Copyright 1992 by the American Psychological Association, Inc. 0882-7974/92/$3.00

Psychology and Aging 1992. Vol. 7. No. 4, 609-621

Assessing Retirement Satisfaction and Perceptions of Retirement Experiences Stephen N. Haynes

Frank J. Floyd Michigan State University

University of Hawaii

Elizabeth Rogers Doll, David Winemiller, Carolyn Lemsky, Tria Murphy Burgy, Mary Werle, and Nancy Heilman

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Illinois Institute of Technology A 2-part investigation developed a measure of retirement as a life transition. Study 1 generated items from interviews with retirees (n = 40) and service providers. Item analysis with recent retirees (n = 86) produced a 51-item Retirement Satisfaction Inventory assessing 6 areas: preretirement work functioning, adjustment and change, reasons for retirement, satisfaction with life in retirement, current sources of enjoyment, and leisure and physical activities. Study 2 examined a heterogeneous sample of men (n = 159) and women (n = 243) retirees. Factor analyses produced internally consistent subscales. Moderate, but acceptable, test-retest reliability was demonstrated. Satisfaction scores correlated with concurrent measures and, together with pre- and postretiremen! experiences, discriminated 4 groups of voluntary and involuntary retirees. Few effects related to gender, socioeconomic status, length of retirement, and part-time employment were found.

Retiring from full-time employment is a milestone that marks the transition into the later stage of life for millions of Americans each year. The retirement experience can lead either to new goals, interests, and activities or to stress, rapid physical deterioration, and depression (Dorfman & Moffett, 1987; Matthews & Brown, 1987; Morse, Dutka, & Gray, 1983; Seccombe & Lee, 1986). As many as one third of retirees have difficulty adjusting to retirement and undergo a decrease in life satisfaction as a result of the transition (Atchley, 1976; Elwell & Maltbie-Crannell, 1981; Harris & Associates, 1979). Compared with workers, retirees report less social support for personal problems and more symptoms of psychological distress (Bosse, Aldwin, Levenson, & Ekerdt, 1987; Bosse, Aldwin, Levenson, & Workman-Daniels, 1990) and are more likely to use medical services (Boaz & Muller, 1989). Also, in some circumstances, retirement is a risk factor for suicide in older persons (Kirsling, 1986; Rothberg, Ursano, & Holloway, 1987). Despite its importance, studies of adult development for older persons frequently neglect the role of the retirement experience or limit their analysis to a single aspect of retirement such as global satisfaction or economic changes (Parnes & Less, 1985). Systematic assessment of the retirement experience is hampered by the lack of a comprehensive, parsimonious, and psychometrically sound assessment instrument. The availability of such an instrument would not only help to stimulate more research about factors affecting satisfaction and adjustment in later adulthood but would also be clinically useful to retirement

This research was funded in part by a grant from the Andrus Foundation. Correspondence concerning this article should be addressed to Frank J. Floyd, Psychology Department, 129 Psychology Research Building, Michigan State University, East Lansing, Michigan 488241117.

609

counselors and mental health professionals in designing interventions and prevention programs for retired individuals. The purpose of our research was to develop a questionnaire for assessing both current retirement satisfaction and perceptions of retirement-related experiences predictive of adjustment and well-being in later life. The content of the questionnaire draws on previous descriptive research on retirement (e.g., Atchley, 1976; Beck, 1982; Parnes et al, 1985) and incorporates the reports of retirees and professionals working with older persons.

Theoretical Framework We used life span development theory and life span transition theory (e.g., Baltes, 1987; Fiske & Chiriboga, 1990) to guide the design of the questionnaire. According to these perspectives, whereas much development in childhood involves progressing through inexorable stages of maturation driven by biological development, personal development in adulthood is multidirectional and consists of a series of transitions brought about by experiences. The most significant experiences for promoting change and development are the major life events that alter the individual's social roles, personal identity, goals, expectations, and sources of rewards. These are the milestones that mark the beginning of a new life stage for the individual and present new demands and opportunities associated with developmental tasks. Unlike childhood developmental milestones, which are relatively invariant, the principle of contextualism proposes that individual experiences surrounding adult transitions determine the meaning and impact of the transitions for different persons. Thus, although transitions such as retirement are "normative" and the event itself is usually predictable and planned, the contexts for different individuals create differences in how this transition comes about, the meaning of the

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

610

FLOYD ET AL.

event, the short-term adjustments required, and the long-term impact of the event on the individual's life. The importance of retirement as a life transition is supported both by cross-sectional studies contrasting groups of retirees and workers and by comprehensive longitudinal research focused on personal and social development in adulthood (e.g., Bosse, Aldwin, Levenson, & Ekerdt, 1987; Bosse, Aldwin, Levenson, & Workman-Daniels, 1990; Fiske & Chiriboga, 1990; Matthews & Brown, 1987). Our work adopts this formulation and draws on the supporting research to design a measure that can be useful both for group studies of retirees and for idiographic analysis to describe the satisfaction and adjustment of individuals in retirement. Assessing retirement as a life transition imposed a temporal perspective on the measure. That is, understanding the impact of the transition involved assessing past experiences and feelings surrounding the transition, present satisfaction in retirement, and prospects for future adjustment. These concerns suggested three general domains to evaluate: retirement experiences that affect the meaning and short-term impact of the event for the person, current satisfaction with life in retirement, and factors associated with long-term adjustment.

Retirement Experiences Similar to chance events, transition points are often experienced as stressful occurrences that provoke disruption and distress for the individual. However, others may experience the transition as a challenge and an opportunity to relinquish unrewarding responsibilities and pursue new interests and sources of pleasure. Accordingly, research and theory on coping with stress (e.g., Lazarus & DeLongis, 1983; McCubbin & Patterson, 1983; Taylor, 1983) emphasize the importance of understanding the individual's appraisal of the event. The appraisal process includes both formulating a subjective meaning of the event as either a threat or a challenge and evaluating one's capacity to meet the demands of the situation. A central issue in determining the subjective meaning of retirement for the individual is the significance of work in the life of the person. Work provides the most central social and psychological framework during adulthood (Henry, 1971). Thus, the loss of work is potentially a crisis point for the individual. To assess the nature of work life and the impact of its loss, we included two sets of items in the questionnaire that evaluate this domain. One section on preretirement work functioning includes items about the significance of work-related activities as rewards before retirement. These are retrospective reports of job satisfaction and job involvement. Another section on adjustment and change evaluates the retiree's own perception of the acute stress associated with the event of retirement. These ratings overcome a limitation with life-events scales (e.g., Holmes & Rahe, 1967) that assume that a standard impact score can be assigned to common life events. Events are judged as threats when they are perceived to be ok^t of the individual's control (Taylor, 1983). For retirement, the issue of controllability involves the retiree's perceptions about the precipitants of retirement. A common finding of research on retirement is that voluntary, as opposed to involuntary retire-

ments are associated with more positive adjustment in retirement (Crowley, 1986; Streib & Schneider, 1971). Although official mandatory retirement policies are increasingly rare, seemingly voluntary retirements are often premature (Ruhm, 1989) and may be prompted by poor health, job stress, or other circumstances not under the retiree's control (Parnes et al., 1985). Furthermore, the precipitants of voluntary retirement have been shown to predict satisfaction in retirement (Levy, 1981; Walker, Kimmel, & Price, 1981). Thus, a third section of the questionnaire assesses perceptions of the retiree's reasons for retirement.

Satisfaction With Life in Retirement The most essential index of subjective well-being for an individual is the person's judgment about quality of life and life satisfaction (Diener, 1984). Global indices of life satisfaction appear to be relevant for all age groups. However, global judgments alone are not sensitive to domain-specific qualities of life (Campbell, Converse, & Rodgers, 1976) and do not elucidate which areas of life are salient for the individual in accounting for current satisfaction and adjustment. To provide both a global measure of overall satisfaction and a sensitive index of specific sources of satisfaction and dissatisfaction, we included a scale requiring subjects to give a global rating of general satisfaction with life in retirement, and we also included several items evaluating satisfaction with specific areas of life in retirement. The specific areas include several that are common to other life satisfaction scales (e.g., Frisch, Cornell, Villanueva, & Retzlaff, 1992), such as economic resources, and interpersonal relationships, along with others that are particularly relevant to retirees, such as the quality of social services and access to transportation. To group these facets into more general scales, we used factor analysis to guide the development of subscales of satisfaction. Thus, the final questionnaire assesses satisfaction with life in retirement at three levels of globality: overall ratings of retirement satisfaction, intermediate subscales for domains of life in retirement, and individual item ratings of specific areas of satisfaction and dissatisfaction.

Rewards and Leisure Activities Although the event of retirement can be an acute stressor, retirement is also a stage of adulthood that continues for the remainder of the individual's life. Most measures of coping with stress (e.g., Lazarus & Folkman, 1984), including a recent measure designed to detect age-related changes in coping styles (Brandtstadter & Renner, 1990), assess skills and resources for adjusting to transitory difficulties (acute stressors) but do not assess factors associated with long-term adaptation to major life changes. Long-term adaptation involves finding ample sources of rewards and gratifications that allow individuals to experience that they are getting enough out of life toward meeting basic developmental goals (Fiske & Chiriboga, 1990). Maintaining life satisfaction during retirement is associated with replacing rewards gained from work activities with rewarding leisure activities so that the retiree maintains a stable ratio of reinforcements from pre- to postretijenwot (Friedman & Or-

RETIREMENT SATISFACTION AND EXPERIENCES bach, 1974; Howard, 1982). Thus, to assess the potential for long-term adaptation, we included a section in the questionnaire that asked respondents to judge sources of enjoyment in retirement. Additionally, because leisure and activities are important ways to counteract the stress of boredom (Fiske & Chiriboga, 1990), the final section of the questionnaire asks about current leisure and physical activities.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Development and Initial Validation This article presents a two-part study focused on the development and initial validation of the questionnaire. The first part (Study 1) consisted of item development and item analyses to select items for the final questionnaire. The second part (Study 2) included initial factor analyses to develop subscales of the three sections assessing reasons for retirement, satisfaction with life in retirement, and sources of enjoyment in retirement. Additionally, we examined internal consistency, test-retest reliability, the intercorrelations among the subscales, the correlations of the questionnaire scores with measures of life satisfaction and marital satisfaction, and the associations between reasons for retirement and satisfaction and adjustment in retirement. An overarching goal was to develop a measure appropriate for use with both women and men and for people from all levels of socioeconomic status (SES). To date, research on retirement has not adequately described the experiences of women retirees (Hatch, 1990), and evidence is mixed about the extent to which men and women differ in retirement satisfaction (Gratton & Haug, 1983; Seccombe & Lee, 1986). Thus, we examined gender differences for all scores, and we explored possible gender effects in all analyses. Regarding SES, previous research suggests that SES not only determines financial security and the quality of services obtainable but also may be associated with different preferences and practices regarding leisure and social relationships in retirement (Rosow, 1985). Thus, in the present research, we carefully sampled across socioeconomic groups, and we evaluated the associations of SES variables with the scores for retirement satisfaction and retirement experiences. Study 1 Study 1 involved several stages of interviews with retirees, interviews with professionals working with older persons, and a review of relevant literature to generate items. These were followed by pilot tests for early versions of the questionnaire to eliminate ambiguities and to select internally consistent sets of items for each section of the questionnaire.

Method

Subjects A total of 126 older persons participated in various stages of Study 1. The majority of the subjects were recruited at meetings of local chapters of the American Association of Retired Persons (AARP), and others came from industry- and university-based retirement organizations and community senior centers. To locate a cross section of people, the subjects were recruited from a large urban area, a medium-

611

sized urban area, and several small towns throughout the Midwest. Retirement was denned as "an age-related reduction in employment," so some of the subjects were employed part time. The sample was roughly equally divided between men and women. Procedure Initial item development. First, a set of potential items was compiled from semistructured interviews with 30 subjects. The interview schedule included six open-ended questions covering current activities, sources of rewards, components of current satisfaction, preretirement work functioning, adaptation and change associated with retirement, and reasons for retirement. The interviewers were trained to use standard follow-up probes, they completed practice interviews, and they were supervised continuously to ensure uniformity in the procedures. The interviews lasted approximately 20-30 min during which the respondents were encouraged to speak freely about their own experiences in retirement. Tape recordings of the interviews were independently reviewed by two members of the research team who generated a questionnaire item to assess each of the points raised by the retirees. Other items suggested by the research literature, but not raised in the interviews, were added to the questionnaire so that at least four items were included to address each of the six areas covered in the interview. To make the questionnaire inclusive but parsimonious, highly specific, idiosyncratic activities and sources of rewards were summarized with questions about more general topics, such as activities with family, activities with friends, and satisfaction with social services. Grammatical conventions were standardized across all items, and a standard 4-point or 6-point Likert-type rating format was adopted. The specific anchors for each of the rating scales matched the content of the particular item, but most items required a rating of either frequency, importance, or degree of satisfaction. The anchors reflected a range from never to often, very unimportant to very important, or very dissatisfied to very satisfied. The general instruction statement and the specific instructions for each section were written to be as concise and unambiguous as possible, and large printing and an uncluttered arrangement were used to make the questions easy to read and thus reduce extraneous sources of error. Pilot testing. An initial version of the questionnaire was pilot tested by administering it to 10 additional retirees who first answered the questionnaire, then completed an individual interview where the questionnaire was reviewed to discover any items that were confusing, difficult to read, or offensive. Each subject also was questioned about whether any other items should be included to adequately assess retirement experiences. A second pilot version, altered on the basis of the retiree's reactions, was then sent to eight psychologists, social workers, and counselors working at agencies that provide counseling to retired people who are experiencing adjustment problems. After reviewing the questionnaire, each professional was interviewed over the phone regarding clarity and the need for additions or deletions. Once again, the suggestions were incorporated into a final pilot version of the questionnaire that included 110 items. Item analysis. The 110-item pilot questionnaire was administered to 86 men and women from AARP and other retirement organizations. To ensure that the questionnaire was sensitive to the experiences of people undergoing the transition into retirement, we recruited only individuals who had been retired for 5 years or less. An item was retained for the final scale if it showed sufficient response variability, where both ends of the rating scale were endorsed by at least 5% of the sample. Also, to ensure internal consistency, the items measuring areas of satisfaction and sources of enjoyment had to correlate at least .65 with a global 6-point rating of retirement satisfaction. These proce-

612

FLOYD ET AL.

dures produced a final 51-item version of the questionnaire that we titled the Retirement Satisfaction Inventory (RSI). The entire questionnaire is given in the Appendix, including the instructions, all items, and all response scales. Study 2 The purposes of Study 2 were to develop a scoring format and initial norms for the RSI and to conduct initial psychometric evaluation of the measure. Method

their experiences regarding retirement. They were encouraged to answer all items, but they were informed that they could skip items if they wished. One item (Item 19) was an open-ended response that allowed subjects to write in reasons for retirement not addressed i n other items. We developed this item for counseling purposes so that the questionnaire would be sensitive to the unique experiences of some retirees, but the item was not included in any of the analyses. Additionally, five of the items regarding satisfaction with marriage, family, and social services included a not applicable response option for subjects who were unmarried or had no experiences in these areas (see Appendix). Because many of the subjects' protocols included occasional unanswered or inapplicable items, analyses were conducted with all available responses to each item, and thus the ns for the analyses that follow vary.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Subjects A sample of 402 retired individuals was recruited from retirement organizations in the same way as the sample for Study 1. For Study 2, the organizations were located in urban, suburban, and rural settings in two states. The subjects were 159 men and 243 women. The mean age of the men was 69.40 years (SD = 5.94 years), and they had been retired a mean of 4.18 years (SD = 2.92 years). The mean age of the women was 67.49 years (SD = 6.20 years), and they had been retired a mean of 4.28 years (SD = 3.07 years). The men reported a mean current family income of $33,630 (SD = $18,010), and the women reported a mean current family income of $28,390 (SD = $18,690). These reports reflected the pooled income from all sources for both spouses and thus were higher than the preretirement earnings of each retiree. Unmarried retirees reported a mean annual income of $18,880 (SD = $10,910). Both the men and the women had retired from occupations ranging from manual laborers to professionals. Scores on the Duncan index of occupational status (Stevens & Featherman, 1981) ranged from 17.7 for assemblers to 87.1 for college teachers, with a mean of 37.39 (SD = 16.82), which is associated with skilled trades and lower management such as technicians, building managers, retail salespersons, and factory supervisors. The sample also represented a wide range of educational backgrounds. Fourteen percent had less than a high school education, 26.1% completed high school only, 31.9% completed trade school or some college, 12.9% obtained bachelor's degrees, and 14.7% obtained advanced degrees. Ninety-five percent of the men and 55% of the women were currently married. The sample consisted of 89% Whites, 10% African Americans, and 1% Asian Americans. Measures All subjects completed the 51 -item RSI and a demographic information sheet. Additionally, to evaluate the concurrent validity of the RSI, all retirees completed the Satisfaction With Life Scale (Diener, Emmons, Larson, & Griffin, 1985), a 5-item measure with a 7-point Likert response format. Married individuals also completed the Marital Satisfaction Questionnaire for Older Persons (Haynes et al., 1992), a 26-item questionnaire about marital adjustment and satisfaction with the marital relationship. Procedure The majority of the questionnaires were distributed at the close of organization meetings and were completed on site or returned by mail. A subgroup of approximately 40 retirees completed the questionnaires in individual meetings that were part of another research project. Another subgroup of 65 subjects participated in a retest session 2 weeks after the initial session. The subjects were informed that the purpose of the study was to develop a questionnaire that would be sensitive to

General Results Factor Analysis To group individual items into subscales, three sets of factor analyses were conducted on the three sections of the questionnaire measuring reasons for retirement (15 items), satisfaction with life in retirement (11 items), and sources of enjoyment (15 items). Principal-components analyses were followed by varimax rotation of factors with eigenvalues greater than one. Separate analyses of the data from the men and the women produced highly consistent solutions; therefore we present only the solutions for the combined sample of men and women. Also, the analyses include only the subjects who answered all items (n = 272), although analyses with the entire sample using mean substitution to estimate missing ratings produced similar factor solutions. Table 1 displays the factor loadings and the composition of the subscales. To form independent subscales, we assigned each item to only one subscale, the one associated with its primary (highest) factor loading. As displayed in Table 1, some of the items showed nearly equal loadings on more than one factor, with no clear primary loading. In most cases, we inspected the separate analyses with the men and the women and assigned the item to the subscale on which it consistently loaded highest. However, when no consistent priority was apparent, we assigned the item on the basis of both its conceptual relevance to other items on the scale and its effect on the alpha coefficients for the scales. Table 1 also lists the eigenvalue for each factor, the percent of variance accounted for, and the alpha coefficient for each scale.

Reasons for Retirement The items assessing reasons for retirement produced four significant factors labeled Job Stress (3 items), Pressure From Employer (4 items), Pursue Own Interests (4 items), and Retirement Due to Circumstances (4 items). Together these factors accounted for 61% of the variance in the reasons-for-retirement items. With one exception, all items showed a primary loading on only one factor, with no secondary loadings above .40. Item 9 loaded similarly on Factors 1 and 2 but loaded slightly higher on Factor 2, so the item was assigned to the second subscale.

613

RETIREMENT SATISFACTION AND EXPERIENCES Table 1 Factor Analyses of Reasons for Retirement, Satisfaction With Life in Retirement, and Sources of Enjoyment Sections of the Retirement Satisfaction Inventory (RSI) Factor

Factor 1

Item

2

3

4

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Eigenvalue % variance Alpha

.81

.19

.17

.11

.77 .74 .32

.24 .38 .75

.05 .12 .02

.11 .12 .23

.20

.73

.16

.02

.20

.67

.11

.34

.49 -.01

.53 .18

.08 .83

.29 .09

.76

.09

.13

-.14 .31

.57

.27

.26

.42

.34

-.05 .48 .16

.21 .10 .35

-.05 .01 .12

.75 .62 .55

-.11

.39

.52

1.63 11 .79

1.11 7 .66

1.04 7 .61

5.31 35 .80

Satisfaction with life in retirement 29. Government services 30. Personal safety 27. Access to transportation

.80 .73 .71

.02 .09 .19

28. Community agency services 21. Financial situation 22. Physical health 26. Level of physical activity 20. Marriage 23. Spouse's health 24. Quality of residence 25. Relations with extended family Eigenvalue % variance Alpha

2

3

4

.02 .21 -.01

.67 .53 .10 .21 .16 -.07 .34

.20 .18 .84 .82 -.12 .25 .41

.04 .30 .11 .05 .82 .60 .46

.20

.39

.38

3.59 33 .78

1.35 12 .72

1.09 10 .57

.73 .72 .68 .63 .55 .55 .45 .27 -.06 .20 .22 . 3? 3£.

-.14 .27 .20 .32 .13 .33 .46 .75 .73 .66 .60

.14 .02 .23 .36 .52 -.01 .33 -.06 .16 .35 .24

Jl

C1

4Sj .*t

-.01 .24 .41

.09 .26 .42

.87 .46 .45

5.95 40 .82

1.38 9 .83

1.05 7 .65

Sources of enjoyment

-.12

.08

.33

1

Satisfaction with life in retirement (continued)

Reasons for retirement 16. Too much stress at work 17. Difficulty with physical demands of job 15. Disliked job 10. Pressured by employer 1 1 . Offered incentives by company 8. Laid off, fired, or hours reduced 9. Difficulties with people at work 12. Wanted time with family 13. Wanted time to pursue own interests 18. Spouse wanted partner's retirement 14. Wanted to make room for younger people 4. Reached mandatory retirement age 5. Poor health 6. Spouse's poor health 7. Could finally afford retirement

Item

50. Can be alone more 4 1 . No boss 43. Less stress 49. More relaxed 48. More time to think 37. Not having to work 47. Being careful 44. Retirement groups 46. Volunteerism 39. More time with friends 38. More time with family 45 Tiitic for activities 36. Freedom to pursue own interests 42. More travel 40. Control over own life Eigenvalue % variance Alpha

Note. N = 272. Numbers in boldface are primary factor loadings and indicate the subscale assignment for the item. See the text for the decision rules when items obtained similar loading on two or more factors. No data are given in the fourth column for the Satisfaction with life in retirement and Sources of enjoyment sections because each of those sections produced only three significant factors.

Satisfaction With Life in Retirement The factor analysis of the items measuring satisfaction with life in retirement produced three significant factors, labeled Satisfaction With Services and Resources (5 items), Satisfaction With Health and Activity (2 items), and Satisfaction With Marriage and Home Life (4 items). These three factors accounted for a total of 55% of the variance in the satisfaction ratings. The factor solution showed relatively good simple structure with the exception of Item 24, "quality of residence," and Item 25, "relations with extended family," each of which obtained only moderate loadings on both Factors 2 and 3. Because both items loaded slightly higher on Factor 3 in the separate analyses with the men and the women and because they significantly reduced the alpha coefficient for the second scale, they

were assigned to the third subscale, Satisfaction With Marriage and Home Life.

Sources of Enjoyment The sources of enjoyment section also produced three significant factors labeled Reduced Stress/Responsibilities (7 items), Social Activities (5 items), and Freedom and Control (3 items), which accounted for a total of 56% of the variance in these items. Although Item 48, "more time to think," showed similar loadings on Factors 1 and 3, it consistently loaded slightly higher on Factor 1 for both the men and the women; therefore it was retained on the first subscale. Similarly, Item 40, "control over own life," loaded only moderately on all three factors but consistently loaded highest on Factor 3 in the separate analyses

614

FLOYD ET AL.

with the men and the women and was assigned to the third subscale. Item 47, "being carefree," loaded similarly on Factors 1 and 2 and was assigned to the first subscale, Reduced Stress/ Responsibility, because it was most similar conceptually to the other items on that scale.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Scoring Scores for the factor-analytically-derived subscales were calculated as the mean of the ratings for the items making up the subscale. Additionally, overall mean scores were calculated for all 11 items measuring satisfaction with aspects of retirement (a = .81). If a subject failed to answer an item, the mean scores were calculated from the remaining items. The individual ratings of current activities (3 items), preretirement work functioning (3 items), and adaptation and change associated with retirement (2 items), along with the global rating of overall satisfaction with retirement, were treated as single-item scores. The mean scores and standard deviations for the men and the women are given in Table 2.

between testings. Test-retest correlations were computed for the overall mean satisfaction scores, the three sets of factor scores, and the other individual items measuring preretirement work functioning, adaptation and change associated with retirement, and participation in activities. The correlations are given in Table 2. In general, the reliability was higher for the factor scores and the mean satisfaction score than for the single-item ratings. The correlations for the two testings ranged from r = .56 to r = .77 (mean r = .68) for the multiple-item scales and from r = .45 to r = .71 (mean r = .62) for the single-item ratings. Although the correlations are moderate, only one of the factor scales, satisfaction with services, and two of the single-item ratings produced correlations below r = .60. The two single items asked the retirees to make complex judgments concerning preretirement expectations about retirement and concerning overall changes in their lives from pre- to postretirement.

Concurrent Validity

Test-Retest Reliability Test-retest reliability was evaluated with the subsample of 65 subjects who completed the RSI twice, with a 14-day interval

Concurrent validity was assessed for five scores from the RSI that assess current satisfaction in retirement: the overall mean

Table 2

Means, Standard Deviations, and Test-Retest Correlations for Retirement Satisfaction Inventory (RSI) Scores Women (« = 243)

Men (« = 159)

Section and variable Retirement satisfaction Services subscale Health subscale Marriage/Home Life subscale Mean of satisfaction ratings Global satisfaction rating Preretirement work functioning Job gratification Job satisfaction Anticipated satisfaction Adjustment and change Initial adjustment Pre-post improvement Sources of enjoyment Reduced Stress and Responsibilities subscale Social Activities subscale Freedom and Control subscale Reasons for retirement Job Stress subscale Pressure From Employer subscale Pursue Own Interests subscale Circumstances subscale Current activities Leisure with friends Leisure with family Physical activities

M

SD

M

SD

Test-retest correlations (ii = 65)

4.93 4.46 4.89 4.82 5.29

0.59 0.95 0.65 0.52 0.85

4.88 4.59 4.98 4.83 5.15

0.75 1.03 0.82 0.69 0.99

.56 .66 .71 .66 .67

4.72 4.76 5.09

1.18 1.18 0.89

4.60 4.77 4.86

1.34 1.23 1.07

.65 .68 .45

4.90** 4.87

1.23 0.92

4.33 4.72

1.56 1.16

.71 .50

2.83 2.93 3.27

0.70 0.66 0.61

3.04* 3.11* 3.34

0.74 0.64 0.65

.64 .64 .70

2.23

1.31

2.03

1.40

.75

1.87

1.20

1.60

1.12

.77

3.04 2.66

1.23 1.21

2.68 2.43

1.38 1.27

.65

3.43 3.27 3.27

0.71 0.73 0.76

3.60* 3.38 3.26

0.58 0.72 0.83

.67 .62 .60

.74

Note. Asterisks indicate significantly higher mean scores based on / tests following significant multivariate effects. * p < .05. ** p < .01, two-tailed.

615

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

RETIREMENT SATISFACTION AND EXPERIENCES score for the satisfaction items, the three subscale scores derived from these items, and the global rating of retirement satisfaction. Validity was indicated by the correlations of the five scores with the Satisfaction With Life Scale and for the married retirees with total satisfaction scores on the Marital Satisfaction Questionnaire for Older Persons. These correlations are presented in Table 3 together with the intercorrelations among the five retirement satisfaction scores for comparison purposes. The correlations for the women are given above the diagonal, and those for the men are below the diagonal. With one exception, all of the validity coefficients (the correlations with the criteria) were highly significant and indicated that the RSI scores share from 9% to 41% of variance with the Satisfaction With Life Scale and the Marital Satisfaction Questionnaire for Older Persons. As shown in Table 3, the validity coefficients were roughly equal for the men and the women on both criteria, and they were somewhat higher for the total satisfaction index and the global rating of retirement satisfaction than for the three subscale scores. Also, the validity coefficients were similar to or greater in magnitude than the intercorrelations among the RSI satisfaction subscale scores. The specificity of the factor scores was supported by the exceptionally high correlation between the Satisfaction With Marriage and Home Life subscale and the Marital Satisfaction Questionnaire for Older Persons (r = .64 and r = .65 for the women and the men, respectively).

Reasons for Retirement To evaluate the validity and utility of the reasons for retirement scales, we assessed whether ratings of reasons for retirement showed expected relationships with other aspects of retirement satisfaction and experiences. We expected that those who rated the Job Stress factor as a particularly important reason for retirement would endorse relatively less preretirement job satisfaction and relatively greater rewards from reduced job stress in retirement than other retirees. Research on involuntary retirement suggested that persons who obtained relatively high scores on the Pressure From Employer factor should report a more difficult transition period, less satisfaction with retire-

ment, fewer sources of enjoyment in retirement, and poorer functioning in physical, social, and leisure activities than people who retired voluntarily. In contrast, the voluntary retirees who obtained relatively high scores on the Pursue Own Interests factor should report an easier transition to retirement, higher satisfaction, more sources of enjoyment, and positive adjustment in retirement. Because most people reported multiple reasons for retirement and because only high scores, but not necessarily low scores, on a factor should be meaningfully related to retirement satisfaction and experiences, we interpreted the factor scores in an ipsative fashion and selected the most important reason for retirement for each subject. Of the 369 respondents with complete data on the reasons for retirement section of the questionnaire, 317 (86%) could be assigned to one of four groups on the basis of having made relatively higher ratings on one of the four factors included in the reasons for retirement section. The group assignment thus indicated that the respondent identified a primary reason for retirement as either escaping from job stress (n = 53), desiring to pursue own interests (n = 154), responding to the employer's pressure to retire (n = 21), or having circumstances (e.g., age and health) mandate retirement (n = 89). The remaining 52 subjects obtained equal factor scores on at least two of the factors and were excluded from these analyses. A series of one-way multivariate analyses of variance (MANOVAs) contrasted the groups on the other five sections of the questionnaire. Significant multivariate effects were followed up with univariate analyses of variance (ANOVAs) and post hoc Scheffe tests. The M ANOVAs revealed significant differences among the four groups of retirees on four of the five sections of the questionnaire: preretirement work functioning, F(9,796) = 9.09, p < .001; adjustment and change, F(6,574) = 3.02, p < .01; satisfaction with life in retirement, F(\5, 853) = 2.26, p < .01; and current sources of enjoyment, F(9, 798) = 4.15, p < .001. The univariate results for these variables are summarized in Table 4. Only the ratings of involvement in activities failed to differ across any of the groups, F(9,870) = 1.11, ns. First, as expected, the group who indicated that they retired

Table 3 Correlations Among Retirement Satisfaction Scores, Satisfaction With Life Scale, and Marital Satisfaction Questionnaire for Older Persons (MSQFOP) for Men (Below Diagonal) and Women (Above Diagonal) Variable

1

2

3

4

5

1. Mean of satisfaction items 2. Satisfaction With Services subscale 3. Satisfaction With Health subscale 4. Satisfaction With Marriage/Home Life subscale 5. Global retirement satisfaction rating 6. Satisfaction With Life Scale 7. MSQFOP*

_ .77 .83

.86 — .50

.74 .46 .38

.37 .29 .30

.57 .47 .35

.47 .34 .11*

.69 .45 .48 .56

.26 .26 .42 .33

.73 .40 — .41 .44 .37 .32

— .31 .30 .65

.38 — .53 .41

.50 .50

.64 .51 .47 —

.56

Note. Unless indicated otherwise, n = 159 men, n = 243 women, and all correlations are significant at p < .001, two-tailed. " n = 143 married men and n = 136 married women. * p > .05.

616

FLOYD ET AL.

Table 4 Group Differences in Retirement Satisfaction and Experiences When Primary Reason for Retirement Involved Job Stress (JS), Employer Pressure (EP), Pursuit of Own Interests (OI), or Circumstances (C)

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Variable Retirement satisfaction Mean of satisfaction items Services subscale Health subscale Marriage/Home Life subscale Global satisfaction rating Preretirement functioning Job gratification Job satisfaction Anticipated satisfaction Adjustment and change Initial crisis Pre-post improvement Sources of enjoyment Reduced Stress and Responsibilities subscale Social Activities subscale Freedom and Control subscale

Ffor group effect 6.90*** 4.58** 4.59** 4.21**

Scheffe contrasts OI > JS, C OI > JS, C

OI>C OI>JS

2.21

19.17*** 20.09***

JS < OI, EP, C JS < OI, EP, C

2.24

3.33* 4.95**

EP>OI

4.08**

JS>EP

EP < JS, OI

2.01

3.86**

OI> EP

Note. JV= 317. All ScheffS contrasts are significant at p < .05. * p < .05. ** p < .01. *** p < .001, two-tailed. primarily to escape job stress reported relatively poor work adjustment but positive retirement experiences. As shown in Table 4, this group was distinguished from the other retirees by reporting the lowest levels of preretirement job gratification (M= 3.73, grand M = 4.72) and job satisfaction (M = 3.64, grand M= 4.73), but they reported the highest level of current enjoyment from reduced job stress (M = 3.21), which was significantly higher than the subscale score for reduced job stress by the group who retired because of pressure from the employer (Af=2.64). The results for those who indicated being pressured into involuntary retirement also were consistent with expectations. These retirees reported the most negative initial crisis period immediately following retirement (M = 3.65), which significantly differed from the ratings of initial crisis by the retirees who chose to pursue their own interests (M= 4.71). The pressured group also reported less positive long-term changes from pre- to postretirement (M= 3.95) as compared with both the job-stress (M = 4.94) and the own-interests groups (M = 4.87), and they rated having freedom and control less positively than the own-interests group (M = 2.99 vs. M = 3.42). In contrast, the voluntary retirees who endorsed the desire to pursue their own interests as the most important reason for retirement reported relatively positive satisfaction with life in retirement. This group obtained an overall mean satisfaction score (M = 4.96) that was significantly higher than the means for the groups who retired because of job stress (M = 4.58) and because of circumstances (M = 4.71). Similarly, as shown in Table 4, the own-pursuits group obtained higher scores on each of the three satisfaction subscales than either or both of the job-stress and circumstances groups.

We had no expectations about either particularly positive or negative experiences for people who retired primarily because of circumstances, and the results showed that this group tended to score consistently with the least extreme groups across all of the domains. The one exception was the significantly lower scores on the Satisfaction with Health subscale for the circumstances group (M = 4.37) versus the own-interests group (M = 4.76), which was consistent with the fact that retirement because of poor health was one of the items used to assign people to the circumstances group. Association With Demographic Characteristics A final set of analyses examined the associations of the RSI factor scores and single-item scores with several demographic variables. First, a series of ANOVAs contrasting the men and the women on all variables showed five significant differences. Compared with the men, the women anticipated less satisfaction in retirement, F(l, 400) = 5.73, p < .05, and they reported more stress immediately following retirement, F(l, 351) = 13.22, p < .001 (see Table 2 for means). However, the women also reported spending more time in leisure activities with their friends, F(l, 373) = 2.46, p < .05, and they rated two sources of enjoyment more important than did the men: reduced stress and responsibilities in retirement, F(\, 406) = 6.52, p < .05, and social activities, F(l, 406) = 5.28, p < .05. Next, the RSI variables were correlated with length of retirement, education, income for the year before retirement, and current annual income. We expected that higher satisfaction and enjoyment would be associated with longer time spent in retirement and higher levels of SES. Eighty correlations each were calculated for the men and the women, and the majority of these correlations were near zero. None of the correlations with length of retirement were significant, but three of the RSI variables showed similar patterns of significant correlations with the SES variables for both the men and the women. As expected, higher scores on the Satisfaction with Services subscale were associated with higher current income for both the men, r(l 58) = .38, p < .01, and the women, r(242) = .35, p < .01, and with higher income before retirement for the women, r(242) = .33, p < .01. However, higher scores for the subscale assessing enjoyment from reduced stress and responsibilities were associated with lower levels of education for both the men, r(158) = -.23, p < .01, and the women, r(242) = -.26, p < .01, and with lower preretirement income, r(242) = —.31, p < .01, and lower current income, r(242) = —.29, p < .01, for the women. Similarly, higher scores on the subscale assessing enjoyment from social interactions were associated with less education for the men, r(158) = -.43, and the women, r(242) = -.37, p < .01; lower income before retirement for the men, Kl 58) = -.27, and the women, r(242) = -.37, p < .01; and lower current income for the men, r(158) = -.28, and the women, r(242) =-.40, p

Assessing retirement satisfaction and perceptions of retirement experiences.

A 2-part investigation developed a measure of retirement as a life transition. Study 1 generated items from interviews with retirees (n = 40) and serv...
1MB Sizes 0 Downloads 0 Views