The Gerontologist

Turner, Shelbie, G;Hooker,, Karen;Geldhof, G, John

doi: 10.1093/geront/gnaa059pmid: 32578845

Abstract Background and Objectives Self-perceptions of aging (SPA)—the appraisals people place on their own aging processes—predict well-being in later life. Researchers are increasingly hypothesizing that the overarching construct of SPA is comprised of two factors—positive SPA and negative SPA—and that SPA are gendered. The purpose of this study was to empirically test the hypothesized two-factor structure of SPA and to analyze how the two-factor structure varies between men and women. Research Design and Methods Data come from the 2012 wave of the Health and Retirement Study (N = 7,029; Mage = 68.08), which includes an 8-item SPA scale. We used confirmatory factor analysis (CFA) to assess (i) the positive SPA and negative SPA two-factor solution for the 8-item scale and (ii) whether the two-factor solution had configural, strong, or weak invariance across men and women. Results CFAs indicated a two-factor latent structure of the 8-item scale, with SPA being comprised of both a positive SPA factor and a negative SPA factor. The latent structure was the same for both men and women. Discussion and Implications Results suggest that SPA is a broader construct made up of positive and negative latent factors. Researchers should consider separating the SPA by positive and negative factors to analyze how each factor uniquely shapes health. Moreover, the two-factor solution was equivalent across men and women, possibly because of the generalized nature of the 8-item scale. Researchers can use the 8-item scale similarly for men and women and should continue to elucidate possible gender differences in SPA. Confirmatory factor analyses, Measurement, Views on aging Self-perceptions of aging (SPA) are the internalized appraisals people have of their own aging processes. SPA impact health behavior and health outcomes in later life, with more positive SPA being associated with greater well-being and more negative SPA being associated with worse well-being. SPA is associated with eating behaviors (Klusmann, Sproesser, Wolff, & Renner, 2017), physical activity (Hooker, Mejía, Phibbs, Tan, & Stevens, 2019), overnight hospitalization rates (Sun, Kim, & Smith, 2017), depressive symptoms (Gum & Ayalon, 2018), and even mortality (Levy, Slade, Kunkel, & Kasl, 2002). Negative SPA carry a yearly health care cost of $33.7 billion (Levy, Slade, Chang, Kannoth, & Wang, 2020). With SPA’s saliency to healthy aging being so well established, gerontological scholars have made efforts to study and improve SPA. The Gerontological Society of America’s Reframing Aging Initiative, for example, seeks to counter the types of social-level ageism that shape SPA (The Gerontological Society of America, n.d.). SPA are malleable (Jung, Cham, Siedlecki, & Jopp, 2019), making interventions to improve them appropriate. Indeed, recent interventions have been successful in decreasing implicit old age bias (Busso, Volmert, & Kendall-Taylor, 2019). Research efforts to strengthen the theoretical and methodological understanding of SPA can help inform such interventions and advance the overall knowledge of the SPA and its impact on well-being in later life. For example, researchers are increasingly theorizing that views on aging are both positive and negative (Davidson & Vaillant, 2016), representing both gains and losses (Diehl & Wahl, 2010). As such, any one person could simultaneously have both high positive SPA and high negative SPA, and indeed, recent analyses include SPA as an overarching construct divided into two separate but interconnected positive and negative factors (Hooker et al., 2019; Sun et al., 2017). Confirming such a two-dimensional structure can inform future studies that aim to elucidate the specific mechanisms through which SPA operates to impact health behavior and health outcomes. Additionally, researchers are calling for the study of how components of SPA function differently between men and women (Rothermund & Kornadt, 2015). Gender is salient to older adults’ experiences of their aging (Calasanti & King, 2018), and age is theorized as a gendered construct, with “men and women probably attach[ing] different social meanings to age, [using] different guidelines to measure the progress of their lives, [and being] subject to different expectations and evaluations” (Settersten & Godlewski, 2016, p. 18). Intersections between age and gender leave older women in disadvantaged positions relative to older men, and such social dynamics offer reference points for women to judge their own aging processes (Calasanti, 2010; Calasanti, Slevin, & King, 2006). Indeed, older women have greater anxiety about aging and greater endorsement of age stereotypes than older men (Lytle, Apriceno, Dyar, & Levy, 2018). As such, certain items used to measure SPA may be variably relevant for men and women. Exploring how a possible two-factor positive and negative SPA structure may differ between men and women is important for advancing theoretical and methodological knowledge of how SPA may operate as a function of gender and can offer insights into accurately considering gender within studies of SPA from a methodological standpoint. In this study, we use the Health and Retirement Study’s (HRS’s; Health and Retirement Study, 2012) 8-item SPA measure to empirically test two SPA-related hypotheses: (i) that SPA is a broader construct comprised of two latent factors—positive SPA and negative SPA—and (ii) that item indicators of SPA and their influence on latent SPA constructs vary by gender. Methods Measures Gender As a part of the in-home interview portion of the 2012 wave of the HRS, interviewers asked respondents to confirm their gender of either male or female. Self-Perceptions of Aging As a part of the Leave Behind Questionnaire of the 2012 wave of HRS, participants completed an eight-item SPA scale comprised of the five-item Attitudes Towards One’s Own Aging subscale within the Philadelphia Geriatric Scale (Lawton, 1975) and three separate items from the Berlin Aging Study (Baltes & Mayer, 2001). For each of the eight items, participants responded on a Likert scale from 1 (strongly disagree) to 6 (strongly agree). Table 1 indicates which four items are associated with positive SPA and which four items are associated with negative SPA. Table 1. Item Mean Scores by Gender . Item . Mean score . . . . Men . Women . Positive self-perceptions of aging 2. I have as much pep as I did last year. 3.41 3.78 4. I am as happy now as I was when I was younger. 3.86 3.8 5. As I get older, things are better than I thought they would be. 3.92 3.95 6. So far, I am satisfied with the way that I am aging. 4.38 4.34 Negative self-perceptions of aging 1. Things keep getting worse as I get older. 3.41 3.34 3. The older I get, the more useless I feel. 2.55 2.47 7. The older I get, the more I have had to stop doing things that I liked. 3.7 3.5 8. Getting older has brought with it many things that I do not like. 3.84 3.86 . Item . Mean score . . . . Men . Women . Positive self-perceptions of aging 2. I have as much pep as I did last year. 3.41 3.78 4. I am as happy now as I was when I was younger. 3.86 3.8 5. As I get older, things are better than I thought they would be. 3.92 3.95 6. So far, I am satisfied with the way that I am aging. 4.38 4.34 Negative self-perceptions of aging 1. Things keep getting worse as I get older. 3.41 3.34 3. The older I get, the more useless I feel. 2.55 2.47 7. The older I get, the more I have had to stop doing things that I liked. 3.7 3.5 8. Getting older has brought with it many things that I do not like. 3.84 3.86 Open in new tab Table 1. Item Mean Scores by Gender . Item . Mean score . . . . Men . Women . Positive self-perceptions of aging 2. I have as much pep as I did last year. 3.41 3.78 4. I am as happy now as I was when I was younger. 3.86 3.8 5. As I get older, things are better than I thought they would be. 3.92 3.95 6. So far, I am satisfied with the way that I am aging. 4.38 4.34 Negative self-perceptions of aging 1. Things keep getting worse as I get older. 3.41 3.34 3. The older I get, the more useless I feel. 2.55 2.47 7. The older I get, the more I have had to stop doing things that I liked. 3.7 3.5 8. Getting older has brought with it many things that I do not like. 3.84 3.86 . Item . Mean score . . . . Men . Women . Positive self-perceptions of aging 2. I have as much pep as I did last year. 3.41 3.78 4. I am as happy now as I was when I was younger. 3.86 3.8 5. As I get older, things are better than I thought they would be. 3.92 3.95 6. So far, I am satisfied with the way that I am aging. 4.38 4.34 Negative self-perceptions of aging 1. Things keep getting worse as I get older. 3.41 3.34 3. The older I get, the more useless I feel. 2.55 2.47 7. The older I get, the more I have had to stop doing things that I liked. 3.7 3.5 8. Getting older has brought with it many things that I do not like. 3.84 3.86 Open in new tab Analyses All analyses were completed in MPlus Version 7.11 (Muthén & Muthén, 2012). Confirmatory Factor Analysis We first ran a confirmatory factor analysis (CFA) to assess the two-factor structure of the eight-item scale. We assessed global model fit through a chi-square test of overall fit, the root-mean-square error of approximation (RMSEA; >.08 = poor fit, .05–.08 = acceptable fit, .00–.05 = close fit), the comparative fit index (CFI; <.90 = poor fit, .90–.95 = acceptable fit, >.95 = good fit), and the Tucker–Lewis index (TLI; <.90 = poor fit, .90–.95 = acceptable fit, >.95 = good fit). We assessed localized model fit via modification indices computed by MPlus; if any of the modification indices were extreme, we made adjustments to the model by allowing the two items associated with the extreme modification index to covary within the model. After each adjustment to the model, we completed a likelihood ratio test (LRT) to evaluate whether the newer model had a better fit. Configural, Weak, and Strong Invariance To test configural invariance (same items measure each factor for men and women), we ran the two-factor latent structure CFA in a multiple-group design by gender. Just as with the initial CFA, we assessed global and local model fit via the chi-square test of overall fit, the RMSEA, the CFI, the TLI, and the modification indices. To test for weak invariance (factor loadings are equivalent for men and women), we equated the item factor loadings across men and women and allowed latent variances to be freely estimated in the women, but not the men. To test for strong invariance (item means are equivalent for men and women), we equated the item intercepts across men and women and allowed latent means to be freely estimated in the women, but not the men. We then (i) conducted an LRT to analyze the goodness of fit differences between the models and (ii) compared the change in CFI between the weak model and the configural model, and then the strong model to the weak model, analyzing whether the change was less than .01 (Cheung & Rensvold, 2002), which would indicate weak invariance. Results Demographics The total sample size included 7,029 participants, with 2,885 men and 4,144 women. Participants ranged in age from 51 to 101 years (Magee = 68.08). Nearly 60% were married; 57% had at least a high school degree; and 50% were white, 37.5% were black, and 12.5% were American Indian, Alaskan Native, Asian, or Pacific Islander. Confirmatory Factor Analysis The model with two latent factors of positive SPA and negative SPA had a good fit [χ 2(19) = 841.08, p < .0001; RMSEA = .078, 90% confidence interval (CI) = (.074, .083); CFI = .95; TLI = .93]. However, modification indices (MI) indicated that Item 7 (“The older I get, the more I have had to stop doing things that I liked”) and Item 8 (“Getting older has brought with it many things that I do not like”) were highly correlated and should be allowed to statistically covary. The correlation between Item 7 and Item 8 is logical, as ceasing to do activities one does enjoy would likely be accompanied by thinking older age has brought on aspects of life one does not like (such as, indeed, ceasing enjoyable activities). Allowing Item 7 and Item 8 to covary improved model fit even further [χ 2(18) = 455.11, p < .0001; RMSEA = .059, 90% CI = (.054, .064); CFI = .97; TLI = .96]; the LRT indicated that the change in model fit was significant [Δχ 2(1) = 385.97, p ≤ .0001]. Modification indices indicated, however, that Item 2 (“I have as much pep as I did last year”) should be allowed to statistically covary with Item 7 (“The older I get, the more I have had to stop doing things that I liked”). Indeed, it is likely that losing one’s “pep” is associated with having to cease activities (and, alternatively, that ceasing enjoyable activities could lead to losing one’s “pep”). The final model (Figure 1), which allowed for Item 7 to covary with both Item 2 and Item 8, had a stronger model fit of χ 2(17) = 296.62, p < .0001; RMSEA = .048, 90% CI = (.044, .053); CFI = .98; TLI = .97, and the LRT indicated that the change in model fit was significant [Δχ 2(1) = 158.49, p ≤ .0001]. It was from this final model that we tested invariance across gender. Item means by gender are denoted in Table 1. Figure 1. Open in new tabDownload slide Path diagram depicting latent structure of 8-item Self-Perceptions of Aging (SPA) scale. Model fit: χ 2(17) = 296.62, p < .0001; root mean-square error of approximation = .048, 90% confidence interval = (.044, .053); comparative fit index = .98; Tucker–Lewis index = .97. The latent covariance and all factor loadings, error variances, and factor covariances are significant at p < .0001. Figure 1. Open in new tabDownload slide Path diagram depicting latent structure of 8-item Self-Perceptions of Aging (SPA) scale. Model fit: χ 2(17) = 296.62, p < .0001; root mean-square error of approximation = .048, 90% confidence interval = (.044, .053); comparative fit index = .98; Tucker–Lewis index = .97. The latent covariance and all factor loadings, error variances, and factor covariances are significant at p < .0001. Invariance Across Gender Configural The CFA comparing the two-factor model between men and women had good global model fit (χ 2(34) = 332.1, p < .0001; RMSEA = .05, 90% CI = [.045, .055); CFI = .982; TLI = .971] and local fit (no extreme modification indices; Figure 2). Figure 2. Open in new tabDownload slide Path diagrams depicting configural invariance of latent structure of 8-item Self-Perceptions of Aging (SPA) scale by gender. Values are standardized estimates. Model fit: χ 2(34) = 332.1, p < .0001; root mean-square error of approximation = .05, 90% confidence interval = (.045, .055); comparative fit index = .982; Tucker–Lewis index = .971. The latent covariance and all factor loadings, error variances, and factor covariances are significant at p < .0001. Figure 2. Open in new tabDownload slide Path diagrams depicting configural invariance of latent structure of 8-item Self-Perceptions of Aging (SPA) scale by gender. Values are standardized estimates. Model fit: χ 2(34) = 332.1, p < .0001; root mean-square error of approximation = .05, 90% confidence interval = (.045, .055); comparative fit index = .982; Tucker–Lewis index = .971. The latent covariance and all factor loadings, error variances, and factor covariances are significant at p < .0001. Weak The model testing weak invariance—with equal factor loadings and free latent variances in the nonreference group—also maintained good global fit [χ 2(40) = 346.88, p < .0001; RMSEA = .047, 90% CI = (.042, .051); CFI = .982; TLI = .974] and local fit (no extreme modification indices). The LRT indicated that the weak model had a significantly worse fit than the configural model [Δχ 2(6) = 14.75, p = .02]. However, comparing the weak model to the configural model, there was no change in CFI, indicating weak invariance (Cheung & Rensvold, 2002; Supplementary Figure 1). Strong The model testing strong invariance—with equal item intercepts and free latent variances in the nonreference group—also maintained good global [χ 2(46) = 391.26, p < .0001; RMSEA = .046, 90% CI = (.042, .05); CFI = .979; TLI = .975] and local fit (no extreme modification indices). The LRT indicated that the strong model was significantly worse than the weak model [Δχ 2(6) = 44.38, p < .0001]. But, the change in CFI between the two models was .003, indicating strong invariance (Cheung & Rensvold, 2002; Supplementary Figure 2). After establishing invariance, we tested latent mean differences between men and women. Compared with men, women had significantly lower negative SPA (p = .0008), but no different positive SPA (p = .66). Discussion Our results offer empirical support for the hypothesized theoretical notion that SPA, as a broad construct, is comprised of positive and negative latent factors. It is possible that the separate factors uniquely shape health behavior and health outcomes. Indeed, Hooker and colleagues (2019) analyzed how positive SPA and negative SPA independently shaped health behaviors and found differential relationships; for example, only negative SPA predicted smoking, and only positive SPA predicted physical activity. Future research should separate SPA by positive and negative factors in order to expand the understanding of how positive SPA and negative SPA operate both independently and in concert to shape well-being in later life. Additionally, each of the eight items in the scale was equally important for men and women, and men and women scored similarly on each item. It is possible that theorized gender differences in SPA are more relevant to domain-specific elements of SPA; interestingly, women scored lower than men on the negative SPA factor. Given gendered social norms and expectations, SPA related to physical appearance, financial well-being, and family relationships may be uniquely salient to each gender. Newer measures of SPA are capturing these various domains (e.g., Kornadt, Hess, & Rothermund, 2020), and further validation of such scales should include testing of invariance across gender. Our results suggest, however, that the eight-item SPA measure is a valid instrument to be used with both men and women to understand SPA at a broad, general level. In conclusion, our study both advances the theoretical understanding of SPA and offers methodological validation for an eight-item SPA scale. Our results support the hypothesis that the SPA is a broader construct that can be divided into two independent factors of positive SPA and negative SPA. Researchers administering and/or analyzing data from the eight-item SPA scale should feel confident in (i) dividing the scale into two subconstructs and (ii) using the scale similarly between men and women. Going forward, understanding positive and negative SPA as two different but related constructs should allow for a more specified awareness of how perceptions of aging shape behaviors and health, and can inform efforts to improve perceptions of aging, allowing interventions to be more relevant and efficient, ultimately improving the process of becoming—and being—an older adult. Funding None reported. Conflict of Interest None reported. References Baltes , P. B. , & Mayer , K. U. (Eds.). ( 2001 ). The Berlin aging study: Aging from 70 to 100 . Cambridge, UK: Cambridge University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Busso , D. S. , Volmert , A., & Kendall-Taylor , N. ( 2019 ). Reframing aging: Effect of a short-term framing intervention on implicit measures of age bias . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 74 ( 4 ), 559 – 564 . doi:10.1093/geronb/gby080 Google Scholar Crossref Search ADS WorldCat Calasanti , T . ( 2010 ). Gender relations and applied research on aging . The Gerontologist , 50 ( 6 ), 720 – 734 . doi:10.1093/geront/gnq085 Google Scholar Crossref Search ADS PubMed WorldCat Calasanti , T. , & King , N. ( 2018 ). The dynamic nature of gender and aging bodies . Journal of Aging Studies , 45 , 11 – 17 . doi:10.1016/j.jaging.2018.01.002 Google Scholar Crossref Search ADS PubMed WorldCat Calasanti , T. , Slevin , K. F., & King , N. ( 2006 ). Ageism and feminism: From “et cetera” to center . NWSA Journal , 18 , 13 – 30 . doi:10.1353/nwsa.2006.0004 Google Scholar Crossref Search ADS WorldCat Cheung , G. W. , & Rensvold , R. B. ( 2002 ). Evaluating goodness-of-fit indexes for testing measurement invariance . Structural Equation Modeling , 9 ( 2 ), 233 – 255 . doi:10.1207/S15328007SEM0902_5 Google Scholar Crossref Search ADS WorldCat Davidson , A. , & Vaillant , G. ( 2016 ). Integrating “positive” and “negative” perspectives on aging . In A. M. Wood & J. Johnson (Eds.), The Wiley handbook of positive clinical psychology (pp. 167 – 179 ). Chichester, UK : John Wiley & Sons . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Diehl , M. K. , & Wahl , H. W. ( 2010 ). Awareness of age-related change: Examination of a (mostly) unexplored concept . Journals of Gerontology Series B: Psychological Sciences and Social Sciences , 65 ( 3 ), 340 – 350 . doi:10.1093/geronb/gbp110 Google Scholar Crossref Search ADS WorldCat Gum , A. M. , & Ayalon , L. ( 2018 ). Self-perceptions of aging mediate the longitudinal relationship of hopelessness and depressive symptoms . International Journal of Geriatric Psychiatry , 33 ( 4 ), 591 – 597 . doi:10.1002/gps.4826 Google Scholar Crossref Search ADS PubMed WorldCat Health and Retirement Study . ( 2012 ). (2012 Biannual Survey) public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740) . Ann Arbor, MI . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hooker , K. , Mejía , S. T., Phibbs , S., Tan , E. J., & Stevens , J. ( 2019 ). Effects of age discrimination on self-perceptions of aging and cancer risk behaviors . The Gerontologist , 59 ( Suppl. 1 ), S28 – S37 . doi:10.1093/geront/gny183 Google Scholar Crossref Search ADS PubMed WorldCat Jung , S. , Cham , H., Siedlecki , K. L., & Jopp , D. S. ( 2019 ). Measurement invariance and developmental trajectories of multidimensional self-perceptions of aging in middle-aged and older adults . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , gbz099:1–13. doi:10.1093/geronb/gbz099 OpenURL Placeholder Text WorldCat Klusmann , V. , Sproesser , G., Wolff , J. K., & Renner , B. ( 2017 ). Positive self-perceptions of aging promote healthy eating behavior across the life span via social-cognitive processes . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 74 ( 5 ), 735 – 744 . doi:10.1093/geronb/gbx139 Google Scholar Crossref Search ADS WorldCat Kornadt , A. E. , Hess , T. M., & Rothermund , K. ( 2020 ). Domain-specific views on aging and preparation for age-related changes—Development and validation of three brief scales . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 75 ( 2), 303 – 307 . doi:10.1093/geronb/gby055 OpenURL Placeholder Text WorldCat Lawton , M. P . ( 1975 ). The Philadelphia Geriatric Center Morale Scale: A revision . Journal of Gerontology , 30 ( 1 ), 85 – 89 . doi:10.1093/geronj/30.1.85 Google Scholar Crossref Search ADS PubMed WorldCat Levy , B. R. , Slade , M. D., Chang , E. S., Kannoth , S., & Wang , S. Y. ( 2020 ). Ageism amplifies cost and prevalence of health conditions . The Gerontologist , 60 ( 1 ), 174 – 181 . doi:10.1093/geront/gny131 Google Scholar Crossref Search ADS PubMed WorldCat Levy , B. R. , Slade , M. D., Kunkel , S. R., & Kasl , S. V. ( 2002 ). Longevity increased by positive self-perceptions of aging . Journal of Personality and Social Psychology , 83 ( 2 ), 261 – 270 . doi:10.1037//0022-3514.83.2.261 Google Scholar Crossref Search ADS PubMed WorldCat Lytle , A. , Apriceno , M., Dyar , C., & Levy , S. R. ( 2018 ). Sexual orientation and gender differences in aging perceptions and concerns among older adults . Innovation in Aging , 2 ( 3 ), igy036 . doi:10.1093/geroni/igy036 Google Scholar Crossref Search ADS PubMed WorldCat Muthén , L. K. , & Muthén , B. O. (1998– 2012 ). Mplus user’s guide (7th ed.). Los Angeles, CA : Author . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Rothermund , K. , & Kornadt , A. E. ( 2015 ). Views on aging: Domain-specific approaches and implications for developmental regulation . Annual Review of Gerontology and Geriatrics , 35 ( 1 ), 121 – 144 . doi:10.1891/0198-8794.35.121 Google Scholar Crossref Search ADS WorldCat Settersten , R. A. , & Godlewski , B. ( 2016 ). Concepts and theories of age and aging . In V. L. Bengtson & R. A. Settersten (Eds.), Handbook of theories of Aging (pp. 9 – 25 ). New York : Springer . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Sun , J. K. , Kim , E. S., & Smith , J. ( 2017 ). Positive self-perceptions of aging and lower rate of overnight hospitalization in the US population over age 50 . Psychosomatic Medicine , 79 ( 1 ), 81 – 90 . doi:10.1097/PSY.0000000000000364 Google Scholar Crossref Search ADS PubMed WorldCat The Gerontological Society of America . ( n.d .). Reframing aging initiative. Retrieved from https://www.geron.org/programs-services/reframing-aging-initiative © The Author(s) 2020. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)