Abstract Objectives We developed brief versions of our questionnaires to assess domain-specific views on aging (age stereotypes and future self-views) and preparation for age-related changes. Methods The brief scales were validated in an online study with N = 301 participants aged 23–88 years. Results Mean values across domains show a differentiated picture for all 3 constructs, yielding evidence for the multidimensionality of views on aging and preparation for age-related changes. Rating profiles for the brief versions were similar to the long versions of the questionnaires, attesting to the equivalence of the brief and long scales. Within-domain correlations between the 3 constructs were also higher than between-domain correlations, further substantiating the claim of domain-specificity with regard to the predictive validity of the brief scales. Discussion The new brief versions of the scales can be recommended for a differentiated assessment of views on aging and preparation for age-related changes when short forms of measurement are required. Age stereotypes, Brief scales, Future self-views, Life domains, Preparation Views on aging are central psychosocial variables in the aging process, predicting health and well-being in later life (Levy, 2009; Westerhof et al., 2014; Wurm, Diehl, Kornadt, Westerhof, & Wahl, 2017). Recently, the argument has been put forward that instead of a unidimensional continuum ranging from positive to negative, views on aging and their behavioral manifestations, such as preparation for age-related changes, are better conceptualized in a domain-specific way, accounting for the inherent multidimensionality of life-span development due to the specific affordances and constraints that characterize different life domains (Diehl et al., 2014; Kornadt & Rothermund, 2015). To account for this multidimensionality, we developed scales that allow us to assess domain-specific age stereotypes (AS), future selves (FS), and age-related preparation (PREP) (Kornadt & Rothermund, 2011, 2012, 2014). Those scales subsume different domains and facets of the respective construct and they have been frequently put to test: Their domain-specificity proved to have incremental validity in explaining, for instance, the timing of internalization and projection effects (Kornadt, Voss, & Rothermund, 2017), heterogeneity in country differences regarding views on aging (Voss, Kornadt, Hess, Fung, & Rothermund, 2018), or differential determinants of preparation for the third and fourth age (Kornadt, Voss, & Rothermund, 2018). The goal of the present study was to develop and validate brief versions of these questionnaires, allowing researchers to capture the multidimensionality of views on aging and their behavioral manifestations when assessment space is limited. Method Sample and Procedure Amazon’s Mechanical Turk (MTurk) was used to recruit a sample of N = 301 adults aged 23–88. To ensure comparability with the North Carolina sample (see below), the Mturk sample was stratified by age group, gender, and nationality via the TurkPrime recruitment tool. Sample size was calculated to guarantee sufficient power (1 − β > .95) to detect medium size effects (f2 = .25). Participants first completed a brief demographic questionnaire, followed by the assessment of domain-specific AS, FS, life satisfaction, and PREP. To avoid missing values, the program required participants to answer all questions to complete the survey. Participants were compensated with $1.50. The sample was balanced with regard to age and gender; for further demographic information, see Supplementary Table 1. To compare our brief scales with full versions of the questionnaires, we used a sample of N = 572 participants aged 26–89 from the Ageing as Future project that was drawn in Wake County, NC (for more information, see Hess et al., 2017, and O’Brien et al., 2017). The MTurk and Wake County samples were comparable regarding sociodemographic characteristics (Supplementary Table 1). Questionnaire Construction Views on aging The original scales measuring AS and FS consisted of 27 items each, covering eight life domains (3–5 items per domain). Our goal was to select one item per domain that would best capture the meaning of the scale and thus reduce the length of the original scales while retaining their domain-specificity. We removed the domain “Religion and Spirituality” because this domain is the only one without clear evaluative content (Kornadt & Rothermund, 2011). Furthermore, the domain “Fitness, Health, and Appearance” was split up into three separate domains, which has proven useful in previous analyses (Kornadt, Voss, & Rothermund, 2015). The resulting domains for the new instruments were “Family,” “Friendships,” “Leisure,” “Personality,” “Finances,” “Work,” “Appearance,” “Fitness,” and “Health.” For each domain, we selected the marker item (i.e., the item with the highest factor loading per scale) from our previous factor analysis (Kornadt & Rothermund, 2011). For domains where two or three items had loadings that were similarly high (Leisure, Personality, Finances), we created a representative new single item by merging the content of these items. As in the original version, each item was preceded by a one-sentence description of the domain, and participants had to rate “Old people” (AS) and “When I am older” (FS) on an 8-point scale in-between two opposite poles, with higher values indicating more favorable evaluations. The final version of the scales can be found in the Supplementary Material. Preparation for age-related changes The original scale consisted of three items in each of nine life domains (Kornadt & Rothermund, 2014). The first item in each domain was concerned with active preparation and illustrated the respective domain content with some examples. The second item was concerned with thoughts and preoccupation with the topic, and the third with gathering and exchanging information. The latter two items were worded exactly the same way for all domains. Items had to be rated on a 4-point scale (not at all—a little—quite a bit—a lot). To achieve our goal of reducing these scales to one item per domain, we slightly changed the wording of the first item for each domain while retaining the respective examples. We then included the content of the other two items (preoccupation, gathering information, discussing it with others) to the questionnaire instructions (Supplementary Material). To ensure better compatibility with the views on the aging questionnaire, we also added two items to assess PREP in the domains “Family” and “Personality.” Therefore, we ended up with a final version of 11 items and domains of PREP—“Financial Situation,” “Emergency Situations,” “Family,” “Fitness,” “Housing,” “Appearance,” “Social Relations,” “Health,” “Leisure,” “Work,” and “Personality.” All items are presented in Supplementary Tables 2–4. Analyses Following our procedure for the long version of the questionnaires (Kornadt & Rothermund, 2011), we first ran three separate mixed-model repeated measurement analyses of variance (ANOVA) in SPSS for each of the three constructs, with participant age group as between-person and domain as within-person factors. Thus, we assessed whether participants rated old persons, themselves in old age, and their preparatory behavior differently depending on life domain. We then computed separate profile correlations for each construct (AS, FS, PREP) to determine the correspondence between domain-specific ratings in the brief and long forms of the questionnaire. A profile correlation is an index of correspondence, with a substantial positive profile correlation indicating that the pattern of means across the different life domains is highly similar for the brief and long scales. To test the general predictive validity of our scales, that is, whether the different constructs (AS, FS, PREP) are related on a cross-construct level, we computed correlations between the scales. To further substantiate the domain-specificity of the scales, we tested whether these predictive relations are also domain-specific in nature. This would be the case if the correlation between AS–FS, AS–PREP, and FS–PREP would be higher for pairs of items referring to the same domain, compared with pairs of items from different domains (Levy & Leifheit-Limson, 2009). We thus estimated the pattern of bivariate correlations between the three constructs for all possible pairwise combinations of domains in Mplus 7 (Muthen & Muthen, 1998–2012). The combinations could either be within the same domain (i.e., match in content [e.g., rAS-family, FS-family]) or between different domains (i.e., did not match in content [e.g., rAS-family, FS-health]). We then estimated and compared two models: One in which correlation coefficients for matching pairs of domains could differ from those of nonmatching domains, and one in which correlations for matching and nonmatching domains were set equal. A worse model fit for the latter model would support the domain-specificity of relations (Voss, Kornadt, & Rothermund, 2017). Results Mean Values For the AS scales, the ANOVA yielded significant effects for domain F(8, 291) = 58.65, p ≤ .001, ηp2 = .62, age group F(2, 298) = 11.94, p ≤ .001, ηp2 = .07, and their interaction F(16, 584) = 5.02, p ≤ .001, ηp2 = .12. For all three constructs, the domain effect was most substantial; we will thus focus on this main effect; results for the interaction effects are presented in Supplementary Figure 1. Mean values for the nine domains are presented in Figure 1 (dark gray bars). AS were most positive in the domains of Personality, Leisure, and Work, and least positive in the domains of Health, Appearance, and Finances. The ANOVA for FS also yielded significant effects for domain F(8, 291) = 40.36, p ≤ .001, ηp2 = .53, and the interaction of age group and domain F(16, 584) = 2.98, p ≤ .001, ηp2 = .08, but there was no main effect of age group, F(2, 298) = 2.48, p = .086. Ratings were most positive in the Family, Leisure, Personality, Work, and Fitness domains, and less positive in the Friends and Finances domain. For PREP, we again found significant effects for domain F(10, 289) = 35.52, p ≤ .001, ηp2 = .55, age group F(2, 298) = 4.18, p = .016, ηp2 = .03, and their interaction F(20, 580) = 2.09, p ≤ .001, ηp2 = .09. Participants reported most PREP in the Family, Fitness, Health, and Personality domains, whereas self-reported PREP was comparatively low in the domains Emergency and Work. The considerable variation between domains for all constructs supports the assumption of domain-specificity of ratings. Figure 1. View largeDownload slide Domain-specific mean values and standard errors for the three constructs for the brief (N = 301) and the long version (N = 572) of the questionnaires. Figure 1. View largeDownload slide Domain-specific mean values and standard errors for the three constructs for the brief (N = 301) and the long version (N = 572) of the questionnaires. Profile Similarity of the Brief and Long Versions Domain-specific mean values for the long versions of the scales that were computed in the comparison sample are displayed in Figure 1 (light gray bars). Profile correlations between the patterns of mean values for the brief scales and the long version were high (Figure 1), with r = .66 for AS, r = .68 for FS and r = .66 for PREP, indicating that the pattern of differences in average ratings for the domains were highly similar for the long and the short versions of the questionnaire. Predictive Validity for Matching and Nonmatching Domains Table 1 displays the mean pairwise correlations between constructs for domains that matched in content (i.e., correlations within the same domain) and those where constructs from different domains were correlated, as well as the results of the χ2-difference tests. When two constructs in the same domain (i.e., AS-family and FS-family) were correlated, these correlations were on average higher than respective nonmatching correlations (i.e., AS-family and FS-health). Setting the correlations for matching and nonmatching domains equal resulted in significant decreases in model fit. This shows that the predictive validity (pairwise correlations between AS-FS, AS-PREP, and FS-PREP) within a domain was higher than between domains, supporting the domain-specificity of the brief scales. Table 1. Average Correlations Between the Constructs for Matching (e.g., Family–Family) and Nonmatching (e.g., Family–Health) Pairs of Domains and Results of χ2-Difference Tests Construct Combinations AS–FS AS–PREP FS–PREP Mean r match .52 .28 .45 Mean r nonmatch .39 .24 .35 χ2-difference (df) 43.24 (1) 11.24 (1) 31.50 (1) P <.00001 .0008 <.00001 Construct Combinations AS–FS AS–PREP FS–PREP Mean r match .52 .28 .45 Mean r nonmatch .39 .24 .35 χ2-difference (df) 43.24 (1) 11.24 (1) 31.50 (1) P <.00001 .0008 <.00001 Note. χ2-difference is Sartorra–Bentler corrected. AS = age stereotypes, FS = future selves, PREP = preparation for age-related changes. View Large Table 1. Average Correlations Between the Constructs for Matching (e.g., Family–Family) and Nonmatching (e.g., Family–Health) Pairs of Domains and Results of χ2-Difference Tests Construct Combinations AS–FS AS–PREP FS–PREP Mean r match .52 .28 .45 Mean r nonmatch .39 .24 .35 χ2-difference (df) 43.24 (1) 11.24 (1) 31.50 (1) P <.00001 .0008 <.00001 Construct Combinations AS–FS AS–PREP FS–PREP Mean r match .52 .28 .45 Mean r nonmatch .39 .24 .35 χ2-difference (df) 43.24 (1) 11.24 (1) 31.50 (1) P <.00001 .0008 <.00001 Note. χ2-difference is Sartorra–Bentler corrected. AS = age stereotypes, FS = future selves, PREP = preparation for age-related changes. View Large Discussion The goal of our study was to derive and validate brief scales from our established, multidimensional measures of AS, FS, and PREP that retain their multidimensional properties while at the same time facilitating a parsimonious assessment. We tested the brief scales in an online sample of adults covering a broad age range and found evidence for the multidimensionality of the scales in differential patterns of mean values across domains. We further found theory-conform relations between the constructs, attesting to their general predictive validity. Importantly, the size of the correlations was significantly larger for matching than for nonmatching domains. Thus, our brief scales revealed distinct patterns of evaluations across domains for both older adults (AS) and the self as an old person (FS). Furthermore, levels of PREP for old age also differed depending on life domain, adding further substance to the claim that preparing for old age is dependent on context (Hershey, Brown, Jacobs-Lawson, & Jackson, 2001; Kornadt & Rothermund, 2014). Comparing the ratings of the brief scale with those from the full version of the questionnaire, we found a high correspondence between the domain-specific rating profiles of both versions. The profile correlations, however, were not perfect, possibly due to the fact that the long version of the instruments contained additional items that had different means (Supplementary Table 5). Item selection for the brief scales was based on which item was most representative for the scale in terms of shared variance with the domain-factor, but this does not necessarily imply similarity with regard to the means. Thus, our findings still attest to the equivalence of the brief and long scales, suggesting that participants applied similar criteria for evaluating older people, themselves, and their levels of preparation for specific domains in the long and in the brief version of the questionnaire. Future studies will have to investigate whether the added value of a domain-specific assessment in explaining heterogeneous results and life-span processes of influence that was found for the long scales can also be observed for the brief version of the questionnaires. Due to the single-item basis of our multidimensional scales, we did not calculate classic indicators of scale functioning (e.g., factor analyses, Cronbach’s alpha). Instead, we applied alternative methods to validate our scale that are in line with the goals of our study. They show that our newly developed scales enable researchers to differentially assess views on aging and preparation in a parsimonious format. The long version of our questionnaires should be the preferred choice when a differentiated, highly reliable assessment is required and when latent variable modeling for the domain-structure is the method of choice. However, when assessment space is limited, such as in large-scale surveys and studies with populations where short measures are indicated, the brief version makes it easier to incorporate a domain-specific assessment of views on aging and their behavioral manifestations, going beyond a unidimensional and simplistic view of lifespan development (Diehl et al., 2014). Supplementary Material Supplementary data are available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online. Funding This work was supported by grants of the VolkswagenStiftung (Az. II / 83 142, Az. 86 758, and Az. 93 272 to K. Rothermund). Conflict of Interest The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article. References Diehl, M., Wahl, H. W., Barrett, A. E., Brothers, A. F., Miche, M., Montepare, J. M.,… Wurm, S. ( 2014). Awareness of aging: Theoretical considerations on an emerging concept. Developmental Review , 34, 93– 113. doi: 10.1016/j.dr.2014.01.001 Google Scholar CrossRef Search ADS PubMed Hershey, D. A., Brown, C. E., Jacobs-Lawson, J. M., & Jackson, J. ( 2001). Retirees’ perceptions of important retirement planning decisions. The Southwest Journal on Aging , 16, 91– 100. Hess, T. M., O’Brien, E. L., Voss, P., Kornadt, A. E., Rothermund, K., Fung, H. H., & Popham, L. E. ( 2017). Context influences on the relationship between views of aging and subjective age: The moderating role of culture and domain of functioning. Psychology and Aging , 32, 419– 431. doi: 10.1037/pag0000181 Google Scholar CrossRef Search ADS PubMed Kornadt, A. E., & Rothermund, K. ( 2011). Contexts of aging: Assessing evaluative age stereotypes in different life domains. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 66, 547– 556. doi: 10.1093/geronb/gbr036 Google Scholar CrossRef Search ADS Kornadt, A. E., & Rothermund, K. ( 2012). Internalization of age stereotypes into the self-concept via future self-views: A general model and domain-specific differences. Psychology and Aging , 27, 164– 172. doi: 10.1037/a0025110 Google Scholar CrossRef Search ADS PubMed Kornadt, A. E., & Rothermund, K. ( 2014). Preparation for old age in different life domains: Dimensions and age differences. International Journal of Behavioral Development , 38, 228– 238. doi: 10.1177/0165025413512065 Google Scholar CrossRef Search ADS Kornadt, A. E., & Rothermund, K. ( 2015). Views on aging: Domain-specific approaches and implications for developmental regulation. Annual Review of Gerontology and Geriatrics , 35, 121– 144. doi: 10.1891/0198-8794.35.121 Google Scholar CrossRef Search ADS Kornadt, A. E., Voss, P., & Rothermund, K. ( 2015). Hope for the best, prepare for the worst? Future self-views and preparation for age-related changes. Psychology and Aging , 30, 967– 976. doi: 10.1037/pag0000048 Google Scholar CrossRef Search ADS PubMed Kornadt, A. E., Voss, P., & Rothermund, K. ( 2017). Age stereotypes and self-views revisited: Patterns of internalization and projection processes across the life span. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 72, 582– 592. doi: 10.1093/geronb/gbv099 Kornadt, A. E., Voss, P., & Rothermund, K. ( 2018). Subjective remaining lifetime and concreteness of the future as differential predictors of preparation for age-related changes. European Journal of Ageing , 15, 67– 76. doi: 10.1007/s10433-017-0426-3 Google Scholar CrossRef Search ADS PubMed Levy, B. ( 2009). Stereotype embodiment: A psychosocial approach to aging. Current Directions in Psychological Science , 18, 332– 336. doi: 10.1111/j.1467-8721.2009.01662.x Google Scholar CrossRef Search ADS PubMed Levy, B. R., & Leifheit-Limson, E. ( 2009). The stereotype-matching effect: Greater influence on functioning when age stereotypes correspond to outcomes. Psychology and Aging , 24, 230– 233. doi: 10.1037/a0014563 Google Scholar CrossRef Search ADS PubMed Muthén, L. K., & Muthén, B. O. ( 1998–2012). Mplus user’s guide ( 7th ed.). Los Angeles, CA: Muthén & Muthén. O’Brien, E. A., Hess, T. M., Kornadt, A. E., Rothermund, K., Fung, H. H. L., & Voss, P. ( 2017). Context influences on the subjective experience of aging: The impact of culture and domains of functioning. The Gerontologist , 57, S127– S137. doi: 10.1093/geront/gnx015 Google Scholar CrossRef Search ADS PubMed Voss, P., Kornadt, A. E., & Rothermund, K. ( 2017). Getting what you expect? Future self-views predict the valence of life events. Developmental Psychology , 53, 567– 580. doi: 10.1037/dev0000285 Google Scholar CrossRef Search ADS PubMed Voss, P., Kornadt, A. E., Hess, T. M., Fung, H. H., & Rothermund, K. ( 2018). A world of difference? Domain-specific views on aging in China, the US, and Germany. Psychology and Aging . Advance online publication. doi:10.1037/pag0000237 Westerhof, G. J., Miche, M., Brothers, A. F., Barrett, A. E., Diehl, M., Montepare, J. M.,… Wurm, S. ( 2014). The influence of subjective aging on health and longevity: A meta-analysis of longitudinal data. Psychology and Aging , 29, 793– 802. doi: 10.1037/a0038016 Google Scholar CrossRef Search ADS PubMed Wurm, S., Diehl, M., Kornadt, A. E., Westerhof, G. J., & Wahl, H-W. ( 2017). How do views on aging affect health outcomes in adulthood and late life? Explanations for an established connection. Developmental Review , 46, 27– 43. doi: 10.1016/j.dr.2017.08.002 Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences – Oxford University Press
Published: May 8, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera