paper. Posters should followthe format that one would use if presenting the results at a major scientific meetings such as the Population Association of America. However, you do not actually need to construct an actualposter. Poster “pages” should be constructed using Powerpoint for presentation to the class.

(30% of grade)

2) Term paper  should aim to be a publishable scientific work suitable for submission to a scientific journal such as Demography. Students should review articles in Demography or other major journals for guidance on presentation styles and organizational format. (50% of grade)

DESCRIPTION: 

This course provides an in-depth introduction to survival methods for the analysis of change in time-dependent discrete dependent variables. The course draws on methodological and empirical research from the social sciences. Special attention is directed at the relationship between theories of social change, life course attainment and dynamic models. Moreover, we will examine alternative data collection strategies to obtain longitudinal data. Examples and homework assignments will draw principally on three data sets: the German Life History Study (GLHS), the National Longitudinal Survey of Older Men, and the National Health Interview Surveys. 

The course begins by focusing on simple stochastic processes for qualitative variables. We do so through the lens of non-parametric, discrete-time life table techniques. The topics include basic life table relations, single- and multiple-decrement life table models, and increment-decrement life table models. The purpose is to learn how to describe increasingly complex dynamic social processes. We then turn to continuous-time, discrete-state models. Parametric and partially parametric models will be introduced which allow for the dependency of rates on both explanatory factors and time. We will briefly examine some special topics including competing risk models, local hazard models, unobserved sources of heterogeneity in hazard models, and continuous state-space hazard models. The purpose of this aspect of the course is to communicate the variety and power of multivariate hazard models for hypothesis testing of social science processes. 

PREREQUISITES: 

Students should have completed the graduate statistics core sequence in sociology (or the equivalent) and have a thorough knowledge of the general linear model. A background in log-linear analysis and logistic regression is important, because hazard models are part of the class of log-linear models. Students must have a solid knowledge of a major spreadsheet package (e.g., EXCEL) for computing and graphing. 

REQUIREMENTS: 

In addition to reading required materials, attending class sessions, and contributing in class, students must complete three sets of written assignments. First, homework occasionally will be assigned. Homework assignments will involve both numerical problem solving and critical writing. The second and third written requirements are the mid-term and final exams. Both exams will be take-home exams involving longitudinal data analysis. The basic aim of the exams is to educate you about how one goes about formulating, estimating, and interpreting event history models. Grades will be based on the two exams (45% each) and the homework 

(10%). I also will evaluate the homework assignments to identify concepts or techniques that are problematic. 

I am also offering the option of writing an empirical paper using event history techniques in lieu of taking the final exam. If you choose this option, you will need to inform me immediately after the mid-term exam. 

READINGS: 

The technical literature on survival analysis is voluminous, spanning a variety of fields. For example, partial likelihood hazard models were developed primarily within biostatistics, while the much of the development of parametric hazard models took place within sociology and economics. Starting in the late 1980s, a literature emerged which places these different traditions within a more general survival analysis framework. A major work in this vein is the book, Event History Analysis (Blossfeld, Hamerle, and Mayer, 1989). This text also shows how to estimate hazard models using a variety of software packages (e.g., SAS). More recently Techniques of Event History Modeling was published (Blossfeld and Rohwer, 1995, 2002); this is a comprehensive introduction to continuous-time event history techniques. The 2002 2nd edition is the core text for this course. 

The textbook has been ordered by the UT bookstore. The full title is: 

Blossfeld, Hans-Peter, and Götz Rohwer. 2002. Techniques of Event History Modeling. Second Edition. Mahway, New Jersey: Lawrence Erlbaum Associates. 

If you are a Stata user and plan to use event history models extensively in your work, you might want to consider ordering a version of this text that uses Stata to generate the examples shown in the text. The text is: 

Blossfeld, Hans-Peter, Katrin Golsch, and Götz Rohwer. 2007. Event History Analysis with Stata. Mahway, New Yersey: Lawrence Erlbaum Associates. 

Course Description

This course provides an in-depth introduction to the major conceptual frameworks and empirical research examining distal and proximate social factors influencing adult health. The underlying question is – “Does the body forget insults and exposures from adverse social as well as beneficial conditions over the life course?” Close attention is paid to the intersection between the etiology of health conditions and the major life-course trajectories of socioeconomic status and family. Of major importance will be the assessment of key social and biological pathways linking social conditions at particular points in the life course with adult health outcomes.

REQUIREMENTS: The course is organized as a seminar. Most of the instruction takes place in the context of guided discussion and exchanges focused on the readings. Students are expected to attend each class and to have read all of the assigned material thoroughly and critically. Reading critically means not only being able to describe the content of an assigned piece, but also to evaluate the scientific issues motivating the research question, the conceptual/theoretical strengths of the research, the adequacy of the study’s methods, the scientific implications of the study’s findings, and the study’s relation to other course material). Students are strongly encouraged to contribute to the discussion both their insights and/or questions from the readings and their own experiences.

There are two major tasks to be accomplished in this graduate seminar: 1) The first task involves preparation of the reading for class presentation and discussion. All students are responsible for all of the assigned readings. Class discussions will be oriented around informal student presentations based on the assigned readings. For each class meeting, a student will be assigned an article, and the student will informally present the scientific factors motivating a study, the key gaps in knowledge being addressed, the major findings, and the conceptual/methodological strengths and weaknesses of the study. An outline of the presentation’s key points (1 page) should be distributed to the class on the Tuesday prior to class. The informal presentations should not be longer than 15 minutes – and 10 minutes is desirable! (20% of grade)

2) The second task is an empirical study on a topic that is relevant to the course’s overall aim. Two assignments are required in accomplishing this task: a poster presentation (30% of grade) and a term paper (50% of grade). The empirical study may overlap with research being done either as part of a student’s thesis/dissertation research, in conjunction with work being done in another course (with the professor’s permission), or as part of a student’s RA assignment (with the professor’s permission). The study should be original research, with the aims of making a scientific contribution to the literature and eventual publication in a scientific outlet.

Given the time constraints imposed by the semester, I recommend that students rely on publicly accessible datasets (e.g., the Health and Retirement Survey, the National Health Interview Surveys, the National Longitudinal Surveys, Americans’ Changing Lives, and Aging, Status, Sense of Control). Other rich datasets are available from NACDA and ICPSR, two major electronic data archives.