Whether planned or unplanned, retirement has major implications for identity and social connectedness. This is because it changes valued relationships that not only inform our self-definition (e.g., as a professional, a grandparent) but also affect the support structures that we invoke to combat stressors, including those associated with retirement. What we have yet to identify are the particular social group processes that predict different adjustment trajectories in this context, and their generalizability to countries where the retirement experience differs. In China, for example, people retire at a younger age than they do in Australia and typically move in with family, changing the nature of social dynamics and potentially the influence of social group factors in adjustment. In the US, pension security is more volatile than in Australia, potentially raising the salience of financial, over social, management at the point of retirement, which may reduce the importance of social factors. This international focus allows us to interrogate the relevance of SIMIC in these different retirement contexts. Accordingly, in Stream 1 we propose two studies that will examine SIMIC-derived hypotheses in retirees from four countries — the United Kingdom, United States, Australia, and China.
Study 1: Social determinants of adjustment to retirement in Australian and British seniors
Datasets and analysis
This study will draw on nationally-representative longitudinal datasets from Australia (Household, Income and Labour Dynamics in Australia; HILDA) and the United Kingdom (English Longitudinal Study of Aging; ELSA) to address two specific questions. HILDA has data from over 13,000 people and will be used to assess the impact of retirement relative to other life transitions on adjustment indexed primarily through mood and well-being. ELSA has data from over 10,000 people and will be used to assess the impact of change in group membership in the retirement transition on adjustment, indexed by cognitive integrity and well-being. Both datasets have information on retirement planning and experiences (HILDA in Waves 3, 7 and 11; ELSA in all waves), social participation (i.e., satisfaction with relationships, community participation, social cohesion), well-being, and cognitive integrity (primarily ELSA).
Multi-level modelling and regression analyses will examine the contribution of social participation to the adjustment outcomes of retirees, cross-sectionally and longitudinally, controlling for a range of factors (i.e.age, gender, class and financial status, perceived physical health) to rule out alternative explanations. Population weighting and multiple imputation will be used where appropriate to reduce the effects of attrition and missing data. We are familiar with these datasets and methods, having used them in published work addressing different research questions in the past (e.g., Cruwys et al., 2013a; 2013b; Haslam et al., 2014a).
Study 2: Social determinants of adjustment to retirement in Australian, Chinese and US seniors
Use of the HILDA and ELSA datasets is limited primarily by the absence of measures specifically indexing social identification and social identity content. This means that they cannot capture fully the nuances of identity change as it relates to our hypotheses. Study 2 addresses this limitation through a longitudinal survey of retirees from Australia, the United States, and China.
We will recruit at least 300 people from each country who are in the process of transitioning to retirement. A sample size of 200 is required to conduct multi-level analyses (with time points nested within persons) with sufficient statistical power to detect small- and medium-sized effects (Scherbaum & Ferreter, 2009). The additional numbers take into consideration attrition rates reported in previous research (Cook et al., 2000). Participants will be surveyed three times: within 6 months of retirement — (Time 1, T1), and then at two further time points at 6 monthly intervals (T2 and T3). Our recruitment strategy will vary across sites and be based on the most feasible method to meet study objectives.
- In Australia participants will be recruited with help from the Online Research Unit (ORU); Australia’s leading panel management and online research company (http://www.theoru.com) that CI Steffens as used in previous research. The ORU owns and manages the largest research-only consumer and business panels in Australia, operates in accordance with the global International Standardization Organization (ISO) standards for market, opinion, and social panel research, and recruits participants offline and on an invitation-only basis to avoid participation of professional online survey respondents.
- In the Unites States, participants will be recruited using the same strategy, but with a US company, Metrixlab (http://www.metrixlab.com/data/), which PI Branscombe has experience accessing.
- In China participants will be recruited from two sources: regional labour unions and the China National Committee on Ageing at provincial and municipal levels (with cities sampled including Beijing, Hefei, Nanchang, Guangzhou and Shenzhen). Both of these units are responsible for the affairs of retiree management. Our named collaborator Prof. Yang has secured a University grant to support research collaboration with the CIs, and this will enable regular travel to China during the project’s duration. Prof. Yang has an established partnership with these units and will manage the data collection from these sites.
To be included, participants must retire within 6 months with no plans to resume full-time employment within the next 18 months. We will, however, record any paid, voluntary, part-time, or bridging work postretirement. As there are no age restrictions for retirement, none will be imposed in the present study, though we anticipate that most participants will be aged between 55 and 70 years of age. Participants will need to consent to participate and be willing to complete follow-up surveys.
At each time point, participants will be asked to complete a survey, online (in Australia, US) or in paper-and pencil format (in China, where this is standard), that comprises demographic, perceived health, class and financial status questions in addition to validated measures of four key constructs: (1) Career and workplace factors (e.g., Job Satisfaction Survey, Employee Commitment Survey); (2) Retirement adjustment, using primary (e.g., Retirement Adjustment Scale, Retirement Planning Questionnaire), and secondary (i.e., wellbeing, Satisfaction with Life Scale; mood, Depression Anxiety Stress Scale; self-efficacy, Retirement Selfefficacy Scale) indicators (3) Social identity, measuring maintained and new identities in addition to identity compatibility (Social Network Diversity, Exeter Identity and Transition Scales), and (4) Individual differences (e.g., Ten-item Personality Inventory). We have experience in using forward- and back translation of similar measures in Chinese organizational contexts (Steffens et al., 2014, 2015).
We will apply the same strategy used in Study 1 to examine the following SIMIC-derived hypotheses.
- The role of multiple group membership at T1 in predicting long-term adjustment over time (H1).
- The mediating roles of maintained (H2, H3) and new (H4) groups at T1-T2 in predicting T3 adjustment.
- The contribution of group membership compatibility to adjustment (H5).
- We will include other factors recognized as influencing adjustment in our modelling. By assessing personal identity, individual differences, and perceived health, in addition to interpersonal and social identity variables, we can determine which has greater power in predicting adjustment (as in Haslam et al., 2014a).