Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Mechanisms of the effect of involuntary retirement on older adults' health and mental health
(USC Thesis Other)
Mechanisms of the effect of involuntary retirement on older adults' health and mental health
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
MECHANISMS OF THE EFFECT OF INVOLUNTARY RETIREMENT ON OLDER
ADULTS‘ HEALTH AND MENTAL HEALTH
by
Min-Kyoung Rhee
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(SOCIAL WORK)
December 2013
Copyright 2013 Min-Kyoung Rhee
ii
ACKNOWLEDGEMENTS
―Whether you think you can or think you can‘t, you‘re right.‖ -Henry Ford-
There have been times when I had doubts about myself in addressing challenges
during my doctoral years. I benefitted from the great fortune of having wonderful people
in my life who believed in me even in the times I thought I could not. I would like to
express my deepest gratitude for their care, love and guidance, and share the joy of
successfully completing my doctoral education.
I would like to give special thanks to my mentor, Dr. Michalle Mor Barak, for her
intellectual guidance and warm support throughout my doctoral years. While working
with her, I was able to expand my view of social work within the global perspective. I am
also grateful to my dissertation committee: Dr. Iris Chi, Dr. Bob Knight, and Dr. William
Gallo. With Dr. Chi‘s continuous encouragement and guidance, I was able to conduct my
first independent study collecting my own data, which was the first step in becoming an
independent researcher. Dr. Bob Knight generously served in my committee both for my
qualifying exam and dissertation, and provided professional insights as a scholar to
advance my work. I met Dr. William Gallo when I was facing one of the greatest
challenges in completing my dissertation, and his unconditional support made me jump
back into it and move forward. It was a great fortune to have him on my committee. I
would like to extend my gratitude to Dr. Aaron Hagedorn for his sincere commitment in
helping me understand the data I used for my dissertation. He kindly shared his time to sit
with me and show me the practicalities of the data.
iii
My years at University of Southern California made me grow both intellectually
and personally. At the heart of my growth, there was my amazing cohort: Dahlia Fuentes,
Ian Holloway, Rohini Pahwa, Ling Xu, and Cindy Hsiao. By sharing hard times and good
times over the last six years, we grew together. Thanks to their friendship, I was able to
come this far.
Finally, my sincere gratitude goes to my family who continuously showed
unconditional love and support. I especially thank my grandmother, Hae Bok Jeon, who
has been praying for my success every single day for years. She has been addressing me
as Dr. Rhee ever since I started the doctoral program, and finally I have become Dr.
Rhee. Now that I have become a parent myself, I have realized how blessed I am to be a
daughter of my parents, Yeon Joon Rhee and Hyang Sook Park. They constantly have
expressed their strong belief in me throughout my life and supported in every way so that
I could achieve my goals. I am truly grateful and happy that they are with me to celebrate
one of the greatest milestones in my life. I also extend my thanks to my sister, Eun-
Kyoung Rhee, and my brother, Dong-Hee Rhee, for their fraternal love and support. I
would like to express my deepest love and gratitude to my husband, Taehyun Cho, who
showed me tremendous love and respect for what I do. I also thank my lovely daughter,
Soomin Cho, who makes me a better person every day.
iv
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ii
LIST OF TABLES ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ vii
LIST OF FIGURES ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ viii
ABSTRACT ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ x
CHAPTER I: INTRODUCTION ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 1
1.1 Problem Statement ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 1
1.2 Purpose of the Dissertation ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 5
CHAPTER II: LITERATURE REVIEW ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 6
2.1 Significance of Differentiating between Voluntary and Involuntary
Retirement ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 6
2.2 Effect of Involuntary Retirement on Self-Rated Health and Mental
Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 7
2.3 Potential Mediators: Financial Control, Positive and Negative Family
Relationships, and Social Integration ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 9
2.4 Theoretical Framework: Toward a Dynamic Conceptual Model ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 12
2.4.1. Conceptual Model ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 14
CHAPTER III: RESEARCH AIMS, QUESTIONS, AND HYPOTHESES ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 16
3.1 Research Aims ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 16
3.2 Research Questions and Hypotheses ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 16
3.2.1. Research Aim 1: To Explore Characteristics of Involuntary
Retirees ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 16
v
3.2.2. Research Aim 2: To Assess Potential Mediators ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 17
CHAPTER IV: METHODS ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 18
4.1 Data Source ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 18
4.2 Sample ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 19
4.3 Measures ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 22
4.3.1 Dependent Variables ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 22
4.3.2 Independent Variable ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 24
4.3.3 Mediating Variables ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 24
4.3.4 Control Variables ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 26
4.4 Statistical Analyses ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 27
4.4.1 Preliminary Data Analyses ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 29
4.4.2 Mediation Analyses ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 30
CHAPTER V: RESULTS ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 42
5.1 Sample Characteristics ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 42
5.2 Results for Research Aim 1 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 42
5.2.1 Results for Research Question 1 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 42
5.2.2 Results for Research Question 2 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 44
5.3 Results for Research Aim 2 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 47
5.3.1 Results for Phase 1 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 48
5.3.2 Results for Phase 2 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 55
5.3.3 Results for Phase 3 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 64
5.4 Summary of Results for Hypotheses Testing ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 71
vi
CHAPTER VI: DISCUSSION ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 73
6.1 Summary of Major Findings ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 73
6.1.1 Prevalence and Characteristics of Involuntary Retirees ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 73
6.1.2 Mechanisms of the Effect of Involuntary Retirement on Self-Rated
Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 75
6.1.3 Mechanisms of the Effect of Involuntary Retirement on Mental
Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 78
6.2 Limitations and Suggestions for Future Studies ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 80
6.3 Implications ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 81
6.3.1 Implications for Retirement Policy and Research ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 81
6.3.2 Implications for Social Work Practice ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 84
6.4 Conclusion ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 85
References ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 87
vii
LIST OF TABLES
Table 1 Sociodemographic and Retirement Characteristics by Retirement Status ∙ 43
Table 2 Characteristics of Key Study Variables by Retirement Status ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 45
Table 3 Complex Sample General Linear Model Coefficients (Retirement
Transition as Binary; n = 1,280) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 49
Table 4 Indirect Effects of Retirement on Self-Rated Health through Financial
Control, Positive Family Relationships, Negative Family Relationships,
and Social Integration (paths ab) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 54
Table 5 Indirect Effects of Retirement on Mental Health through Financial Control,
Positive Family Relationships, Negative Family Relationships, and Social
Integration (paths ab) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 54
Table 6 Complex Sample General Linear Model Coefficients (Retirement
Transition as Multiple Categories; n = 1,280) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 56
Table 7 Relative Specific Indirect Effects of Voluntary and Involuntary Retirement
on Health through Financial Control, Positive Family Relationships,
Negative Family Relationships, and Social Integration (paths ab) ∙∙∙∙∙∙∙∙∙∙ 60
Table 8 Relative Specific Indirect Effects of Voluntary and Involuntary Retirement
on Mental Health through Financial Control, Positive Family
Relationships, Negative Family Relationships, and Social Integration
(paths ab) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 64
Table 9 Complex Sample General Linear Model Coefficients (Retirement
Transition as Binary; n =429) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 66
Table 10 Indirect Effects of Involuntary Retirement on Health through Financial
Control, Positive Family Relationships, Negative Family Relationships,
and Social Integration (paths ab) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 68
Table 11 Indirect Effects of Involuntary Retirement on Mental Health through
Financial Control, Positive Family Relationships, Negative Family
Relationships, and Social Integration (paths ab) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 71
viii
LIST OF FIGURES
Figure 1 Conceptual Model of the Mechanisms of the Effect of Involuntary
Retirement on Health and Mental Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 15
Figure 2 Sample Selection Flow Chart ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 21
Figure 3A Total Effect of X on Y when IV is Binary (Phase 1) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 34
Figure 3B A Multiple Mediator Model when IV is Binary (Phase 1) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 34
Figure 4A Total Effect of X on Y when IV is Multicategorical (Phase 2) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 38
Figure 4B A Multiple Mediator Model when IV is Multicategorical (Phase 2) ∙∙∙∙∙∙∙ 38
Figure 5A Total Effect of X on Y when IV is Binary (Phase 3) ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 40
Figure 5B A Multiple Mediator Model when IV is Multicategorical (Phase 3) ∙∙∙∙∙∙∙ 40
Figure 6A Total Effect of Retirement on Self-Rated Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 50
Figure 6B Direct and Indirect Effects of Retirement on Self-Rated Health ∙∙∙∙∙∙∙∙∙∙∙∙∙ 50
Figure 7A Total Effect of Retirement on Mental Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 52
Figure 7B Direct and Indirect Effects of Retirement on Mental Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 52
Figure 8A Total Effects of Voluntary and Involuntary Retirement on Self-Rated
Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 57
Figure 8B Direct and Indirect Effects of Voluntary and Involuntary Retirement
on Self-Rated Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 57
Figure 9A Total Effects of Voluntary and Involuntary Retirement on Mental Health∙62
Figure 9B Direct and Indirect Effects of Voluntary and Involuntary Retirement on
Mental Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 62
Figure 10A Total Effect of Involuntary Retirement on Self-Rated Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 67
ix
Figure 10B Direct and Indirect Effects of Involuntary Retirement on Self-Rated
Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 67
Figure 11A Total Effect of Involuntary Retirement on Mental Health ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 69
Figure 11B Direct and Indirect Effects of Involuntary Retirement on Mental Health ∙ 69
x
ABSTRACT
The purpose of the present study is to provide in-depth information on older
adults‘ experience of involuntary retirement by examining mechanisms of the effect of
involuntary retirement on self-rated health and mental health among adults aged 50 years
or older. Although it is a normative expectation to perceive retirement as a phase of life
in which older adults become disengaged from paid work, volunteer, and enjoy leisure
time with family, not everyone has the luxury of maintaining control over their retirement
decisions. Approximately one third of retirees perceive their retirement as forced rather
than voluntary. Involuntary retirees are likely to face greater challenges than voluntary
retirees during their postretirement adjustment period because they have multiple burdens
of health, mental health, and job displacement issues that may have partially led them to
involuntarily retire. The prevalence of involuntary retirement is likely to increase because
older adults are expected to work longer due to increased financial responsibility despite
the challenge of securing or maintaining employment. Although an expanding body of
research has addressed various topics of retirement including preretirement planning and
decision making, relatively few studies on retirement have focused on the voluntariness
of retirement or its varied contexts. Research that has explored health and mental health
outcomes after retirement without accounting for voluntariness has reported mixed
findings. Considering the fact that retirement has become a more complex and diverse
life transition, it is critical to consider the nature of retirement as well as its contexts.
To fill this gap of knowledge in research and practice, this study had two specific
aims: (1) to explore the prevalence of involuntary retirement among older adults and the
xi
extent to which the characteristics of involuntary retirees are different from voluntary
retirees or those who did not retire, and (2) to investigate the mechanisms of the health
and mental health effect of involuntary retirement by examining the potential mediating
effects of financial control, positive and negative family relationships, and social
integration. The research questions and hypotheses were formulated based on the life
course perspective and latent deprivation theory. Using two waves of longitudinal data
extracted from Health and Retirement Study (2006 and 2010), a final sample of 1,280
individuals working for pay at baseline who responded to a lifestyle questionnaire in both
waves was selected. Univariate, bivariate, and regression-based path analyses were
conducted using SPSS 18.0. This study employed a multiple mediation model that
considered four mediators simultaneously and the model was estimated in three phases.
Results of the study found that 29.3% (n = 429) of the sample retired between
2006 and 2010, and 37.2% (n = 155) of those individuals reported that they retired
involuntarily. Results of Phase 1, a multiple mediator model that did not account for
voluntariness of retirement using a binary independent variable (retired or not), indicated
that there was no significant direct or indirect effects of retirement on self-rated health
and mental health outcomes. In Phase 2, in which the same model was estimated using a
multicategorical independent variable (involuntarily retired, voluntarily retired, not
retired), involuntary retirement had an direct adverse health effect compared to not
retiring, whereas voluntary retirement had an indirect positive health effect via financial
control. In terms of mental health outcomes, the positive mental health effect of voluntary
retirement was mediated by financial control, whereas involuntary retirement had no
xii
significant effect. Results of Phase 3, a model that considered retirees only, revealed
direct adverse health effects of involuntary retirement compared to voluntary retirement.
Involuntary retirement also had an indirect effect on mental health via financial control.
Findings of this study indicate the significance of specifying the nature of
retirement when conducting retirement research and the need to pay more attention to
potential detrimental effects of involuntary retirement. Implications of the findings are
discussed with regard to retirement policy, research, and social work.
1
CHAPTER I: INTRODUCTION
1.1. Problem Statement
Individuals 65 years old or older comprise 13% (approximately 40 million) of the
U.S. population, a figure projected to grow at a fast rate, reaching 19.3% (72.1 million)
by 2030 (Administration on Aging, 2011; Werner, 2011). Along with their growth in
number, millions of older adults are expected to transition to retirement throughout the
coming decades as the baby boomer generation continue to retire. Retirement is one of
the significant life course transitions for older adults either completely or partially
withdrawing from their main work. Although it has been a normative expectation to
consider retirement as a one-time exit from the labor force in the past, retirement has
become a more complex and gradual process in recent years. For example, instead of
taking a traditional retirement path of withdrawing from the labor force and never
returning, approximately 20% of retirees continue to work in their retirement and more
than 50% plan to work beyond their retirement age (Brown, Aumann, Pitt-Catsouphes,
Galinsky, & Bond, 2010; Collinson, 2012; Johnson, Butrica, & Mommaerts, 2010).
Moreover, although retirement is often perceived as a phase of life during which older
adults become disengaged from paid work, volunteer, and enjoy leisure time with family,
not everyone has the luxury of maintaining control over their retirement decisions. Thus,
retirement can be very heterogeneous depending on its type (complete or partial) and
nature (voluntary or involuntary), as well as the context in which the decision to retire
was made.
2
Despite the heterogeneity of retirement experiences and an expanding body of
retirement research addressing various topics related to preretirement planning and
decision making (for reviews, see Ekerdt, 2010; Wang & Shultz, 2010), relatively few
studies have focused on the voluntariness of retirement or its varied contexts. According
to previous empirical studies, approximately one third of retirees retire involuntarily
(Lachance & Seligman, 2008; Szinovacz & Davey, 2004) and 50% of retirees leave the
workforce earlier than planned due to health problems, employer factors such as
company closure or layoffs, and personal reasons such as taking care of family members
(Helman, Greenwald, Copeland, & VanDerhei, 2012). Once older adults leave the labor
force, it becomes very challenging for them to secure employment again. Even if they do
become reemployed, they experience substantial salary reductions (Johnson & Butrica,
2012).
Involuntary retirement also can be very stressful, particularly in the current
context in which older adults are expected to work longer not only because of their
improved health status and the increasing number of less physically demanding jobs, but
also because of increased financial needs. The existing literature consistently has stated
that the primary motivation for postponing retirement or working for pay during
retirement is financial in nature (Brown, 2012; CareerBuilder, 2011). This is partially
because external conditions regarding retirement have become restrictive and thus more
financial responsibility is imposed on individuals. For example, the minimum age to
receive full Social Security benefits has increased from 65 to 67 depending on birth year
and the monthly benefit amount decreases between 7% and 30% when it is claimed
3
earlier than the full retirement age (Social Security Administration, 2013). Another
significant change in retirement income is the shift of employer pensions from traditional
defined-benefit retirement plans to defined-contribution plans. More financial burden is
imposed on the individual because the defined-contribution plan does not guarantee
retirement benefits and only fixed contributions are made to retirement accounts by both
employers and employees during employment. Some evidence has suggested that
individuals with defined-contribution plans retire later than those with defined-benefit
plans (Friedberg & Webb, 2005). Thus, leaving the labor market unexpectedly and earlier
through involuntary retirement does not allow an individual to prepare for a stable
retirement transition. In addition, involuntary retirees can be at higher risk of financial
vulnerability because they are more likely to have reductions in Social Security benefits
and have less time to accrue employment-based retirement pensions. Financial
vulnerability coupled with causes of involuntary retirement such as health, company
closure, or family may affect various dimensions of the lives of older adults. Thus,
unexpected and involuntary withdrawal from the labor force can create various
challenges for older adults in managing financial stability and maintaining a healthy life.
Research that has explored health and mental health outcomes after retirement without
accounting for voluntariness has reported mixed findings. Some studies reported that
retirement significantly increases the risk of health problems (Behncke, 2012), whereas
other studies reported no evidence of negative health effects (Bound & Waidmann, 2007)
or positive mental health effects (Johnston & Lee, 2009; Mein, Martikainen, Hemingway,
Stansfeld, & Marmot, 2003) related to retirement. In contrast, the few studies that have
4
accounted for retirement voluntariness have presented consistent evidence of the negative
effects of involuntary retirement on health and mental health (Calvo, Haverstick, & Sass,
2007; Dave, Rashad, & Spasojevic, 2007; Gallo, Bradley, Siegel, & Kasl, 2000;
Szinovacz & Davey, 2004). More specifically, involuntary retirees experienced a more
negative initial crisis following retirement compared to voluntary retirees (Floyd et al.,
1992) and had more adjustment problems as well as less retirement satisfaction (van
Solinge & Henkens, 2008). This emphasizes the importance of specifying the nature of
retirement.
Although the adverse health and mental health effects of involuntary retirement
have been consistently reported, knowledge of the mechanisms or pathways of
involuntary retirement to health and mental health remains scarce. The few studies that
have viewed retirement or unemployment in their particular contexts have reported that
financial, family, and social resources are potential mechanisms through which retirement
or unemployment shapes somatic and mental health (Kessler, Turner, & House, 1987;
Kim & Moen, 2002). Considering the complexity of relations between retirement and its
potential influence on other dimensions of older adults‘ lives, it is reasonable to examine
the mechanisms of involuntary retirement‘s effect on health and mental health. Thus, to
address this gap in the retirement literature, this study examined the mechanisms of the
effect of involuntary retirement compared to voluntary retirement and no transition to
retirement using longitudinal panel data.
5
1.2. Purpose of the Dissertation
The overall purpose of this dissertation was to provide in-depth information about
involuntary retirement among older adults and how it affects their health and mental
health. More specifically, this study aimed to explore the prevalence of involuntary
retirement among older adults and to what extent the characteristics of involuntary
retirees are different from those who retire voluntarily and those who do not transition to
retirement. In addition, this study aimed to investigate the mechanisms of the health and
mental health effects of involuntary retirement by examining the potential mediating
effects of financial control, positive family relationship, negative family relationship, and
social integration. This study also aimed to highlight the significant need to distinguish
between involuntary and voluntary retirement by examining the potential unobserved
differences among the mechanisms of both types of retirement. To achieve the purpose of
the study, a conceptual model was developed based on the life course perspective (Elder,
1974, 1994) and latent deprivation theory (Jahoda, 1982). The research questions and
hypotheses that were formulated based on the conceptual model were analyzed using
secondary longitudinal data from the Health and Retirement Study (HRS).
6
CHAPTER II: LITERATURE REVIEW
2.1. Significance of Differentiating between Voluntary and Involuntary Retirement
The aging workforce and its fiscal implications have become one of the major
political contentions in the United States during the last couple of decades, and the trend
of early retirement is no longer a reality (Ekerdt, 2010). As previously described,
retirement policies are geared toward having older adults stay in the labor force longer by
gradually extending the full-benefit age for Social Security retirement pensions from 65
to 67 (Social Security Administration, 2013). Empirical evidence also has showcased the
dramatic increase in the number of people enrolled in defined-contribution plans, from 8%
in 1980 to 31% in 2008 (Wheaton & Crimmins, 2013). Moreover, Claxton and colleagues
(2008) reported a 50% decrease in the availability of employer-sponsored retiree health
insurance among companies with 200 or more employees that offered health insurance
(as cited in Wheaton & Crimmins, 2013, p. 30). Changes in both government and
employer policies regarding retirement have imposed greater constraints and financial
pressure on older adults who are close to retirement, ultimately leading them to continue
to work beyond retirement age.
Despite environmental conditions that require a longer stay in the labor force,
some individuals have little or no control over their retirement decisions. Unlike those
who voluntarily retire due to a preference for leisure (Dorn & Sousa-Poza, 2007), a
considerable number of older adults experience involuntary retirement due to health
issues, employment constraints such as job loss or company closure, and family
responsibilities (Helman et al., 2012; Lachance & Seligman, 2008). According to
7
Lachance and Seligman‘s (2008) study, approximately 27% of retirees experience
involuntary retirement. The results of that study also indicated that factors that motivate
voluntary retirement are very different than factors that determine involuntary retirement.
For example, although poor health and job loss were two major causes of involuntary
retirement, factors such as a desire to do other things or spend more time with family
were important retirement reasons for voluntary retirees (Lachance & Seligman, 2008).
In addition, although pensions, Social Security, savings, and work preferences were
positively associated with voluntary retirement, the same factors had a much weaker
impact on involuntary retirement (Lachance & Seligman, 2008). Thus, it is critical to
consider involuntary retirement separately from voluntary retirement because they may
lead to different postretirement adjustment processes as well as different effects on health
and mental health.
2.2. Effect of Involuntary Retirement on Self-Rated Health and Mental Health
A substantial amount of research that examined the effects of retirement on health
and mental health outcomes has reported mixed findings. Some studies indicated that
retirement significantly increased the risk of chronic conditions and severe cardiovascular
diseases (Behncke, 2009), worsened self-assessed health (Behncke, 2012), increased
difficulties related to mobility and performing daily activities, and led to mental health
decline (Dave et al., 2007). In contrast, some studies reported no evidence of negative
health effects related to retirement (Bound & Waidmann, 2007) and other studies
reported positive mental health effects such as lower anxiety and distress among retirees
(Drentea, 2002; Johnston & Lee, 2009; Mein et al., 2003). However, most of these
8
studies did not distinguish between involuntary and voluntary retirement despite the fact
that voluntary and involuntary retirement can be two significantly different experiences.
On the other hand, studies that have explored the voluntariness of retirement have
presented consistent evidence of the negative effects of involuntary retirement on health
and mental health (Calvo et al., 2007; Dave et al., 2007; Floyd et al., 1992; Szinovacz &
Davey, 2004). These adverse health and mental health effects are even more strongly
documented in the unemployment literature. Involuntary retirement is very similar to
involuntary job loss or unemployment in that decisions regarding retirement are often
strongly restricted by the lack of employment options. Empirical evidence has suggested
that displaced older workers retire at substantially higher rates than nondisplaced workers,
and among displaced workers, older workers are substantially more likely than younger
workers to exit the labor force (Chan & Stevens, 2002; Farber, 2005). The negative
impact of unemployment or involuntary job loss on physical health, mental health, and
well-being is well documented in the literature (Gallo et al., 2000; McKee-Ryan, Song,
Wanberg, & Kinicki, 2005; Paul & Moser, 2009). In addition, the adverse health and
mental health effects of involuntary job loss in older age tend to last longer. According to
a 10-year follow-up study on the impact of involuntary job loss on physical health, the
risk of myocardial infarction and stroke among those who experienced involuntary job
loss increased more than twice over 10 years compared to those who were working
(Gallo, Teng, et al., 2006). In terms of mental health, evidence also has suggested that job
displacement in older age is associated with long-term depressive symptoms for
individuals with limited wealth (Gallo, Bradley, et al., 2006). Thus, perceptions of the
9
retirement transition as chosen or forced can be significantly associated with retiree
health and mental health.
2.3. Potential Mediators: Financial Control, Positive and Negative Family
Relationships, and Social Integration
Retirement is an individual process that may lead to different health and mental
health outcomes depending on various contextual indicators such as wealth, family
relationships, and social engagement. However, most existing research on the impact of
retirement or unemployment on health and mental health has focused on investigating
direct effects. A few studies that viewed retirement as an experience occurring in
particular contexts reported that financial, family, and social resources serve as
mechanisms through which retirement shapes retiree health and mental health (Kessler,
Turner, & House, 1987; Kim & Moen, 2002).
One of the most important contextual factors related to retirement is financial
status, including adequacy of retirement income such as Social Security and private
pensions (Coile & Gruber, 2007; Schmitt & McCune, 1981). In addition, having adequate
postretirement income and being more prepared for retirement are positively associated
with retirement confidence, retirement adjustment, and life satisfaction (Kim, Kwon, &
Anderson, 2005; Noone, O‘Loughlin, & Kendig, 2013; Taylor & Doverspike, 2003).
Older adults who are close to retirement currently face two contradicting environments:
financial disincentives built into Social Security and private pensions require continued
work, yet it is more challenging for them to stay in the labor force or regain employment
compared to younger workers. Thus, it can be difficult for older adults to retire with
10
adequate financial resources, especially involuntary retirees who experience abrupt
discontinuity from work without sufficient planning. Although previous studies reported
mixed findings that financial assets of older adults increase (Coile & Milligan, 2009) or
decrease (Haveman, Holden, Wolfe, & Romanov, 2005; Kim & Moen, 2002) after
retirement, research has consistently indicated that involuntary retirees or those who
experience job displacement in older age experience greater wealth inadequacy and have
less perceived financial control compared to those who do not experience job
displacement or who retire voluntarily (Baxter, 2010; Chan & Stevens, 2002; Ozturk &
Gallo, 2013). Taking into account the fact that a lack of personal control is a key factor
influencing retiree mental health or well-being (Kim & Moen, 2002) and that relative
income deprivation is associated with health status (Subramanyam, Kawachi, Berkman,
& Subramanian, 2009; Wilkinson & Pickett, 2006), financial control may be a potential
mediator in the relationship between involuntary retirement and health and mental health.
Another potential mediator is family resources and support. Considerable
empirical evidence has indicated the spillover effect of unemployment on family
members. For instance, unemployment has many negative consequences on families such
as increased marital dissolution and propensity for divorce, poor school performance
among children, and worsened well-being of spouse and children (Hansen, 2005; Rege,
Telle, & Votruba, 2011; Ström, 2003). In addition, retirement may lead to decreased
marital quality (Kim & Moen, 2002). However, families should not just be considered
victims of unemployment or retirement because they adapt to challenges. In fact, there
are indications in the literature that a strong marital relationship softens the effect of
11
economic hardship (Elder, Conger, Foster, & Ardelt, 1992). Family members can provide
social and psychological continuity to retirees as they adjust to retirement or may
promote social participation and increase well-being (Reitzes & Mutran, 2004). Thus,
both positive and negative relationships with family members may mediate the
relationship between involuntary retirement and the health and mental health of older
adults.
Employment typically features formal and informal relationships that allow older
adults to fully participate in society, and retirement can cut off that social participation
(Moen, Fields, Quick, & Hofmeister, 2000). On the other hand, although it can be
challenging for older adults to remain socially engaged as they age due to the loss of
social roles and poor health, retirement may allow more free time to engage in socially
integrating activities (Cornwell, Laumann, & Schumm, 2008; Peppers, 1976). In terms of
the relationship between social integration and health and mental health, previous
research has suggested that being socially disconnected or isolated is associated with
poorer physical and mental health (Cornwell & Waite, 2009), whereas participating in
meaningful activities improves both physical and mental health (Moen, 1995). The
impact of retirement on social integration and well-being can differ significantly
depending on whether older adults perceive their retirement as voluntary or involuntary
(Moen et al., 2000). Considering the fact that individuals who planned more for their
retirement participated in social activities more (Rosenkoetter & Garris, 2001), people
who experience involuntary retirement may have relatively restricted social integration
that may in turn affect their health and mental health.
12
2.4. Theoretical Framework: Toward a Dynamic Conceptual Model
Potential mediators of the relationship between involuntary retirement and health
and mental health have not been reviewed in empirical research. To fully understand the
mechanisms of the effect of involuntary retirement on health and mental health, it is
important to take these potential mediators into consideration because they may shed
light on how retirement is experienced by individuals.
In terms of developing a conceptual model of this study, insights from two
theories were adopted as the guiding framework: latent deprivation theory and the life
course perspective. Jahoda‘s (1979, 1981, 1982) latent deprivation theory is one model
that provides a theoretical explanation of why unemployment may negatively affect an
individual‘s well-being. Latent deprivation theory specifically focuses on the manifest
functions (financial rewards) and latent functions (time structure, social contact, the
linking of individuals in a shared collective effort or purpose, the provision of social
identity or status, and regular enforced activity) of employment (Jahoda, 1981). The main
assumption of this theory is that people have strong psychological needs for the
satisfaction of latent functions and that these needs are best met by paid employment.
Therefore, when individuals are deprived of both manifest and latent functions of
employment, meeting the needs of latent psychological functions is more important to
well-being. This theory provides a theoretical explanation of the effect of involuntary
retirement on health and mental health in the current study. In addition, it provides
theoretical support to the potential role of manifest functions (financial control) and one
13
latent function (social integration) as mediators of the relationship between involuntary
retirement and health and mental health.
The second theory employed in this study is the life course perspective. The life
course perspective is a useful framework that provides important insight regarding the
mechanisms and consequences of retirement because it attempts to understand long-term
patterns of changes during the life course in a social and historical context (Moen, 1996).
The life course perspective has four dominant themes: (1) interplay of human lives and
historical time, (2) timing of lives, (3) linked or interdependent lives, and (4) human
agency in making choices (Elder, 1994). First, in terms of human lives and historical time,
the life course perspective emphasizes that the individual life course is historically
embedded and that a historical event can influence multiple dimensions of a person‘s life
(George, 2007). Secondly, the life course perspective focuses on the timing of lives; that
is, the ages at which specific life events or transitions occur. Based on social norms or
shared experiences, the timing of a transition may be considered ―off time‖ or ―on time,‖
and the personal impact of any change depends on an individual‘s life status and context
at the time of change (Elder, 1994; George, 1993). Therefore, involuntary retirement at
age 50 may be considered an off-time transition, whereas voluntary retirement at age 65
may be considered on time, and these two transitions will have different effects on each
individual. Third, one of the central principles of the life course perspective is linked or
interdependent lives. The life course perspective posits that individuals are embedded in
social relationships with family, friends, and coworkers across the life span and
emphasizes the interdependence of their lives (Elder, 1994). This theory allows
14
researchers to describe the context in which transitions take place more accurately than
considering an individual as an independent actor (George, 2007). The last principle is
human agency, which refers to the use of personal power to achieve goals. The life
course perspective acknowledges people‘s strength and capacity for change (Hutchison,
2005). These principles of the life course perspective provide theoretical explanations of
the effect of involuntary retirement on health and mental health through mediating roles
of positive and negative family relationships.
2.4.1. Conceptual Model
Based on the theoretical framework as well as empirical studies discussed in the
previous section, this study involved developing a multiple mediation model that
estimated the indirect effects of retirement transition on health and mental health through
four mediators: financial control, positive family relationships, negative family
relationships, and social integration (Figure 1).
15
Figure 1. Conceptual Model of the Mechanisms of the Effect of Involuntary Retirement
on Health and Mental Health.
Retirement
Transition
Financial Control
Positive Family
Relationships
Social Integration
Mental Health
Health
Negative Family
Relationships
16
CHAPTER III. RESEARCH AIMS, QUESTIONS AND
HYPOTHESES
The major aim of this study was to examine the mediating effects of financial
control, positive family relationships, negative family relationships, and social integration
on the relationship between retirement and health and mental health among older adults
in the United States. Specific research aims, research questions, and hypotheses were
developed based on the conceptual model of the study, which was derived from the life
course perspective and latent deprivation theory.
3.1. Research Aims
The current study had two specific research aims. The first aim of the study was
to explore characteristics of individuals who transitioned to involuntary retirement
compared to those who retired voluntarily or didn‘t retire. The second aim of the study
was to assess the potential mediating effects of financial control, positive family
relationships, negative family relationships, and social integration on the relationship
between involuntary retirement and health and mental health among older adults.
3.2. Research Questions and Hypotheses
The research aims described above were accompanied by the following research
questions and hypotheses.
3.2.1. Research Aim 1: To Explore Characteristics of Involuntary Retirees
Research question 1. Among people aged 50 or older, how prevalent is the
experience of involuntary retirement and what are the characteristics of involuntary
retirees in terms of sociodemographics as well as key variables of the study (financial
17
control, positive family relationships, negative family relationships, social integration,
health, and mental health)?
Research question 2. To what extent do involuntary retirees differ from voluntary
retirees or those who didn‘t retire in terms of sociodemographic variables as well as key
study variables (financial control, positive family relationships, negative family
relationships, social integration, health, and mental health)?
3.2.2. Research Aim 2: To Assess Potential Mediators
Hypothesis 1. Transition to involuntary retirement would have a negative effect
on self-rated health and mental health after controlling for sociodemographics and
baseline outcome variables.
Hypothesis 2. Transition to involuntary retirement would predict financial control,
positive family relationships, negative family relationships, and social integration after
controlling for sociodemographics and baseline outcome variables.
Hypothesis 3. Transition to involuntary retirement would predict self-rated health
and mental health, and those relationships would be mediated by financial control,
positive family relationships, negative family relationships, and social integration after
controlling for sociodemographics and baseline outcome variables.
18
CHAPTER IV: METHODS
4.1. Data Source
The data used for this study were extracted from the Health and Retirement Study
(HRS), a longitudinal household panel study sponsored by the National Institute of Aging
(grant number NIA U01AG009740) and conducted by the Institute for Social Research at
the University of Michigan. Since its launch in 1992 (Wave 1), the HRS has surveyed a
nationally representative sample of more than 27,000 individuals older than 50 years and
followed them and their spouses every 2 years through 2010 (Wave 10). During the
baseline survey, face-to-face interviews were conducted with the initial sample (the HRS
cohort, born between 1931 and 1941) that consisted of 12,652 individuals representing
7,702 households. To maintain the national representativeness of the sample, additional
cohorts were added in 1998 and 2004. In 1998, the War Babies cohort (born between
1942 and 1947) was added and consisted of 2,701 respondents and spouses. In 2004, the
Early Boomers cohort (born between 1948 and 1953) was added and consisted of 3,256
respondents and spouses. Because the HRS was specifically designed to explore the
health and economic experiences of older adults during their transition from work to
retirement, it provides extensive information on the economic circumstances,
employment, retirement, health and health care, cognition, living arrangement, family
relationships, and demographics of older Americans (National Institute on Aging, 2007).
To examine the research questions and hypotheses of this study, two waves of
HRS data (2006 and 2010) merged with the RAND HRS data set (Version L, 2010) were
used. HRS 2008 data were additionally used to identify participants who transitioned to
19
retirement between 2006 and 2010. RAND HRS is a user-friendly longitudinal database
that was created by the RAND Center for the Study of Aging. It contains a subset of the
most frequently used HRS variables that have been cleaned and processed (St. Clair et al.,
2011). The reasons why this study specifically focused on HRS data from 2006 and
beyond are twofold. One is that the Early Boomers cohort, which is currently in the
process of transitioning from work to retirement, was added in 2004. Examining that
cohort‘s retirement experiences is expected to result in more meaningful implications for
future retirement policy and practice. The other reason is that longitudinal data for some
of the psychosocial measures used in this study were made available in 2006 due to a
leave-behind questionnaire (LBQ).
The LBQ collected psychosocial data biannually beginning 2004. With the
purpose of expanding its assessment of psychosocial issues, HRS piloted a new feature
for data collection in 2004, leaving self-administered questionnaires with respondents
who completed an in-person interview. In 2006, the updated and revised psychosocial
questionnaire was randomly administered to 50% of participants. In 2008, the
questionnaire was given to the other 50% of HRS respondents who had not completed it
in 2006. In 2010, the same 50% subsample that was eligible for the questionnaire in 2006
completed the LBQ again. Thus, the first longitudinal psychosocial data available at 4-
year intervals were from 2006 to 2010 (HRS, 2013).
4.2. Sample
To obtain a nationally representative sample, HRS generated a multistage area
probability sample of households using a design that featured four stages of sample
20
selection. First, primary stage units (PSUs) were sampled through probability
proportionate to size selection of U.S. metropolitan statistical areas (MSAs) and non-
MSA counties. Second, area segments were selected within the PSUs and a complete list
of all housing units within those segments was created. Third, a sample of housing units
was systematically selected from the list. Finally, people within housing units who were
age-eligible were selected. In addition, HRS oversampled African Americans (1.86:1),
Hispanics (1.72:1), and people in Florida (2:1; HRS, 2008).
The sample for this study consisted of individuals who had transitioned to
retirement either voluntarily or involuntarily between 2006 and 2010, as well as a
comparison group of workers who did not retire during this period. To be included in this
study, individuals had to meet the following criteria: (1) age 50 or older; (2) responded to
the LBQ both in 2006 (T
1
) and 2010 (T
2
); and (3) were working for pay in 2006 (T
1
). As
presented in Figure 2, a sample of 2,023 participants who met the eligibility criteria for
this study was selected from the total sample in 2006 (n = 18,469). Of these 2,023
potential participants, individuals meeting the following criteria were excluded: (1) those
who reported that they were either partially or completely retired at baseline (n = 571;
missing = 22); (2) those who didn‘t provide information about their retirement status in
2010 (n = 15); (3) those who provided contradicting information regarding their
retirement status and retirement year (n = 82; i.e., those who stated that they were not
retired in 2006 but listed a retirement year before 2006); (4) those who reported no
information regarding retirement voluntariness or who answered that their retirement was
half forced and half wanted (n = 36); and (5) those who retired sometime between T
1
and
21
Figure 2. Sample Selection Flow Chart.
Selected Sample
≥ 50
Answered LBQs at T
1
&
T
2
Working for pay in 2006
(n = 2,023)
Total Sample in 2006
(n = 18,469)
Excluded (n = 16,446)
Didn‘t answer either or both
LBQs in 2006 & 2010 (n =
13,334)
Under age 50 (n = 166)
Were not working for pay in
2006 (n = 2,945; missing = 1)
Excluded (n = 743)
Reported either complete or
partial retirement in 2006 (n =
571; missing = 22)
Didn‘t report retirement status
in 2010 (n = 15)
Contradicting information on
retirement status and retirement
year (n = 82)
No information on retirement
voluntariness or who answered
that their retirement was half
forced and half wanted (n = 36)
Reentered labor force and
reported themselves not retired
(n = 17)
Final Analytic Sample
(n = 1,280)
22
T
2
but had reentered labor force and described themselves as not retired at T
2
(n = 17).
Thus, a final sample of 1,280 individuals was included in the analysis of this study.
The sample selection procedure excluded a considerable number of people who
didn‘t respond to the LBQ at both time points. Thus, those who were excluded from the
analyses due to not responding to the questionnaire and were otherwise eligible (n = 473)
were compared with the analytic sample using t-tests and chi-square tests. Participant
characteristics did not differ at statistically significant levels except for education,
race/ethnicity, and mental health. Those who were excluded from the study were less
educated (p < .01) and more likely to not be White (p < .001) compared to the study
sample. Those who were excluded due to nonresponse to the psychosocial questionnaire
had significantly higher mental health problems at T
2
(p < .01); there was no statistically
significant difference in self-rated health.
4.3. Measures
This section describes the measures employed in this study, including measures of
mental health, self-rated health, financial control, positive family relationships, negative
family relationships, social integration, retirement transition, and control variables. Most
of the variables used in this study were measured both at baseline and T
2
for the purpose
of conducting a longitudinal study.
4.3.1. Dependent Variables
Dependent variables in this study were mental health and health at T
2
, that is, after
the transition from work to retirement.
23
Mental health. The mental health of participants was measured using an 8-item
shortened form of the 20-item Center for Epidemiologic Studies Depression Scale (CES-
D; Radloff, 1977). The eight items included six negative indicators that reflect the
presence of depressive symptoms and two positive indicators that refer to the absence of
depressive symptoms. The negative indicators measured whether the respondent felt
depressed, felt everything was an effort, experienced restless sleep, could not get going,
felt alone, and felt sad much of the time during the previous week. The positive indicators
measured whether the respondent enjoyed life and was happy much of the time during the
previous week. These eight indicators were dichotomized (0 = no, 1 = yes) and the two
positive indicators were reverse coded. A higher CES-D score indicated more depressive
symptoms. The shortened 8-item CES-D measure has been proven to be valid and
reliable (Gallo et al., 2000; Mandal & Roe, 2008). Cronbach‘s alpha coefficients at
baseline and T
2
for the study sample were .77 and .80, respectively.
Self-rated health. Physical health was measured using a single item of self-rated
general health status. HRS asked respondents, ―Would you say your health is excellent,
very good, good, fair, or poor?‖ Answer options ranged from 1 (excellent) to 5 (poor).
The codes were reversed so that higher scores represented better health. This single-item
measure of self-rated health is commonly used to assess general health status. Although
its test–retest reliability is known to be modest (Zajacova & Dowd, 2011), studies have
reported that the measure has been a consistently strong predictor of physical health
indicators such as mortality (DeSalvo, Bloser, Reynolds, He, & Muntner, 2006; Idler &
Benyamini, 1997).
24
4.3.2. Independent Variable
Transition to retirement. Transition to retirement was measured as whether or not
participants retired between T
1
and T
2
. The sample featured respondents who were
working for pay and did not view themselves as retired in 2006. Anyone who reported
retiring between 2006 and 2010 were coded as retired. The transition of retirement was
measured as a binary variable (0 = retired, 1 = not retired) for the first phase of the
analyses, during which retirement was of interest. In Phase 2, during which the nature of
retirement was of interest, retirement was recoded into three groups (0 = involuntary, 1 =
voluntary, 2 = not retired). To specify whether the retirement transition was voluntary or
not, the following question was used: ―Thinking back to the time you (partly/completely)
retired, was that something you wanted to do or something you felt you were forced into?‖
Lastly, for the third phase of the study, during which a direct comparison was made
between voluntary and involuntary retirement, retirement was coded as a binary variable
(0 = involuntary, 1 = voluntary).
4.3.3. Mediating Variables
In this study, four factors—control over financial situation, positive family
relationships, negative family relationships, and social integration—were examined as
potential mediators of the relationship between transition to involuntary retirement and
older adults‘ health and mental health. Mediating variables for this study were measured
both at baseline and T
2
, and baseline measures were used as control variables.
Financial control. Financial control was measured using a single item asking
respondents to indicate on a scale ranging from 0 to 10 the amount of control they had
25
over their current financial situation. This measure was based on the fact that two
individuals with the same amount of financial resources can have different perceptions of
their financial control depending on their financial obligations. Previous research also
suggested that perceptions of financial situation rather than an objective measure are
more highly correlated with well-being (Ullah, 1990).
Positive and negative family relationships. Both positive and negative family
relationships were assessed with a battery of questions on relationship quality. HRS
leave-behind questionnaires include a 3-item scale regarding positive relationships as
well as a 4-item scale regarding negative relationships with respondent‘s spouse, children,
and other immediate family members. The positive family relationship scale includes
items such as ―How do they (spouse, children, or other family members) understand the
way you feel about things?‖ (1 = a lot, 2 = somewhat, 3 = a little, 4 = not at all). The
negative family relationship scale uses the same answer options and includes items such
as ―How much do they (spouse, children, or other family members) criticize you?‖ All
items were reverse coded so that higher scores corresponded to more positive and
negative family relationships. Scores within each relationship category were averaged to
create an index of positive and negative relationships with spouse, children, and
immediate family members. Cronbach‘s alpha coefficients for positive relationships with
spouse, children, and other immediate family members in this study sample were .82, .83,
and .87, respectively. Cronbach‘s alpha coefficients for negative relationships with
spouse, children, and other immediate family members were .80, .76, and .76,
26
respectively. All relationship categories were combined to create overall scores for
positive and negative family relationships.
Social integration. Social integration was measured by a single-item question
asking about respondents‘ frequency of participation in nonreligious activities. Because
the LBQ used slightly different words and answer options at baseline and T
2
, answers
were recoded to be comparable at both time points. For example, social integration at
baseline was measured by asking, ―Not including attendance at religious service, how
often do you attend meetings or programs of groups, clubs, or organizations that you
belong to?‖ The answer options were 1 = more than once a week, 2 = once a week, 3 = 2
or 3 times a month, 4 = about once a month, 5 = less than once a month, and 6 = never.
On the other hand, social integration at T
2
was measured by asking, ―How often do you
attend meetings or non-religious organizations such as political, community or other
interest groups?‖ and the answer options were 1 = daily, 2 = several times a week, 3 =
once a week, 4 = several times a month, 5 = at least once a month, 6 = not in the last
month, 7 = never/not relevant. Thus, the answer categories were recoded to make both
questions comparable and reverse coded so that higher scores represented more social
integration. Answer options were recoded as 1 = not at all, 2 = one or more times a year,
3 = two or three times a month, 4 = once a week, and 5 = more than once a week.
4.3.4. Control Variables
In the analyses of this study, sociodemographic characteristics, retirement-related
characteristics, and outcome variables at baseline were controlled. Sociodemographic
characteristics included age at T
2
, gender, education, marital status at T
2
, race, and
27
current work status. Age was measured in years and gender was a binary variable (0 =
female, 1 = male). Education was measured as 1 = less than high school, 2 = high school
graduate/GED, 3 = some college, and 4 = college and above. Marital status was recoded
as a binary variable (0 = married or partnered, 1 = other). Race was also recoded as a
binary variable (0 = White, 1 = other). Current work status was a binary variable
measuring whether the individual was working for pay in 2010 (0 = working for pay, 1 =
not working for pay). Coding was adjusted for complex sample general linear modeling,
setting the highest category as the reference group by default. Retirement-related
characteristics included retirement duration and age at retirement. Retirement duration
was measured in months by subtracting retirement month and year from the month and
year when the T
2
interview was conducted. Age at retirement was measured in years and
computed by subtracting months of retirement from each individual‘s age at T
2
. Because
these two variables were specifically related to retirees, they were used only in the third
phase of the analyses. Lastly, baseline mediator variables and dependent variables were
considered as control variables.
4.4. Statistical Analyses
The current study was a secondary data analysis of existing data without personal
identification of study participants, and was approved by the Institutional Review Board
of the University of Southern California as exempt from full human subjects review. Data
analysis of this study was conducted with SPSS 18.0 mainly using the SPSS Complex
Samples procedure. SPSS Complex Samples is an add-on module that provides statistics
28
and standard errors that are corrected for complex sampling design such as stratified,
clustered, or multistage sampling.
HRS features a complex sample that employed a multistage area probability
sampling design with oversampling of African Americans, Hispanics, and Florida
residents. In general, this complex sample design has several properties that are different
from simple random sampling. First, unlike simple random sampling in which each
individual sampling unit is selected at random and thus has the same probability of being
chosen, complex sample design may have unequal probability of selection due to the
incorporation of clusters and strata (West, 2008). Second, cluster sampling often
employed in complex sample designs randomly selects clusters of individuals within
primary sampling units, and this results in a lack of independence of the individuals
within clusters because they tend to be more similar than individuals randomly selected
from the population (Osborne, 2011; West, 2008). Third, although estimates based on
complex samples can have fewer errors compared to simple random sampling because of
stratification, it can have more errors related to clustering (West, 2008). Therefore, it is
important to consider the features of complex samples during the analysis to compute
unbiased estimates of population parameters. In fact, Osborne (2011), in a study
comparing four analyses (not applying any weights; applying weights only; applying
scaled weights; and applying the weight, primary sampling unit, and cluster information
provided in the data) in modeling complex samples, emphasized how unweighted
samples can misestimate parameters as well as standard errors. Thus, it is important to
take sampling effects into consideration. To adjust for the complex sampling design of
29
HRS, data were weighted by stratification (STRATUM), clustering (SECU), and
respondent weight (KLBWGTR) information provided by HRS. For the respondent
weight of the 2006 leave-behind questionnaire subsample, KLBWGTR was used to
adjust the sample selection because only 50% of the HRS sample was randomly selected
to answer the LBQ. KLBWGTR is the product of the HRS respondent-level weight in
2006 and a nonresponse adjustment factor (see HRS, 2008, 2013; Ofstedal, Weir, Chen,
& Wagner, 2011 for further details on HRS sampling weights).
4.4.1. Preliminary Data Analyses
Prior to conducting main analyses, data were observed using univariate and
bivariate analyses to better understand the data and to assess whether basic assumptions
were met for the planned analyses. Univariate analyses were conducted for all study
variables and normality of continuous variables was assessed using Q-Q plots. Because
mental health at both time points was skewed, the variable was transformed into natural
log to improve normality. The amount of missing variables was also examined. All
variables had missing information less than 5% and thus were not imputed. In addition,
bivariate analyses such as complex sample cross-tabulation and general linear modeling
were conducted to examine the relationships among study variables.
The first aim of the study was to develop an overall understanding of involuntary
retirement by exploring the characteristics of involuntary retirees. Descriptive statistics
were conducted to address the first research question, which was to examine the
prevalence of involuntary retirement among older adults and their characteristics.
Frequencies and descriptives were calculated for sociodemographic and key study
30
variables (financial control, positive family relationships, negative family relationships,
social integration, health, and mental health) by retirement category to compare the
characteristics of involuntary retirees and those of voluntary retirees and participants who
didn‘t retire. Weighted percentages and unweighted frequencies are reported for
categorical variables. For continuous variables, weighted means and standard errors are
reported.
The second research question was to examine differences among involuntary
retirees, voluntary retirees, and those who didn‘t retire in terms of sociodemographic
variables and the main study variables (financial control, positive family relationships,
negative family relationships, social integration, health, and mental health). For
categorical variables, weighted cross-tabulations were conducted. In terms of significance,
an adjusted F is reported, which is a variant of the second-order Rao-Scott adjusted chi-
square statistic. The second-order Rao-Scott chi-square is used for categorical analyses of
complex survey designs because the Pearson chi-square can yield type I errors when used
with complex survey designs (Rao & Scott, 1984; Thomas & Rao, 1987). For interval-
level variables, complex sample general linear modeling was conducted and Wald F
statistics are reported.
4.4.2. Mediation Analyses
To address the second aim of the study, which was to examine mechanisms of the
effect of involuntary retirement on older adults‘ health and mental health by testing
potential mediating effects, regression-based path analyses were conducted. A mediation
model examines whether an independent variable exerts an effect on a dependent variable
31
through one or more mediators (Hayes, 2009). The mediation model designed for this
study was a multiple-mediation model (Preacher & Hayes, 2008) that simultaneously
considered four mediators (financial control, positive family relationships, negative
family relationships, and social integration). According to Preacher and Hayes (2008), a
single multiple-mediation model has more benefits when testing multiple mediators
compared to conducting a simple mediation model multiple times. Some of the benefits
described by Preacher and Hayes (2008) include: (1) the mediating effects of multiple
mediators as a set can be examined by testing the total indirect effect; (2) the mediating
effects of a specific mediator can be tested while controlling for other mediators; (3)
parameter bias due to omitted variables is reduced; and (4) the relative magnitude of the
specific indirect effect can be tested. Thus, a multiple-mediation model can be a more
appropriate approach to examining the complex mechanisms of involuntary retirement
with regard to health and mental health.
To assess the indirect effects of the mediators proposed in the current study, path
analytic approaches suggested by Preacher & Hayes (2008) were employed. Although
one of the most common approaches for testing mediation models is the causal steps
approach suggested by Baron and Kenny (1986), it has been criticized for its limitations.
One criticism is that it requires a significant total effect (relationship between an
independent and dependent variables) as one of the prerequisites for testing mediation.
This approach does not allow further exploration for evidence of an indirect effect when
there is no significant relationship between an independent variable and a dependent
variable. However, this often prevents researchers from detecting the potential indirect
32
effect of an independent variable on a dependent variable via a mediator (Hayes, 2009).
In addition, researchers who have suggested not requiring a significant total effect before
testing indirect effects explained that total effects can be attenuated or contaminated
when other variables that may affect the relationship are not considered or when indirect
effects in opposite direction are present in the model (Hayes, 2009; MacKinnon, Krull, &
Lockwood, 2000; Mathieu & Taylor, 2006).
To assess the mechanisms of involuntary retirement on older adults‘ health and
mental health, several models were estimated in multiple phases. The first phase of the
analysis estimated a multiple-mediator model with a binary independent variable (retired
or not) to compare with a later model in the second phase, which included a
multicategorical independent variable (0 = involuntarily retired, 1 = voluntarily retired, 2
= not retired). This approach allowed a comparison of the mechanisms of the effect of
retirement on older adults‘ self-rated health and mental health versus the mechanisms of
the effect of involuntary retirement as well as voluntary retirement compared to that of no
transition to retirement. The third phase of the analysis excluded those who didn‘t retire
and considered only those who transitioned to retirement to examine the mechanisms of
the effect of involuntary retirement directly compared to voluntary retirement. This final
phase also explored whether the mechanism of the effect of involuntary retirement
changed when other variables specific to retirement (age at retirement and duration of
retirement) were considered.
33
Phase 1. Total, Direct, Total Indirect, and Specific Indirect Effects of
Retirement on Self-Rated Health and Mental Health (n = 1,280)
In this phase, the total effect of retirement on health and mental health was examined first
(Figure 3A). In addition, both direct and indirect effects of retirement through multiple
mediators of financial control, positive family relationships, negative family relationships,
and social integration were assessed. Figure 3B represents the multiple-mediator model
based on a binary retirement variable (retired or not). The paths labeled a
1
to a
4
represent
the effects of retirement on each mediator (M1–M4), and those labeled b
1
to b
4
are the
effects of each mediator on self-rated health and mental health, controlling for the effect
of retirement. Direct effect refers to the effect of retirement on self-rated health and
mental health that is independent of the pathway through various potential mediators and
is denoted as c′ (Hayes, 2009; Preacher & Hayes, 2008). Indirect effect refers to mediated
effects of retirement on self-rated health and mental health, and was computed as the
product of a and b (Hayes, 2009; Preacher & Hayes, 2008). The total effect (path c in
Figure 3A) of retirement on self-rated health and mental health is the sum of the direct
and indirect effects, which can be denoted as c = c′ + ab (Hayes, 2009; Preacher & Hayes,
2008). When using a multiple-mediator model as in this study, the total effect refers to
the sum of the direct effect and indirect effects through each mediator. Thus, the total
effect is denoted as c = c′ + a
1
b
1
+ a
2
b
2
+ a
3
b
3
+ a
4
b
4
. In a multiple-mediator model, the
indirect effect of retirement on self-rated health and mental health through each mediator
is called a specific indirect effect, and the sum of each specific indirect effect is called the
total indirect effect (Hayes, 2009; Preacher & Hayes, 2008). To examine the total, direct,
34
Figure 3A. Total Effect of X on Y when IV is Binary (Phase 1).
Figure 3B. A Multiple Mediator Model when IV is Binary (Phase 1).
X: Retirement
(retired vs. not)
M1: Financial
Control
Y1: Health T
2
Y2: MH T
2 _LOG
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
a
1
b
1
a
2
a
3
a
4
b
2
b
3
b
4
c′
X: Retirement
(retired vs. not)
Y1: Health T
2
Y2: MH T
2 _LOG
c
35
specific indirect, and total indirect effects of retirement on self-rated health and mental
health, the following models were estimated using complex sample general linear
modeling in SPSS 18.0.
Total effect of IV on DV (c path)
(1) Health at T
2
= constant + c (retired) + covariates + error
IV to mediators (a paths)
(2) Financial control = constant + a
1
(retired) + covariates + error
(3) Positive family relationships = constant + a
2
(retired) + covariates + error
(4) Negative family relationships = constant + a
3
(retired) + covariates + error
(5) Social integration = constant + a
4
(retired) + covariates + error
Direct effects of mediators on DV (b paths) and direct effect of IV on DV (c′ path)
(6) Health at T
2
= constant + c′(retired) + b
1
(financial control) + b
2
(positive
family relationships) + b
3
(negative family relationships) + b
4
(social
integration) + covariates + error
The significance of the total indirect effect as well as specific indirect effects was
tested using Monte Carlo confidence intervals. Monte Carlo is used to make inferences
about indirect effects that are comparable with bootstrap confidence intervals (Hayes &
Preacher, 2013; Preacher & Selig, 2012). Because both bootstrap and Monte Carlo
confidence intervals cannot be constructed using the SPSS Complex Sample procedure, R
was used. To construct Monte Carlo confidence intervals, R codes provided by Selig and
Preacher (2008) were used for specific indirect effects and R codes provided by Preacher
36
and Selig (2012) were used for the total indirect effect. The same analytic procedure was
repeated with mental health as a dependent variable.
Phase 2. Relative Total, Direct, Total Indirect, and Specific Indirect Effects of
Voluntary and Involuntary Retirement on Self-Rated Health and Mental Health (n =
1,280)
In Phase 2, the same procedure used in Phase 1was followed. The only difference
was that the independent variable was multicategorical (involuntarily retired, voluntarily
retired, not retired) instead of binary (retired, not retired). When an independent variable
is multicategorical, equations in Phase 1 cannot be used because there is more than one a
path and c′ path. Hayes and Preacher (2013) have described different ways to estimate the
direct and indirect effects of a model with a multicategorical independent variable using a
general linear modeling approach. This approach allowed an examination of how
mechanisms of the effect of involuntary retirement and voluntary retirement on self-rated
health and mental health were different from no transition to retirement without the need
to collapse the data to a dichotomous independent variable or drop some data to compare
only two groups at a time. This study employed a dummy-coding strategy to represent
groups as suggested by Hayes and Preacher (2013); this approach was automatically
generated in SPSS. The reference category for the Phase 2 analyses was participants who
did not retire.
In a model with a multicategorical independent variable, slightly different terms
are used to indicate different effects: relative total effect, relative direct effect, relative
specific indirect effect, and relative total indirect effect. Although the basic definitions
37
are the same as described in the previous section, these effects refer to the effects of one
group relative to a reference group. Thus, relative direct effect refers to the direct effect
of being in one group on a dependent variable relative to the reference group (Hayes &
Preacher, 2013). Relative specific indirect effect is the specific indirect effect of being in
one group on a dependent variable relative to the reference group (Hayes & Preacher,
2013). Relative total indirect effect is the sum of each relative specific indirect effect and
relative total effect is the relative direct effect plus the relative total indirect effect (Hayes
& Preacher, 2013).
Relative total, direct, specific indirect, total indirect effects of involuntary
retirement and voluntary retirement on self-rated health and mental health compared to
no transition to retirement were estimated using SPSS 18.0. Equations used in the model
are as follows and Figure 4A and Figure 4B illustrate these equations.
Total effect of IV on DV (c path)
(1) Health at T
2
= constant + c
1
(voluntary retirement) + c
2
(involuntary retirement)
+ covariates + error
IV to mediators (a paths)
(2) Financial control = constant + a
11
(voluntary retirement) + a
21
(involuntary
retirement) + covariates + error
(3) Positive family relationships = constant + a
12
(voluntary retirement) +
a
22
(involuntary retirement) + covariates + error
(4) Negative family relationships = constant + a
13
(voluntary retirement) +
a
23
(involuntary retirement) + covariates + error
38
Figure 4A. Total Effect of X on Y when IV is Multicategorical (Phase 2).
Figure 4B. A Multiple Mediator Model when IV is Multicategorical (Phase 2).
Dummy 1:
Voluntary
Retirement
M1: Financial
Control
Y: HealthT
2
Y2: MH T
2 _LOG
Dummy 2:
Involuntary
Retirement
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
c′
1
c′
2
a
11
a
12
a
13
a
14
a
21
a
22
a
23
a
24
b
1
b
2
b
3
b
4
Dummy 1:
Voluntary
Retirement
Y: Health T
2
Y2: MH T
2 _LOG
c
1
Dummy 2:
Involuntary
Retirement
c
2
39
(5) Social integration = constant + a
14
(voluntary retirement) + a
24
(involuntary
retirement) + covariates + error
Direct effects of mediators on DV (b paths) and direct effects of IV on DV (c′ path)
(6) Health at T
2
= constant + c′
1
(voluntary retirement) + c′
2
(involuntary retirement)
+ b
1
(financial control) + b
2
(positive family relationships) + b
3
(negative
family relationships) + b
4
(social integration) + covariates + error
Again, the significance of relative total indirect effect as well as relative specific
indirect effects was tested with Monte Carlo confidence intervals using R. The same
analytic procedure was repeated with mental health as a dependent variable.
Phase 3. Total, Direct, Total Indirect, and Specific Indirect Effects of
Involuntary Retirement on Self-Rated Health and Mental Health (n = 429)
Lastly, the third phase of the mediation analysis was to examine the mechanisms
of the effect of involuntary retirement compared with voluntary retirement. For this
analysis, only participants who had transitioned to either type of retirement were included
in the analysis. In addition, variables that were specific to only those who had retired,
such as age at retirement and retirement duration, were used as covariates in this model.
Equations used in the model are as follows and Figure 5A and Figure 5B illustrate these
equations.
Total effect of IV on DV (c path)
(1) Health at T
2
= constant + c (involuntary retirement) + covariates + error
IV to mediators (a paths)
(2) Financial control = constant + a
1
(involuntary retirement) + covariates + error
40
Figure 5A. Total Effect of X on Y when IV is Binary (Phase 3).
Figure 5B. A Multiple Mediator Model when IV is Binary (Phase 3).
X: Involuntary
Retirement
M1: Financial
Control
Y: Health T
2
Y2: MH T
2 _LOG
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
a
1
b
1
a
2
a
3
a
4
b
2
b
3
b
4
c′
X: Involuntary
Retirement
Y: Health T
2
Y2: MH T
2 _LOG
c
41
(3) Positive family relationships = constant + a
2
(involuntary retirement) +
covariates + error
(4) Negative family relationships = constant + a
3
(involuntary retirement) +
covariates + error
(5) Social integration = constant + a
4
(involuntary retirement) + covariates + error
Direct effects of mediators on DV (b paths) and direct effects of IV on DV (c′ path)
(6) Health at T
2
= constant + c′(involuntary retirement) + b
1
(financial control) +
b
2
(positive family relationships) + b
3
(negative family relationships) +
b
4
(social integration) + covariates + error
Significance tests for the total indirect effect as well as specific indirect effects
were conducted using R. In addition, the same analytic procedure was repeated using
mental health as a dependent variable.
42
CHAPTER V: RESULTS
5.1. Sample Characteristics
Sociodemographic and retirement characteristics of the study sample by
retirement status are presented in Table 1. The gender composition of the study sample (n
= 1,280) was 52% men and 48% women. Average age at T
2
was 61.89 years, ranging
from 53 to 88 years. In terms of education, the sample was relatively highly educated,
with 34% of participants having a college degree or above and 29% reporting some
college education. The sample was predominantly White (88.8%), married or partnered
(75.7%), and working for pay at T
2
(74.5%). In terms of retirement-specific variables,
only those who had transitioned either to voluntary or involuntary retirement were
considered (n = 429). The average age of retirement was 62.2 years, ranging from 51 to
85 years, and respondents reported being retired for an average of 22.8 months.
5.2. Results for Research Aim 1
5.2.1. Results for Research Question 1
To address the first research question of the study, the overall prevalence of
involuntary retirement among retirees was explored as well as the main characteristics of
involuntary retirees in terms of sociodemographic, retirement-related, and key study
variables. Between 2006 and 2010, about 29.3% (n = 429) of the sample made the
transition to retirement and among those retirees, 37.2% (n = 155) reported that they
retired involuntarily. Involuntary retirees were evenly divided by gender and their mean
age at T
2
was 63.13 years, with a range of 53 to 83 years. The education level of
involuntary retirees was lower compared to the total sample, with more participants
43
reporting less than high school (11.1% vs. 6.2%, respectively) and fewer participants
reporting having college degrees or above (19.0% vs. 34.0%, respectively). Most of the
older adults who involuntarily retired were White (90.4%), married or partnered (66.2%),
and not working for pay at T
2
(83.8%). Their average retirement age was 61.15 years,
ranging from 51 to 85 years. Their average retirement duration was 22.68 months.
Table 1
Sociodemographic and Retirement Characteristics by Retirement Status
a
Variable Total
n (%)
Involuntary
Retirees
n (%)
Voluntary
Retirees
n (%)
Not Retired
n (%)
Adjusted
F
b
(N = 1,280) (n = 155) (n = 274) (n = 851)
Gender
Female
Male
708 (48)
572 (52)
92 (50.3)
63 (49.7)
155 (46.4)
119 (53.6)
461 (48.1)
390 (51.9)
.214
Education
Less than high school
High school grad/GED
Some college
College degree and above
94 (6.2)
405 (30.9)
357 (29.0)
424 (34.0)
20 (11.1)
58 (37.7)
45 (32.1)
32 (19.0)
19 (5.4)
100 (38.5)
60 (21.9)
95 (34.3)
55 (5.6)
247 (27.8)
252 (30.4)
297 (36.2)
4.52
**
Race
White
Other
Marital status (T
2
)
Married/partnered
Other
Work status (T
2
)
Working for pay
Not working for pay
Age (T
2,
yrs), M (SD)
Retirement duration (month)
Age at retirement (yrs)
1082 (88.8)
198 (11.2)
952 (75.7)
328 (24.3)
912 (74.5)
367 (25.5)
61.89 (0.122)
22.80 (0.684)
62.20 (0.224)
124 (90.4)
31 (9.6)
101 (66.2)
54 (33.8)
27 (16.2)
127 (83.8)
63.13 (0.312)
22.68 (1.275)
61.15(0.297)
234 (87.6)
40 (12.4)
215 (82.2)
59 (17.8)
87 (34.2)
187 (65.8)
64.69 (0.324)
22.87 (0.829)
62.84 (0.327)
724 (88.8)
127 (11.2)
636 (75.5)
215 (24.5)
798 (94.0)
53 (6.0)
60.97 (0.149)
n/a
n/a
.41
3.98
*
260.41
***
57.18
***
0.014
12.80
**
a
Percentages of categorical variables and means and standard errors of continuous variables presented in
this table are weighted. Frequency is not weighted.
b
The adjusted F is a variant of the second-order Rao-Scott adjusted chi-square statistic. Significance is
based on the adjusted F and its degrees of freedom. For continuous variables, Wald F is reported.
*
p < .05,
**
p < .01,
***
p < .001
44
The characteristics of key study variables by retirement status are presented in
Table 2. Although the baseline measures were used only as control variables, they are
included in the table to present the change over four years. Compared to the total sample,
those who involuntarily retired had relatively lower financial control; slightly less
positive relationships with family members including spouse, children, and immediate
family; slightly more negative relationships with family members; and less social
integration at T
2
. In terms of self-rated health, involuntary retirees had lower self-rated
health at both time points compared to the total sample. Their mental health scores
measured at both time points were higher compared to the overall sample, meaning that
they had more mental health problems. Although their depressive symptoms were well
below the cutoff point of 4 or more suggested by HRS, it is notable that their CES-D
scores at both time points were about twice as high as the total sample.
5.2.2. Results for Research Question 2
The second research question involved the extent to which involuntary retirees
were different from voluntary retirees and those who didn‘t retire in terms of
sociodemographic, retirement-related, and key study variables of financial control,
positive family relationships, negative family relationships, social integration, self-rated
health, and mental health. The results presented in Tables 1 and 2 show that these
variables were significantly different across the three groups, including education (F[5,
276] = 4.52, p < .01), marital status (F[2, 103] = 43.98, p =.024), working status at T
2
(F[2, 110] = 260.41, p < .001), and age (F[2, 54] = 57.18, p < .001).
45
Table 2
Characteristics of Key Study Variables by Retirement Status
a
Variable Total
M (SE)
Involuntary
Retirees
M (SE)
Voluntary
Retirees
M (SE)
Not Retired
M (SE)
Wald F
(N = 1,280) (n = 155) (n = 274) (n = 851)
Financial control (T
1
)
Financial control (T
2
)
Positive family relationships (T
1
)
Positive family relationships (T
2
)
Negative family relationships (T
1
)
Negative family relationships (T
2
)
Social integration (T
1
)
Social integration (T
2
)
Self-rated health (T
1
)
Self-rated health (T
2
)
Mental health (T
1
)
Mental health (T
2
)
7.24 (0.076)
7.02 (0.087)
3.10 (0.023)
3.11 (0.025)
1.81 (0.017)
1.88 (0.015)
2.05 (0.040)
1.60 (0.030)
3.71 (0.031)
3.51 (0.037)
1.14 (0.063)
0.96 (0.050)
6.85 (0.213)
6.15 (0.223)
2.96 (0.069)
2.99 (0.071)
1.84 (0.053)
1.92 (0.040)
1.83 (0.091)
1.40 (0.044)
3.09 (0.085)
2.69 (0.100)
2.10 (0.194)
1.93 (0.192)
7.52 (0.151)
7.49 (0.168)
3.14 (0.041)
3.16 (0.039)
1.78 (0.037)
1.85 (0.031)
2.04 (0.074)
1.64 (0.049)
3.59 (0.065)
3.53 (0.054)
1.08 (0.135)
0.72 (0.074)
7.23 (0.086)
7.04 (0.094)
3.11 (0.028)
3.12 (0.029)
1.82 (0.020)
1.87 (0.017)
2.08 (0.051)
1.63 (0.037)
3.83 (0.038)
3.63 (0.047)
1.00 (0.059)
0.87 (0.049)
4.06
*
12.61
***
2.56
2.28
.48
1.10
3.05
10.89
***
30.41
***
35.37
***
17.31
***
17.68
***
a
Means and standard errors of continuous variables presented in this table are weighted.
*
p < .05,
**
p < .01,
***
p < .001
Post hoc tests were conducted to examine which pairs were significantly different,
and Bonferroni correction was made to p-values to adjust for potential alpha inflation due
to multiple comparisons. In terms of education, voluntary retirees (F[3, 129] = 4.78, p
=.012) and those who didn‘t retire (F[3, 141] = 5.30, p = .009) had significantly higher
education levels than involuntary retirees. In addition, the proportion of participants with
a higher education level was significantly higher for those who didn‘t retire compared to
those who voluntarily retired (F[3, 150] = 3.80, p = .042). The proportion of participants
who were married or partnered was significantly lower for involuntary retirees compared
to voluntary retirees (F[1, 46] = 9.48, p = .012), whereas there was no significant
difference compared with those who didn‘t retire. In terms of current working status, the
46
proportion of those who were not working for pay at T
2
was highest for involuntary
retirees compared to voluntary retirees (F[1, 46] = 12.91, p = .003) or those not retired
(F[1, 54] = 451.75, p < .001). The average age of involuntary retirees was about 1.5 years
lower than that of voluntary retirees (F[1, 46] = 11.44, p < .01) and about 2.2 years
higher than that of those who had not retired (F[1, 54] = 32.90, p < .001). In terms of
retirement-related characteristics, involuntary retirees retired about 1.7 years earlier than
voluntary retirees (F[1, 45]) = 12.80, p < .01). Involuntary retirees retired at the average
age of 61.15 compared to 62.84 years for voluntary retirees. There was no significant
difference in retirement duration between involuntary and voluntary retirees.
Group differences in terms of the key study variables are presented in Table 2.
Results of complex sample general linear modeling showed that financial control at both
time points (T
1
: F[2, 53] = 4.06, p = .016; T
2
: F[2, 54] = 12.61, p < .001), social
integration at T
2
(F[2, 54] = 10.89, p < .001), self-rated health at both time points (T
1
: F[2,
54] = 30.41, p < .001; T
2
: F[2, 54] = 35.37, p < .001), and mental health at both time
points (T
1
: F[2, 54] = 17.31, p < .001; T
2
: F[2, 54] = 17.68, p < .001) differed
significantly by the nature of retirement. At baseline, involuntary retirees had
significantly less financial control compared to voluntary retirees (t = 2.83, df = 46, p
= .007), whereas there was no significant difference compared to those not retired.
However, average financial control among involuntary retirees at T
2
was significantly
lower than both voluntary retirees (t = 5.26, df = 45, p < .001) and participants who had
not retired (t = 4.03, df = 54, p < .001). Also, involuntary retirees had significantly less
social integration at T
2
compared to voluntary retirees (t = 3.61, df = 46, p < .01) and
47
those not retired (t = 4.48, df = 54, p < .001). In terms of self-rated health, involuntary
retirees reported significantly lower self-rated health at both time points compared to
voluntary retirees (T
1
: t = 4.91, df = 46, p < .001; T
2
: t = 7.77, df = 46, p < .001) and those
not retired (T
1
: t = 7.84, df = 54, p < .001; T
2
: t = 8.40, df = 54, p < .001). In addition,
involuntary retirees reported relatively more depressive symptoms at both time points
compared to voluntary retirees (T
1
: t = -5.52, df = 46, p < .001; T
2
: t = -6.42, df = 46, p
< .001) and nonretired respondents (T
1
: t = -6.52, df = 54, p < .001; T
2
: t = -6.67, df = 54,
p < .001).
5.3. Results for Research Aim 2
The second aim of the study was to examine mechanisms of the effect of
involuntary retirement on older adults‘ self-rated health and mental health by testing the
mediating effects of financial control, positive family relationships, negative family
relationships, and social integration in multiple phases. The same multiple-mediator
model was tested using a binary independent variable (retired or not) in Phase 1 and a
multicategorical independent variable in Phase 2 (voluntary retirement, involuntary
retirement, and no retirement). In Phase 3, the mechanisms of involuntary retirement
were examined among only those participants who retired between baseline and T
2
. After
conducting mediation analyses in each phase, the significance of total and specific
indirect effects as well as relative total and relative specific indirect effects of financial
control, positive family relationships, negative family relationships, and social integration
were tested.
48
5.3.1. Results for Phase 1: Total, Direct, Total Indirect, and Specific Indirect
Effects of Retirement on Self-Rated Health and Mental Health (n = 1,280)
The analyses conducted in Phase 1 were preliminary to the main analyses of
hypotheses testing in Phases 2 and 3.The main purpose of the analyses in this phase was
to compare the results of the same model when the nature of retirement was not specified
as voluntary or involuntary but considered as a whole. Thus, in this phase, the mechanism
of the effect of retirement on self-rated health and mental health was examined. Results
of multivariate analyses estimating the effect of retirement on older adults‘ health are
shown both in Table 3 and Figures 6A and 6B. Model 5 in Table 3 presents the result of
testing the total effect of retirement on health. The results indicated that whether or not an
individual transitioned to retirement was not associated with follow-up self-rated health
when sociodemographic factors, baseline health, and mental health at T
2
were controlled
(Figure 6A). Factors significantly associated with health at T
2
were being female (p < .01)
and education (p < .05). In addition to those two factors, baseline health, which was
measured before participants transitioned to retirement, was positively associated with
postretirement health (p < .001), whereas mental health status at T
2
was negatively
associated with postretirement self-rated health at T
2
(p < .001).
Models 1 to 4 in Table 3 present the results of multivariate regression analyses
testing paths a
1
–a
4
for each mediator. Four models were estimated with each mediator as
a dependent variable, controlling for sociodemographic factors as well as other baseline
mediators. Results indicated that retirement was not associated with any of the four
mediators: financial control, positive family relationships, negative family relationships,
49
Table3
Complex Sample General Linear Model Coefficients (Retirement Transition as Binary; n = 1,280)
a
a
Unstandardized path coefficients and standard errors are reported in this table.
*
p < .05,
**
p < .01,
***
p < .001
Mediators (T
2
) DV1 (T
2
) DV2 (T
2
)
Fin Control Pos Relation Neg Relation Social Int Self-Rated Health Mental Health
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
Intercept 1.197 (.865) 1.136
***
(.232) 1.078
***
(.177) 1.002
**
(.291) .682
*
(.310) .685 (.391) 1.003
***
(.218) 1.047
***
(.251)
Retirement .264 (.221) .026 (.037) -.031 (.031) .051 (.074) -.024 (.081) -.033 (.083) .002 (.044) .009 (.042)
Female .112 (.105) .068
*
(.032) .030 (.021) -.026 (.050) .134
**
(.043) .151
**
(.046) .056 (.033) .060 (.034)
Married/partnered -.126 (.147) .042 (.038) .050 (.029) -.009 (.051) .018 (.047) .045 (.050) -.121
**
(.039) -.128
**
(.037)
White -.338 (.228) -.136
*
(.051) -.106
**
(.036) -.065 (.074) .027 (.068) .058 (.073) .043 (.045) .030 (.047)
Currently working .698
*
(.260) .003 (.035) -.052 (.034) .095 (.079) .164 (.086) .131 (.088) -.025 (.053) -.012 (.051)
Age T
2
.033
**
(.011) -.001 (.003) .001 (.002) -.003 (.004) .008 (.005) .006 (.005) -.002 (.003) .000 (.003)
Education -.030 (.073) .026 (.017) -.020 (.013) .096
***
(.024) .049
*
(.022) .048 (.024) -.013 (.017) -.009 (.017)
Fin control T
1
.500
***
(.039) - - - - - - -
Pos relation T
1
- .649
***
(.034) - - - - - -
Neg relation T
1
- - .494
***
(.022) - - - - -
Social int T
1
- - - .242
***
(.027) - - - -
Health T
1
- - - - .577
***
(.027) .564
***
(.025) - -
MH T
2
_log - - - - -.356
***
(.038) -.297
***
(.040) - -
Health T
2
- - - - - - -.155
***
(.021) -.132
***
(.022)
MH T
1
_log - - - - - - .346
***
(.034) .306
***
(.035)
Fin control T
2
- - - - - .036
**
(.011) - -.033
***
(.007)
Pos relation T
2
- - - - - .027 (.031) - -.052 (.027)
Neg relation T
2
- - - - - -.134 (.074) - .100
*
(.040)
Social int T
2
- - - - - .009 (.030) - -.017 (.017)
R
2
.255 .404 .357 .147 .482 .493 .297 .324
50
Figure 6A. Total Effect of Retirement on Self-Rated Health.
Significant path
Nonsignificant path
Reference group: those who did not retire
Figure 6B. Direct and Indirect Effects of Retirement on Self-Rated Health.
X: Retirement
(retired or not)
M1: Financial
Control
Y: Health T
2
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
a
1
= .264
b
1
= .036
a
3
= - .031
a
4
= .051
b
3
= -.134
b
2
= .027
c′ = -.033
X: Retirement
(retired or not)
Y: Health T
2
c = -.024
a
2
=.026
b
4
=.009
51
and social integration. Thus, paths a
1
–a
4
were not significant (Figure 6B). Model 6 was
estimated to examine the direct effect of retirement (c′) as well as the direct effect of each
mediator on health (paths b
1
–b
4
). In this model, being female and financial control were
positively associated with self-rated health at T
2
(p < .01). In addition, the direction and
significance of baseline self-rated health and mental health status at T
2
remained the same
even after the mediators were entered (p < .001). This suggests that preretirement health
and postretirement mental health had direct effects on postretirement health, whereas the
retirement transition itself didn‘t have a direct effect on postretirement health.
Second, the same model was estimated with log-transformed mental health at T
2
as a dependent variable. Similar to the self-rated health model, transition to retirement did
not have a total or direct effect on postretirement mental health (Figures 7A and 7B).
Model 7 in Table 3 indicates that self-rated health at T
2
was negatively related with
postretirement mental health, whereas baseline mental health was positively associated
with postretirement mental health (p < .001). In addition, being married or partnered was
associated with having fewer postretirement mental health problems (p < .01). In terms of
paths a
1
–a
4
, the results were identical to the self-rated health model because the same
model was estimated for each mediator as a dependent variable. That is, retirement was
not associated with any of the mediators (Figure 7B). In Model 8 of Table 3, in which the
indirect effects of retirement on postretirement mental health through financial control,
positive family relationships, negative family relationships, and social integration were
examined, financial control and negative family relationships had direct effects on
postretirement mental health. Thus, those who had less financial control (p < .001) and
52
Figure 7A. Total Effect of Retirement on Mental Health.
Significant path
Nonsignificant path
Reference group: those who did not retire
Figure 7B. Direct and Indirect Effects of Retirement on Mental Health
X: Retirement
(retired or not)
M1: Financial
Control
Y: Mental
Health T
2_LOG
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
b
1
= -.033
a
3
= - .031
a
4
= .051
b
2
= -.052
b
3
= .100
b
4
= -.017
X: Retirement
(retired or not)
Y: Mental
Health T
2_LOG
c = .002
a
1
= .264
a
2
= .026
c′ = .009
53
more negative relationships with family (p < .05) had more depressive symptoms.
However, in terms of the indirect effects of retirement on mental health through
mediators, none of the mediators influenced the effect of retirement on postretirement
mental health. Among the control variables, being married or partnered, baseline mental
health, and self-rated health at T
2
had direct effects on postretirement mental health.
The statistical significance of the indirect effects of the proposed mediators both
for health and mental health models were examined by constructing 95% Monte Carlo
confidence intervals. Unlike the causal step approach in which the significance of paths a
and b is examined to determine a mediation effect, the direction and size of indirect
effects are considered important in determining a mediation effect using confidence
intervals. Thus, although the indirect paths for both health and mental health models were
not significant (Figures 6B and 7B), indirect effects were tested using 95% Monte Carlo
confidence intervals to confirm the significance of the indirect effects.
As shown in Tables 4 and 5, none of the specific indirect effect or total indirect
effect was significant because 95% Monte Carlo confidence intervals for all indirect
effects included zero. Thus, results of Phase 1 analyses suggest that whether or not an
individual retires is not associated with follow-up health or mental health. In addition, the
effects of retirement on postretirement health and mental health were not mediated
through financial control, positive family relationships, negative family relationships, or
social integration.
Rather, an individual‘s previous self-rated health and postretirement mental health
were associated with postretirement health, whereas previous mental health and
54
postretirement health were associated with postretirement mental health. Thus, regardless
of retirement status, participants continued to have better health or mental health if they
had better health or mental health prior to retirement.
Table 4
Indirect Effects of Retirement on Self-Rated Health through Financial Control, Positive
Family Relationships, Negative Family Relationships, and Social Integration (paths ab)
Point Estimate
Monte Carlo 95% CI
Lower Upper
Retirement → Fin Control → Health 0.0095 -0.0060 0.0292
Retirement → Pos Relation → Health 0.0007 -0.0049 0.0020
Retirement → Neg Relation → Health 0.0042 -0.0044 0.0170
Retirement → Social Int → Health 0.0005 -0.0051 0.0072
Total Indirect Effect 0.0149 -0.0041 0.0377
Table 5
Indirect Effects of Retirement on Mental Health through Financial Control, Positive
Family Relationships, Negative Family Relationships, and Social Integration (paths ab)
Point Estimate
Monte Carlo 95% CI
Lower Upper
Retirement → Fin Control → MH -0.0087 -0.0250 0.0056
Retirement → Pos Relation → MH -0.0014 -0.0066 0.0028
Retirement → Neg Relation → MH -0.0031 -0.0115 0.0031
Retirement → Social Int → MH -0.0009 -0.0056 0.0025
Total Indirect Effect -0.0141 -0.0328 0.0021
55
5.3.2. Results for Phase 2: Relative Total, Direct, Total Indirect, and Specific
Indirect Effects of Voluntary and Involuntary Retirement on Self-Rated Health and
Mental Health (n = 1,280)
All the analytic procedures in Phase 2 were identical to Phase 1, except that a
multicategorical independent variable was applied. The independent variable in this
model was treated as a dummy variable, with participants who didn‘t retire functioning as
the reference group. Thus, separate estimates were created in this model: one for the
effect of voluntary retirement on health and mental health relative to no retirement
transition, and the other for the effect of involuntary retirement on health and mental
health relative to no retirement transition. The results of the analyses in this phase
allowed for a comparison of the mechanisms of the effect of involuntary and voluntary
retirement on self-rated health and mental health relative to no transition to retirement.
Results of multivariate analyses estimating the effects of voluntary and involuntary
retirement on self-rated health relative to no retirement transition are presented in Table 6
and Figures 8A and 8B. After controlling for sociodemographic variables, baseline health,
and log-transformed mental health at T
2
, involuntary retirement had a relative total effect
on self-rated health at T
2
, whereas voluntary retirement did not have a relative total effect
on self-rated health relative to no retirement transition (Model 5 in Table 6, Figure 8A).
In other words, participants who involuntarily retired had poorer health at T
2
than those
who did not retire (p < .01), whereas health at T
2
among voluntary retirees was not
significantly different than those who didn‘t retire. This result shows a clear difference in
terms of the total effect of retirement when it is considered as a whole and when its
56
Table 6
Complex Sample General Linear Model Coefficients (Retirement Transition as Multiple Categories; n = 1,280)
a
a
Unstandardized path coefficients and standard errors are reported in this table.
*
p < .05,
**
p < .01,
***
p < .001
Mediators (T
2
) DV1 (T
2
) DV2 (T
2
)
Fin Control Pos Relation Neg Relation Social Int Self-Rated Health Mental Health
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
Intercept 1.752 (.915) 1.161
***
(.229) 1.054
***
(.179) 1.079
**
(.303) .892
**
(.315) .904
*
(.390) .928
***
(.214) .971
***
(.244)
Invol. retirement -.334 (.264) -.002 (.062) -.002 (.046) -.044 (.072) -.296
**
(.107) -.279
*
(.108) .091 (.070) .090 (.067)
Vol. retirement .536
*
(.252) .038 (.040) -.044 (.030) .092 (.085) .094 (.078) .077 (.081) -.036 (.042) -.026 (.041)
Female .107 (.106) .068
*
(.032) .030 (.021) -.028 (.050) .131
**
(.043) .150
**
(.046) .056 (.033) .059 (.033)
Married/partnered -.187 (.147) .039 (.038) .053 (.029) -.017 (.051) -.003 (.047) .023 (.050) -.114
**
(.039) -.121
**
(.037)
White -.294 (.223) -.133
*
(.051) -.108
**
(.036) -.058 (.076) .049 (.065) .075 (.070) .035 (.045) .024 (.047)
Currently working .592
*
(.256) -.002 (.034) -.047 (.035) .078 (.077) .124 (.083) .095 (.085) -.013 (.053) -.001 (.051)
Age T
2
.027
*
(.011) -.001 (.003) .001 (.002) -.004 (.005) .005 (.005) .004 (.005) -.001 (.003) .001 (.003)
Education -.046 (.074) .025 (.018) -.019 (.013) .094
***
(.024) .046
*
(.022) .046 (.023) -.012 (.017) -.009 (.017)
Fin control T
1
.494
***
(.039) - - - - - - -
Pos relation T
1
- .648
***
(.034) - - - - - -
Neg relation T
1
- - .494
***
(.021) - - - - -
Social int T
1
- - - .241
***
(.027) - - - -
Health T
1
- - - - .569
***
(.026) .558
***
(.025) - -
MH T
2
_log - - - - -.335
***
(.038) -.279
***
(.041) - -
Health T
2
- - - - - - -.149
***
(.022) -.127
***
(.022)
MH T
1
_log - - - - - - .343
***
(.034) .303
***
(.035)
Fin control T
2
- - - - - .033
**
(.010) - -.032
***
(.007)
Pos relation T
2
- - - - - .025 (.030) - -.052 (.027)
Neg relation T
2
- - - - - -.135 (.073) - .101
**
(.040)
Social int T
2
- - - - - .005 (.031) - -.015 (.017)
R
2
.264 .404 .358 .149 .491 .501 .300 .326
57
Figure 8A. Total Effects of Voluntary and Involuntary Retirement on Self-Rated Health.
Significant path
Nonsignificant path
Reference group: those who did not retire
Figure 8B. Direct and Indirect Effects of Voluntary and Involuntary Retirement on
Self-Rated Health.
Dummy 1:
Voluntary
Retirement
M1: Financial
Control
Y: Health T
2
Dummy 2:
Involuntary
Retirement
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
c′
1
= .077
c′
2
= -.279
a
11
=.536
a
12
=.038
a
13
= - .044
a
14
= .092
a
21
= -.334
a
22
= -.002
a
23
= - .002
a
24
= - .044
b
1
= .033
b
2
= .025
b
3
= - .135
b
4
= .005
Dummy 1:
Voluntary
Retirement
Y: Health T
2
Dummy 2:
Involuntary
Retirement
c
1
= .094
c
2
= - .296
58
nature is specified as voluntary or involuntary (Figures 6A and 8A). On the other hand,
control variables that were significantly associated with postretirement health were the
same across Phase 1 and Phase 2. Being female (p < .01), education (p < .05), baseline
self-rated health (p < .001), and mental health at T
2
(p < .001) remained significantly
associated with postretirement health.
In terms of paths a
1
–a
4
, two estimates were created for each mediator, one for
voluntary retirement and the other for involuntary retirement. Models 1 to 4 in Table 6
features paths a
1
–a
4
for each mediator in the study, and the results are visualized in
Figure 8B. As shown in Figure 8B, although participants who voluntarily retired had
more financial control relative to those who didn‘t retire (p < .05), financial control
among involuntary retirees did not significantly differ compared to participants who
didn‘t retire. All three other mediators (positive family relationships, negative family
relationships, and social integration) were not predicted by voluntary or involuntary
retirement. The direct effects of voluntary retirement (c′
1
) and involuntary retirement (c′
2
)
on health as well as the direct effects of each mediator (paths b
1
–b
4
) were estimated
(Model 6 in Table 6). Results indicated that involuntary retirement had a direct effect on
postretirement health even after control variables and mediators were controlled (p < .05),
whereas transition to voluntary retirement did not have a direct effect on postretirement
health. Among mediators, only financial control had a direct effect on postretirement self-
rated health (p < .01). These results indicate that mechanism of the effect of retirement on
health is different depending on whether the nature of retirement is voluntary or
involuntary. As shown in Figure 8B, although transition to voluntary retirement exerted
59
its effect on postretirement health through financial control (p < .05), transition to
involuntary retirement exerted its effect directly on postretirement health (p < .05). In
other words, those who voluntarily retired had significantly more financial control after
their retirement relative to those who did not retire, and this difference led to better health
among voluntary retirees at T
2
. On the other hand, those who retired involuntarily had
poorer postretirement health compared to those who didn‘t retire, and their health was
directly affected by involuntary retirement rather than through another mechanism or
mediator.
The size and the direction of relative specific indirect effects as well as relative
total indirect effects of voluntary and involuntary retirement via each mediator are shown
in Table 7. The results of 95% Monte Carlo confidence intervals indicated that the
relative specific indirect effect of voluntary retirement on health through financial control
was significant after controlling for other mediators (p < .05) and the relative total
indirect effect of voluntary retirement on health through all four mediators as a set was
significant (p < .05). The three other mediators did not mediate the relationship between
voluntary retirement and health because the Monte Carlo confidence interval did not
include zero. These results suggest that financial control as well as all four mediators as a
set mediated the relationship between voluntary retirement and postretirement health. On
the other hand, none of the mediators examined in this study mediated the relationship
between involuntary retirement and health. These results not only indicate that the
mechanisms of voluntary and involuntary retirement differ in terms of their effects on
postretirement health, but also clearly illustrate that potential mechanisms can be
60
unobserved when retirement is considered as a whole without specifying its nature as
voluntary or involuntary (Figures 6B and 8B).
Table 7
Relative Specific Indirect Effects of Voluntary and Involuntary Retirement on Health
through Financial Control, Positive Family Relationships, Negative Family Relationships,
and Social Integration (paths ab)
Point Estimate
Monte Carlo 95% CI
Lower Upper
Voluntary → Fin Control → Health 0.0177
*
0.0001 0.0040
Voluntary → Pos Relation → Health 0.0010 -0.0021 0.0057
Voluntary → Neg Relation → Health 0.0059 -0.0023 0.0192
Voluntary → Social Int → Health 0.0005 -0.0078 0.0093
Total Indirect Effect 0.0251
*
0.0019 0.0520
Involuntary → Fin Control → Health 0.0110 -0.0313 0.0062
Involuntary → Pos Relation → Health -0.0001 -0.0056 0.0051
Involuntary → Neg Relation → Health 0.0003 -0.0146 0.0159
Involuntary → Social Int → Health -0.0002 -0.0060 0.0054
Total Indirect Effect 0.0110 -0.0372 0.0131
*
p < .05
The results of multivariate analyses estimating the effects of voluntary and
involuntary retirement on log-transformed mental health are shown in Table 6 and
Figures 9A and 9B. Model 7 in Table 6 shows that neither voluntary nor involuntary
retirement had a relative total effect on postretirement mental health after controlling for
sociodemographic variables, log-transformed baseline mental health, and self-rated health
at T
2
(Figure 9A). Being married or partnered (p < .01), log-transformed baseline mental
health (p < .001) and health at T
2
(p < .001) were significantly associated with log-
61
transformed mental health at T
2
, and these results were the same as the mental health
model in Phase 1.
In terms of paths a
1
–a
4
for mental health, the results were identical to the previous
health model in Phase 2 (Models 1–4 in Table 6), indicating that only voluntary
retirement transition was positively associated with financial control (p < .05). Model 8
of Table 6 estimated the relative specific indirect effects as well as direct effects of
voluntary and involuntary retirement on mental health relative to those who didn‘t retire.
As shown in Figure 9B, neither voluntary nor involuntary retirement had direct effects on
postretirement mental health. Among mediators, both financial control (p < .001) and
negative relationships with family (p < .01) were associated with postretirement mental
health.
Thus, in terms of the relative specific indirect effect of each mediator, only
financial control mediated the relationship between voluntary retirement and
postretirement mental health after controlling for the three other mediators (p < .05). That
is, those who retired voluntarily had more financial control after retirement relative to
those who do not retire, and this difference was associated with fewer mental health
problems among voluntary retirees. Although negative family relationships were
positively associated with more mental health problems, they did not mediate the effect
of voluntary retirement or involuntary retirement on postretirement mental health. In
addition, being married or partnered, log-transformed baseline mental health, and health
at T
2
were significant predictors of log-transformed mental health at T
2
. The results of the
mediation analyses in this phase suggested that the mechanisms of the effect of voluntary
62
Figure 9A. Total Effects of Voluntary and Involuntary Retirement on Mental Health.
Significant path
Nonsignificant path
Reference group: those who did not retire
Figure 9B. Direct and Indirect Effects of Voluntary and Involuntary Retirements on
Mental Health.
Dummy 1:
Voluntary
Retirement
M1: Financial
Control
Y: Mental
HealthT
2
_
LOG
Dummy 2:
Involuntary
Retirement
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
c′
1
= - .026
c′
2
= .90
a
11
=.536
a
12
=.038
a
13
= - .044
a
14
= .092
a
21
= -.334
a
22
= -.002
a
23
= - .002
a
24
= - .044
b
1
= - .032
b
2
= - .052
b
3
= .101
b
4 =
- .015
Dummy 1:
Voluntary
Retirement
Dummy 2:
Involuntary
Retirement
Y: Mental
HealthT
2
_
LOG
c
1
= - .036
c
2
= .091
63
and involuntary retirements on postretirement mental health were different (Figure 9B).
Although the positive effect of voluntary retirement on mental health compared to those
who had not retired occurred via increased financial control, involuntary retirement did
not exert any direct or indirect effect on postretirement mental health relative to
nonretired participants. Moreover, mechanisms of the effect of retirement were clearly
different when its nature was specified as voluntary or involuntary compared to when it
was not specified (Figure 7B and 9B).
The results of relative specific indirect effects and relative total indirect effects
obtained in the mediation analyses were further confirmed by testing their significance.
Results shown in Table 8 indicate that the relative specific indirect effect of voluntary
retirement on mental health through financial control was significant after controlling for
other mediators (p < .05). In addition, the relative total indirect effect of voluntary
retirement on mental health via all four mediators as a set was significant (p < .05).
However, none of the mediators examined in the study, neither individually nor as a set,
mediated the relationship between involuntary retirement and mental health at T
2
. These
results suggest that financial control as well as all four mediators as a set mediated the
relationship between voluntary retirement and postretirement mental health. On the other
hand, none of the mediators mediated the relationship between involuntary retirement and
mental health.
64
Table 8
Relative Specific Indirect Effects of Voluntary and Involuntary Retirement on Mental
Health through Financial Control, Positive Family Relationships, Negative Family
Relationships, and Social Integration (paths ab)
Point Estimate
Monte Carlo 95% CI
Lower Upper
Voluntary → Fin Control → MH -0.0172
*
-0.0364 -0.0009
Voluntary → Pos Relation → MH -0.0020 -0.0079 0.0022
Voluntary → Neg Relation → MH -0.0044 -0.0128 0.0013
Voluntary → Social Int → MH 0.0014 -0.0024 0.0073
Total Indirect Effect -0.0222
*
-0.0450 -0.0030
Involuntary → Fin Control → MH 0.0107 -0.0059 0.0292
Involuntary → Pos Relation → MH 0.0001 -0.0076 0.0079
Involuntary → Neg Relation → MH -0.0002 -0.0109 0.0098
Involuntary → Social Int → MH -0.0007 -0.0051 0.0024
Total Indirect Effect 0.0099 -0.0119 0.0326
*
p < .05
5.3.3. Results for Phase 3: Total, Direct, Total Indirect, and Specific Indirect
Effects of Involuntary Retirement on Self-Rated Health and Mental Health (n = 429)
Although the results of Phase 2 provided informative contrasts of mechanisms of
the effects of voluntary and involuntary retirement on health and mental health, these
mechanisms were examined relative to those who did not retire. Thus, direct comparison
between those who retired voluntarily and involuntarily was not made. The third phase of
analyses examined the effect of involuntary retirement on health and mental health
relative to voluntary retirement. To achieve this purpose, only those who transitioned to
retirement were considered and retirement-specific variables such as age at retirement
and retirement duration were controlled by including them in the model.
65
Models estimating the mechanisms of the effect of involuntary retirement on
postretirement health relative to voluntary retirement are presented in Table 9 and Figures
10A and 10B. Model 5 in Table 9 shows that involuntary retirement had a total effect on
postretirement health after controlling for sociodemographics, retirement-related
variables, baseline health, and log-transformed mental health at T
2
(p < .001).
Postretirement health among involuntary retirees was approximately 0.36 units lower
than that of voluntary retirees. In addition, education (p < .05), baseline health (p < .001),
and mental health at T
2
(p < .001) were significant predictors of postretirement health.
Models 1 to 4 in Table 9 represent paths a
1
–a
4
for each mediator after controlling for
sociodemographics, retirement-related variables, and other baseline mediators. The
results showed that involuntary retirement only predicted postretirement via financial
control (p < .01) and not the other three mediators. In Model 6 (Table 9), in which direct
effects of involuntary retirement as well as all four mediators were estimated, involuntary
retirement (p < .001) and negative family relationships (p < .05) had direct effects on
postretirement health (Figure 10B). In terms of control variables, being female (p < .05),
education (p < .05), baseline self-rated health (p < .001) and mental health at T
2
(p < .001)
were significantly associated with postretirement self-rated health. On the other hand,
none of the retirement-related variables such as age at retirement and retirement duration
had significant relationships with postretirement health. These results indicate that
although involuntary retirees had less perceived financial control after retirement
compared to voluntary retirees, it did not lead to lower self-rated health after retirement.
Instead, involuntary retirement affected postretirement health more directly rather than
66
Table 9
Complex Sample General Linear Model Coefficients (Retirement Transition as Binary; n = 429)
a
a
Unstandardized path coefficients and standard errors are reported in this table.
*
p < .05,
**
p < .01,
***
p < .00
Mediators (T
2
) DV1 (T
2
) DV2 (T
2
)
Fin Control Pos Relation Neg Relation Social Int Self-Rated Health Mental Health
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
Intercept 3.281 (1.824) 1.877
***
(.323) 1.190
***
(.268) .703 (.539) 1.162
*
(.548) 1.677
**
(.601) .686 (.499) 1.171
*
(.490)
Invol. retirement -1.025
**
(.307) -.088 (.059) .045 (.037) -.122 (.080) -.360
***
(.086) -.341
***
(.090) .145
*
(.063) .120 (.060)
Female -.297 (.238) .034 (.043) .055 (.050) -.013 (.091) .084 (.072) .168
*
(.074) .013 (.049) .006 (.049)
Married/partnered -.167 (.267) .073 (.064) .014 (.060) .054 (.094) .082 (.099) .130 (.100) -.180
*
(.075) -.156
*
(.071)
White -.284 (.381) -.097 (.082) -.068 (.070) -.112 (.133) -.110 (.117) -.092 (.122) .014 (.094) -.031 (.097)
Currently working -.168 (.268) -.069 (.050) .025 (.041) .090 (.088) .174 (.117) .170 (.117) .069 (.065) .045 (.062)
Age at retirement .025 (.023) -.009 (.005) .000 (.004) .001 (.007) .006 (.008) .003 (.008) .004 (.006) .005 (.006)
Ret duration -.007 (.007) .000 (.001) .000 (.001) .004 (.002) -.001 (.003) -.001 (.002) .001 (.002) .001 (.002)
Education -.021 (.123) -.034 (.018) -.039
*
(.019) .112
**
(.037) .095
*
(.038) .085
*
(.039) -.023 (.027) -.025 (.030)
Fin control T
1
.455
***
(.080) - - - - - - -
Pos relation T
1
- .634
***
(.047) - - - - - -
Neg relation T
1
- - .452
***
(.041) - - - - -
Social int T
1
- - - .249
***
(.051) - - - -
Health T
1
- - - - .500
***
(.057) .480
***
(.058) - -
MH T
2
_log - - - - -.371
***
(.084) -.322
**
(.090) - -
Health T
2
- - - - - - -.150
***
(.039) -.123
**
(.038)
MH T
1
_log - - - - - - .326
***
(.055) .283
***
(.056)
Fin control T
2
- - - - - .031 (.018) - -.040
**
(.012)
Pos relation T
2
- - - - - -.055 (.076) - -.088 (.054)
Neg relation T
2
- - - - - -.194
*
(.092) - .010 (.086)
Social int T
2
- - - - - -.020 (.049) - -.051 (.034)
R
2
.238 .456 .345 .196 .526 .529 .366 .398
67
Figure 10A. Total Effect of Involuntary Retirement on Self-Rated Health.
Significant path
Nonsignificant path
Reference group: those who retired voluntarily
Figure 10B. Direct and Indirect Effects of Involuntary Retirement on Self-Rated Health.
X: Retirement
(involuntary vs.
voluntary)
M1: Financial
Control
Y: Health T
2
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
a
1
= - 1.025
b
1
= .031
a
2
= - .088
a
3
= .045
a
4
= -.122
b
2
= -.055
b
3
= -.194
b
4
= -.020
X: Retirement
(involuntary vs.
voluntary )
Y: Health T
2
c = -.360
c′ = -.341
68
through other mechanisms, even after controlling for sociodemographics, retirement-
related variables, baseline health measure, and mental health at T
2
.
Significance tests of the indirect effects of involuntary retirement on health
through financial control, positive family relationships, negative family relationships, and
social integration were conducted and the results indicated that all four mediators, either
independently or as a set, did not mediate the relationship between involuntary retirement
and postretirement health (Table 10).
Table 10
Indirect Effects of Involuntary Retirement on Health through Financial Control, Positive
Family Relationships, Negative Family Relationships, and Social Integration (paths ab)
Point Estimate
Monte Carlo 95% CI
Lower Upper
Involuntary → Fin Control → Health -0.0318 -0.0812 0.0036
Involuntary → Pos Relation → Health 0.0048 -0.0103 0.0263
Involuntary → Neg Relation → Health -0.0087 -0.0301 0.0057
Involuntary → Social Int → Health 0.0024 -0.0115 0.0198
Total Indirect Effect -0.0333 -0.0884 0.0137
Total, specific indirect, and total indirect effects of involuntary retirement on log-
transformed mental health were estimated and the results are given in Table 9 and Figures
11A and 11B. Similar to the health model, involuntary retirement had a total effect on
postretirement mental health (p < .05). That is, participants who involuntarily retired had
more mental health problems compared to those who voluntarily retired after controlling
for sociodemographics, retirement variables, log-transformed baseline mental health, and
health at T
2
(Figure 11A). Although the total effect of retirement on mental health was
69
Figure 11A. Total Effect of Involuntary Retirement on Mental Health.
Significant path
Nonsignificant path
Reference group: those who retired voluntarily
Figure 11B. Direct and Indirect Effects of Involuntary Retirement on Mental Health.
X: Involuntary
Retirement
M1: Financial
Control
Y: Mental
Health T
2_LOG
M2: Positive
Family
Relationships
M3: Negative
Family
Relationships
M4: Social
Integration
b
1
= -.040
a
2
= - .088
a
3
= .045
b
2
= -.088
b
3
= .010
b
4
= - .051
c′ =.120
X: Involuntary
Retirement
Y: Mental
Health T
2_LOG
c =.145
a
4
= -.122
a
1
= - 1.025
70
not significant when its nature was not specified as voluntary or involuntary as well as
when involuntary retirement was compared to participants who had not retired, there was
a significant total effect of involuntary retirement on mental health when it was compared
to voluntary retirement. In addition, being married or partnered (p < .05), log-transformed
baseline mental health (p < .001), and health at T
2
(p < .001) were significantly associated
with mental health at T
2
.
Model 8 in Table 9 shows the estimates of direct effects of involuntary retirement
as well as four mediators on postretirement mental health, and they are visualized in
Figure 11B. The results indicate that only financial control was a significant predictor of
mental health at T
2
(p < .01) and that the three other mediators were not associated with
postretirement mental health. As shown in Figure 11B, involuntary retirement exerted its
effect on postretirement mental health through financial control (p < .01). That is, those
who experienced involuntary retirement tended to have less financial control after
retirement compared to voluntary retirees, and this perception of less financial control
was associated with having more mental health problems after retirement. In terms of
control variables, being married or partnered (p < .05), log-transformed baseline mental
health (p < .001), and self-rated health at T
2
(p < .01) were significantly associated with
postretirement mental health.
Again, significance tests were conducted for all potential indirect paths using 95%
Monte Carlo confidence intervals. As shown in Table 11, the results of significance tests
of specific indirect effects and the total indirect effect indicate that the indirect effects of
involuntary retirement on mental health through financial control and all four mediators
71
as a set were significant (p < .05). Positive family relationships, negative family
relationships, and social integration did not mediate the relationship between involuntary
retirement and postretirement mental health. Also, it is worthwhile to note that the
significant total effect of involuntary retirement on postretirement mental health became
nonsignificant when the mediators were entered in the model.
Table 11
Indirect Effects of Involuntary Retirement on Mental Health through Financial Control,
Positive Family Relationships, Negative Family Relationships, and Social Integration
(paths ab)
Point Estimate
Monte Carlo 95% CI
Lower Upper
Involuntary → Fin Control → MH 0.0410
*
0.0117 0.0806
Involuntary → Pos Relation → MH 0.0082 -0.0033 0.0272
Involuntary → Neg Relation → MH 0.0005 -0.0098 0.0114
Involuntary → Social Int → MH 0.0062 -0.0030 0.0218
Total Indirect Effect 0.0559
*
0.0183 0.1007
*
p < .05
5.4. Summary of Results for Hypotheses Testing
With regard to examining mechanisms of the effect of involuntary retirement on
both health and mental health, three hypotheses were constructed. First, it was
hypothesized that transitioning to involuntary retirement would have a total effect on self-
rated health and mental health after controlling for sociodemographics and baseline
outcome variables. Hypothesis 1 was partially supported in that transition to involuntary
retirement had a total effect on self-rated health, both compared to participants who had
72
not retired and those who retired voluntarily. However, in terms of mental health,
transition to involuntary retirement had a total effect on postretirement mental health only
compared to those voluntarily retired but not to those who didn‘t retire.
In Hypothesis 2, it was hypothesized that transition to involuntary retirement
would predict financial control, positive family relationships, negative family
relationships, and social integration after controlling for sociodemographics and baseline
outcome variables. Hypothesis 2 was also partially supported. Transition to involuntary
retirement predicted financial control when compared to voluntary retirees. However,
involuntary retirement did not predict financial control when compared to those who did
not retired. In addition, involuntary retirement did not predict any of the other three
mediators: positive family relationships, negative family relationships, and social
integration.
Lastly, Hypothesis 3 hypothesized that transition to involuntary retirement would
predict self-rated health and mental health, and these relationships would be mediated by
financial control, positive family relationships, negative family relationships, and social
integration after controlling for sociodemographics and baseline outcome variables.
Hypothesis 3 was partially supported in this study. In terms of postretirement self-rated
health, the effect of involuntary retirement was not mediated by any of the mediators
examined in the study. However, in terms of postretirement mental health, the effect of
involuntary retirement relative to voluntary retirement was mediated by financial control.
Thus, all three hypotheses of this study were partially supported.
73
CHAPTER VI: DISCUSSION
6.1. Summary of Major Findings
Using nationally representative longitudinal data from the Health and Retirement
Study, the current study examined two specific research aims: (1) to explore
characteristics of involuntary retirees in comparison with voluntary retirees and those
who did not retire, and (2) to examine the mechanisms of the effect of involuntary
retirement on health and mental health in comparison with voluntary retirees and those
who did not retire. These research aims were accomplished by conducting univariate,
bivariate, and regression-based path analyses using the SPSS Complex Sample procedure
to correct for the complex sampling design of HRS.
6.1.1. Prevalence and Characteristics of Involuntary Retirees
Between 2006 and 2010, approximately 29.3% (n = 429) of the sample
transitioned to retirement. Among those retirees, 37.2% (n = 155) perceived that their
retirement was involuntary. This indicates that a substantial number of older adults who
retire do not have control over their retirement decisions and that involuntary retirement
is fairly prevalent. The proportion of involuntary retirees reported in this study is
relatively higher compared to previous studies. In their study examining the prevalence of
involuntary retirees using HRS, Lachance and Seligman (2008) reported that about 27%
of retirees experienced involuntary retirement. This discrepancy may be due to different
sample selection processes as well as use of different waves of data. However,
considering the fact that Lachance and Seligman (2008) used HRS data from before 2004
and this study used HRS data from 2006 and beyond, it is possible that the prevalence of
74
involuntary retirees increased due to the economic recession that occurred near the end of
2007.
Overall observations of the characteristics of involuntary retirees indicate that
involuntary retirees had less human capital even in preretirement compared to those who
retired voluntarily or those who didn‘t retire. For example, involuntary retirees‘
educational level was significantly lower than that of voluntary retirees and those who
didn‘t retire, and the proportion of involuntary retirees who were married or partnered
was also significantly lower compared to voluntary retirees. In addition, baseline
measures of financial control, health, and mental health for involuntary retirees were
significantly different compared to the other two groups. Preretirement financial control
was significantly lower for involuntary retirees compared to voluntary retirees, and their
baseline health and mental health were also significantly worse compared to both
voluntary retirees and those who had not retired. In addition, there were fewer
involuntary retirees who were working for pay despite the fact that they were relatively
younger than voluntary retirees. Involuntary retirees also retired almost 2 years earlier
than voluntary retirees.
These findings are in accordance with previous research on involuntary retirement
as well as involuntary job loss. Research reporting the effect of human capital on
retirement or labor force participation has suggested that those with lower levels of
education are more likely to perceive their retirement as forced (Szinovacz & Davey,
2004) and more likely to be unemployed or out of labor force (Flippen & Tienda, 2000).
Flippen and Tienda (2000) also reported that men who are not married or partnered have
75
higher rates of unemployment and nonparticipation in the labor market, whereas women
who are unmarried, divorced, or widowed are more likely to stay in the labor force and
not retire compared to those who are married or partnered. However, the effect of human
capital weakened when health and job characteristics were accounted for, indicating that
those with less human capital were pushed out of the labor force indirectly due to poor
health and unstable work histories (Flippen & Tienda, 2000; Szinovacz & Davey, 2004).
Characteristics of involuntary retirees in the present study also indicate the potential
possibility of them being pushed out of the labor force due to poor health and mental
health. Despite the fact that less preretirement financial control implied a greater need to
maintain employment, involuntary retirees left the labor force about 2 years earlier than
voluntary retirees and were less likely to be working for pay. Considering the fact that
involuntary retirees had poorer health and more depressive symptoms at baseline
compared to voluntary retirees and those who had not retired, it is possible that they were
pushed out of the labor force involuntarily due to their poor health and mental health and
were not able to return to the labor force as easily as voluntary retirees.
6.1.2. Mechanisms of the Effect of Involuntary Retirement on Self-Rated
Health
Involuntary retirement had a significant total effect on postretirement health
compared to no transition to retirement and voluntary retirement after controlling for
sociodemographic and retirement-related variables, baseline health, and mental health at
T
2
. Health after retirement for involuntary retirees was about 0.30 and 0.36 units lower
than nonretired participants and voluntary retirees, respectively. This finding is indicative
76
of the significant negative health effect of involuntary retirement and is consistent with
previous findings regarding the adverse health effects of involuntary job loss,
unemployment, and involuntary retirement (Gallo et al., 2000; Strully, 2009; van Solinge,
2007). Although the adverse health effects of involuntary job loss and involuntary
retirement has been consistently reported, results regarding the effect of retirement have
been relatively mixed. Some studies reported that retirement increases the risk of various
health problems (Behncke, 2009), whereas others reported that retirement does not harm
health or has a positive effect (Bound & Waidmann, 2007; Neuman, 2008). This
discrepancy may be partly due to the failure to specify the nature of retirement because
evidence has suggested that lack of control over retirement is an important factor that has
adverse effects on health even after controlling for health as a reason for retirement (van
Solinge, 2007). This was supported by the results of Phase 1 of this study, in which the
voluntariness of retirement was not considered and retirement did not have any
significant adverse health effects.
In Phase 2, the mechanisms of voluntary and involuntary retirements were
compared to those who did not make the transition into retirement, and this provided a
useful contrast of mechanisms of voluntary and involuntary retirement. Although the
effect of voluntary retirement on health was indirect through financial control, the effect
of involuntary retirement on health was direct. Compared to those who did not retire,
voluntary retirees perceived more financial control after their retirement and that
financial control was associated with better health. On the other hand, transitioning to
77
involuntary retirement affected health directly rather than through the mediators
examined in the study.
Differences in the mechanisms of voluntary and involuntary retirement can be
partially understood based on the context in which older adults make decisions to retire.
Health and financial reasons are known to be key factors of retirement decisions (Barnes-
Farrell, 2003; Hatcher, 2003; Topa, Moriano, Depolo, Alcover, & Morales, 2009).
Because individuals may decide to retire when they feel that they have accumulated
enough financial resources, it is more likely for individuals with more financial resources
to retire earlier or voluntarily (Feldman, 1994; Hatcher, 2003). In addition, individuals
with higher income or wealth—whether it be absolute or relative—are known to have
better health outcomes because they have greater resources to promote their health status
(Bloom & Canning, 2000; Subramanian & Kawachi, 2006; Subramanyam et al., 2009).
Thus, individuals may retire voluntarily when they perceive that they have sufficient
financial resources for their retirement years and they may use the resources to maintain
their health after retirement.
The direct effect of involuntary retirement on health can also be understood
within the context of the retirement decision-making process. The results of this study
show that the adverse health effects of involuntary retirement remained dominant even
after controlling for sociodemographics, retirement-specific variables, baseline health,
and mental health at T
2
. Although we can conclude that involuntary retirement harms
health, it is crucial not to ignore the fact that baseline health was a strong predictor of
postretirement health across the analyses of this study. This suggests the possibility of
78
health selection effects—that those in poor health are more likely to be forced to retire
rather than vice versa. This perspective is also supported by research indicating that poor
health is an important predictor of labor force withdrawal and early retirement (Bound,
Stinebrickner, & Waidmann, 2010; Dwyer & Mitchell, 1999; van den Berg, Elders, &
Burdorf, 2010) and earlier life experiences are associated with health outcomes in older
age (Hayward & Gorman, 2004). Thus, in terms of the direct effect of involuntary
retirement on postretirement health, health selection effects as well as health causation
effects seem to exist. This finding is consistent with Gallo and colleagues‘ (2000) study,
in which they reported that the effect of involuntary job loss on health at older age is a
combination of health selection and health causality effects.
6.1.3. Mechanisms of the Effect of Involuntary Retirement on Mental Health
Results regarding the mechanisms of the effect of involuntary retirement on
mental health indicate that involuntary retirement has an adverse mental health effect
compared to voluntary retirement after controlling for sociodemographics, retirement-
specific variables, baseline mental health, and health at T
2
. This result echoes previous
empirical findings that involuntary job loss or retirement has negative effects on various
psychosocial outcomes such as depression (Brand, Levy, & Gallo, 2008; Gallo et al.,
2000), well-being (Calvo & Sarkisian, 2011), and life satisfaction (Hershey & Henkens,
2013). Control over retirement is one of the key factors associated with the happiness of
older adults during retirement (Calvo et al., 2009). In fact, involuntary retirement can be
profoundly different from voluntary retirement in that it often occurs under very
restrictive conditions such as poor health, job loss, and family responsibilities, whereas
79
voluntary retirement represents opportunities such as seeking new experiences and
spending more time with family (Lachance & Seligman, 2008). Thus, the fact that older
adults not only retired involuntarily but also the negative and challenging reasons that
forced them into retirement may aggravate their mental health problems.
Whereas the effect of involuntary retirement on self-rated health was direct, the
effect of involuntary retirement on mental health was transmitted through financial
control. This effect was only present when compared to voluntary retirees. As described
earlier in reference to the health model, financial resources are a key factor that allows
individuals retire voluntarily (Feldman, 1994). Thus, involuntary retirees who often
experience abrupt and unplanned retirement transitions tend to have greater economic
vulnerability (Baxter, 2010; Lachance & Seligman, 2008). According to McKee-Ryan
and colleagues (2005), financial resources can be significant coping resources that
facilitate well-being and mental health during unemployment. In addition, Gallo, Bradley,
and colleagues (2006), in their study of depression among older workers who
experienced involuntary job loss, found that the mental health effect of involuntary job
loss was significant only among those who had below-median net worth. Thus,
considering the relative financial vulnerability of involuntary retirees as well as the
negative impact of financial hardship on mental health (Boes & Winkelmann, 2006;
Vinokur, Price, & Caplan, 1996), the adverse mental health effect of involuntary
retirement may operate through lower financial control that is accompanied by
involuntary retirement, rather than involuntary retirement itself.
80
6.2. Limitations and Suggestions for Future Studies
The current study has several limitations that must be acknowledged in terms of
interpreting its results. First, causal relationships among the study variables need to be
drawn with caution. Although this study attempted to strengthen causal inferences by
controlling for baseline measures, both mediators and dependent variables of the study
were measured at T
2
, which limits drawing causal inferences among study variables. This
was because the measures of the mediators were from the leave-behind questionnaire,
which had only two data points, 2006 and 2010. Thus, in terms of examining mechanisms
or pathways of the effect of involuntary retirement on health and mental health, future
studies should use longitudinal data with three or more repeated measurements to allow
more rigorous statistical tests. Second, major study variables such as retirement, health,
and mental health as well as the four potential mediators were measured based on self-
reported data. Although widely used, self-reported data can potentially have reporting
bias and it is difficult for researchers to know whether the construct is being measured
consistently across participants due to subjectivity. Thus, future studies may complement
self-reported measures with more objective measures. For example, along with self-
assessed retirement status, researchers may consider measuring work hours, earnings, and
receipt of retirement benefits in terms of defining retirement. In addition, a wider variety
of objective health measures should be employed to assess physical health. Third, the
potential role of external environmental factors such as the financial crisis between 2007
and 2009, as well as gender or cohort differences, was not examined in this study. This
limitation can be addressed in future studies by conducting subgroup analyses to examine
81
if the mechanisms of the effect of involuntary retirement on health and mental health are
different among different groups of people. Fourth, because the study sample was largely
of White racial/ethnic descent, further exploration of mechanisms among different
ethnicities was limited. It is recommended that future studies address this limitation by
replicating the study with underrepresented populations whose members may experience
involuntary retirement more frequently than individuals of other races and ethnicities.
6.3. Implications
6.3.1. Implications for Retirement Policy and Research
This study attempted to shed light on the adverse health and mental health effects
of involuntary retirement by examining their mechanisms, and its findings provide
important implications for retirement policy and research. First, retirement researchers
need to expand their investigation of the various aspects of involuntary retirement.
Previous research on involuntary retirement, including this study, found that a substantial
number of older adults transition to involuntary retirement, ranging from 27% to 37.2%
(Lachance & Seligman, 2008; Szinovacz & Davey, 2004). This indicates that one in three
retirees perceive their retirement as forced. However, retirement has been often
considered a voluntary decision and thus not much attention has been paid to various
aspects of the experiences of involuntary retirees. Although recent research on
involuntary retirement revealed considerable information about its negative health and
mental health effects (Hershey & Henken, 2013; Shultz, Morton, & Weckerle, 1998; van
Solinge, 2007), little is known about the mechanisms or pathways of how involuntary
82
retirement leads to poor health or mental health. Thus, more in-depth exploration of
involuntary retirement is needed.
Second, retirement policies that incorporate the diverse forms and nature of
retirement need to be developed instead of a one-size-fits-all approach. Analysis of the
characteristics of involuntary retirees in this study indicates that older adults who are
forced to retire are different than those who retire voluntarily mainly in terms of their
human capital resources. That is, older adults who experience involuntary retirement are
more likely to be less educated, not have a spouse, have less financial control both before
and after retirement, and have worse health and mental health both before and after
retirement compared to voluntary retirees. In addition, evidence has suggested that the
motivation for retirement is clearly different (Lachance & Seligman, 2008; Shultz et al.,
1998). Therefore, when policy makers do not consider these differences when developing
retirement policies and create monolithic policies, they are not able to efficiently serve
and support involuntary retirees. Researchers also need to incorporate the diversity of
retirement experiences and consider differences in the nature of involuntary and
voluntary retirement. The Phase 1 results of this study, in which the negative health and
mental health effects of involuntary retirement as well as their mechanisms were
unobserved, are a good example of how potential effects or mechanisms can be masked
or unobserved when the nature of retirement is not specified.
Third, government and private organizations need to develop policies and
programs that reduce the shock as well as emotional and financial burdens that can be
caused by involuntary retirement. Although retirement decisions may appear to be made
83
at the individual level, retirement policies of governments and organizations exert
influence on individual retirement decisions. For example, due to a series of recent
changes in retirement policies such as an increase in the minimum retirement age to
receive full Social Security benefits, a dramatic decrease in the provision of defined-
benefit employer pension programs, and limited availability of employer-sponsored
retiree health insurance (Social Security Administration, 2013; Wheaton & Crimmins,
2013), individuals have chosen to stay in the labor force longer by continuing to work
beyond retirement age. Under these circumstances, in which individuals have more
responsibility than in the past to secure financial resources, being pushed out of the labor
force can create even greater financial hardship for involuntary retirees. Because older
adults are forced to retire several years before the traditional retirement age, they may not
be eligible for benefit programs provided by government or employers. On the other hand,
despite their greater financial needs, they face challenges in finding employment due to
ageism, poor health, and lack of human capital (Dennis & Thomas, 2007; Flippen &
Tienda, 2000; Lahey, 2005). Therefore, both government and organizations need to
manage these older employees who are at higher risk of experiencing involuntary
retirement so that they can promote their health and plan their retirement. In addition,
more proactive attempts need to be made by government and organizations in terms of
eradicating ageism in the labor market and providing flexible work arrangements that
may allow involuntary retirees to maintain or regain employment despite limited health
and human capital. In terms of research, more studies should focus on the macro level so
that the impact of retirement policies on involuntary retirement can be better understood.
84
6.3.2. Implications for Social Work Practice
Findings of this study also provide important implications for social work practice.
Given that older adults who are pushed into retirement tend to be a vulnerable group with
multiple issues of poor health and mental health, lack of financial and human capital
resources, and unemployment, there is great potential for social work practitioners to
engage with them to provide needed support and services. It is important for social work
practitioners to understand the characteristics of involuntary retirees and the potential
mechanisms of the effect of involuntary retirement on self-rated health and mental health.
Understanding these mechanisms can provide more insight for social workers regarding
when and how to intervene.
This study indicates that the adverse health effect of involuntary retirement is
direct, whereas financial control is a significant mediator of the effect of involuntary
retirement on mental health. Thus, to address the direct effect of involuntary retirement
on detrimental health, social work interventions that improve human capital resources
among older adults can be implemented. For instance, development of health promotion
interventions within communities or workplaces can enhance older adults‘ health by
teaching them how to better manage their health. Because health is one of the key factors
that can be both an antecedent and consequence of involuntary retirement, significant
emphasis needs to be placed on promoting older adults‘ health. In addition, services that
enhance financial control among involuntary retirees need to be provided. For example,
financial literacy programs for people with relatively less human and financial capital can
be developed to provide information on how to plan their finances and address financial
85
hardship after involuntary retirement. Because those who involuntarily leave the
workforce have greater needs for reemployment or bridge employment (Wang & Schultz,
2010), programs that enhance their job skills or services that help them seek jobs need to
be redesigned specifically for older involuntary retirees. Through these various programs,
social work practitioners can help older adults avoid experiencing involuntary retirement
and economic vulnerability. Lastly, social work practitioners need to pay attention to
psychosocial aspects of involuntary retirement and address them appropriately because
involuntary retirees often experience a lack of control over their employment situation,
which may lead to poorly managing their health and mental health.
6.4. Conclusion
This study contributes to retirement research by showing the significant need to
specify the nature of retirement as voluntary or involuntary and by shedding light on the
mechanisms of voluntary and involuntary retirement in terms of their effects on health
and mental health. During the course of this study, more in-depth information regarding
the experience of involuntary retirement was demonstrated, and this information can be
useful for policy makers, employers, and social work practitioners seeking to address
issues faced by involuntary retirees. Although the decision to retire appears to be an
individual choice, not everyone has control over this decision. Thus, it is important to pay
more attention to individuals who are forced out of the labor market because they are
more likely to suffer from detrimental aftereffects of involuntary retirement. Proactive
and multidimensional collaboration among policy makers, employers, and social work
86
practitioners is required to address the involuntary retirement issues discussed in this
paper.
87
References
Administration on Aging. (2011). A profile of older Americans: 2011. Retrieved from
http://www.aoa.gov/Aging_Statistics/Profile/2011/docs/2011profile.pdf
Barnes-Farrell, J. L. (2003). Beyond health and wealth: Attitudinal and other influences
on retirement decision-making. In G. A. Adams & T. A. Beehr (Eds.), Retirement:
Reasons, processes, and results (pp. 159–187). New York, NY: Springer.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in
social psychological research: Conceptual, strategic, and statistical consideration.
Journal of Personality and Social Psychology, 51, 1173–1182. doi:10.1037/0022-
3514.51.6.1173
Baxter, L. E. (2010). Income and life satisfaction among voluntary vs. involuntary
retirees (Master‘s thesis, University of Tennessee). Retrieved from
http://trace.tennessee.edu/utk_psycpubs/3/
Behncke, S. (2009). How does retirement affect health? (IZA Discussion Paper No.
4253). Retrieved from Institute for the Study of Labor website:
http://ftp.iza.org/dp4253.pdf
Behncke, S. (2012). Does retirement trigger ill health? Health Economics, 21, 282–300.
doi:10.1002/hec.1712
Bloom, D. E., & Canning, D. (2000). The health and wealth of nations. Science, 287,
1207–1209. doi:10.1126/science.287.5456.1207
88
Boes, S., & Winkelmann, R. (2006). The effect of income on positive and negative
subjective well-being (Working Paper No. 0605). Zurich, Switzerland: University
of Zurich, Socioeconomic Institute.
Bound, J., Stinebrickner, T., & Waidmann, T. (2010). Health, economic resources and the
work decisions of older men. Journal of Econometrics, 156, 106–129.
doi:10.1016/j.jeconom.2009.09.010
Bound, J., & Waidmann, T. (2007). Estimating the health effects of retirement (Working
Paper 2007-168). Ann Arbor, MI: University of Michigan, Michigan Retirement
Research Center. Retrieved from
http://www.mrrc.isr.umich.edu/publications/papers/pdf/wp168.pdf
Brand, J. E., Levy, B. R., Gallo, W. T. (2008). Effects of layoffs and plant closings on
depression among older workers. Research on Aging, 30, 701–721.
doi:10.1177/0164027508322574
Brown, M., Aumann, K., Pitt-Catsouphes, M., Galinsky, E., & Bond, J. T. (2010).
Working in retirement: A 21st century phenomenon. Retrieved from Families and
Work Institute website:
http://familiesandwork.org/site/research/reports/workinginretirement.pdf
Brown, S. K. (2012). What are older workers seeking?: An AARP/SHRM survey of 50+
workers. Washington, DC: AARP. Retrieved from
http://www.aarp.org/content/dam/aarp/research/surveys_statistics/econ/2012/Wha
t-Are-Older-Workers-Seeking-An-AARP-SHRM-Survey-of-50-Plus-Workers-
AARP.pdf
89
Calvo, E., Haverstick, K., & Sass, S. (2007). What makes retirees happier: A gradual or
‘cold turkey ’ retirement? Chestnut Hill, MA: Center for Retirement Research at
Boston College.
Calvo, E., & Sarkisian, N. (2011). Retirement and well-being: Examining the
characteristics of life course transitions (IPP Documento de Trabajo No. 2).
Retrieved from Instituto de Politicas Públicas website:
http://www.politicaspublicas.udp.cl/media/publicaciones/archivos/263/Document
o_Completo.pdf
CareerBuilder. (2011). Fewer workers age 60 and up postponing retirement, finds new
CareerBuilder survey (Press release). Retrieved from
http://www.careerbuilder.com/share/aboutus/pressreleasesdetail.aspx?id=pr617&s
d=1%2f26%2f2011&ed=1%2f26%2f2011&siteid=cbpr&sc_cmp1=cb_pr617_
Chan, S., & Stevens, A. H. (2002). How does job loss affect the timing of retirement?
(NBER Working Paper 8780). Retrieved from National Bureau of Economic
Research website: http://www.nber.org/papers/w8780
Claxton, G., DiJulio, B., Finder, B., Jarlenski, M., McHugh, M., Hawkins, S., … Gabel, J.
(2008). Employer health benefits: 2008 annual survey. Menlo Park, CA: Kaiser
Family Foundation.
Coile, C., & Gruber, J. (2007). Future Social Security entitlements and the retirement
decision. Review of Economics and Statistics, 89, 234–246.
doi:10.1162/rest.89.2.234
90
Coile, C., & Milligan, K. (2009). How household portfolios evolve after retirement: The
effect of aging and health shocks. Review of Income and Wealth, 55, 226–248.
doi:10.1111/j.1475-4991.2009.00320.x
Collinson, C. (2012). Redefining retirement: The new ‘retirement readiness ’: The 13th
annual Transamerica Retirement Survey. San Francisco, CA: Transamerica
Center for Retirement Studies. Retrieved from https://www.ta-
retirement.com/resources/TCRS%2013th%20Annual%20Thematic%20Report%2
0Final%205-14-12.pdf
Cornwell, B., Laumann, E. O., & Schumm, L. P. (2008). The social connectedness of
older adults: A national profile. American Sociological Review, 73, 185–203.
doi:10.1177/000312240807300201
Cornwell, E. Y., & Waite, L. J. (2009). Social disconnectedness, perceived isolation, and
health among older adults. Journal of Health and Social Behavior, 50, 31–48.
doi:10.1177/002214650905000103
Dave, D., Rashad, I., & Spasojevic, J. (2007). The effects of retirement on physical and
mental health outcomes (Working Paper 07-35). Retrieved from Andrew Young
School of Policy Studies website:
http://aysps.gsu.edu/publications/2007/downloads/07-
35%20DaveRashadSpasojevic-TheEffectsofRetirement.pdf
Dennis, H., & Thomas, K. (2007). Ageism in the workplace. Generations, 31, 84–89.
DeSalvo, K. B., Bloser, N., Reynolds, K., He, J., & Muntner, P. (2006). Mortality
prediction with a single general self-rated health question: A meta-analysis.
91
Journal of General Internal Medicine, 21, 267–275. doi:10.1111/j.1525-
1497.2005.00291.x
Dorn, D., & Sousa-Poza, A. (2007). ‗Voluntary ’ and ‘involuntary ’ early retirement: An
international analysis (IZA Discussion Paper No. 2714). Retrieved from Institute
for the Study of Labor website: http://ftp.iza.org/dp2714.pdf
Drentea, P. (2002). Retirement and mental health. Journal of Aging and Health, 14, 167–
194. doi:10.1177/089826430201400201
Dwyer, D. S., & Mitchell, O. S. (1999). Health problems as determinants of retirement:
Are self-rated measures endogenous? Journal of Health Economics, 18, 173–193.
doi:10.1016/S0167-6296(98)00034-4
Ekerdt, D. J. (2010). Frontiers of research on work and retirement. Journals of
Gerontology Series B: Psychological Sciences and Social Sciences, 65B, 69–80.
doi:10.1093/geronb/gbp109
Elder, G. H., Jr. (1974). Children of the Great Depression: Social change in life
experience. Chicago, IL: University of Chicago Press.
Elder, G. H., Jr. (1994). Time, human agency, and social change: Perspectives on the life
course. Social Psychology Quarterly, 57, 4–15.
Elder, G. H., Jr., Conger, R. D., Foster, E. M., & Ardelt, M. (1992). Families under
economic pressure. Journal of Family Issues, 13, 5–37.
doi:10.1177/019251392013001002
Farber, H. S. (2005). What do we know about job loss in the United States? Evidence
92
from the Displaced Workers Survey, 1984–2004. Economic Perspectives, 29, 13–
28.
Feldman, D. C. (1994). The decision to retire early: A review and conceptualization.
Academy of Management Review, 19, 285–311.
doi:10.5465/AMR.1994.9410210751
Flippen, C., & Tienda, M. (2000). Pathways to retirement: Patterns of labor force
participation and labor market exit among the pre-retirement population by race,
Hispanic origin, and sex. Journals of Gerontology Series B: Psychological
Sciences and Social Sciences, 55, S14–S27. doi:10.1093/geronb/55.1.S14
Floyd, F. J., Haynes, S. N., Doll, E. R., Winemiller, D., Lemsky, C., Burgy, T. M., …
Heilman, N. (1992). Assessing retirement satisfaction and perceptions of
retirement experiences. Psychology and Aging, 7, 609–621. doi:10.1037/0882-
7974.7.4.609
Friedberg, L., & Webb, A. (2005). Retirement and the evolution of pension structure.
Journal of Human Resources, 40, 281–308.
Gallo, W. T., Bradley, E. H., Dubin, J. A., Falba, T. A., Teng, H.-M., Kasl, S. V., &
Jones, R. N. (2006). The persistence of depressive symptoms in older workers
who experience involuntary job loss: Results from the Health and Retirement
Study. Journals of Gerontology Series B: Psychological Sciences and Social
Sciences, 61B, S221–S228.
Gallo, W. T., Bradley, E. H., Siegel, M., & Kasl, S. V. (2000). Health effects of
involuntary job loss among older workers: Findings from the Health and
93
Retirement Survey. Journals of Gerontology Series B: Psychological Sciences
and Social Sciences, 55, S131–S140. doi:10.1093/geronb/55.3.S131
Gallo, W. T., Teng, H. M., Falba, T. A., Kasl, S. V., Krumholz, H. M., & Bradley, E. H.
(2006). The impact of late career job loss on myocardial infarction and stroke: A
10 year follow up using the Health and Retirement Study. Occupational &
Environmental Medicine, 63, 683–87. doi:10.1136/oem.2006.026823
George, L. K. (1993). Sociological perspectives on life transitions. Annual Review of
Sociology, 19, 353–373. doi:10.1146/annurev.so.19.080193.002033
George, L. K. (2007). Age structures, aging and the life course. In J. M. Wilmoth & K. F.
Ferraro (Eds.), Gerontology: Perspectives and issues (pp. 203–222). New York,
NY: Springer.
Hansen, H.-T. (2005). Unemployment and marital dissolution: A panel data study of
Norway. European Sociological Review, 21, 135–148. doi:10.1093/esr/jci009
Hatcher, C. (2003). The economics of the retirement decision. In G. A. Adams & T. A.
Beehr (Eds.), Retirement: Reasons, processes, and results (pp. 136–158). New
York, NY: Springer
Haveman, R., Holden, K., Wolfe, B., & Romanov, A. (2005). Assessing the maintenance
of saving sufficiency over the first decade of retirement (CESifo Working Paper
No. 1567). Retrieved from CESifo Group Munich website:
http://www.ifo.de/portal/pls/portal/docs/1/1188496.PDF
94
Hayes, A. F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new
millennium. Communication Monographs, 76, 408–420.
doi:10.1080/03637750903310360
Hayes, A. F., & Preacher, K. J. (2013). Statistical mediation analysis with a
multicategorical independent variable. Retrieved from
http://www.afhayes.com/public/hpcatx.pdf
Hayward, M. D., & Gorman, B. K. (2004). The long arm of childhood: The influence of
early-life social conditions on men‘s mortality. Demography, 41, 87–107.
doi:10.1353/dem.2004.0005
Health and Retirement Study. (2008). Sample evolution: 1992-1998. Retrieved from
http://hrsonline.isr.umich.edu/sitedocs/surveydesign.pdf
Health and Retirement Study. (2013). Psychosocial and Lifestyle Questionnaire 2006-
2010: Documentation Report Core Section LB. Ann Arbor, MI: University of
Michigan, Survey Research Center.
Helman, R., Greenwald, M., Copeland, C., & VanDerhei, J. (2012). The 2012 Retirement
Confidence Survey: Job insecurity, debt weigh on retirement confidence, savings.
Retrieved from Employee Benefit Research Institute website:
http://www.ebri.org/pdf/briefspdf/EBRI_IB_03-2012_No369_RCS2.pdf
Hershey, D. A., & Henkens, K. (2013). Impact of different types of retirement transitions
on perceived satisfaction with life. Gerontologist. Advance online publication.
doi:10.1093/geront/gnt006
95
Hutchison, E. D. (2005). The life course perspective: A promising approach for bridging
the micro and macro worlds for social work. Families in Society, 86, 143–152.
doi:10.1606/1044-3894.1886
Idler, E. L., & Benyamini, Y. (1997). Self-rated health and mortality: A review of
twenty-seven community studies. Journal of Health and Social Behavior, 38, 21–
37. doi:10.2307/2955359
Jahoda, M. (1979). The impact of unemployment in the 1930s and 1970s. Bulletin of the
British Psychological Society, 32, 309–314.
Jahoda, M. (1981). Work, employment, and unemployment: Values, theories, and
approaches in social research. American Psychologist, 36, 184–191.
doi:10.1037/0003-066X.36.2.184
Jahoda, M. (1982). Employment and unemployment: A social-psychological analysis.
London, England: Cambridge University Press.
Johnson, R. W., & Butrica, B. (2012). Age disparities in unemployment and
reemployment during the Great Recession and recovery. Retrieved from Urban
Institute website: www.urban.org/publications/412574.html
Johnson, R. W., Butrica, B. A., & Mommaerts, C. (2010). Work and retirement patterns
for G.I. generation, silent generation, and early boomers: Thirty years of change.
Retrieved from Urban Institute website:
http://www.urban.org/publications/412175.html
Johnston, D. W., & Lee, W.-S. (2009). Retiring to the good life? The short-term effects
96
of retirement on health. Economics Letters, 103, 8–11.
doi:10.1016/j.econlet.2009.01.015
Kessler, R. C., Turner, J. B., & House, J. S. (1987). Intervening processes in the
relationship between unemployment and health. Psychological Medicine, 17,
949–961. doi:10.1017/S0033291700000763
Kim, J., Kwon, J., & Anderson, E. A. (2005). Factors related to retirement confidence:
Retirement preparation and workplace financial education. Financial Counseling
and Planning, 16, 77–89.
Kim, J. E., & Moen, P. (2002). Retirement transitions, gender, and psychological well-
being: A life-course, ecological model. Journals of Gerontology Series B:
Psychological Sciences and Social Sciences, 57, P212–P222.
doi:10.1093/geronb/57.3.P212
Lachance, M.-E., & Seligman, J. S. (2008, November). Involuntary retirements:
Prevalence, causes, and impacts. Paper presented at the annual meeting of the
Association for Public Policy Analysis and Management, Los Angeles, CA.
Lahey, J. N. (2005). Do older workers face discrimination? (Brief No. 33). Retrieved
from Center for Retirement Research at Boston College website:
http://crr.bc.edu/wp-content/uploads/2005/07/ib_33.pdf
National Institute on Aging. (2007). Growing older in America: The Health & Retirement
Study. Bethesda, MD: Author.
97
MacKinnon, D. P., Krull, J. L., & Lockwood, C. M. (2000). Equivalence of the mediation,
confounding and suppression effect. Prevention Science, 1, 173–181.
doi:10.1023/A:1026595011371
Mandal, B., & Roe, B. (2008). Job loss, retirement and the mental health of older
Americans. Journal of Mental Health Policy and Economics, 11, 167–176. doi:
10.2139/ssrn.991134
Mathieu, J. E., & Taylor, S. R. (2006). Clarifying conditions and decision points for
mediational type inferences in organizational behavior. Journal of Organizational
Behavior, 27, 1031–1056. doi:10.1002/job.406
McKee-Ryan, F., Song, Z., Wanberg, C. R., & Kinicki, A. J. (2005). Psychological and
physical well-being during unemployment: A meta-analytic study. Journal of
Applied Psychology, 90, 53–76. doi:10.1037/0021-9010.90.1.53
Mein, G., Martikainen, P., Hemingway, H., Stansfeld, S., & Marmot, M. (2003). Is
retirement good or bad for mental and physical health functioning? Whitehall II
longitudinal study of civil servants. Journal of Epidemiology & Community
Health, 57, 46–49. doi:10.1136/jech.57.1.46
Moen, P. (1995). A life course approach to postretirement roles and well-being. In L. A.
Bond, S. J. Culter, & A. Grams (Eds.), Promoting successful and productive
aging (pp. 230–257). Thousand Oaks, CA: Sage.
Moen, P. (1996). A life course perspective on retirement, gender, and well-being. Journal
of Occupational Health Psychology, 1, 131–144. doi:10.1037/1076-8998.1.2.131
98
Moen, P., Fields, V., Quick, H. E., & Hofmeister, H. (2000). A life-course approach to
retirement and social integration. In K. Pillemer, P. Moen, E. Wethington, & N.
Glasgow (Eds.), Social integration in the second half of life (pp. 75–107).
Baltimore, MD: Johns Hopkins University Press.
Neuman, K. (2008). Quit your job and get healthier? The effect of retirement on health.
Journal of Labor Research, 29, 177–201. doi:10.1007/s12122-007-9036-8
Noone, J., O‘Loughlin, K., & Kendig, H. (2013). Australian baby boomers retiring
‗early‘: Understanding the benefits of retirement preparation for involuntary and
voluntary retirees. Journal of Aging Studies, 27, 207–217.
doi:10.1016/j.jaging.2013.02.003
Ofstedal, M. B., Weir, D. R., Cehn, K.-T., & Wagner, J. (2011). Updates to HRS sample
weights. Ann Arbor, MI: University of Michigan, Survey Research Center.
Osborne, J. W. (2011). Best practices in using large, complex samples: The importance of
using appropriate weights and design effect compensation. Practical Assessment
Research & Evaluation, 16, 12.
Ozturk, G. B., & Gallo. W. T. (2013). Effect of job loss on wealth accumulation of older
workers. Retrieved from U.S. Department of Labor website:
https://dol.gov/asp/evaluation/reports/WealthEffectsofJobLoss.pdf
Paul, K. I., & Moser, K. (2009). Unemployment impairs mental health: Meta-analyses.
Journal of Vocational Behavior, 74, 264–282. doi:10.1016/j.jvb.2009.01.001
Peppers, L. G. (1976). Patterns of leisure and adjustment to retirement. Gerontologist, 16,
441–446. doi:10.1093/geront/16.5.441
99
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for
assessing and comparing indirect effects in multiple mediator models. Behavior
Research Methods, 40, 879–891. doi:10.3758/BRM.40.3.879
Preacher, K. J., & Selig, J. P. (2012). Advantages of Monte Carlo confidence intervals for
indirect effects. Communication Methods and Measures, 6, 77–98.
doi:10.1080/19312458.2012.679848
Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the
general population. Applied Psychological Measurement, 1, 385–401.
doi:10.1177/014662167700100306
Rao, J. N. K., & Scott, A. J. (1984). On chi-squared tests for multiway contingency tables
with cell proportions estimated from survey data. Annals of Statistics, 12, 46–60.
doi:10.1214/aos/1176346391
Rege, M., Telle, K., & Votruba, M. (2011). Parental job loss and children‘s school
performance. Review of Economic Studies, 78, 1462–1489.
doi:10.1093/restud/rdr002
Reitzes, D. C., & Mutran, E. J. (2004). The transition to retirement: Stages and factors
that influence retirement adjustment. International Journal of Aging and Human
Development, 59, 63–84. doi:10.2190/NYPP-RFFP-5RFK-8EB8
Rosenkoetter, M. M., & Garris, J. M. (2001). Retirement planning, use of time, and
psychosocial adjustment. Issue in Mental Health Nursing, 22, 703–722.
doi:10.1080/01612840120432
100
Schmitt, N., & McCune, J. T. (1981). The relationship between job attitudes and the
decision to retire. Academy of Management Journal, 24, 795–802.
doi:10.2307/256177
Shultz, K. S., Morton, K. R., & Weckerle, J. R. (1998). The influence of push and pull
factors on voluntary and involuntary early retirees‘ retirement decision and
adjustment. Journal of Vocational Behavior, 53, 45–57.
doi:10.1006/jvbe.1997.1610
Selig, J. P., & Preacher, K. J. (2008). Monte Carlo method for assessing mediation: An
interactive tool for creating confidence intervals for indirect effects. Retrieved
from http://www.quantpsy.org/medmc/medmc.htm
Social Security Administration. (2013). Retirement planner: Full retirement age.
Retrieved from http://www.ssa.gov/retire2/retirechart.htm
St. Clair, P., Bugliari, D., Campbell, N., Chien, S., Hayden, O., Hurd, M., …
Zissimopoulos, J. (2011). RAND HRS data documentation, version L. Retrieved
from RAND website:
http://www.prgs.edu/content/dam/rand/www/external/labor/aging/dataprod/randhr
sL.pdf
Ström, S. (2003). Unemployment and families: A review of research. Social Service
Review, 77, 399–430. doi:10.1086/375791
Strully, K. W. (2009). Job loss and health in the U.S. labor market. Demography, 46,
221–246. doi:10.1353/dem.0.0050
101
Subramanian, S. V., & Kawachi, I. (2006). Being well and doing well: On the importance
of income for health. International Journal of Social Welfare, 15, S13–S22.
doi:10.1111/j.1468-2397.2006.00440.x
Subramanyam, M., Kawachi, I., Berkman, L., & Subramanian, S. V. (2009). Relative
deprivation in income and self-rated health in the United States. Social Science &
Medicine, 69, 327–334. doi:10.1016/j.socscimed.2009.06.008
Szinovacz, M. E., & Davey, A. (2004). Retirement transitions and spouse disability:
Effects on depressive symptoms. Journals of Gerontology Series B:
Psychological Sciences and Social Sciences, 59, S333–S342.
doi:10.1093/geronb/59.6.S333
Taylor, M. A., & Doverspike, D. (2003). Retirement planning and preparation. In A.
Adams & T. A. Beehr (Eds.), Retirement: Reasons, processes, and results (pp.
53–82). New York, NY: Springer.
Thomas, D. R., & Rao, J. N. K. (1987). Small-sample comparisons of level and power for
simple goodness-of-fit statistics under cluster sampling. Journal of the American
Statistical Association, 82, 630–636. doi:10.1080/01621459.1987.10478476
Topa, G., Moriano, J. A., Depolo, M., Alcover, C.-M., & Morales, J. F. (2009).
Antecedents and consequences of retirement planning and decision-making: A
meta-analysis and model. Journal of Vocational Behavior, 75, 38–55.
doi:10.1016/j.jvb.2009.03.002
102
Ullah, P. (1990). The association between income, financial strain and psychological
well-being among unemployed youths. Journal of Occupational Psychology, 63,
317–330. doi:10.1111/j.2044-8325.1990.tb00533.x
van den Berg, T. I. J., Elders, L. A. M., & Burdorf, A. (2010). Influence of health and
work on early retirement. Journal of Occupational and Environmental Medicine,
52, 576–583. doi:10.1097/JOM.0b013e3181de8133
van Solinge, H. (2007). Health change in retirement: A longitudinal study among older
workers in the Netherlands. Research on Aging, 29, 225–256.
doi:10.1177/0164027506298223
van Solinge, H., & Henkens, K. (2008). Adjustment to and satisfaction with retirement:
Two of a kind? Psychology and Aging, 23, 422–434. doi:10.1037/0882-
7974.23.2.422
Vinokur, A. D., Price, R. H., & Caplan, R. D. (1996). Hard times and hurtful partners:
How financial strain affects depression and relationship satisfaction of
unemployed persons and their spouses. Journal of Personality and Social
Psychology, 71, 166–179. doi:10.1037/0022-3514.71.1.166
Wang, M., & Shultz, K. S. (2010). Employee retirement: A review and recommendations
for future investigation. Journal of Management, 36, 172–206.
doi:10.1177/0149206309347957
Werner, C. A. (2011). The older population: 2010. Retrieved from U.S. Census Bureau
website: http://www.census.gov/prod/cen2010/briefs/c2010br-09.pdf
103
West, B. T. (2008). Statistical and methodological issues in the analysis of complex
sample survey data: Practical guidance for trauma researchers. Journal of
Traumatic Stress, 21, 440–447. doi:10.1002/jts.20356
Wheaton, F., & Crimmins, E. M. (2013). The demography of aging and retirement. In M.
Wang (Ed.), The Oxford handbook of retirement (pp. 22–41). New York, NY:
Oxford University Press.
Wilkinson, R. G., & Pickett, K. E. (2006). Income inequality and population health: A
review and explanation of the evidence. Social Science & Medicine, 62, 1768–
1784. doi:10.1016/j.socscimed.2005.08.036
Zajacova, A., & Dowd, J. B. (2011). Reliability of self-rated health in US adults.
American Journal of Epidemiology, 174, 977–983. doi:10.1093/aje/kwr204
Abstract (if available)
Abstract
The purpose of the present study is to provide in-depth information on older adults‘ experience of involuntary retirement by examining mechanisms of the effect of involuntary retirement on self-rated health and mental health among adults aged 50 years or older. Although it is a normative expectation to perceive retirement as a phase of life in which older adults become disengaged from paid work, volunteer, and enjoy leisure time with family, not everyone has the luxury of maintaining control over their retirement decisions. Approximately one third of retirees perceive their retirement as forced rather than voluntary. Involuntary retirees are likely to face greater challenges than voluntary retirees during their postretirement adjustment period because they have multiple burdens of health, mental health, and job displacement issues that may have partially led them to involuntarily retire. The prevalence of involuntary retirement is likely to increase because older adults are expected to work longer due to increased financial responsibility despite the challenge of securing or maintaining employment. Although an expanding body of research has addressed various topics of retirement including preretirement planning and decision making, relatively few studies on retirement have focused on the voluntariness of retirement or its varied contexts. Research that has explored health and mental health outcomes after retirement without accounting for voluntariness has reported mixed findings. Considering the fact that retirement has become a more complex and diverse life transition, it is critical to consider the nature of retirement as well as its contexts. ❧ To fill this gap of knowledge in research and practice, this study had two specific aims: (1) to explore the prevalence of involuntary retirement among older adults and the extent to which the characteristics of involuntary retirees are different from voluntary retirees or those who did not retire, and (2) to investigate the mechanisms of the health and mental health effect of involuntary retirement by examining the potential mediating effects of financial control, positive and negative family relationships, and social integration. The research questions and hypotheses were formulated based on the life course perspective and latent deprivation theory. Using two waves of longitudinal data extracted from Health and Retirement Study (2006 and 2010), a final sample of 1,280 individuals working for pay at baseline who responded to a lifestyle questionnaire in both waves was selected. Univariate, bivariate, and regression-based path analyses were conducted using SPSS 18.0. This study employed a multiple mediation model that considered four mediators simultaneously and the model was estimated in three phases. ❧ Results of the study found that 29.3% (n = 429) of the sample retired between 2006 and 2010, and 37.2% (n = 155) of those individuals reported that they retired involuntarily. Results of Phase 1, a multiple mediator model that did not account for voluntariness of retirement using a binary independent variable (retired or not), indicated that there was no significant direct or indirect effects of retirement on self-rated health and mental health outcomes. In Phase 2, in which the same model was estimated using a multicategorical independent variable (involuntarily retired, voluntarily retired, not retired), involuntary retirement had an direct adverse health effect compared to not retiring, whereas voluntary retirement had an indirect positive health effect via financial control. In terms of mental health outcomes, the positive mental health effect of voluntary retirement was mediated by financial control, whereas involuntary retirement had no significant effect. Results of Phase 3, a model that considered retirees only, revealed direct adverse health effects of involuntary retirement compared to voluntary retirement. Involuntary retirement also had an indirect effect on mental health via financial control. ❧ Findings of this study indicate the significance of specifying the nature of retirement when conducting retirement research and the need to pay more attention to potential detrimental effects of involuntary retirement. Implications of the findings are discussed with regard to retirement policy, research, and social work.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Life transitions, leisure activity engagement, and cognition among older adults
PDF
The effects of wisdom-related personality traits on caregivers’ health: an application of the resilience model
PDF
The effects of time and space on health status of older adults in China
PDF
Impacts of caregiving on wellbeing among older adults and their spousal caregivers in the United States
PDF
Intergenerational support between grandparents and grandchildren in rural China and its effect on the psychological well-being of older adults
PDF
Family relationships and their influence on health outcomes over time among older adults in rural China
PDF
The effects of familism and cultural justification on the mental and physical health of family caregivers
PDF
My Life, My Wishes, a death education program for community-dwelling Chinese older adults: development and pilot testing
PDF
A study of depression disclosure among Latino older adults in Los Angeles
PDF
Examining the longitudinal influence of the physical and social environments on social isolation and cognitive health: contextualizing the role of technology
PDF
Is stress exposure enough? Race/ethnic differences in the exposure and appraisal of chronic stressors among older adults
PDF
Mental Health First
PDF
Neighborhood context and adolescent mental health: development and mechanisms
PDF
Mechanisms of stress effects on learning and decision making in younger and older adults
PDF
Three essays on health, aging, and retirement
PDF
Examination of the long-term psychosocial and functioning effects of a healthy living, life management behavior intervention for older adults
PDF
Integrative care strategies for older adults experiencing co-occurring substance use and mental health disorders (I-CARE)…
PDF
Mental health service use by children and youth in the child welfare system: a focus on need and predisposing factors and caregiver type
PDF
Investigating the effectiveness of a social work intervention on reducing hospital readmissions among older adults
PDF
The roles of social isolation and secondary caregiver support in the stress process among primary caregivers of older adults
Asset Metadata
Creator
Rhee, Min-Kyoung
(author)
Core Title
Mechanisms of the effect of involuntary retirement on older adults' health and mental health
School
School of Social Work
Degree
Doctor of Philosophy
Degree Program
Social Work
Publication Date
12/02/2013
Defense Date
06/04/2013
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
depressive symptoms,financial control,involuntary retirement,OAI-PMH Harvest,self-rated health
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Mor Barak, Michàlle E. (
committee chair
), Chi, Iris (
committee member
), Gallo, William T. (
committee member
), Knight, Bob G. (
committee member
)
Creator Email
minkyour@gmail.com,minkyour@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-355168
Unique identifier
UC11296668
Identifier
etd-RheeMinKyo-2198.pdf (filename),usctheses-c3-355168 (legacy record id)
Legacy Identifier
etd-RheeMinKyo-2198.pdf
Dmrecord
355168
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Rhee, Min-Kyoung
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
depressive symptoms
financial control
involuntary retirement
self-rated health