| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Department of Sociology, University of Kansas, Lawrence
b Department of Gerontology, University of Nebraska at Omaha
c Department of Social Sciences, University of Maryland Eastern Shore, Princess Anne
David J. Ekerdt, Gerontology Center, 3050 Dole Center, University of Kansas, Lawrence, KS 66045-2930 E-mail: dekerdt{at}ukans.edu.
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. Using survey responses among workers in the 1992 and 1994 waves of the Health and Retirement Study, we described the prevalence of nonsubstantive answers to questions about the expected form and timing of retirement (e.g., "don't know," "haven't thought about it"). We tested explanations for this uncertainty as an artifact of the survey process, but also as an outcome of the opportunity structure for retirement planning.
Results. Survey procedure did generate some of these noncommittal responses. Depending on question type, approximately 10% to 40% of workers did not state when or how they would retire, and such responses were less prevalent across age and time. In addition, categorical uncertainty about form and timing was theoretically predictable in a framework that supposed that workers less subject to a socially attended life—at work or away—would be more undecided about the future.
Discussion. Uncertainty is an authentic, meaningful stance toward retirement that theory and research design should not ignore. Just as actual transitions to retirement can be ambiguous or blurred, the expectation of retirement, as well, can be untidy.
PROMPTED by survey reports that some workers do not know or cannot report their plans for retirement, this study examined the assumption that retirement seems eventual to all workers. We tested explanations for uncertain plans as artifacts of the survey process and as outcomes of circumstances that weaken social attentiveness to the intentions of prospective retirees.
Every theory of retirement behavior—or decision making for that behavior—is also a theory about older workers' stance toward the future. Regardless of the way that retirement behavior is theorized, there is a metamodel of older workers that assumes they are on a path to retirement, holders of plans and preferences. They expect and are expected to retire, brought to this consciousness by advancing age and by features of social structure at the societal, firm, and family levels that bear in to them the contemporary habit, and perhaps the norm, of retirement. This consciousness of life's next turns arises at other stages of the life course, as when adolescents normally anticipate a nexus of education, work, and marital decisions, or when older people normally contemplate their final arrangements and legacy (
Heckhausen and Schulz 1995
).
Theories of retirement decision making express the metamodel in different ways but they revolve, nonetheless, around the idea of the planful, deliberative preretiree. One strain of theory views retirement as an individual choice: Workers prospectively evaluate the utility of retirement, then act at some point. In the dynamic models of economists, workers age through the hazard of retirement, choosing in each successive period whether to work or retire based on the relative advantage of the competing statuses (
Berkovec and Stern 1991;
Stock and Wise 1990). In another economic model of the retirement process, workers design their route to retirement, some making a gradual withdrawal from the labor force using one or more "bridge jobs" to acquire reduced or flexible hours (
Quinn and Kozy 1996;
Ruhm 1990). Psychologists see workers evaluating retirement for its utility in reaching valued personal goals (
Talaga and Beehr 1989;
Taylor and Shore 1995). Another strain of theory views retirement as a normative program that workers carry out. In the life-course perspective of sociologists, workers attend to the (institutionally) scheduled "givenness" of retirement and devise actions toward it, subject to the opportunities and constraints of their situation (
Ekerdt, DeViney, and Kosloski 1996;
Kohli 1986). Whether retirement is an individual choice or a normative program, what all these approaches share in common is a view that older workers engage themselves farsightedly in plans, choices, or serial decisions related to retirement.
The metamodel—every worker consciously on a path to retirement—also undergirds policy discussions about retirement savings. For example, one of the proposed reforms for the Social Security retirement program would institute a system of mandatory personal savings accounts (
Rosenbaum 1999). A percentage of the pay of each American worker would be deposited in a self-directed savings account to accumulate income for one's (presumably) inevitable retirement. As another example, numerous surveys report subjective and objective evaluations of Americans' savings readiness for retirement (
Americans Discuss Social Security 1998;
Employee Benefit Research Institute 1999). In these reports, the conscientious saver becomes the exemplar against which all other working adults are measured.
Nonetheless, there is a stubborn suggestion in prior research that some older workers are not on a clear path to retirement. This is not uncertainty about the details of the transition (pension amounts, living arrangements) but rather about the basic facts of retirement—when it will happen or how it will occur. We have reported elsewhere that more than 40% of workers in their 50s will not or cannot project the form that their retirement will take (
Ekerdt et al. 1996). Recent surveys show that anywhere from 5% to 7% of adults cannot cite the age at which they will retire, and another 5% to 8% say that they will never retire (
Americans Discuss Social Security 1998;
Employee Benefit Research Institute 1999;
National Public Radio 1999).
We allow that the absence of settled preferences or personal schedules for retirement does not necessarily mean that a worker is ignoring the subject. Perhaps a worker entertains real, conceivable alternatives, but has not yet committed to any one of them. Not knowing how or when retirement will occur could be, thus, a temporary state, perhaps even a strategy for keeping one's options open.
At the same time, the general literature on survey design warns that "nonsubstantive responses" such as "don't know", "not sure", and "no opinion" can be meaningful, can identify a certain type of respondent, and can provide information about the conceptual relevance and clarity of a topic (
Brody 1986;
Duncan and Stenbeck 1988;
Francis and Busch 1975). The nonresponse frequencies cited above suggest that the preretiree with uncertain plans could be an authentic type of older worker, every bit as worthy of theoretical consideration as the other types conveyed in theory, such as the rational chooser or the bridge-job strategizer. Rather than drop cases from analysis when they offer nonsubstantive responses to questions about retirement plans (and thus bias results in unknown ways), don't-know responses deserve to be understood (
Dannefer 1988).
To investigate whether uncertainty is an authentic individual stance, item nonresponse about retirement plans or intentions could be explained in three ways. First, the prevalence of uncertain retirement plans could be an artifact of the way in which survey questions are asked. Survey questions that pose greater cognitive difficulty tend to generate more don't-know responses (
Converse 1976;
Schuman and Presser 1981). Questions about the form of retirement appear to call for more complex judgment and future projection than do questions about timing. Perhaps, then, question wording exaggerates the prevalence of uncertainty.
Second, uncertain plans could be a function of age or the temporal salience of retirement. Surveys may be asking for specifics too soon relative to the typical occurrence of the event. There are cultural ideas and workplace norms about the usual age for retirement (
Ekerdt 1998;
Settersten and Hagestad 1996), and workers may simply need closer proximity to these ages before crossing the threshold of serious retirement consideration. We have reported a weak but positive correlation (.08) between age and a composite measure of involvement with retirement (thinking and talking about it) in the HRS (
Ekerdt, Kosloski, and DeViney 2000). If age salience is a reason for uncertain plans, we could expect to find lower levels of uncertainty at older ages. We would also expect to see less uncertainty over time in any cohort of continuing workers. Advancing age presumably brings adults closer to normative retirement ages, brings them more information, and subjects them to stronger social expectations about the transition, reducing their ability to avoid or defer questions about how and when. All of these dynamics should reduce the prevalence of uncertain plans.
Even if uncertainty abates with age or time, a third focus of explanation would regard the "don't know" response as proceeding from an opportunity structure for retirement planning—that is, a set of circumstances that makes future alternatives for retirement more or less conceivable to an individual. Research by ourselves and others has identified various correlates of workers' specific intentions for retirement, as well as its topicality (
Ekerdt et al. 1996;
Ekerdt et al. 2000;
Fronstin 1999;
Hall and Johnson 1980;
Henkens and Tazelaar 1997; Kosloski, Ekerdt, & DeViney, in press;
Shaw 1984). These include pension eligibility and the availability of retiree health insurance, longer job tenure, larger workplaces, union membership, and whether one is self-employed.
Ekerdt and colleagues 2000 summarized the first set of factors by noting that they jointly indicate a worker's participation in bureaucratic work that has rationalized grade, promotion, and pension systems that, in turn, would orient one to a timetable for job exit. Other correlates of retirement intentions are older age, being male, being married, and greater education, income, and health.
It is possible to make a more comprehensive summary of both sets of factors: Together, they indicate the extent to which a worker has a life that is socially "attended," where the root meaning of attend is to look after or accompany. As noted, workers in bureaucratic employment have their tenures regulated or administered in ways that the self-employed do not. Also as noted, advancing age subjects persons to a higher topicality of retirement from coworkers, friends, and family. Likewise, ill health and disability raise the surveillance of others over one's life choices. Spouses are also surveillants of each other, so that marriage promotes mutual review and discussion of retirement plans (
Henkens 1999). In all, when it is more likely that one's retirement intentions are a public matter—on the job, at home, among friends—one can be said to have an "attended life" that is the context or opportunity structure for anticipating retirement. This proposition is consistent with emerging research in behavioral economics suggesting that social networks have a key role in the formulation of retirement behavior (
Aaron 1999).
We expect that uncertain plans for retirement are more prevalent when lives are less socially attended. Other known correlates of retirement plans—sex, education, income—also happen to be consistent predictors of the "don't know" response in surveys generally, where women and persons of lower education and income are more likely to have "no opinion" (
Francis and Busch 1975). Their nonsubstantive response supposedly stems from structurally reduced access to information. In the context of retirement planning, we likewise expect these factors to predict uncertainty because they are covariates of a socially attended life, especially of rationalized, bureaucratic employment and the information and material base it provides for formulating plans.
In the following analysis, we evaluate whether nonsubstantive response about retirement plans is more than a survey artifact and perhaps is an authentic preretirement stance. Following the three explanations for uncertainty, we first assess the prevalence of uncertain plans under different question regimes in the Health and Retirement Study. We next test whether, as expected, age and time reduce the prevalence of uncertain plans. Temporal salience notwithstanding, we also examine whether uncertain plans follow the contours of a life that is, for questions of retirement, less socially attended, both in cross-section and across time.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). Total sample size for the first (1992) wave of this biennial longitudinal survey was 12,652 persons in some 7,000 households. This design affords a sample population that represents the late career and retirement experience of both men and women. The study population for the present analysis included all persons aged 51 to 61 (born 1931 to 1941) in 1992. Because the HRS oversampled Blacks, Hispanics, and Floridians, we used sample weights to adjust for unequal selection probabilities, taking the full complement of 9,824 age-eligible cases to 7,492 for all statistical procedures. This number was further reduced to 4,921 by including only current workers who also claimed they had not completely retired. The 1992 study population was divided between 2,575 men and 2,346 women, was 82.1% White, and had a mean age of 55.6 years. By 1994, 40 persons had died and 360 were not reinterviewed, but 92% (4,521) of the 1992 preretirees were followed up. Of these, 3,888 were current workers at the time of the 1994 survey and thus compose our study population of continuing workers for the analysis of changing plans.
Dichotomous Measures of Uncertain Plans
Workers' intentions about retirement (in this case, nonresponse about intentions) can be read three ways from different items in the HRS survey (Table 1 ). One is a question about the form of retirement. This question followed another about "usual retirement age" on one's main job. The interviewer next asked: "Are you currently planning to stop working altogether or work fewer hours at a particular date or age, to change the kind of work you do when you reach a particular age, have you not given it much thought, or what?" When the question was repeated in 1994, respondents were led in with: "Now I want to ask you about your retirement plans." Responses to this somewhat open-ended question were coded into these categories: stop work altogether, never stop work, not given much thought, no current plans, work fewer hours, change kind of work, work for myself, and other.
|
). Essentially, workers were grouped in the category "No Current Plans" only if they said that they had not given retirement much thought or had no current plans. A second item that came later in the 1992 survey asked workers about the timing of retirement. Respondents were asked whether they considered themselves partly, completely, or not retired at all. If not retired, they were next asked, "When do you think you will retire?" (or "...will retire completely?" in the case of partly retired persons). Most respondents offered some expected age or date for retirement, but the HRS also accepted the volunteered responses "never" and "don't know." The survey item was not repeated in 1994.
A third way to assess uncertainty is based on two other HRS items that asked workers for their surety about two specific retirement intentions: the chances that they would work full time past age 62 and past age 65. Workers chose their responses on a 0–10 scale, where 0 equals "absolutely no chance" and 10 equals "absolutely certain." For this study, we grouped responses in the middle of that range—4, 5, or 6—as indicating ambivalence about that specific intention.
Honig 1996 notes that persons making such midscale responses "may not have made a decision or may face greater uncertainty" (p. 375).
When considering the prevalence of uncertain plans among workers aged 51–61, it is well to remember that this age group had already experienced early retirements, presumably of some workers who were more planful about their futures. Retirements between 1992 and 1994 may also have removed more planful preretirees from the cohort of continuing workers.
Other Measures
We selected several variables to represent the opportunity structure for retirement planning, which we have broadly summarized as a socially attended life. These are listed in Table 2 (left-hand column) with their means or percentage distributions. Age, because of the life course "givenness" of retirement, brings greater social scrutiny of the older worker. Sex, education, and race are structural factors that provide access to bureaucratic employment and that bring information and resources for planning (
Dannefer 1987;
O'Rand 1996).
|
Health difficulties should raise the salience of retirement (
Ekerdt et al. 2000), and so we included an indicator of whether health limits paid employment. Finally, the familial supervision of retirement plans should be greater among married workers, here a binary variable.
In addition to these predictors, which represent the opportunity structure for retirement planning, regression models also controlled for partial retirement and part-time employment. Workers were "partly retired" according to explicit self-report about their retirement status. They were coded as having part-time employment if "usually" working less than 35 hours per week.
The HRS panel data allow us to explore whether a changed opportunity structure for retirement planning affects follow-up responses about retirement intentions. A changed opportunity structure was represented by job, marital, and health changes between 1992 and 1994. Workers changing jobs could have their retirement prospects clouded or clarified. Of the 3,888 continuing workers between 1992 and 1994, 10.5% were not on the same job at follow-up. Marital status changes, likewise, could disrupt or resolve plans about the form of retirement. Altogether, 3.8% of continuing workers had any sort of marital change (exit or reentry). This small number does not permit examination of more specific transitions. Job and marital change were introduced into the regression models as binary variables. Following cross-sectional evidence, we venture the hypothesis that health changes for the worse should focus planning and reduce uncertainty. To measure health, we used a 1994 self-reported evaluation of health as better or worse than 2 years ago, with responses on a 5-point scale (much worse health = 5; M = 3.0 ± 0.6).
Method of Analysis
We have proposed three explanations for uncertain retirement plans. To consider the first—whether uncertain plans are an artifact of question type—we examined responses on different items in the HRS survey. To consider the second—uncertainty as a function of age-salience—we examined responses by age in cross-section, and also examined the stability of uncertain plans between 1992 and 1994. To consider the third—that uncertainty arises in an opportunity structure—we regressed the report of uncertain plans in 1992 and 1994 on variables that delineate a socially attended life. Weighted regression models were estimated using Stata routines for complex survey data (
Statacorp 1999).
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
The simpler question about the timing of retirement (Table 1 B) yielded less uncertainty about retirement intentions than did the question about form. Rather than cite some expected age or date, 12% of workers volunteered the answer "don't know" or "haven't thought about it."
(Both questions also generated a "never" response. What is significant about this answer is that workers volunteered it—it was not a response alternative that was suggested by the interviewer. On the form question, 7.2% [1992] and 7.1% [1994] would "never stop work." On the 1992 timing question, 13.4% would "never" retire. In this report, we will not analyze the "never" response as an indication of uncertainty because, in our reading, it seems to be the voluntary expression of a definite intention. Yet, grouping the no-plans and never responses together for each question indicates that nearly half of these workers are not on the path to a specific form of retirement and about one quarter are not citing a time for retirement. These are large, nonignorable proportions of workers in their 50s who are not entertaining an explicit retirement intention.)
The form and timing questions (A and B in Table 1 ) ask about different types of retirement intentions that nevertheless may coincide. A cross-classification of the 1992 responses (not shown) showed that those uncertain about form still tended to cite a time for retirement, but those uncertain about timing were also uncertain about form. Of those with no current plans about form, 66.5% still offered some age or date for retirement, 13.6% said "never," and 19.6% "don't know" about timing. This is evidence that the majority of those who cannot say how they will retire can still address the simpler question about when. Reading the comparison in reverse, for those who "don't know" about timing, fully 70.6% also had No Current Plans as to form. Jointly, 8.5% of the total group of workers in 1992 were uncertain about both form and timing. We regard this as the most conservative estimate of uncertain retirement plans, encompassing 1 out of 12 workers.
Rounding out the frequencies in Table 1 , the HRS also included two questions about specific intentions to work full time past age 62 or 65 (Table 1 C). For the age-62 scenario, 19% of workers chose a midscale response between the absolutely-no-chance and absolutely-certain endpoints. (Most of these actually chose the value of 5 on the 0–10 scale.) For the age-65 scenario, 15.9% placed themselves in the middle. The proportions of midscale responses were similar in 1994.
Thus, the framing of questions about retirement plans does make a difference: About 40% of workers were uncertain about the form of retirement, 12% were uncertain about the less complex topic of timing, and 16–19% were undecided about the likelihood of specific scenarios. And, about 1 in twelve workers were uncertain about both form and timing. These are not negligible proportions who are unsure of their futures.
For the remaining analyses, we used the responses about form and timing only. The proportions undecided about specific scenarios are informative, but they do not provide the analysis with explicit don't-know responses.
Age and Temporal Salience
The prevalence of uncertain plans could be explained by the prematurity of the questions about retirement. In this case, we would expect less uncertainty at older ages and a reduction in uncertainty over time. To a limited extent, both these patterns were observed.
The three panels of Fig. 1 show the percentage of workers with uncertain plans by age for the 1992 questions about form and timing. The proportion with No Current Plans as to the form of retirement (left panel) was slightly lower at older ages. If this response were coded as a binary variable (no plans vs any plan), the correlation with age would be –.08 (p < .001). There was no age pattern for don't-know responses on the timing question (middle panel), or for joint occurrence of uncertainty on the form and timing questions (right panel).
|
|
These aggregate percentages by age and year ( Fig. 2) do not speak to intra-individual continuity of plans. Persons may retain uncertain plans over time, or they may shift toward or away from specific intentions. Overall, of workers with No Current Plans in 1992, 51.2% repeated this response in 1994. If we consider all continuing workers, then 23.5% had No Current Plans both at baseline and at follow-up. Fig. 3 displays the percentage of workers by age who had No Current Plans at baseline and follow-up. The bars again show that older workers were not quite as likely to repeat the response. If continuity of No Current Plans were coded as a binary variable, the correlation with age would be –.10 (p < .001). Between 1992 and 1994, persons could also shift from some specific intention into No Current Plans. Overall, 15.7% did just that, but this percentage did not increase across age.
|
Opportunity Structure
If uncertain plans are not wholly an artifact of survey design—proceeding from questions that are too difficult or not relevant—then they can be regarded as meaningful and perhaps explainable by circumstances that cloud or clarify the prospect of retirement. Table 2 shows the regression of uncertain plans on measures of such circumstances, controlling also for partial retirement and part-time employment. Uncertainty was measured by binary variables in three alternate ways. First, we grouped workers with No Current Plans about form in 1992 (42.8% of respondents) versus workers with any plan. Second, we compared the 1992 "don't know" response about timing (12.0% of respondents) to other responses. Third, we combined these two 1992 responses to indicate uncertainty about both form and timing (8.5% of respondents). The available number of cases for each regression model was reduced by 6% due to listwise deletion for missing data. Collinearity was not a problem in these regression models, as indicated by an average variance inflation factor of 1.19 (
Fox 1991).
Two of the three dependent variables in Table 2 are variations on the form-of-retirement question, and so it is not surprising that the independent variables tend to have consistent signs across models. Findings are also quite similar between the second and third models because workers uncertain about timing also are the majority of those uncertain about both form and timing.
Older age predicted less uncertainty as to form, a pattern already observed in Fig. 1 Fig. 2 Fig. 3. Controlling for age, the two strongest predictors of all types of uncertainty were sex and pension eligibility, where women and those not enrolled in pension plans were less sure of the form and timing of retirement. Unmarried persons were also less sure of form and timing. Workers with fewer years of education and with smaller workplaces were less sure as to form. Part-time work, while a covariate in these models, could also indicate workers who are further removed from the rationalized, bureaucratic employment that focuses retirement planning. This group of workers was consistently more likely to offer substantive nonresponses about form and timing.
These variables predict form and timing alike, with effects in the same direction. Two variables, however, have an inconsistent direction of odds ratios between the 1992 models for form and timing (first two models in Table 2 ). Self-employment predicted less uncertainty about form (odds ratio [OR] = .804; p < .05), but predicted greater uncertainty about timing (OR = 1.252; p < .10). The self-employed thus tended to report what they will do, but tended not to say when they will do it. (From other analyses [
Ekerdt et al. 1996] we know that the self-employed disproportionately favor partial retirement and "never" retiring over other options.) In a parallel way, health limitations predicted less uncertainty about form (OR = .754; p < .01) but greater uncertainty about timing (OR = 1.273; p < .10). These findings suggest a particular variety of uncertainty: commitment to an intention but indecision about the timing for its implementation.
Follow-up administration of the form-of-retirement question in 1994 allowed us to examine the stability of uncertain plans as well as switches toward or away from this response. As noted earlier, 23.5% of continuing workers had No Current Plans in 1992 and 1994, 21.1% switched from No Current Plans to some specific plan, 15.7% switched from some specific plan to No Current Plans (were newly uncertain), and 39.7% on both surveys stated specific plans for the form of retirement. These categories served as the dependent variable for two multinomial logistic regression models predicting change in retirement plans, with the reference category being workers with some specific plan both in 1992 and 1994 (Table 3 ). Of particular interest are workers who retain the no-plans response and workers who adopt the no-plans response (left- and right-hand columns in Table 3 ). Listwise deletion for missing data reduced the number of available cases by 4%.
|
Model II in Table 3 included indicators of interim events in the opportunity structure, as well as other baseline covariates (results for covariates not shown). These findings show that the adoption of the no-plans response was more likely when having made a job change, perhaps because new employment presents new retirement possibilities that take time to sort out. Results were inconclusive regarding marital status change. Further research could well discriminate among types of marital status change (remarried, divorced, widowed) and their effect on changed retirement plans. Finally, the adoption of uncertain plans was significantly less likely with a worse opinion of one's health. The persistence of uncertain plans was also less likely with worse health, though not statistically significant (p < .10). Consistent with other results here, worse or worsening health focuses retirement intentions. Altogether, the findings in Model II are some evidence that new circumstances contribute instability to retirement intentions.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
The prevalence of uncertain plans depended on the type of question that was asked. About 40% of workers aged 51–61 could not state how they would exit their job, and about 20% altogether were undecided about this at successive survey waves in 1992 and 1994. There was less uncertainty about the timing of retirement, where 12% said they "don't know" the date or age. The core of uncertainty is probably the 8.5% of workers who had nonresponses about the form and the timing of retirement. Altogether, these figures describe the explicit prevalence of uncertainty, but it could be more pervasive than that. Among HRS respondents who stated plans to retire at this age or in that manner, many of these intentions could well be tentative.
Nonresponse to the more complex question about the form of retirement abated across age and time, indicating that the salience of retirement rises with its approach (
Ekerdt et al. 2000). Even so, sizable numbers of older workers were unclear about a retirement strategy even in their early 50s. We infer from these findings that that there is a point in adulthood when it is too soon to profitably ask workers to state their retirement intentions. Nevertheless, it is a common practice for opinion surveys to plumb retirement topics among workers in their 30s and 40s and to promote the results as heralding the future of retirement (
Americans Discuss Social Security 1998;
National Public Radio 1999).
Although some uncertainty was probably an artifact of survey design, our regression results were consistent with more general research on nonsubstantive responses that has viewed them as meaningful expressions. Categorical uncertainty about form and timing was theoretically predictable in a framework that supposed that workers less subject to a socially "attended" life—at work or away—would be more undecided about the future. Besides age, which raises the social salience and topicality of retirement plans, pension availability was a prominent predictor in our regression models. Pensions focus retirement decision making, as they are institutionally designed to do (
Quadagno and Hardy 1996). No pension, less focus. Female workers had greater uncertainty, confirming observations about their reduced access to an information and material base for retirement planning (
Calasanti 1993); also their retirement plans may await the plans of others with whom their lives are linked (
Elder 1998). On this last point, marriage nevertheless seemed to benefit the making of retirement plans, something long suspected in the literature (
Szinovacz and DeViney 2000). In all, the opportunity structure for retirement planning identified factors that favor uncertainty and may point to persons at risk of disorderly or ill-considered retirement transitions, or who fail to save or prepare adequately.
Indifferent plans, we maintain, flow from social contexts in which such intentions matter less to others. At the same time, it is possible that some don't-know responses may be tactical—held by workers who as yet are uncommitted to available options rather than really don't know (see
Duncan and Stenbeck 1988). The findings on age and time suggest this possibility, as does the finding that workers in better health, which is certainly no disadvantage, were more likely to hold or adopt a don't-know stance about the form of retirement. At any rate, the distinction between "don't-know" and "won't say" is not something that can be pursued with these survey data.
The theoretical story of retirement at the microlevel is morally reassuring. Older workers, as preretirees, exercise choices, options, and strategies in pursuit of their preferences for retirement. Workers control their transition to retirement, less bound by the structural imperatives of the past that had channeled them out of the labor force (
Han and Moen 1999). But it is well to ask how purposive and planful preretirees are. Although the anticipation of retirement and its topicality seem normative for older workers (
Ekerdt et al. 2000), the responses of workers in the HRS indicate that many do not know what they will do. They may be farsighted, but with no end in view. One need only read qualitative treatments of retirement behavior to understand how quirky and self-surprising the transition can be (
Savishinsky 2000). The heterogeneity of retirement expectations, and authentic uncertainty in particular, is unfortunately often hidden in data analyses that delete cases with a don't-know response, or sequester them in the residual category of a dichotomy, such as "plan early retirement" versus "other." Such maneuvers bias analysis samples and reduce the precision of findings. Just as actual transitions to retirement can be ambiguous or blurred (
Mutchler, Burr, Pienta, and Massagli 1997), the expectation of retirement can be untidy, too. For theoretical and design adequacy, uncertainty about prospective retirement is a circumstance that should not be ignored.
|
Acknowledgments |
|---|
Received for publication May 26, 2000. Accepted for publication November 21, 2000.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. M. Raymo and M. M. Sweeney Work-family conflict and retirement preferences. J. Gerontol. B. Psychol. Sci. Soc. Sci., May 1, 2006; 61(3): S161 - S169. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. J. Ekerdt and J. K. Hackney Workers' Ignorance of Retirement Benefits Gerontologist, August 1, 2002; 42(4): 543 - 551. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
