SRC Summer Internship Program 3rd Annual Symposium

1
SRC Summer Internship Program3rd Annual Symposium

• Tuesday, July 25, 2006
• Noon-200 p.m.
• ISR Building, Room 6050

The Survey Research Center is an equal
opportunity employer who values diversity in the
workplace.
2
Agenda

• Welcome
• Coordinators
• Background
• Overall Purpose of Symposium
• 10 Minute Presentations (wide spectrum of
  topics)
• Symposium Format
• General Q/A

3
Acknowledgements

• Sponsors
• Health and Retirement Study
• Life Course Development Program (2)
• Social Economic and Health Program (2)
• Quantitative Methodology Program
• Partners
• Senior Staff Advisory Committee
• SRC Diversity Committee
• Summer Institute
• Survey Research Operations
• Inter-university Consortium for Political and
  Social Research
• ISR and SRC Human Resources
• SRC Computing
• SRC Directors Office

4
Joslyn M. GainesAir Force Study

• Penny Pierce, PhD, RN, Col, USAFR
• Amiram Vinokur, PhD
• Social Environment and Health

5
Ohio welcomes back the 211th Maintenance
Company upon their return from Iraq - Google Video
6
Womens Veterans Project

• Sample consists solely of Air Force women who
  were deployed to various military operations
  since March of 2003.
• Purpose of the study is to determine the effects
  of deployment experiences as well as civilian
  work and family on the womens physical and
  mental health and their likelihood to remain in
  the Air Force.
• Commonly referred to as OIF for Operation Iraqi
  Freedom. About half of the sample was deployed to
  OIF and half were deployed elsewhere.

7
Work, Family, and Stress Deployment, Resilience,
and Retention

• Very similar to OIF with the exception that men
  are included in the sample. Also this sample
  tracks Air Force personnel who have been deployed
  since October of 2001.
• Half of the sample are personnel deployed to the
  theatre of war and half are deployed elsewhere.
• Commonly referred to as READI.
• Similar UK study being done, but there are
  significant differences.

8
Current Procedure
9
Problems

• Military personnel are very mobile.
• Contact info can be outdated very quickly.
• Deployments increase response problems.
• Respondents simply wont answer the phone.
• Families may be wary of releasing information or
  even speaking about enlisted individuals.

10
Solutions

• Various Tracking Methods
• Insight Collect.
• Internet Searches.
• Base Locators.
• Determined Callers
• Nag Calling.
• Calling Methods.
• Theres always someone at ISR.
• Dana, Isabella, Jessica, and Mona.

11

• The trouble with research is that it tells you
  what people were thinking about yesterday, not
  tomorrow. Its like driving a car using a
  rear-view mirror.
• Bernard Loomis

12
Real-Time Research

• Anthrax Shot - end it! People have gotten really
  ill. People have not said anything out of fear of
  confidentiality.
• Mandatory anthrax shots were very controversial
  studies should be sure to look into the issue
  because a lot of people got out of the military
  to protest it huge health issue of this time
  period."

13
Real-Time Research
14
Real-Time Research

• The phrase, If the Air Force wanted you to have
  a family, it would issue you one! has been shown
  increasingly to be true, where it is clear that
  if you have a family and give them priority in
  your life, you cant continue to serve.
• Reservists are not getting the help they need
  and deserve, not the same treatment as Active
  Duty members.

15
Real-Time Research
16
Real-Time Research

• I have trouble reconnecting with my children
  (ages 3 and 7) especially the 3 year old. I left
  when she was 1 year old.
• I have major concerns about how my deployment
  will affect my kids later in life. I hope that
  they will always see the positive side of it
  mom defended our country in war.

17
Real-Time Research
18
The True Value of Survey Research
19
Thank You!

• SRC Summer Internship Program
• George Myers, Anita Johnson, Fellow Interns
• SEH Staff
• Amiram Vinokur, Penny Pierce, Susan Clemmer, Lisa
  Lewandowski-Romps, Lillian Berlin, Elli Georgal,
  Brianne Ott, Jumoke Johnson, Courtney Baarman

20
Jionglin Wu

• Tailoring Treatments to Individual Patients

Quantitative Methodology Program Advisor Susan
Murphy, PhD
21

• Main Points
• Motivation heterogeneity of treatment effects to
  individual patients.
• Statistical learning and Q-learning algorithm.
• Computer simulations.

22
Motivation Heterogeneity of Treatment Effects

• Dimensions
• Difference in baseline risk/ risk without
  treatment.
• Treatment may only be worthwhile for patients
  with poor prognosis.
• Responsiveness to the treatment.
• Absorb a drug rapidly, metabolize it slowly , or
  have a high concentration of functional drug
  receptor.
• Vulnerability to adverse side effects.
• Related to intrinsic biological characteristics
  of different patients.
• Utilities for different outcomes.
• Patients vary on how they trade off side effects
  vs. reduction in symptoms.

23
Our Practical Goal

• Individualized treatment to patient is a process
  involving multiple time points with a combination
  of various medications.
• The data we use are from randomized trials that
  randomly assign people to treatments whenever
  they are altered.
• Develop an estimation method that use those data
  to determine tailoring of individualized
  treatments.

24
Statistical Learning

• Q-learning is a statistical algorithm that is
  frequently used to tackle maximization problems
  involving multiple decision points whenever
  treatments are altered.
• We use Q learning to construct individualized
  treatments with the goal of maximizing the
  expected patient benefit.
• Q learning use regression for every decision
  points.
• Statistical learning tools Test sample,
  Cross-validation, Training sample, Bootstrap are
  used to produce results that are reproducible.

25
Simulation Goal

• Each regression in Q-learning requires a model.
• We want to find models that maximize expected
  patient benefit.
• Find some reliable statistics to help determine
  whether any of our models are maximizing the
  expected patient benefit.

26
Computer Simulation

• We test statistics using simulated data.
• Generative model We generate 3,000 set of
  patient response and characteristics data with
  each of size 300.
• Fitted model We engineered 2 false models and
  the correct model for contrast.

27
True Expected Benefit Under 3 different models
using 100000 Test sample
Expected Benefit
28
Estimated Expected Benefit Under 3 different
models using Cross-validation
Expected Patient Benefit
29
Using mean and variance to choose the best model
30
Future Work

• Multiple Time Points.
• Integration of information such as mean and
  variance.

31
Acknowledgements

• Susan Murphy, Daniel Almirall, and Hasan Cheema
• George Myers and Anita Johnson
• All the ISR interns

32
Jennifer Swayne
Jail Recidivism
Social Environment and Health Program
33
Introduction

• The goal of this research project is to answer
  the following overarching questions about how
  jail recidivism operates within the penal system
• Does jail overcrowding affect sentence length?
• Do longer sentences make recidivism more or less
  likely?

34
History of Criminology

• Deterrence Theory influenced by two theoretical
  frameworks
• Classical Criminology
• Role of legislatures.
• Role of judges.
• Seriousness of crime and subsequent punishment.
• Punishment should be prompt and certain.
• Laws should be structured to prevent crime from
  happening.

35
History of Criminology

• Positivist Criminology
• - Emerged as a response to classical
  criminology.
• - Changes in punishment policies alone would not
  change crime.
• - Classical Code does not allow for individual
  differences such as mental condition, age,
  repeats in offenses, and other extenuating
  circumstances.

36
Criminal Justice Process

Preliminary Exam
Plea Bargain
Pretrial
Arraignment
Trial
1 night jail
Sentencing
Crime committed
Review Hearings
37
Jail Data Collection

• The sample consists of data that either
  correspond to five overcrowding emergencies or
  five comparison time periods in 2002.
• 1) Jail Database collection of information
  from the Washtenaw County Jail.
• 2) Prosecutors Data- collection of information
  on the same individuals listed in the jail
  database, as well as offenders not included in
  the jail database.

38
Jail Project Measures

• Selection of Offenders
• - Domestic Violence Assault and Battery
  (misdemeanor and felony offenses).
• - Drug Use, Possession, Delivering,
  Manufacturing, etc. (misdemeanor and felony drug
  crimes).
• - Non-support (Child support payments).

39
Jail Data Method

• Instrumental Variables (IV) Method
• - quasi-experimental method where variable Z
  (overcrowding) is an IV for the causal effect of
  X (sentence length) on Y (recidivism).
• X (sentence length) Y (recidivism)
• Z (Overcrowding)

40
Next Steps

• Learned that we have to examine more than short
  vs. long sentences because
• Short vs. long is not random.
• Those who have longer sentences are more likely
  to recidivate.
• What differences are there between people who
  commit more severe crimes and those who commit
  less severe crimes?

41
Implications

• Causes of recidivism
• Effect of jail on various people
• Provide better research

42
Conclusion

• - Further study is needed to determine the
  relationship between overcrowding and sentence
  length on recidivism.
• - The researchers on this project will continue
  assessing the various associations between
  overcrowding, sentence length, and recidivism.

43
Closing and Thanks

• SRC Summer Internship Program
• - George Myers
• - Anita Johnson
• Jail Data Project
• - Jeffrey Morenoff
• - Ben Hansen
• - Sarah Jirek
• -Washtenaw County Jail and Prosecutors Office

44
Corina Mommaerts

• Health Insurance Trends Among the Near Elderly
  1998-2004

Health and Retirement Study
45
Presentation Overview

• Background on the Health and Retirement Study
• Classwork and Officework
• Health Insurance Trends

46
Health and Retirement Study http//hrsonline.isr.
umich.edu/

• The leading resource for data on health and
  economic circumstances for Americans over 50.
• Sponsored by National Institute on Aging.
• Longitudinal survey of over 20,000 every two
  years.
• Data since 1992.
• Recently awarded 70 million, the largest single
  research award in UM history!

47
Officework

• ADAMS database
• The Aging, Demographics, and Memory Study.
• First population-based study of dementia 856 HRS
  participants aged 70.
• Helped creation of codebook.
• 2005 Prescription Drug Study and Participant
  Lifestyle Questionnaire
• Mail Questionnaires.
• PLQ first to use scanning techniques.

48
Classwork

• Introduction to Survey Research Techniques
• 8 weeks, created a survey over the course of it.
• Health and Retirement Study Workshop
• Weeklong.
• Morning lecture, afternoon lab.
• Used this knowledge to experiment with data and
  come up with health insurance project.

49
Health Insurance Trends

• What are the current trends in health insurance
  coverage among the near elderly?
• Important to examine due to aging population,
  rising costs of healthcare, and the worry of
  decreasing employer-based coverage.
• Complex question, much more research must be
  done.
• Data from 1998, 2000, 2002, 2004 HRS waves
• Used ages 55-64 in each wave.

50
Health Insurance Trends
New cohorts added to existing sample in 1998 and
2004.
51
Health Insurance Trends

• Each wave
• Public health insurance programs (Medicare,
  Medicaid, military plans).
• Private insurance (including employer-based,
  individually purchased, and retiree coverage
  populations).
• Among married couples, looked into spouse
  coverage by gender.
• Uninsured.

52
Health Insurance Trends
Totals 1998 7648 2000 6833 2002 6183 2004
5512
53
Health Insurance Trends
54
Health Insurance Trends
Married women tend to obtain health insurance
through their husbands much more often than men
obtain health insurance through their wives.
55
Conclusions

• The general trend is going away from private
  insurance into public and no insurance.
• Women seek insurance from their spouse more than
  men.
• Little evidence of dropping spouse coverage.

56
Future Directions

• Other analyses of the uninsured population
• Who is uninsured?
• Why do people become uninsured?
• How employer-based insurance is changing
• Prescription drug insurance

57
Acknowledgements

• George Myers and Anita Johnson
• Fellow Interns
• Lynette Hoelter and Emilia Peytcheva
• Gwen Fisher, Jessica Faul, Mary Beth Ofstedal,
  David Weir, and all of the HRS staff
• Thank you.

58
Kristina Hartman
Sibling Rivalry After All These Years
Life Course Development Program
59
Internship
Survey Research Center ? Summer Institute in
Survey Research Techniques ? Introduction to
Survey Research Techniques. ? Life Course
Development Program ? Social networks over
the lifespan and well-being. ? Assisted with
several research projects. ? Secondary Data
Analysis.
60
Sibling Rivalry After All These Years
Education Discrepancy Sibling Relationship
Quality Across the Adult Lifespan

• Acknowledgements Kira Birditt and Toni Antonucci

61
Unique Relationship

• ? Sibling relationship is unique
• ? Shared genetics, environment.
• ? Longest lasting relationship, often lifelong.
• ? Highest negativity, conflict.

62
Sibling Rivalry Background

• Rivalry the condition or fact of competing
  with somebody
• ? Sibling rivalry in childhood
• ? Individuation (Raffaelli, 1992)
• ? Parents Affection (Adler, 1931)
• ? Sibling rivalry in adulthood ? Success
• (Adams, 1968 Blood Blood, 1978)
• ? Parents
• (Aldous, Klaus, Klein, 1985 Brackbill
  Kitch, Noffsinger, 1988)

63
Importance

• ? Prior research focus on sibling relationship in
  childhood, but less is known about adult
  siblings.
• ? Sibling relationship is a resource of social
  support (Antonucci, 2001).
• ? Building from Adams, 1968, to include
• ? parental context.
• ? status of both siblings in the relationship.

64
Research Questions

• ? Does sibling relationship quality vary by
  education level of respondent and their siblings?
• ? Is the association between education level (of
  respondent and sibling) and sibling relationship
  quality moderated by the quality of parental
  relationship?

65
Method

• ? Data are from the Social Relations and
  Mental Health 10 Years Later survey conducted
  in 2005 (Antonucci).
• ? Participants include 1,076 people (433
  male, 643 female), ages 20-100 (mean51.8) from
  the Detroit metropolitan area.

66
Measures

• ? Demographics
• ? Education Level (0high school or less
  1some college or higher)
• Respondents reported their own as well as
  their siblings education level.

67
Measures

• ? Scales measuring perceptions of positive
  quality of sibling and parent relationships
• ? Positive Quality (12 items)
• Examples
• ? I enjoy being with my mother/father/sibling.
  
• ? I think that my relationship with
  my mother/father/sibling is a good one.
• Scale from 1 (strongly disagree) to 5 (strongly
  agree)

68
Results

• ? Does sibling relationship quality vary by
  education level of respondent and sibling?
• ? Analysis Strategy
• ? Two univariate ANOVAs with education level of
  respondent and sibling as independent variables
  and positive relationship quality with siblings
  as the dependent variable.

69
Sibling Education Level
When respondent has college or more education and
sibling has high school or less education, the
relationship has the lowest positivity.
70
Results

• ? Is the association between education level (of
  respondent and sibling) and sibling relationship
  quality moderated by the quality of parental
  relationship?
• ? Analysis Strategy
• ? Regression models estimated with respondent
  education, sibling education, and parental
  positivity as independent variables, as well as
  all possible interactions, predicting
  the dependent variable of sibling positive
  quality.
• ? Median Respondent-Parent Positivity 4.67

71
Low Parental Relationship PositivitySibling
Positivity as a Function of Sibling and
Respondent Education
Sibling Education Level
Sibling relationship least positive when there is
a discrepancy in level of education.
72
High Parental Relationship PositivitySibling
Positivity as a Function of Sibling and
Respondent Education
Respondent with some college education or more
and sibling with high school or less have lowest
quality relations with sibling.
73
Findings

• ? Does sibling relationship quality vary by
  education level of respondent and sibling?
• ?As hypothesized, the sibling relationship is
  significantly less positive when there is a
  discrepancy in education level, in which the
  respondent has a higher level of education than
  the sibling.

74
Findings

• ? Is the association between education level (of
  respondent and sibling) and sibling relationship
  quality moderated by the quality of parental
  relationship?
• ? Yes. When the respondent-parent relationship
  has high positivity, a college educated
  respondent and high school educated sibling have
  a significantly less positive relationship.

75
Implications Directions

• ? Education differences between siblings appear
  to affect the quality of the sibling
  relationship.
• ? Siblings may compare themselves to each
  other.
• ? The parent-adult child relationship can affect
  the relationship between that adult child and
  their sibling.
• ? Future research directions
• ? Marital status/quality of adult siblings.
• ? Income of adult siblings.
• ? Employment status of adult children.
• ? Better discrimination of relationship
  positivity rating.

76
Thank You

• ? Kira Birditt
• ? Toni Antonucci Life Course Development Staff
• ? George Myers, Anita Johnson, SRC Intern
  Program
• ? Lynette Hoelter , Emilia Peytcheva, SRC
  Summer Institute
• ? SRC Interns

77
Vontrese Deeds
Family Transitions Following the Birth of a
Sibling
Family Transition Study
78
Background

• Nearly 80 of Americas children have at least
  one sibling.
• It is not uncommon for parents to report greater
  marital conflict, depression, and physical
  exhaustion during the months following an
  infants birth.
• If mature adults experience increased stress with
  the birth of an infant, one wonders how firstborn
  children, who may in fact be toddlers themselves,
  react.

79
1 New York Times Bestseller
80
Uniqueness of Our Study

• In opposition to previous research on sibling
  rivalry, our study is
• Long-term longitudinal.
• A large sample size (200 families).
• In opposition to previous research in the field
  of child development, our study
• Focuses on the husband/father.

81
Goals of the Study

• Examine the adjustment of the older child
• Identify different change trajectories in the
  firstborn childrens first-year adjustment.
• Examine change in parental well-being and family
  relationships
• Identify the interrelations between these
  changes.
• Predict different developmental trajectories from
  prenatal assessments
• Assessments of contextual, child, and parent
  characteristics
• Address the socioemotional development of the
  infant throughout the transition to siblinghood
• Examine the consequences of the older siblings
  adjustment on the younger sibling.

82
Study Design

• Study each family longitudinally at five
  different time points
• Last trimester before siblings birth
• 1 month
• 4 month
• 8 month
• 12 month

83
Methods

• Observational
• Parent Interviews / Questionnaires
• Child Assessments
• Lab Visits

84
Observational Methodology

• Examples of video taped sessions
• Mother and baby with older sibling present.
• Father and baby with older sibling present.
• Both parents and baby with older sibling present.
• Siblings with parents present.
• Every session is coded to see the older childs
  reaction.
• Jealousy.
• Attention seeking .

85
Questionnaires/ Parent Interviews

• Questionnaires
• Marital relationship quality
• Personality
• Depression
• Parent Interviews
• Demographic information
• Division of labor
• Reports on infant

86
Child Assessments

• Most children experience the birth of a sibling
  before they are 4 years old, a significant time
  frame for the development of
• Ability to regulate behavior and emotion.
• Understanding of others emotions and minds.
• Internalization and development of conscience.
• Through different tasks and stories, we are able
  to assess the older siblings
• IQ.
• Theory of Mind.
• Emotional understanding.

87
IQ

• Task example

Point to the foot
Point to parallel
88
Theory of Mind

• Task example

89
Emotional Understanding

• Task example

Can you tell me why Jack feels both happy and sad?
90
Method Summary

• Observational
• Examine sibling jealousy.
• Parent Questionnaires/ Interviews
• Examine changes in parental well-being and family
  relationship functioning.
• Child assessment
• Examine older siblings social and emotional
  understanding.
• Lab Visits
• Examine developing infants secure attachments.

91
Hypothesis

• All firstborn children will initially experience
  some disruption over the transition period.
• The transition period will bring about change in
  the parents well-being and family relationship
  functioning.
• Greater adjustment difficulty over time will be
  associated with parent and family relationships
  which display increasing marital conflict,
  depression symptoms, and harsh punishment.

92
Implications

• Knowledge gained from this study will be
  important to our understanding of how young
  children adjust and adapt to siblinghood.
• Before the transition to siblinghood, we may be
  able use prenatal assessments to identify those
  families most at-risk for negative changes.
• This information can help create preventative
  efforts and recommendations for health care
  professionals as they work with parents expecting
  their second child and jealous older siblings.

93
Acknowledgements

• Dr. Brenda Volling
• Anne Lock, Jenn Lindsay, Kathy Speregen, Lauren
  Rosenberg
• Undergraduates involved in Dr. Vollings lab
• George Myers / Anita Johnson
• SRC Interns

94
Meet the Interns