| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 School of Public Health, University at Albany, Rensselaer, New York.
2 School of Public Health, University of Michigan, Ann Arbor.
Address correspondence to Benjamin A. Shaw, Department of Health Policy, Management, and Behavior, University at Albany School of Public Health, One University Place, Rensselaer, NY 12144. E-mail: bashaw{at}albany.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. We used hierarchical linear modeling with data from a nationwide survey of 1,103 elders who were interviewed up to four times over a 10-year period.
Results. Age-related changes in social relations varied across the different dimensions, and significant interpersonal differences existed in these trajectories. Emotional support was relatively stable with advancing age, whereas other types of received support (i.e., tangible and informational) increased with age and levels of provided support declined. Furthermore, the findings revealed declines in contact with friends, support satisfaction, and anticipated support. These changes were not uniform throughout the sample, as indicated by significant random effects with respect to the intercepts and slopes in virtually each model. Gender and socioeconomic status accounted for some of this variation.
Discussion. These findings highlight the dynamic nature of social relationships in late life. In addition, the findings both provide evidence of older adults managing their social ties to meet the challenges of aging and suggest the importance of the interplay between giving and receiving support.
AN abundance of research findings reveal that the quality and quantity of older adults' social ties are closely linked with a variety of health outcomes, including mortality (Eng, Rimm, Fitzmaurice, & Kawachi, 2002
), functional disability (Mendes de Leon et al., 1999), cognitive impairment (Seeman, Lusignolo, Albert, & Berkman, 2001), psychological distress (Matt & Dean, 1993), and disease risk and recovery (for a review, see Seeman, 2000). Indeed, maintaining social connections and having access to social support are key markers of successful aging (Rowe & Kahn, 1998).
Despite compelling evidence that social support is associated with health, however, there is still a great deal researchers do not know about the social support process. Most research on social support and social networks has utilized cross-sectional data. Moreover, analyses of longitudinal data over an extended period of time are rare. As a result, experts know relatively little about how social support changes as people age. When researchers have examined change, most of their studies have focused on aggregate-level, rather than individual-level, change (e.g., Field & Minkler, 1988). Studying within-person change is important because, as research by Krause (1999) revealed, there may be substantial individual variations in levels of support over time.
Describing patterns of change as well as variations in these patterns, however, is only a first step. Researchers also must explain, or account for, interpersonal differences in intra-individual change. Consistent with findings from cross-sectional studies (e.g., Turner & Marino, 1994), the current study focused on social structural differences in individual change. More specifically, we assessed whether gender (Antonucci & Akiyama, 1987), socioeconomic status (Krause & Borawski-Clark, 1995), and race (Peek & O'Neill, 2001) explain change in support over time within individuals.
Aging and Social Relationships
In order to lay the theoretical foundation for this research, we first outline key perspectives regarding the nature of age-related change in social relationship resources. Early work in this area relied upon disengagement theory to explain the hypothesized decline in social interaction among aging adults (Cummings & Henry, 1961). Other scholars, however, argued that aging should not be viewed simply as a period of disengagement, but rather as a time when older adults should be striving to maintain their social activities and social roles (Neugarten, Havighurst, & Tobin, 1968). Still others emphasized older adults' needs for continuity over time in their patterns of social relationships (Atchley, 1989).
More recently, socioemotional selectivity theory has provided a compelling explanation for how social relationships change with advancing age (Carstensen, 1992). This theory suggests that aging adults deliberately allow for reductions of certain types of social ties while striving to maintain others. The social ties that are disbanded are most likely to be of a nonintimate and peripheral nature, whereas those that are maintained are likely to be the most intimate ties in one's network (see also Tornstam, 1997). The reasoning behind this selective reduction in social ties is that as individuals get older and begin to perceive that time is limited, short-term rather than long-term goals become increasingly important. Proponents of this theory consider emotional well-being to be one of the most salient short-terms goals for older adults, and maintaining intimate social ties appears to offer the best opportunities for satisfying this particular goal (M. M. Baltes & Carstensen, 1999). Thus, socioemotional selectivity theory suggests that age-related changes in social relationships are not uniform with respect to all dimensions of support: Elders may replace losses in social contact and support from peripheral ties with gains in contact and emotional support from the more intimate network members that remain. This pattern of replacement of social ties over the life course is consistent with Kahn and Antonucci's (1981) social convoy model as well as P. B. Baltes's (1997) theory of successful aging known as selection, optimization, and compensation.
Empirical Evidence of Age-Related Changes in Social Relationships
Empirical support for the social transitions described by socioemotional selectivity theory is growing, but it suffers from methodological limitations. Most notably, several studies examining social relationship changes in late life have been based on cross-sectional data. For example, studies by Morgan (1988) and Lang and Carstensen (1994) found that currently older respondents reported smaller social networks than did presently younger respondents. Additional cross-sectional studies found stability in levels of support across different age groups (Antonucci & Akiyama, 1987), whereas others found age differences in sources of support, with older adults reporting most of their support as coming from family members and younger adults reporting more support from friends (Levitt, Weber, & Guacci, 1993).
Although findings from cross-sectional studies are informative, evaluating age differences in social relationships across, rather than within, individuals makes it impossible to distinguish between true age effects and those that are due to cohort or period differences across the sample. Therefore, some investigators have used longitudinal data in an attempt to more accurately represent the experiences of individuals as they progress through adulthood. Most of these studies corroborated findings from cross-sectional studies with respect to an overall decline in network size or social contact among older adults (e.g., Barnes, Mendes de Leon, Bienias, & Evans, 2004; Krause, 1999). Nevertheless, others found considerable stability in social contact with advancing age (Martire, Schulz, Mittelmark, & Newsom, 1999). Again, it appears that different patterns of change are evident across different aspects of social relationships. For example, as suggested by socioemotional selectivity theory, Field and Minkler's (1988) findings showed that the bulk of reductions in social contact were associated with social ties that were beyond the family (see also van Tilburg, 1998). Additional studies suggested that at least some types of support from social network members increase with age (e.g., Cornman, Lynch, Goldman, Weinstein, & Lin, 2004), whereas others found levels of social support to be stable over time (Bergeman, Neiderhiser, Pedersen, & Plomin, 2001), even in the face of declining social roles (Bosse, Aldwin, Levenson, Spiro, & Mroczek, 1993).
Although these findings appear to provide some support for the basic tenants of socioemotional selectivity theory, it is uncertain if they represent the experiences of the general population of older adults as these individuals age over extended periods of time. The current study aimed to address this issue by estimating change in a full complement of social support and social network measures over an extended period of time, using data from a nationally representative sample of older adults. Specifically, we hypothesized that only those aspects of social relationships that helped older adults to meet their relatively immediate needs, such as emotional and tangible support, would increase with advancing age. In contrast, we hypothesized decreasing levels of social relationship dimensions that were unhelpful in meeting immediate needs, such as provided support or negative interaction, or that involved peripheral rather than intimate ties.
Heterogeneity in Age-Related Trajectories of Social Relationships
The general age-related changes in social relationships described by socioemotional selectivity theory apply to the population as a whole; however, the notion of aged heterogeneity (Dannefer & Sell, 1988) predicts substantial variation around these general trends. Indeed, compelling empirical support for such heterogeneity in age-related social relationship changes currently exists (Krause, 1999; Stoller & Pugliesi, 1991; Taylor & Lynch, 2004). Presently, however, there is a clear need to explain this heterogeneity by identifying factors that are associated with variability in these individual trajectories. Consistent with the notion of a social stratification of social relationship resources, and given the strong cross-sectional evidence for gender, socioeconomic, and race differences in social relationships resources, it makes sense to begin trying to account for variation in social relationship trajectories by focusing on these same social status indicators. Accordingly, in light of evidence that women have relatively strong interpersonal orientations compared to men (Turner & Marino, 1994), we hypothesized less age-related decrement in social relationship resources among aging women than among men. In addition, because most research seems to suggest that social networks in lower socioeconomic groups do not function as well as those in upper socioeconomic groups (Krause & Borawski-Clark, 1995), we hypothesized greater decrement in the social relationship trajectories of older adults of low socioeconomic status compared with their counterparts of high socioeconomic status. Finally, although some controversy exists in the literature (Kaniasty & Norris, 2000), the bulk of current evidence seems to indicate that social networks are more well developed among minority elders. Thus, we hypothesized advantages in the social relationship trajectories of older minorities compared to Whites.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). Researchers conducted the baseline survey in 1992/1993, at which time they defined the study population as all household residents aged 65 or older who were not institutionalized, who spoke English, and who were retired (i.e., not working for pay). The survey excluded residents of Alaska and Hawaii, as well as elderly people who did not have a Social Security number and those who were 100 years or older at baseline. A total of 1,103 respondents completed interviews at baseline. Researchers collected additional waves of data from 605 study participants in 1996/1997, 530 participants in 1998/1999, and 269 respondents in 2002/2003.
To minimize the loss of respondents due to missing items or attrition, we used multiple imputation (Rubin, 1987). Specifically, we imputed missing data on social support for any nonrespondent at a given wave by using data collected at baseline as well as repeated measures of social support up to the time when that individual became a nonrespondent. We imputed six complete data sets with the NORM software developed by Schafer (1997), and we ran analyses on each of these six data sets. For all analyses, we averaged each estimate across the six imputations to generate a single point estimate. We then calculated standard errors using a formula that combined the average of the squared errors of the estimates and the variance of the parameter estimates across the six samples (Rubin, 1987). Having employed these multiple imputation procedures, we obtained sample sizes at each wave as follows: 1,103 at Wave 1; 1,103 at Wave 2; 930 at Wave 3; and 854 at Wave 4.
The data analytic procedures described in the section Data Analysis (hierarchical linear modeling) allowed us to base estimates on all 1,103 original respondents. Of these, 39% were men, and 92% were White. The average age at baseline was 74.49 years (SD = 6.74). Also at baseline, these respondents reported having completed an average of 11.39 years (SD = 3.58) of schooling.
Measures
The survey contained measures of 11 dimensions of social networks and support in each of the four waves. Table 1 presents the individual items that contributed to each scale, along with their response options. We selected these measures to represent each of the social relationship dimensions identified by Barrera (1986): (a) social embeddedness, which is assessed with measures of the frequency of contact with family and friends; (b) enacted support, which is measured with items that estimate the amount of emotional, tangible, and informational support received from others in the year prior to the interview; and (c) perceived support, which is measured by indicators of satisfaction with support exchanges, anticipated support (i.e., the belief that assistance would be forthcoming in the future should the need arise), and negative interaction. The survey also included measures of the amount of emotional, tangible, and informational support that respondents provided to others in the year prior to the survey. The origins of each of these scales, as well as their internal consistency reliability estimates, have been described previously (Krause, 1999). As revealed in that study, all of the reliability estimates are acceptable. Table 2 presents the means and standard deviations for each scale at each wave.
|
|
Data Analysis
The data analytic strategy employed to make full use of these longitudinal data involved estimating individual social relations trajectories, or growth curves, using hierarchical linear modeling (Raudenbush & Bryk, 2002). The analysis of individual growth curves is considered hierarchical in that the four occasions of measurement of social relations are nested within individuals. This enabled us to examine how social relations changed with increasing age within individuals, rather than just across age groups as is proposed with cross-sectional data.
We carried out the statistical procedures for examining individual growth curves in two stages. In the descriptive stage, we modeled social support as a function of age (centered on the sample mean) over the course of the four waves of data collection for each individual as follows (Level 1):
|
|
0i is the intercept, which, because the variable Age is centered on the sample mean, is located at the mean age level across all observations (78.69); 1i is the rate of change (slope) in social support for individual i with increasing age; and 
ij represents random error for individual i at age j.
In order to control for potential cohort effects, we included two dummy variables representing birth cohort in the Level 2 models as follows:
|
|
|
|
, p. 56). Thus, the fixed effects ß00 and ß10 represent the average level of social support and the average growth rate for social support with increasing age for the sample taken as a whole, after controlling for the effects of birth cohort. The random effects u0i and u1i represent the degree to which respondents' social support at the intercept and growth trajectories deviates from the mean. We used these to determine whether social support levels and rates of change were uniform throughout the sample, or alternatively, whether the data included between-subjects variance with regard to trajectories of social support. We should note that in addition to the linear model presented in Equations 1 through 3, we tested a nonlinear model by adding a quadratic term (2iAgeij2) to Equation 1 and decomposing the parameter 2i into fixed and random effects exactly as we had done in Equations 2 and 3.
In the second stage of analysis, we entered the explanatory variables (gender, education, and race) into the Level 2 models (Equations 2 and 3) in order to explain individual differences in the intercept and slope parameters. We can represent these explanatory models by combining the Level 1 and 2 models as follows:
|
|
). |
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
|
Analysis of Attrition
Before turning to the analyses testing the extent to which gender, education, and race explained individual differences in the intercept and slope parameters, it is necessary to call attention to the fact that those individuals with complete data across all four waves of this study were likely to differ from those who had dropped out or died during the study period. Although the fixed effects presented in Tables 3 through 6 were based on data from all 1,103 original participants, because these effects were weighted composites of each individual trajectory, those with complete data were given more weight than those with incomplete data. Thus, it was possible that these fixed effects masked systematic differences in individual growth curves between participants with complete data and those who had dropped out.
In order to investigate this possibility, we formed a new variable that distinguished between respondents with complete data after the imputation (n = 854) and those with incomplete data (n = 249). Following the work of Mroczek and Spiro (2005), we then added this variable as a predictor in the Level 2 models in order to see if the distinction between remaining in the study and dropping out was associated with interpersonal variation in the intercept or slope parameters of any of the social support trajectories. The results of these analyses showed that the trajectories of participants who had remained in the study did not differ widely from those of the participants who had dropped out. However, there were two exceptions to this trend. As shown in Table 7, elders who had remained in the study had a significantly larger intercept for emotional support provided than those who had dropped out. In addition, satisfaction with support exchanges declined more precipitously for respondents who had remained in the study compared to those who had dropped out.
|
Table 4 presents the results of the explanatory models predicting enacted support. These results showed that, at the average age of the sample, men received less of emotional support (B = –.602, p <.001), tangible support (B = –.775, p <.001), and informational support (B = –.224, p <.05) than did women. In addition, these results showed that, at the average age of the sample, older adults with higher levels of education received less tangible support than did those with less education (B = –.078, p <.001). Two of the multiplicative terms in these models were significant. In particular, these coefficients indicated greater age-related increases in both emotional support received and tangible support received among men compared to women. Taken together, these findings suggest that deficits in received social support experienced by elderly men tend to attenuate with increasing age, at least with respect to emotional and tangible support (see Figure 1).
|
Table 6 presents the findings of models predicting perceived support and negative interaction. These results showed that elderly men, at the average age of the sample, had higher levels of satisfaction with support exchanges than did elderly women (B =.121, p <.05). Results from the explanatory model predicting anticipated support showed a significant multiplicative term indicating that the general age-related decline in anticipated support exchanges observed in the sample as a whole was more severe among more highly educated older adults (B = –.024, p <.05). Additionally, a significant multiplicative term in the model predicting negative interaction indicated that the age-related increase in negative interaction was greater among men than among women (B =.197, p <.05).
Finally, we should highlight one additional result from these explanatory models. In particular, it is important to note that in most of the explanatory models, the random effects associated with both the intercept and slope remained significant. This means that additional interpersonal variation in these parameters—beyond that which can be explained by gender, education, and race—remained unaccounted for in these models.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
), the normative trajectories derived in the current study are particularly informative because they are based upon a much longer period of follow-up than was previously examined. These results show, for example, that with increasing age, older adults tend to report substantially less contact with friends but relatively stable levels of contact with family. This decline in contact with friends, but not family, is consistent with a socioemotional selectivity perspective (Carstensen, 1992), suggesting that as people grow older, they invest increasingly scarce resources in maintaining relationships with more intimate social ties.
Other normative trajectories found in the current study highlight the importance of the interplay between giving and receiving support. Our findings show that older adults provide less emotional, tangible, and informational support as they age. At the same time, they report receiving progressively more tangible and informational support. This combination of increased receiving and decreased giving is consistent with Antonucci's (1985) notion of the support bank. According to this perspective, young and middle-aged adults provide more support than they receive. This imbalance, however, allows them to "bank" credits to be used in later life when, as older adults, they may need more from their social network members than they provide. Thus, equity in social exchanges is achieved when viewed over the entire life course.
One normative trajectory that is somewhat difficult to interpret is that of emotional support received. Whereas levels of tangible and informational support received increase with age, levels of emotional support received are fairly stable. This continuity in levels of emotional support received may initially appear to be at odds with socioemotional selectivity theory (Carstensen, 1992). However, further consideration suggests otherwise. In particular, the observation of stability in levels of emotional support in the face of declining rates of contact with friends, but unwavering levels of contact with family, suggests that individuals may be adapting to old age by relinquishing some of the emotional support provided by friends while garnering more emotional support from a smaller set of family ties.
The second contribution of our study involves the level of interpersonal heterogeneity found in age-related trajectories of social relations. For 10 of the 11 dimensions of social relations examined in the current study, individuals varied significantly with respect to both levels (i.e., intercepts) as well as rates of change (i.e., slopes). These findings are consistent with the aged heterogeneity hypothesis (Dannefer & Sell, 1988). Viewed more generally, these results show that even though normative trajectories of social support and social networks are informative, they may conceal a number of distinct subtrajectories exhibited by different individuals. For example, support may increase for some, decrease for others, and remain the same for yet other subgroups of older adults. Because we examined 11 different dimensions of support in this study, it was not possible to identify individual subtrajectories of change in each dimension. Nevertheless, we laid the groundwork for doing so by establishing the existence of substantial individual variation in the way in which support changes in late life.
The third major contribution of this study arises from our efforts to explore social structural correlates of change in social relations. Previous work in this area (Krause, 1999) focused solely on describing change. The additional waves of data in the current study allowed us to take advantage of more sophisticated multilevel modeling procedures in order to begin accounting for interindividual variation in these patterns of change. These results provide some support for the view of social relations as resources that are subject to unique social stratifications in later life. For example, the current findings show that elderly men generally receive and provide less support than do elderly women, but that men are also more satisfied than are women with their support exchanges. In addition, gender differences in age-related support trajectories emerging from this study help to specify finer nuances of socioemotional selectivity theory. In particular, the data suggest that levels of emotional support increase with age, but only among older men. This may provide evidence that gender roles change in late life such that men become increasingly concerned with interpersonal issues (Gutman, 1987).
The findings from this study also provide some insight into how socioeconomic stratification may influence social ties in late life. More specifically, the data suggest that older adults with less education experience persistent disadvantages over time with regard to levels of social contact with friends and provided support compared to older people with more schooling. Whereas some researchers have found evidence of a late-life convergence of education-based differences in support (House, Lepkowski, & Kinney, 1994), and others have found evidence of enlarging differences (Schieman, 2001), the current findings suggest that for at least some social relationship factors, stratification across education levels simply endures.
However, the data further reveal that socioeconomic differences in social support are not always straightforward. In particular, the findings suggest that levels of received tangible support are higher, and levels of anticipated support decline less precipitously, among individuals with low levels of education compared to their more educated counterparts. Perhaps the increase in tangible support received is a response to greater material needs among less educated individuals, whereas less decline in anticipated support may be a direct consequence of receiving relatively abundant tangible support.
It was somewhat surprising to find that there were no significant associations between race and either the intercepts or slopes of the social relationship trajectories. Previous research also found a lack of association between race and social support slopes (Barnes et al., 2004); however, we did not anticipate the lack of race differences in levels of social support. This lack of race differences may be partially due to racial differences in education. That is, education may have mediated racial differences in social support (Guralnik, Land, Blazer, Fillenbaum, & Branch, 1993). In addition, because non-Whites represented only 8% of the total sample, inadequate statistical power may be partially responsible for the lack of significant race effects. Finally, this lack of race differences could be due to the fact that we compared Whites to all non-Whites, rather than to a single group such as Blacks. We did this because of the relatively small number of cases in specific minority groups. This may be have created problems because research has revealed that some minority group members receive less support than do Whites, whereas other minority group members receive more (e.g., Kaniasty & Norris, 2000).
As readers review these findings, they should keep in mind a number of limitations of this study. First, although the data for this study were collected at four different observation periods over a 10-year period, the current analyses would have benefited from additional waves of data collection. For instance, use of additional waves of data may make it possible to detect patterns of nonlinear change in support over time. Future research should also examine potential social status differences in the curvatures of social relations trajectories.
An additional limitation arises from our use of age, rather than time since baseline, as an indicator of the passage of time in our models. Although the use of age in this way allowed us to examine how social relations changed with increasing age within individuals, the normative trajectories that were produced from these analyses did not fully distinguish between aging and cohort effects. Although we attempted to control for potential cohort effects with relatively coarse measures of birth cohorts, it is likely that the trajectories presented in this study still reflect some combination of aging and cohort effects. We should note, however, that using time since baseline as the time scale would have had its own problems: It would have failed to distinguish between aging and period effects. That is, we could not have been sure whether any changes observed during a particular time period would have been due to the aging process or due to other influences unique to that particular time period. Given the limitations of both approaches, we based our choice of age as the time indicator on our primary substantive interest in examining changes in social relations with increasing age rather than changes over a certain 10-year period. Incidentally, in a supplemental set of (unpublished) analyses, we estimated each of our normative trajectories using time since baseline as the time scale and found only two major differences: The slopes of contact with friends and informational support provided, which had demonstrated significant change (p <.05) in the age-based models, were no longer statistically significant.
Finally, the current analyses do not include any time-varying predictors that might help to explain age-related changes in various measures of social relations. For example, given the positive association between being married and receiving social support (Turner & Marino, 1994), along with the increasing likelihood of becoming widowed in late life, future analyses should include marital status as a time-varying predictor in order to test whether marital transitions mediate the effects of age on social support and interaction.
Despite these limitations, the findings from this study clearly highlight the existence of tremendous diversity in the nature of the social ties that people maintain with others as they grow older. We hope that these findings encourage other investigators to enter the complex, but largely unexplored, study of change in social relationships during late life. Bringing aspects of change to the foreground of research on social support and social networks should lead to an improved ability to represent and understand the dynamic facets of older adults' changing lives.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication January 24, 2006. Accepted for publication August 2, 2006.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B. Cornwell, L. P. Schumm, E. O. Laumann, and J. Graber Social Networks in the NSHAP Study: Rationale, Measurement, and Preliminary Findings J Gerontol B Psychol Sci Soc Sci, November 1, 2009; 64B(suppl_1): i47 - i55. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. A. Ertel, M. M. Glymour, and L. F. Berkman Social networks and health: A life course perspective integrating observational and experimental evidence Journal of Social and Personal Relationships, February 1, 2009; 26(1): 73 - 92. [PDF]
|
|
|
|
 |
|
 B. Cornwell Network Bridging Potential in Later Life: Life-Course Experiences and Social Network Position J Aging Health, February 1, 2009; 21(1): 129 - 154. [Abstract] [PDF]
|
|
|
|
|
|
K. S. Rook Gaps in social support resources in later life: An adaptational challenge in need of further research Journal of Social and Personal Relationships, February 1, 2009; 26(1): 103 - 112. [Abstract] [PDF]
|
|
|
|
 |
|
 N. C. Franklin and C. A. Tate Lifestyle and Successful Aging: An Overview American Journal of Lifestyle Medicine, January 1, 2009; 3(1): 6 - 11. [Abstract] [PDF]
|
|
|
|
 |
|
 R. A. Ward Multiple Parent-Adult Child Relations and Well-Being in Middle and Later Life J. Gerontol. B. Psychol. Sci. Soc. Sci., July 1, 2008; 63(4): S239 - S247. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. F. Mendes de Leon Aging and the Elapse of Time: A Comment on the Analysis of Change J. Gerontol. B. Psychol. Sci. Soc. Sci., May 1, 2007; 62(3): S198 - S202. [Full Text] [PDF]
|
|
|
|
|
|
B. A. Shaw, N. Krause, J. Liang, and J. Bennett Age Versus Time Since Baseline as the Time Scale in the Analysis of Change J. Gerontol. B. Psychol. Sci. Soc. Sci., May 1, 2007; 62(3): S203 - S204. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
