FACE-TO-FACE AND FACEBOOK ACROSS THE LIFESPAN: SOCIAL INTERACTIONS BUFFER STRESS ON MOOD, BUT ONLY FOR ADOLESCENTS

Abstract Social support serves as a protective factor, buffering stress in both adolescents and adults, however Socioemotional Selectivity Theory suggests developmental differences in stress reactivity and social support. It is unclear how modern forms of social contact, such as social media buffer stress, and the extent to which this differs across the lifespan. We utilized ecological momentary data to examine the moderating effects of age and two distinct types of social contacts the person had experienced in prior hours (frequency of face-to-face, or social media contacts) on the association between daily stress and momentary mood. Participants were recruited initially through Amazon.com’s Mechanical Turk (adolescents referred by a parent). A total of 119 adolescent (n = 44; Agemean= 15.73) and middle-aged/older adult participants (n = 75; Agemean= 59.67) provided momentary data three times a day, on three consecutive days, every two weeks, for up to 12 weeks. Multi-level models showed significant 3-way interactions between stress appraisal of avoiding an argument, age group, and frequency of social contact via face-to-face (β = 1.698, se = 0.542, p = .002) and social media (β = 3.341, se = 0.984, p = .001). Older adults experienced better mood than adolescents. When avoiding an argument was appraised as more stressful, both age groups displayed worse mood. Whereas high levels of recent social contact (both face-to-face and social media) seemed to exacerbate the impact of this stressor on poorer mood for older persons, high levels of recent social contact, particularly social media, had stress-buffering benefits for adolescents.

and outreach, NACDA has amassed data and metadata covering a wide array of studies worldwide that address the aging lifecourse. Because our collections are multinational, we share these data at no cost to interested users worldwide

OLDER ADULTS' HEALTH AND SOCIAL ENGAGEMENT: THE MEDIATING ROLE OF SUBJECTIVE LIFE EXPECTANCY
Dahee Kim, 1 and Kyuho Lee 2 , 1. Iowa State University, Ames,Iowa,United States,2. Daegu University,Gyeongsan,Republic of Korea Older adults' mental and physical health is likely to limit social engagement, but their perception of how much time they have left, according to the socio-emotional selectivity theory, might influence it as well. The aim of the research is to investigate the mediating effect of subjective life expectancy (SLE) on the pathways from older adults' mental health and functional limitation to volunteering and contacts with close relationships. The current research used data of 5,285 older adults aged 50 to 75 from the Health and Retirement Study collected in 2014. Structural equation modeling was performed to investigate the direct effect of older adults' depressive symptoms and functional limitation on volunteering and contact with close relationships. Predictors' indirect effects via SLE was also assessed. The results indicated that older adults' higher depressive symptoms and functional limitations significantly decreased volunteering time and frequency of contacts with close relationships. Older adults' SLE attenuated the effects of depressive symptoms and functional limitations on their volunteering time and frequency of contact with close relationships. The findings describe the mechanism of how older adults engage in volunteering and contact with close relationships through their perception of remaining time. Further, this research highlights SLE as a motivator for encouragement of older adults' social engagement.

PHYSICAL AND SOCIAL ACTIVITIES AND COGNITIVE CHANGES AMONG OLD-OLD ADULTS: RESULTS FROM 3-YEAR FOLLOW-UP PERIOD
Yoshiko L. Ishioka, 1 Midori Takayama, 1 and Ikuko Sugawara 2 , 1. Keio University,Yokohama,Kanagawa,Japan,2. University of Tokyo,Bunkyo,Tokyo,Japan The association between activity engagement and latelife cognitive function is considered to depend on the characteristics of the activity, the cognitive processes it involves, and the life stage of participants. A better understanding of this association is required to comprehend cognitive function in old age. The present study examined the association between baseline activity engagement and cognitive changes across a 3-year period among old-old adults. We extracted data for 873 Japanese communitydwelling participants from data of the Keio-Kawasaki Aging Study. We assessed cognitive performance thrice (at baseline, 1.5-year follow-up, and 3-year follow-up) using a short version of the Mini-Mental Status Examination. For the subsequent analyses, we used three measures of cognitive function: total score, orientation, and concentration, which showed diverse individual differences. We measured the frequency of physical activity and social group participation at baseline. Using conditional latent growth curve models, we examined which baseline activity was associated with the three measures of cognitive function over 3 years. Greater physical activity was significantly related to higher rate of orientation, after adjusting for age and education (β = −.261, p < .001). Social activity was significantly related to rates of higher total cognitive score (β = −.276, p < .001) and higher orientation (β = −.207, p < .001). These findings suggest that the association between activity engagement and late-life cognitive function among old-old adults varies by activity type and cognitive domain. Leisure is a major context within which older couples interact, and researchers have recently called for more longitudinal data analysis exploring how leisure-related couple interactions change over the life course. Several waves of the Midlife in the United States (MIDUS) study include a single-item question asking respondents how much they disagree with their spouse or partner about leisure activities. Given the longitudinal nature of MIDUS, the variable offers great utility to explore shifts in leisurerelated couple interactions over the life course. Utilizing longitudinal data from Wave 1 (1995-1997), 2 (2004-2006), and 3 (2013-2015) of the MIDUS study, we explored how leisure-related partner disagreement changed with increased age (age range = 20-93). We first ran an unconditional multilevel model, which revealed that 68% of the variation in leisure-related spousal disagreement was attributed to within-person differences over time, justifying our analysis of longitudinal within-person change. An age-based growth curve model then revealed that leisure-related partner disagreements decreased linearly over the life course (Estimate = -0.01, SE = 0.001, p<.0001). Men reported more leisure-related partner disagreements than women at age 20 (p = 0.002). But men's reported disagreements decreased over the life course at a faster rate than did women's reported disagreements (p = 0.03), so that from ages 70-93, men reported less disagreements than women. To our knowledge, this is the first longitudinal study to explore leisure-related couple disagreements over an extended period of time (20 years). The significance of our results sheds light on the value of longitudinal research on leisure. Social support serves as a protective factor, buffering stress in both adolescents and adults, however Socioemotional Selectivity Theory suggests developmental differences in stress reactivity and social support. It is unclear how modern forms of social contact, such as social media buffer stress, and the extent to which this differs across the lifespan. We utilized ecological momentary data to examine the moderating effects of age and two distinct types of social contacts the person had experienced in prior hours (frequency of face-to-face, or social media contacts) on the association between daily stress and momentary mood. Participants were recruited initially through Amazon.com's Mechanical Turk (adolescents referred by a parent). A total of 119 adolescent (n = 44; Agemean= 15.73) and middle-aged/older adult participants (n = 75; Agemean= 59.67) provided momentary data three times a day, on three consecutive days, every two weeks, for up to 12 weeks. Multi-level models showed significant 3-way interactions between stress appraisal of avoiding an argument, age group, and frequency of social contact via face-to-face (β = 1.698, se = 0.542, p = .002) and social media (β = 3.341, se = 0.984, p = .001). Older adults experienced better mood than adolescents. When avoiding an argument was appraised as more stressful, both age groups displayed worse mood. Whereas high levels of recent social contact (both face-to-face and social media) seemed to exacerbate the impact of this stressor on poorer mood for older persons, high levels of recent social contact, particularly social media, had stress-buffering benefits for adolescents.

Institute for Social Research, University of Michigan, Ann Arbor, Michigan, United States, 2. University of Michigan, Ann Arbor, Michigan, United States
The reminiscence bump is a well-documented autobiographical memory phenomenon characterized by middleaged and older adults reporting a disproportionate number of memories from adolescence and early adulthood (Rubin, Wetzler, & Nebes, 1986). It is typically assessed through either cue word or important memory techniques. The Life History Mail Survey (LHMS) in the Health and Retirement Study affords unique data to investigate this phenomenon among a representative US sample of older adults. At the beginning of the LHMS, participants (N=3088, M age=70, range 50-107) completed a calendar noting the important things that happened to them in seven life decades, starting with ages 0-9 and ending by ages 70-79 (or their actual age). For each life period, we coded the number of events respondents reported. We observed significantly more memories reported for the age decade 20-29, compared with other life periods (80% vs 47-66%). Our results are consistent with previous findings in the autobiographical memory literature. Follow-up analyses evaluated existing theoretical accounts of the bump, such as cultural life script theory which suggests that life events occur in a specific order and are characterized by a prototypical life course. For example, we determined whether respondents' sociodemographic characteristics, such as age cohort, gender, marital and educational histories (information available in LHMS) influenced the size and temporal location of the reminiscence bump. We also analyzed the content of reported important life events to investigate whether types of events included in each decade of life are consistent with the cultural life script account of the phenomenon.

EARLY-LIFE ADVERSITY AND ADULT MENTAL AND BEHAVIORAL HEALTH OUTCOMES: AN INTERSECTIONAL APPROACH
Chioun Lee, 1 Lexi Harari, 1 and Jennifer Coons 2 , 1. University of Riverside,California,United States,2. California State University,Fullerton,California,United States There is much research confirming the link between experiencing early life adversities (ELAs) and adverse health outcomes in adulthood. Further, experiencing co-occurring ELAs rather than one ELA appears to be the rule rather than the exception, and leads to poorer health outcomes. However, information on how ELAs cluster and differentially harm the health of different intersectional groups is lacking. The stress process model suggests that intersectional configurations of race and gender are differentially exposed to ELAs and as a result, some groups suffer worse outcomes than others. We examine the risks of experiencing different clusters of ELAs-low childhood socioeconomic status (SES), family instability, and child abuse-among four intersectional groups (white men, white women, Black men, Black women). We also investigate whether experiencing these ELAs is responsible for the association between having a particular racial and gender configuration and adverse mental and behavioral health outcomes. Data come from a subsample of the Midlife in the U.S. Study (n = 2,076). Black men and women have the highest risk of experiencing all three ELA configurations. Men, regardless of race, have a higher risk of experiencing low SES while women are more vulnerable to low SES/family instability and all three ELAs. Black men and women appear to suffer the worst mental health outcomes, while Black and white men experience more drug/alcohol abuse than their female counterparts. ELAs, especially the co-occurrence of all three ELAs, partially mediate most of these associations, but more so for women than men.

INFLUENCING FACTORS OF SUBJECTIVE AGE: FINDINGS FROM THE KUSATSU LONGITUDINAL STUDY ON AGING AND HEALTH
Tomoko Ikeuchi, 1 Satoshi Seino, 1 Yu Taniguchi, 1 Miki Narita, 1 Takumi Abe, 1 Hidenori Amano, 1 Akihiko Kitamura, 1 and Shoji Shinkai 1 , 1. Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan Background: Subjective age (SA) has been found to be a biopsychosocial marker of aging, yet little is known about factors that influence SA development. This study examined factors influencing SA using longitudinal data of community-dwelling older Japanese. Methods: Data drawn from the Kusatsu Longitudinal Study were collected during annual health check-ups in 2017 and 2018 from participants (aged 65-95) who completed all the measurement items used for this analysis (N=981). SA was indexed by asking participants to specify in years how old they felt. Proportional discrepancy scores ((subjective age -chronological age)/chronological age ×100) were calculated to indicate younger or older SAs and used as a dependent variable. As influencing factors of SA, chronological age, sex, years of schooling, history of smoking, cognitive function (using MMSE scores, range 14-30 at baseline), depressive symptoms, physical function (gait speed), and social function (employment status) were examined. Analyses were