SRC Summer Internship Program 5th Annual Symposium

1
SRC Summer Internship Program5th Annual Symposium

• Tuesday
• July 29, 2008
• Noon-200 p.m.
• ISR Building, Room 6050

The Survey Research Center is an equal
opportunity employer that 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
• Economic Behavior Program (2)
• Family and Demography Program
• Social Environment and Health 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
Factors Related to Role and Emotional Functioning
of Air Force Personnel
Social Environment and Health Penny Pierce, PhD,
Col., USAFR Amiram Vinokur, PhD
Douglas Roehler University of Michigan
5
Overview

• Background
• Objective
• Population
• Role and Emotional Functioning
• Results
• Key Findings
• Similarities
• Differences
• Future Research

6
Background

• Work, Family, and Stress Study assesses readiness
  and deployability of Air Force servicemen and
  women
• Models of stress, coping, resource conservation
  (gains/losses), and retention tested
• Interactive effects of job and family,
  deployment-related, and organizational stressors
  were all studied

7
Project Description

• Objective
• Gains, losses, and social support were
  investigated to see if they are predictive of
  role and emotional functioning in Air Force
  personnel.
• Relationships were explored for Air Force
  personnel that are highly committed to the
  service and for those reporting a low commitment
  to the service.

(photo courtesy of the Spring City Chronicle,
October 14, 2006)
8
Sample
9
Role and Emotional Functioning
RE Functioning determines the capacity for daily
life management. Questions included (Caplan et
al., 1984)
10
Variables

• Control Variables
• Sex
• Deployment Status
• Dependent Child Status
• Rank
• Theater of War

• Predictor Variables
• Social Support
• Gains
• Losses

11
Results
12
Similarities

• For both high and low commitment groups
• increased social support related to increased
  role and emotional functioning

13
Results
14
Similarities

• For both high and low commitment groups
• increased social support related to increased
  role and emotional functioning

• increased reports of losses related to decreased
  role and emotional functioning

15
Results
16
For the high commitment group, greater reports of
gains were related to increased role and
emotional functioning.
17
Results
18
Questions for Future Research

• Why are gains unrelated to role and emotional
  functioning in the low commitment group?

• What factors influence Air Force personnel to
  develop and/or sustain commitment to the service
  and what erodes commitment?

19
Acknowledgements

• Special thanks to
• Col. Penny Pierce
• Dr. Amiram Vinokur
• Dr. Lisa Lewandowski-Romps
• Mrs. Lillian Berlin
• Mrs. Susan Clemmer
• Mrs. Elli Georgal
• George Myers Anita Johnson
• This research is supported by the Tri-Service
  Nursing Research Program

20
July 29, 2008
Money Resource Allocation Child Quality
Xuanzhong Wang Frank Stafford, PhD Panel Study
for Income Dynamics
21
Overview

• Project Description
• Research Questions
• Primary Results
• Methodology/Procedure
• Results
• Conclusion

22
Research Questions

• What has affected the amount of money resource
  allocated to children?
• How parents allocate money resource when there
  are more than one child in the FU?
• Age, Sex, Ability?

Primary Result

• Strong association between childs ability and
  school related expenditure

23
Analytic Sample
24
Data

• Child data CDS-I CDS-II
• Standardized Woodcock Johnson Test (WJR) Score
  (CDS-I CDS-II)
• School related expenditure sum of school cost,
  private lessons, school supplies (CDS-I CDS-II)
• Total expenditure sum of school related cost and
  all other cost (food, clothes, vacation and etc.)
  (CDS-II)
• Family Income data PSID Core
• Family income in 2002
• Variation in family income over the past 10-15
  years

25
Methodology/Procedure

• Child data were merged with Family Income data
  using FIMS
• Sibling information gathered for the subsample
• Complex Survey features incorporated with the
  STATA svy command
• Standard error estimation using linearization

26
Assumptions

• Ability of a child can be measured
• Standardized WJR score is an estimate of ability
• No imputation for missing data
• i.e. case wise deletion in analysis

27
Four Ability Brackets(Based on 2002 WJR Score
Distribution)

• 1 Least Capable
• (Standardized WJR Score in the 1st Quartile)
• 4 Most Capable
• (Standardized WJR Score in the 4th Quartile)

28
Mean Expenditure Comparison
Most Capable
Moderately Capable
Entire Population
Less Capable
Least Capable
29
Regression Results
30
Results

• Strong association between childs ability and
  school related expenditure
• Hard to conclude causation either way
• - use WJR score in 1997 as a predictor
• (No significant effect)
• - use of instrumental variables
• (No significant effect)
• - regress WJR score(02) on expenditure(97)
• (No significant effect)
• Other factors might have an effect too

31
Consider Siblings Ability
32
Consider Siblings Ability
33
Regression Results
34
Conclusion Further Issues

• Parents prefer more equal development among
  children
• - Ability of children in the same family is
  usually correlated
• - Choice of having less children
• Expenditure by family is only part of the money
  resource allocated to children
• Cost of Living

35
Many Thanks to
36
Many Thanks to

• Professor Frank P. Stafford
• Steven Heeringa, Patricia Berglund Brady West
• George and Anita
• Fellow Interns
• All of You )

37
Questions ?
If you have any further questions, please feel
free to E-mail me wangxz_at_umich.edu
38
Retirement Timing and Factors Leading to
Premature Retirement

• Presenter Fan Fei
• Sponsor Professor Charles C. Brown
• Health and Retirement Study (HRS),
• Economic Behavior Program

39
Outline

• Background
• Financial loss due to premature departure
• Factors leading to premature departure
• Size of their pensions worth the wait?
• Lack of knowledge about their own pensions
• Health status unwilling departures?
• Early out windows?

40
Health and Retirement Study(HRS)

• Begun in 1992, a nationally representative study
  of over 22,000 individuals age 50 and older and
  their spouses
• Longitudinal design, conducted every 2 years,
  tracked the respondents until their refusals or
  deaths
• Current director Prof. David Weir
• http//hrsonline.isr.umich.edu/

41
Defined-Benefit Pension Plans

• Employer-provided social security
• Pension benefit f (salary, years of service)
• Normal Retirement Age (NR)
• Early Retirement Age (ER)
• Pension benefits greatly increase at ER, creating
  strong incentive for people to leave at or past
  ER, and not before!

42
Data and Sample

• HRS (Health and Retirement Study) core data
• Identify those with pension on their jobs in 1992
  HRS cohort and track them until they leave their
  employers
• 1992 HRS Pension Present Value Database
• Based on employer-provided pension descriptions
  and respondents reports of salary and years with
  employer
• Contain calculations of present value of pensions
  at certain ages, workers early retirement age
  (ER) age and normal retirement (NR) age
• ONLY defined-benefit plan owners, with
  non-missing values for key measures

43
Financial Loss from Leaving Before ER

• N 249
• Approximate losses as we dont have PV at every
  age

44
Why do people leave before ER?Explanations of
Premature Departures

• Are their pensions smaller than those workers who
  leave at/after ER?
• Do they understand their pensions?
• Pension type
• Early retirement age (ER)
• Did they have health problems that might lead to
  involuntary premature departure?
• Did they leave because of the early out windows
  provided by employers?

45
Factor 1 Size of Pension

• Size of the pensions are the magnitude of
  incentives.

46
Factor 2 Knowledge of Pension

• Compare the survey answers and employer-provided
  pension information.
• Mismatching is a strong indicator of ones lack
  of basic knowledge of his/her own pension.

47
Factor 2 Knowledge of Pension

• Not knowing ER correctly can lead to unwise
  retirement timing .
• Compare 1) Workers reports of their ERs in
  survey AND
• 2) More reliable ER calculated from
  employers pension descriptions
• The early-leaving group showed
• Significantly-lower percentage in exact
  matching
• 20.46 vs 29.80
• Significantly-higher percentage in group -5, -1
  (worker-reported ERs were one to five years too
  low than actual ERs)
• 32.83 vs 14.07
• Caveats

48
Factor 3 Health Status

• Health problems might force people out
  prematurely.
• Focus on the health status and health change in
  the wave people left their 1992 employers.
• Controlling for age, we found those left before
  ER reported significantly poorer health status.
• Few significant differences found in health
  change (in past two years) measures.

49
Factor 4 Early Out Window

• EOW Special financial reward packages to
  stimulate retirement. Offered when firms want to
  downsize.
• From the sample of 1433 individuals, we found
  many EOW takers were already past their ERs. So
  we lack evidence to claim early out windows
  stimulate premature departures.

50
Factor 4 Early Out Window
Percentage of retirees accepting EOW among
total retirees in each wave
51
Limitations and Future Studies

• Limitations
• Sample size
• The limitations of the present value database
• Future studies
• Study of measurement error in present value
  database
• Involuntary departurelayoffs?
• Linking with spouses whether workers leave
  prematurely because their spouses pensions are
  so big that they can disregard their own.
• Patterns of later cohorts

52
Acknowledgements

• Prof. Charles Brown, Prof. David Weir, Prof.
  Helen Levy
• Mary-Beth Ofstedal, Jessica Faul, Miles Putnam,
  David Knapp, Ken Kashiwase
• Carol Bowen, Joyce Sisung, Janet Keller, Becky
  Bahlibi and other HRS staff
• Fellow interns
• Special thanks to George Myers III and Anita
  Johnson

53
How We Age Examining Psychological Profiles in
Midlife and Old Age

• Frank Infurna
• Jacqui Smith, PhD
• Health and Retirement Study

54
Overview

• Theory
• Systemic-Wholistic perspective
• Methodology
• Health and Retirement Study (HRS)
• Domains of functioning in the analysis
• Results
• Midlife
• Old Age
• Conclusions

55
Systemic-Wholistic Perspective

• Individual is an integrated whole comprised of
  multiple domains that are interrelated
• Cluster Analysis
• Form of classifying people on a multidimensional
  level
• Empirical bottom-up classification
• Individuals are grouped by similar
  characteristics
• Theoretical interpretation- desirable vs.
  undesirable
• Objective Use this approach, coupled with HRS
  data to examine psychosocial profiles

56
Health and Retirement Study

• A nationally representative study of 22,000
  persons age 50 and older and their spouses
• Longitudinal Study begun in 1992 and conducted
  every 2 years
• 2006- Introduced Psychosocial Questionnaire
• New questions included.
• Personality, Self-Rated Beliefs, Social
  Relationships, Lifestyle, Well-Being, Work

57
Variables

• Psychological Domains

• External Correlates

• Cognition
• Total Recall
• Serial 7s
• Personality
• Neuroticism
• Extraversion
• Conscientiousness
• Self-Related Beliefs
• Internal/External Control
• Purpose in Life
• Social Relationships
• Positive/Negative Support
• Loneliness
• Social Network

• Age
• Gender
• Education
• Well-Being
• Life Satisfaction
• Subjective Health
• Health
• Health Conditions
• Total Limitations

58
Sample
59
Profiles

• Cluster 1 (28.5, 55.81)- General Positive
  Profile- Successful Agers, excelling in all
  domains
• Cluster 2 (33, 51.95)- Moderately Positive
  Profile- Cognitively fit, in control, purpose
  driven, little constraints, socially engaged
• Cluster 3 (19, 47.63)- Average Profile- Poor
  cognition, easing through life, average support
• Cluster 4 (19.5, 43.78)- General Negative
  Profile- Average cognition, but withdrawn,
  lonely, not in control of life

60
Cluster CompositionMidlife
Post hoc interpretation of profile desirability
61
External Correlates
62

• Cluster 1 (15.7, 56)- Successful Agers
  Excelling in all domains
• Cluster 2 (12.1, 54)- Poor cognition,
  psychological vitality, supported and engaged
• Cluster 3 (14, 52)- Cognitively fit, reserved,
  no constraints, supported and engaged
• Cluster 4 (4.3, 50)- Average profile, perceived
  constraints, supported and engaged
• Cluster 5 (14.8, 50)- Average profile, poor
  cognition, easing through life, contented
• Cluster 6 (10.2, 49)- Easing through life,
  withdrawn, lonely
• Cluster 7 (11.5, 48)- Average profile, average
  cognition, neurotic, not in control
• Cluster 8 (12.8, 44)- Poor cognition, withdrawn,
  not in control, but supported
• Cluster 9 (4.6, 40)- Average cognition, little
  control, negative support, lonely

63
Cluster CompositionOld Age
Post hoc interpretation of profile desirability
64
External Correlates
65
Conclusions

• Midlife and Old Age profiles have similarities
  and differences
• Greater heterogeneity in old age than midlife
• Least desirable profiles in Midlife and Old Age
  associated with poor well-being and health
  outcomes
• External correlates help to verify psychological
  profiles
• Similar to Berlin Aging Study (BASE)
• Final Thought Interesting to see if there is
  mobility between profiles over time Survival
  outcomes?

66
Acknowledgements

• Special Thanks to..
• Dr. Jacqui Smith
• Dr. Lindsay Ryan
• Aneesa Buageila
• George Myers and Anita Johnson
• SRC Summer Interns

67
Questions?

• Feel free to contact me if you have further
  questions
• fji102_at_psu.edu

68
Chronic Illness, Stress, and the SelfAre
Chronically Ill People Also Chronically Stressed?
Leslie Rott Life Course Development Program
69
Why Study Chronic Illness Across the Life Span?

• Illnesses that are prolonged, do not resolve
  spontaneously, and are rarely cured completely
• E.g. Heart disease, cancer, diabetes, arthritis
• 1 out of every 10 Americans lives with chronic
  illness
• Leading cause of death and disability in U.S.
  one-third of mortality before age 65

70
Previous Research Findings

• Findings
• Chronic Illness
• Daily Stress
• Personal Resources

71
Previous Research Cont.

• Main contributions of this study
• Many chronic illnesses
• Young and older adults
• Chronic illness predicts stress
• Focus on personal resources

72
Buffering Model
Personal resources
Chronic illness
Stress
73
Research Questions

• 1. Does chronic illness predict concurrent
  levels of stress?
• H1 chronic illness gt stress
• 2. Does chronic illness predict stress 12 years
  later?
• H2 chronic illness gt stress longitudinally
• 3. Do personal resources buffer the influence of
  chronic illness on stress concurrently and 12
  years later?
• H3 self-esteem and self efficacy will
  buffer the influence of chronic
  illness on stress

74
Participants
Social Relations and Mental Health Over the Life
Course Study - Waves 1 and 2 (1992 and 2005)
75
Measures

• Chronic Illness
• 0 (Not Ill), 1 (Ill)
• Stress Bothered by
• Interpersonal parents, child, relative, friend,
  co-worker
• Physical - health, sex life, appearance, physical
  ability
• Work job, workload, job security, job goals
• Money cost of living, unexpected expenses,
  investments, economizing
• Environmental political/social/legal issues,
  group affiliations
• Personal Resources
• Self-esteem 10-item measure using Rosenberg
  scale
• Self efficacy 23-item measure
• Controls Gender, Race, Age

76
Chronic Illness
Percentage of Sample Reporting Specific Type of
Illness 15.6 High Blood Pressure 15.4 Arthri
tis 10.1 Heart Problems 7.9 Lung
Problems 6.6 Eye Problems 6.5 Stomach
Problems 6.1 Bone Problems 5.3 Diabetes
The analysis does not include those reporting
injuries, mental illness, or allergies
This only includes illnesses reported by 5 or
more of the chronically ill sample
77
Does stress vary by chronic illness?




plt.01
78
Does stress over 12 years vary by chronic illness?


plt.05 plt.01
79
Buffering effect of self-esteem for interpersonal
stress and chronic illness
plt.10
80
Buffering effect of self-esteem for physical
stress and chronic illness
plt.01
81
Buffering effect of self-esteem for environmental
stress and chronic illness
plt.10
82
Buffering effect of self efficacy for
interpersonal stress and chronic illness
plt.05
83
Buffering effect of self efficacy for work stress
and chronic illness 12 years later
plt.10
84
Implications

• Significant interactions between chronic illness
  and personal resources
• Self-esteem buffers the effect of chronic
  illness only for physical stress in wave 1
• Buffering was not shown for other types of
  stress why?
• Chronic illness trumps personal resources as a
  more accurate predictor of certain types of
  stress

85
Areas for Future Research

• More frequent assessments
• Qualitative analyses
• Physician reported disease status
• Specific Diseases and Co-morbidities
• Age issues
• Other psychological outcomes

86
Acknowledgements

• Dr. Kira Birditt
• Dr. Toni Antonucci
• Life Course Development Staff
• George Myers and Anita Johnson
• SRC Summer Interns

87
Who are we calling today? Examining Demographics
of Attrited and On-Time Respondents
Michele Dunsky Dr. Jennifer Barber, Ph.D. Dr.
Bill Axinn, Ph.D. Family and Demography Program
88
Outline

• Project Description
• Data Collection
• Research Questions
• Data Analysis
• Preliminary Findings
• Select Cross-tabulations Chi-Squared tests
• Select Logistic Regressions
• Conclusions

89
The Michigan Study of Young Women (MSYW)
Description

• Investigate factors affecting both intended and
  unintended pregnancies and birth rates
• Administer surveys covering diverse topics to
  increase understanding of unintended pregnancies

90
Data Collection

• Approximately 1,000 (18 and 19 year old) women
  sampled in a Michigan county
• Selected from public records
• Face-to-face baseline enrollment interview
• Longitudinal weekly surveys (30 months)
• Semi-structured face-to-face interview

91
Unique Study

• Weekly survey
• Reduce retrospective reporting
• Mixed methods
• Face-to-face survey
• Telephone survey
• Online survey
• Qualitative aspects
• Semi-structured interviews with subset of Rs who
  experience a pregnancy subset who do not

92
Daily Reminder Calls

• Days 7-9 Three automated reminders
• Day 10-11 Two lab (SSL) calls
• Day 12 First FamDem staff contact
• Day 19 Second FamDem staff contact
• Day 24 Third FamDem staff contact
• Day 30 Refusal packets sent
• Every 12 weeks Reward packets sent

93
Research Questions

• Among those who have completed the first survey,
  who drops off?
• Among those who have completed two or more
  surveys AND have not dropped off, who always does
  surveys on time?

94
Data Analysis

• Baseline (BL) and journal (J) datasets
• Dependent variables
• 20 days late (1/0)
• All surveys on time (1/0)
• Independent variables
• Sociodemographic characteristics (BL)
• Sex, contraception, and pregnancy (BL)
• Relationship information (BL)
• Interviewer id and observations (BL)
• Contact information provided and reminder mode
  (J1)
• Other (e.g., depression, stress, partners race)

95
Preliminary findings

• 17.9 dropped off
• N 234
• 34.3 all surveys completed on time
• N118

96
Cross-tabulations Chi-Square testsdropped off
respondents
p.01
N234
97
Cross-tabulations Chi-Square testsdropped off
respondents
p.004
N234
98
Cross-tabulations Chi-Square testson-time
respondents
p.025
N234
99
Cross-tabulations Chi-Square testson-time
respondents
p.049
N234
100
Select logistic regression results (coefficients)
101
Select logistic regression results (coefficients)
102
Conclusion

• In summary
• Most respondents (82.1) are continuing to
  complete surveys
• Many respondents (34.3) are completing surveys
  on time
• Next steps include
• Investigating changes in respondents lives
• Analyzing new variables

103
Implications

• Refine surveys to prevent Rs from dropping off
• Alternative methods for Rs to access surveys
• New strategies for interviewers to contact Rs
• Redefine on-time to 75 of surveys completed
  on-time

104
Acknowledgments

• Jennifer Barber
• Bill Axinn
• Yasamin Kusunoki
• Heather Gatny
• Latasha Robinson
• George Myers
• Anita Johnson

105
Questions?
If you have any further questions feel free to
contact me mbdunsky_at_umich.edu
106
Meet the Internshttp//www.isr.umich.edu/sip