UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Alone but not lonely? distinct types, antecedents, and correlates of older and younger adults' daily… Lay, Jennifer Christina 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_september_lay_jennifer.pdf [ 2.14MB ]
Metadata
JSON: 24-1.0368723.json
JSON-LD: 24-1.0368723-ld.json
RDF/XML (Pretty): 24-1.0368723-rdf.xml
RDF/JSON: 24-1.0368723-rdf.json
Turtle: 24-1.0368723-turtle.txt
N-Triples: 24-1.0368723-rdf-ntriples.txt
Original Record: 24-1.0368723-source.json
Full Text
24-1.0368723-fulltext.txt
Citation
24-1.0368723.ris

Full Text

        ALONE BUT NOT LONELY? DISTINCT TYPES, ANTECEDENTS, AND CORRELATES OF OLDER AND YOUNGER ADULTS’ DAILY LIFE SOLITUDE EXPERIENCES IN TWO CULTURAL CONTEXTS  by JENNIFER CHRISTINA LAY  A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Psychology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  June 2018  © Jennifer Christina Lay, 2018    ii     The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Alone but not lonely? Distinct types, antecedents, and correlates of older and younger adults’ daily life solitude experiences in two cultural contexts   submitted by Jennifer Lay  in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Psychology  Examining Committee: Dr. Christiane Hoppmann, Department of Psychology Supervisor  Dr. Anita DeLongis, Department of Psychology Supervisory Committee Member  Dr. Peter Graf, Department of Psychology Supervisory Committee Member Dr. Elizabeth Dunn, Department of Psychology University Examiner Dr. Mark Beauchamp, Department of Kinesiology University Examiner  Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member     iii     Abstract    Solitude (the absence of social interaction, whether in-person or electronic) is a ubiquitous yet understudied experience, often confused with loneliness, but sometimes sought out in daily life. This research program aimed to better understand the negative and positive aspects of solitude, drawing on data from three samples: 50 university students in Vancouver, 100 community-dwelling adults aged 50+ in Vancouver (including 51% East Asian immigrants), and 56 community-dwelling adults aged 50+ in Hong Kong. Participants completed approximately 30 repeated daily life assessments over 10 days on their current thoughts, affect, location, activities, social situation, and desire for solitude. Study 1 used latent profile analysis to identify distinct types of solitude experiences from the everyday thought-affect patterns of younger and middle-aged/older adults in Vancouver, and examined for whom and under what circumstances solitude may have positive or negative connotations. Two distinct types of solitude experiences were identified. Overall desire for solitude and social self-efficacy were associated with positive solitude experiences, and self-rumination and self-reflection with negative solitude experiences. Study 2 specifically examined solitude desire and its location and affective correlates among middle-aged and older adults living in Vancouver. At most occasions, solitude happened by individuals’ own choosing. Older adults were more likely to go to locations that matched their desired social context, and solitude-seeking had less negative affective associations for them as compared to middle-aged adults. East/Southeast Asian participants reported more loneliness than European/North American participants. Study 3 combined two data sets, from Vancouver and Hong Kong, to disentangle the roles of culture, immigration, and acculturation on solitude-loneliness associations among adults aged 50+. Participants high in   iv     acculturation to the local (host) culture or who desired solitude at that moment showed no association between being in solitude and feeling lonely. Taken together, these studies show that solitude and loneliness are distinct constructs with different predictors, correlates, and consequences. This research identified key individual difference, life phase, and social contextual factors associated with seeking solitude and thriving during solitude.     v     Lay Summary    Solitude (time spent without social interaction) is a common but poorly understood phenomenon. This research program sought to understand how people experience solitude in daily life, and when and for whom solitude may be experienced positively versus negatively. Participants completed 30 questionnaires over a 10-day period that captured their lived experiences. Study 1 identified two distinct types of solitude experiences (one positive, one negative) and tied these to key individual difference characteristics. Study 2 showed that middle-aged and older adults regularly seek solitude, and that they differ in how they feel and where they go when seeking solitude. Study 3 combined data from Vancouver and Hong Kong, showing that mainstream culture acculturation helps individuals thrive in solitude. These studies show that solitude need not feel lonely, and identify key motivational, situational, and person-level characteristics conducive to thriving in solitude.     vi     Preface   In collaboration with my advisor, Dr. Christiane Hoppmann, I was responsible for the design of this PhD research program. I was the study coordinator for the core project involving community-dwelling middle-aged and older adults in Vancouver, titled Health and intergenerational Activities Research Project (HARP). I also coordinated an affiliated study involving undergraduate students in Vancouver, titled Social Wellbeing and Everyday Engagement Study (SWEET). I was responsible for data collection, data management, and leading a team of research assistants for these two studies. In collaboration with Prof. Helene Fung, a similar study was conducted with community-dwelling middle-aged and older adults and university students in Hong Kong (HARP HK). I was responsible for training the research team to collect and manage the data for this project. The work presented in this dissertation draws on data from all three studies.   I am the primary author and contributor to this dissertation (Chapters 1–5). Chapters 2, 3, and 4 are based on manuscripts in press or in preparation for publication (details below). Because the three manuscripts address different aspects of an overarching solitude model that I developed, and because they use data from the same project, these chapters contain some overlapping material and hence some repetition of key conceptual and study design information.   A version of Chapter 2 has been accepted for publication as: Lay, J. C., Pauly, T., Graf, P., Biesanz, J. C., & Hoppmann, C. A. (in press). By myself and liking it? Predictors of distinct types of solitude experiences in daily life. Journal of Personality. Christiane Hoppmann and Peter Graf were responsible for study conception. I was responsible for research design (with Christiane Hoppmann), data analysis and interpretation, and manuscript composition. Theresa Pauly, Peter Graf, Jeremy Biesanz, and Christiane Hoppmann assisted with interpretation and   vii     manuscript revisions.   A version of Chapter 3 has been accepted for publication as: Lay, J. C., Pauly, T., Graf, P., Mahmood, A., & Hoppmann, C. A. (2018). Choosing solitude: Age differences in situational and affective correlates of solitude-seeking in midlife and older adulthood. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences. Advance online publication. doi: 10.1093/geronb/gby044. Christiane Hoppmann, Peter Graf, and Atiya Mahmood were responsible for study conception. I was responsible for research design (with Christiane Hoppmann), data analysis and interpretation, and manuscript composition. Theresa Pauly, Peter Graf, Atiya Mahmood, and Christiane Hoppmann assisted with interpretation and manuscript revisions.   A version of Chapter 4 is in preparation for submission for publication as: Lay, J. C., Fung, H. H., Jiang, D., Mahmood, A., Graf, P., & Hoppmann, C. A. (in preparation). Solitude in context: On the role of culture, immigration, acculturation, and solitude desire in the experience of time to oneself. Helene Fung, Christiane Hoppmann, Da Jiang, and I were responsible for study conception. I was responsible for research design (with Christiane Hoppmann), data analysis/interpretation, and manuscript composition. Helene Fung, Da Jiang, Atiya Mahmood, Peter Graf, and Christiane Hoppmann are assisting with interpretation and manuscript revisions.   The research presented in this dissertation was approved by the University of British Columbia Behavioural Research Ethics Board (BREB), approval number H12-03117. Project title: “HARP (Health and intergenerational Activities Research Project)”. This UBC BREB approval covers both the Vancouver and the Hong Kong portions of the project. The Hong Kong study was also approved by the Chinese University of Hong Kong Survey and Behavioural Research Ethics Committee (SBREC), approval number N/A (approved October 6, 2015).   viii     Project title: “Can Being Alone Benefit the Mental Health of Older Adults? The Role of Culture and Migration”.    ix     Table of Contents Abstract ............................................................................................................................... iii Lay Summary ......................................................................................................................... v Preface ................................................................................................................................. vi Table of Contents ................................................................................................................. ix List of Tables ....................................................................................................................... xiii List of Figures ...................................................................................................................... xiv List of Abbreviations ............................................................................................................. xv Acknowledgments ............................................................................................................... xvi Chapter 1 Introduction ...................................................................................................... 1 1.1 What is solitude? ..................................................................................................... 1 1.2 Solitude in daily life: Context specificity ................................................................... 5 1.2.1 Facets of solitude in daily life: Momentary affect and thoughts during solitude ........ 5 1.2.2 Desire for solitude and solitude experiences ............................................................... 8 1.2.3 Other time-varying situational correlates of solitude experiences ............................. 9 1.3 Individual differences in solitude experiences ........................................................ 11 1.3.1 Social and personal resources and vulnerabilities shaping solitude experiences ..... 11 1.3.2 Solitude in the second half of life ............................................................................... 13 1.3.3 Solitude across cultures .............................................................................................. 15 1.4 Overview of research program ............................................................................... 18 Chapter 2 By myself and liking it? Predictors of distinct types of solitude experiences in daily life (STUDY 1) .............................................................................................................. 22 2.1 Introduction .......................................................................................................... 22 2.1.1 Different types of solitude experiences in everyday life ............................................ 23 2.1.2 Stable individual difference correlates and time-varying motivational correlates of everyday solitude experiences .............................................................................................. 25 2.1.3 Current study .............................................................................................................. 26 2.2 Method ................................................................................................................. 26 2.2.1 Participants ................................................................................................................. 27 2.2.2 Procedure ................................................................................................................... 27 2.2.3 Measures and basic descriptive data ......................................................................... 29 2.2.4 Statistical analyses ...................................................................................................... 32   x     2.3 Results ................................................................................................................... 35 2.3.1 Descriptive findings .................................................................................................... 35 2.3.2 Different types of solitude experiences in everyday life: Latent profile analysis results .................................................................................................................................... 35 2.3.3 Stable individual difference correlates and time-varying motivational correlates of everyday solitude experiences: Latent class regression ....................................................... 40 2.4 Discussion .............................................................................................................. 44 2.4.1 How is solitude experienced in daily life? Distinct types of solitude experiences ..... 44 2.4.2 For whom and under what circumstances is solitude likely to be a negative or positive experience? .............................................................................................................. 46 2.4.3 Limitations and future directions ............................................................................... 49 2.4.4 Conclusions ................................................................................................................. 50 Chapter 3 Choosing solitude: Age differences in situational and affective correlates of solitude-seeking in midlife and older adulthood (STUDY 2) ................................................... 52 3.1 Introduction .......................................................................................................... 52 3.1.1 Solitude-seeking and situation selection .................................................................... 53 3.1.2 Solitude-seeking and affective experiences ............................................................... 54 3.1.3 Solitude-seeking in midlife and older adulthood ....................................................... 54 3.1.4 Current study .............................................................................................................. 55 3.2 Method ................................................................................................................. 56 3.2.1 Participants ................................................................................................................. 56 3.2.2 Procedure ................................................................................................................... 56 3.2.3 Measures .................................................................................................................... 58 3.2.4 Statistical analyses ...................................................................................................... 59 3.3 Results ................................................................................................................... 60 3.3.1 Descriptive statistics ................................................................................................... 60 3.3.2 Solitude-seeking and situation selection in midlife and older adulthood ................. 61 3.3.3 Solitude-seeking and affective experiences in midlife and older adulthood ............. 65 3.4 Discussion .............................................................................................................. 69 3.4.1 Solitude-seeking in midlife and older adulthood ....................................................... 69 3.4.2 Solitude-seeking and situation selection in midlife and older adulthood ................. 70 3.4.3 Solitude-seeking and affective experiences in midlife and older adulthood ............. 71 3.4.4 Limitations .................................................................................................................. 73 3.4.5 Conclusions and future directions .............................................................................. 74   xi     Chapter 4 Solitude in context: On the role of culture, immigration, acculturation, and solitude desire in the experience of time to oneself (STUDY 3) .............................................. 76 4.1 Introduction .......................................................................................................... 76 4.1.1 Culture and solitude-loneliness associations ............................................................. 77 4.1.2 Immigration and solitude-loneliness associations ..................................................... 78 4.1.3 Host culture acculturation and solitude-loneliness associations ............................... 79 4.1.4 Solitude desire and solitude-loneliness associations ................................................. 80 4.1.5 Current study .............................................................................................................. 81 4.2 Method ................................................................................................................. 81 4.2.1 Participants ................................................................................................................. 81 4.2.2 Procedure ................................................................................................................... 84 4.2.3 Measures .................................................................................................................... 85 4.2.4 Statistical analyses ...................................................................................................... 86 4.3 Results ................................................................................................................... 87 4.3.1 Descriptive findings .................................................................................................... 87 4.3.2 Momentary solitude-loneliness associations ............................................................. 88 4.4 Discussion .............................................................................................................. 92 4.4.1 Culture and solitude-loneliness associations ............................................................. 93 4.4.2 Immigration and solitude-loneliness associations ..................................................... 94 4.4.3 Acculturation and solitude-loneliness associations ................................................... 94 4.4.4 Desire for solitude and solitude-loneliness associations ........................................... 95 4.4.5 Further exploratory findings ....................................................................................... 96 4.4.6 Limitations and future directions ............................................................................... 97 4.4.7 Conclusions ................................................................................................................. 98 Chapter 5 General discussion .......................................................................................... 99 5.1 Synthesis ............................................................................................................... 99 5.1.1 Momentary affect and thoughts during solitude ..................................................... 101 5.1.2 Desire for solitude and solitude experiences ........................................................... 103 5.1.3 Other time-varying situational correlates of solitude experiences ......................... 104 5.1.4 Social and personal resources and vulnerabilities shaping solitude experiences ... 106 5.1.5 Solitude in the second half of life ............................................................................. 107 5.1.6 Solitude across cultures ............................................................................................ 108 5.2 Contributions ....................................................................................................... 110 5.2.1 Disentangling solitude from related constructs ....................................................... 110 5.2.2 Time-sampling approach to examine lived experiences of solitude ........................ 113   xii     5.3 Limitations and future directions ......................................................................... 114 5.3.1 Other approaches to measuring solitude and solitude-seeking .............................. 114 5.3.2 Participant sample limitations, generalizability, and replicability ........................... 116 5.4 Final remarks ....................................................................................................... 119 References......................................................................................................................... 120 Appendices ........................................................................................................................ 141 Appendix A: Study 1 multilevel latent profile analysis procedure and results .............. 141 Appendix B: Study 1 multilevel latent class regression procedure ............................... 149 Appendix C: Study 1 variable descriptive information ................................................ 152 Appendix D: Study 2 data analytic approach and descriptive findings ......................... 155 Appendix E:     Study 3 multilevel model equations and variables .................................. 160      xiii     List of Tables TABLE 2-1: OVERALL SAMPLE DESCRIPTIVES FOR MOMENTARY AFFECT AND THOUGHT DIMENSIONS AND CLASS-SPECIFIC MEANS, STANDARD DEVIATIONS, AND STANDARDIZED MEAN CLASS DISTANCES FOR THE FINAL 2-CLASS MODEL FROM LATENT PROFILE ANALYSIS (N = 2944 SOLITUDE EPISODES) ............................................... 38 TABLE 2-2: MULTILEVEL LATENT CLASS REGRESSION PREDICTING LOG-ODDS OF EXPERIENCING POSITIVE SOLITUDE OVER NEGATIVE SOLITUDE (N = 150 INDIVIDUALS, N = 2944 SOLITUDE EPISODES) USING MAXIMUM LIKELIHOOD ESTIMATION WITH ROBUST STANDARD ERRORS ................................................................. 43 TABLE 3-1: LOGISTIC MULTILEVEL MODELS: CURRENT LOCATION BY SOLITUDE DESIRE. N = 95 INDIVIDUALS, N = 3058 MOMENTARY ASSESSMENTS ................................................................................................. 62 TABLE 3-2: MULTILEVEL MODELS: CURRENT AFFECT OUTCOMES BY SOLITUDE DESIRE. N = 95 INDIVIDUALS, N = 3058 MOMENTARY ASSESSMENTS. ................................................................................................ 66 TABLE 4-1: PARTICIPANT SAMPLE CHARACTERISTICS AND PERSON-LEVEL VARIABLE INTERCORRELATIONS .............. 83 TABLE 4-2: FIXED EFFECTS FOR MULTILEVEL MODEL PREDICTING CURRENT LONELINESS (N = 151 INDIVIDUALS, N = 4654 MOMENTARY ASSESSMENTS) ................................................................................................ 89 TABLE 5-1: SUMMARY OF STUDY CHARACTERISTICS AND FINDINGS ............................................................. 100 TABLE A-1: IDENTIFYING A SET OF DISTINCT TYPES OF SOLITUDE EXPERIENCES USING LATENT PROFILE ANALYSIS: CLASS PROPORTIONS, MODEL FIT INDICES, RESIDUALS, AND CLASSIFICATION INDICES FOR CANDIDATE MODELS ...... 147 TABLE C-1: INTERCORRELATIONS OF PERSON-LEVEL VARIABLES AND PERSON-AVERAGED MOMENTARY VARIABLES (N = 150 INDIVIDUALS) .................................................................................................................. 153 TABLE D-1: SITUATION-LEVEL (WITHIN-PERSON) VARIABLE DESCRIPTIVES BY SOLITUDE SITUATION AND BY SOLITUDE DESIRE (N = 3195 MOMENTARY ASSESSMENTS) .............................................................................. 157 TABLE D-2: INTERCORRELATIONS OF PERSON-LEVEL VARIABLES AND OF PERSON-AVERAGED SITUATION-LEVEL VARIABLES (N = 90-100 INDIVIDUALS) ......................................................................................... 158 TABLE D-3: SITUATION-LEVEL VARIABLE DESCRIPTIVES BY AGE GROUP (N = 3195 MOMENTARY ASSESSMENTS, N = 95 INDIVIDUALS) ...................................................................................................................... 159      xiv     List of Figures FIGURE 1-1: RESEARCH PROGRAM OVERVIEW .......................................................................................... 19 FIGURE 2-1: CONCEPTUAL MODEL AND DATA ANALYTIC STAGES: IDENTIFICATION OF DISTINCT SOLITUDE CLASSES BASED ON AFFECT/THOUGHT DIMENSIONS (LATENT PROFILE ANALYSIS) AND PREDICTION OF SOLITUDE CLASS MEMBERSHIP FROM SITUATION- AND PERSON-LEVEL CHARACTERISTICS (LATENT CLASS REGRESSION)............ 33 FIGURE 2-2: TWO TYPES OF SOLITUDE EXPERIENCES: CLASS-SPECIFIC MEANS OF MOMENTARY AFFECT AND THOUGHT DIMENSIONS FOR FINAL 2-CLASS MODEL FROM LATENT PROFILE ANALYSIS (N = 150 INDIVIDUALS, N = 2944 SOLITUDE EPISODES) .................................................................................................................... 39 FIGURE 3-1: ASSOCIATIONS OF MOMENTARY SOLITUDE DESIRE WITH CURRENT LOCATION (HOME, OUTSIDE) AND HIGH AROUSAL POSITIVE AFFECT IN MIDLIFE AND OLDER ADULTHOOD..................................................... 64 FIGURE 4-1. ASSOCIATIONS BETWEEN CURRENT SOLITUDE AND CURRENT LONELINESS AS A FUNCTION OF ACCULTURATION TO HOST CULTURE (A) AND CURRENT SOLITUDE DESIRE (B) ............................................ 92 FIGURE 5-1: CONCEPTUAL MODEL OF SOLITUDE EXPERIENCES: KEY FINDINGS FROM THE PRESENT RESEARCH PROGRAM ............................................................................................................................... 101 FIGURE A-2: PERSON-LEVEL DISTRIBUTION OF SOLITUDE EXPERIENCE CLASSES (N = 150 INDIVIDUALS) ............. 148 FIGURE C-1: DISTRIBUTIONS OF MOMENTARY AFFECT AND THOUGHT DIMENSIONS (N = 2944 SOLITUDE EPISODES) ............................................................................................................................................. 154      xv     List of Abbreviations  AIC  Aikike Information Criterion  AvePP average posterior probability BIC  Bayesian Information Criterion HANA  high arousal negative affect HAPA  high arousal positive affect ICC   intraclass correlation coefficient LANA   low arousal negative affect LAPA  low arousal positive affect LCR  latent class regression LPA   latent profile analysis OCC  odds of correct classification R1R   within-person scale reliability coefficient RKF   between-person (person-level score) reliability coefficient SMD  standardized mean distance     xvi     Acknowledgments    I owe my deepest thanks to my graduate supervisor, Christiane Hoppmann, for so many things. For sharing my passion for both quantitative methods and community-based research, and for being a role model for developing strong collaborative relationships across disciplinary boundaries and outside of academia. Thank you for encouraging me to celebrate successes when I wasn’t feeling them, for tolerating my writing process, and for continuing to be there as my mentor at every stage. Your support has made everything possible!   I also owe much thanks to my graduate committee members: Anita DeLongis, for your mentorship and support over the years, and Peter Graf, for your thought-provoking questions and insights throughout the writing process. Thank you to Helene Fung, Atiya Mahmood, Sandra Petrozzi, and Valerie Pruegger giving me opportunities to explore new research ideas through your mentorship and collaboration. Thank you also to Jeremy Biesanz and Victoria Savalei for training me to be a better quantitative researcher. All of you have taught me so much.   Thank you to my labmates and research assistants at the UBC Health and Adult Development Lab and the CUHK Motivation and Emotion Lab for your enthusiasm and countless hours spent meeting participants, coding, organizing, translating, and helping make sense of all the data. Special thanks to Lu Minjie for being my exchange partner and for coordinating the study in Hong Kong, and to Helene Fung for making this possible.    I am grateful for stipend and travel support from the Social Sciences and Humanities Research Council of Canada (Vanier CGS program and Joseph-Armand Bombardier CGS program), the UBC Faculty of Arts scholarship program, the UBC Department of Psychology Quinn Exchange Fellowship program, and the CUHK Global Scholarship Programme for   xvii     research excellence, which have allowed me to dedicate my time to research.   So much thanks to my grad school friends/colleagues, including (in no particular order) Julie Chang, Ellen Stephenson, Jenn Campbell, Jordan Brace, Marlise Hofer, Daniel Sude, Theresa Pauly, Victoria Michalowski, Pavel Kozik, Yvette Graveline, and Spooky Ho, for sharing this adventure and keeping me sane over the past six years.   Finally, thank you to Mom, Dad, Heath, and Sean, for not questioning my life choices and for reminding me who I am. Without you, there would be no “Why?”        The research presented in Chapters 2-4 was supported by a Vancouver Foundation planning grant and a 3-year grant awarded to Christiane Hoppmann, Sandra Petrozzi, Atiya Mahmood, and Peter Graf, and a University of British Columbia Faculty of Arts grant awarded to Christiane Hoppmann. Additional funding for the research presented in Chapter 2 was provided by a University of British Columbia Alpha Mater Society grant awarded to Jennifer Lay. Additional funding for the research presented in Chapter 4 was provided by a research grant from the South China Programme, Chinese University of Hong Kong to Helene Fung, Christiane Hoppmann, Da Jiang, and Jennifer Lay. We thank Sandra Petrozzi, Kitsilano Neighbourhood House; the University of British Columbia Learning Exchange; and Dr. Atiya Mahmood of Simon Fraser University for their contributions to this research.   1     Chapter 1 Introduction  1.1 What is solitude?  “There are days when solitude is a heady wine that intoxicates you with freedom, others when it is a bitter tonic, and still others when it is a poison that makes you beat your head against the wall.”  - Sidonie Gabrielle Colette (1966)    Solitude is multifaceted. On the one hand, solitude (time spent without social interaction) is often associated with being lonely and maladjusted (Hawkley, & Cacioppo, 2010; Jylhä & Saarenheimo, 2010). Loneliness has well-documented links with poor health and wellbeing, notably, depression, cardiovascular disease risk, cognitive decline, and all-cause mortality (Chen, Hicks, & While, 2014; Hawkley & Cacioppo, 2010; Perissinotto, Cenzer, & Covinsky, 2012). Older adults may be at particular risk of loneliness. An increasing number of older adults in both Western and East Asian countries are living alone and with few family contacts (Hays, 2002; Liu & Guo, 2007). In a recent response to a report by the Jo Cox Commission on Loneliness (2017), the U.K. government has appointed a Minister of Loneliness to tackle this well-recognized social problem across the lifespan. Social isolation and loneliness have become pressing concerns in political and in academic circles (Harrison, 2018).   However, despite a basic need for social connectedness and an aversion to loneliness (Cohen, 2004), people across the adult lifespan spend a substantial proportion of their daily lives alone (Averill & Sundararajan, 2014; Larson, 1990). Individuals sometimes, in fact, choose to   2     spend time alone rather than with others (Chua & Koestner, 2008; Burger, 1995: Leary, Herbst, & McCrary, 2003). Philosophers, artists, spiritual masters, and writers in both Western and East Asian traditions have long recognized the benefits of solitude (Koch, 1989; Long & Averill, 2003). Free from social stimuli and constraints, people may also be better able to explore different ways of understanding the world and themselves – solitude may thus also be good for creativity, problem-solving, or simply to get work done (Dahlberg, 2007; Koch, 1994; Long & Averill, 2003; Pedersen, 1997, 1999; Rokach, 2004). A growing research literature also suggests solitude can provide space for relaxation, self-attunement, and emotional renewal (Burger, 1995; Larsen et al., 1982; Long & Averill, 2003; Matias, Nicolson, & Freire, 2011).   This presents a puzzle: how can time in solitude be both lonely and nourishing, both avoided and sought out? Here, I outline some of the conceptual issues in the literature on solitude, and explain how my research addresses these issues to help solve the puzzle. Despite preliminary scientific evidence of the benefits of solitude, most psychological research on solitude focuses on loneliness and its negative connotations (Burger, 1995; Long & Averill, 2003). Moreover, there is a lack of conceptual clarity in much of the literature, as previous research often does not distinguish between solitude and loneliness (Cornwell & Waite, 2009; Jylhä & Saarenheimo, 2010). Solitude is the absence of social interaction, whether in-person or electronic (Burger, 1995): an objectively-defined state or situation without any specific emotional connotations. Loneliness, in contrast, is a negative emotional response to a “discrepancy between one’s desired and achieved levels of social relations” (Perlman & Peplau, 1981, p. 32). Hence, loneliness is an affective experience that stems from a negative appraisal of one’s current social situation, and that can occur when a person is alone or in the company of others (Larson, 1990). The conflation of solitude with loneliness, particularly in research on   3     older adults (Cornwell & Waite, 2009; Jylhä & Saarenheimo, 2010), obscures a better understanding of this important phenomenon. For research on solitude to move forward, making a conceptual distinction between solitude and loneliness is crucial.   It is also important to distinguish between solitude (the absence of social interaction) and aloneness (having no one physically present). Previous research has not always kept these two ideas separate. Some work has completely merged the definitions of solitude and aloneness, defining solitude as simply no one being physically present at a particular moment (e.g. Matias et al., 2011). This, however, leaves open the possibility that social interaction may still be occurring on the phone or online when alone. Most studies on solitude account for this possibility, but treat aloneness as a prerequisite for solitude. That is, they define solitude as times when no one is physically present and no social interaction is occurring through other means (Larson, 1990; Pauly, Lay, Nater, Scott, & Hoppmann, 2017; Pauly, Lay, Scott, & Hoppmann, in press). Still other work has included in the definition of solitude those instances when others are nearby but no social interaction is occurring: for example, reading a book in the middle of a busy coffee shop (e.g. Long & Averill, 2003; Long, Seburn, Averill, & More, 2003). Given these inconsistencies in the solitude literature, definitional clarity is needed. In this research, I examine solitude experiences in daily life, using a well-circumscribed and theory-driven definition of solitude (the absence of social interaction) that disentangles solitude from both loneliness and being alone.    As I explain further in Section 1.2, the literature on solitude is fragmented. In studies examining daily life processes across the adult lifespan, solitude has been linked with loneliness and other negative affective states, but also with self-attunement and feelings of calm (e.g. Chui, Hoppmann, Gerstorf, Walker, & Luszcz, 2014; Larson, 1990; Pauly et al., 2017). Other work   4     based on retrospective reports from university students has identified different types of solitude experiences, ranging from lonely to productive to spiritual (Long et al., 2003). There is some indication that motivation matters: Solitude that is desired may feel less lonely and be more beneficial than solitude that is unwanted (Averill & Sundararajan, 2014; Burger, 1995; Long & Averill, 2003). However, it is unclear how the previously described affective states, thought patterns, and motivations interact to produce solitude experiences, as previous research is not always well connected. In this dissertation, I aimed to develop a conceptual framework that would enable solitude to be investigated more systematically. Using this conceptual model, I aimed to examine the multifacetedness of solitude to arrive at a better understanding of under what circumstances and for whom solitude may be associated with negative experiences (e.g. loneliness) versus positive experiences (e.g. calm).    This research also focuses on diverse samples. As explained in Section 1.3, most previous work on solitude has focused on university student samples and has been conducted from a Western perspective (Averill & Sundararajan, 2014). Older adults spend more time in solitude, and may also be more likely to thrive during solitude, than their younger and middle-aged counterparts (Larson, Zuzanek, & Mannell, 1985; Pauly et al., 2017). By examining the experiences of both middle-aged and older adults of varying socioeconomic backgrounds, the present research taps into more diverse experiences of solitude than previous research conducted on relatively homogeneous university student samples. The present work focuses on community-dwelling adults age 50+ (“middle-aged and older adults”), as they may provide unique insight into how solitude is structured and experienced. Moreover, solitude may be interpreted and experienced differently across cultures, for example, in North American versus East Asian cultures (Averill & Sundararajan, 2014; Jiang et al., in press; Wang, 2006). By drawing on   5     samples of middle-aged and older adults living in Vancouver (51% of whom were East Asian immigrants) and in Hong Kong, the present work examined solitude among individuals living in two different cultures. To summarize, solitude needs to be understood both from an adult lifespan perspective (Larson, 1990) and from a cross-cultural perspective (Averill & Sundararajan, 2014), and I aimed to do that.  1.2 Solitude in daily life: Context specificity   Solitude is a normal part of everyday life (Larson, 1990), hence, it is important to understand solitude experiences in this daily life context. When in solitude, we often engage with our inner selves (Burger, 1995; Long & Averill, 2003). This solitude may be experienced positively and negatively to varying degrees, depending on what an individual is feeling and thinking, and depending on the context. The present work captures solitude experiences as they occur in daily life, and examines time-varying or situation-specific factors that may shape how solitude is experienced.  1.2.1 Facets of solitude in daily life: Momentary affect and thoughts during solitude   Previous research has provided initial evidence that solitude can be experienced negatively and positively (Long, 2000; Long et al., 2003; Wang, 2006). This evidence of multifaceted solitude experiences, however, comes from retrospective reports. When individuals are asked to recall their past experiences of solitude (what they were thinking and feeling), their reports may be shaped by pre-conceived notions of what solitude is or should be like, by the idea that one is an introvert or extravert, or by any number of other beliefs and knowledge sources   6     (Lay, Gerstorf, Scott, Pauly, & Hoppmann, 2017; Robinson & Clore, 2002). One way to reduce such retrospective self-report discrepancies is to use time-sampling methods (Almeida, 2005; Bolger, Davis, & Rafaeli, 2003; Hoppmann & Riediger, 2009). These methods capture participants’ experiences as they go about their daily lives, maximizing ecological validity and enabling researchers to examine how experiences vary across time and across situational contexts. Importantly, time-sampling methods minimize lag time between the experience and the self-report, thereby reducing the influence of retrospective report biases. The present work uses time-sampling to examine solitude experiences: affective states and thoughts when individuals happen to be in solitude.   Affective experiences during daily life solitude may vary across time and across situations. Time-sampling methods have been used to examine affective experiences during solitude, using varying definitions of solitude that are in fact closer to “aloneness” (e.g., Lang & Baltes, 1997; Larson et al., 1985; Larson, Csikszentmihalyi, & Graef, 1982). This work has found that time spent alone (as compared to time spent with others) is associated with lower positive affect, higher negative affect, lower energy levels, and poorer subjective well-being (Chui et al., 2014; Larson, 1990; Larson & Csikszentmihalyi, 1978; Larson et al., 1982, 1985; Matias et al., 2011; Pauly et al., 2017, in press). On the other hand, some of these studies have also found positive affective correlates, linking time alone with emotional renewal, calm, and lower self-consciousness (Larson, 1990; Larson et al., 1982; Pauly et al., 2017). This lends empirical support to the idea that time alone or in solitude can also have emotional benefits (Burger, 1995; Long & Averill, 2003). Other work suggests that positive experiences of solitude may even be more frequent than negative solitude experiences (Long, 2000; Long et al., 2003). In general, solitude is complex and seems to hold the potential for both negative and positive   7     affective experiences.   In addition to affective experiences, solitude has been linked with a wide variety of thought processes including creativity, problem-solving, self-exploration, and concentration (Koch, 1994; Larson & Csikszentmihalyi, 1978; Larson et al., 1982; Long, 2000; Long & Averill, 2003; Long et al., 2003; Rokach, 2004). Researchers have argued that the absence of social stimuli may free up attentional resources, which can then be used for self-reflection and mindful or focused attention (Baer, Smith, Hopkins, Krietemeyer, & Toney, 2006; Koch, 1994, Long & Averill, 2003; Trapnell & Campbell, 1999). Engaging in self-reflection may be an important way in which individuals derive benefit from solitude, as it has been shown to promote self-discovery, self-rejuvenation, and a sense of tranquility when in solitude (Burger, 1995; Dahlberg, 2007; Trapnell & Campbell, 1999). Mindfulness (nonjudgmental attentiveness to the present moment) may also be beneficial in solitude, to the extent that it fosters autonomy, self-control, and calm (Averill & Sundararajan, 2014; Baer et al., 2006; Brown & Ryan, 1992; Huffziger et al., 2013). Indeed, time-sampling research has shown that when in solitude, individuals report greater ease of concentration (Larson et al, 1982). Inward-focused thought, however, has its pitfalls. Falling into self-rumination (repetitive, uncontrollable thoughts) during solitude may magnify loneliness and perpetuate an unpleasant experience (Cacioppo & Hawkley, 2009; Huffziger et al., 2013; Trapnell & Campbell, 1999). Time-sampling research examining such cognitions during solitude in daily life is sparse (e.g. Larson et al., 1982), however, the broader literature on solitude suggests that thoughts involving self-reflection and mindfulness may be associated with positive solitude experiences, and thoughts involving self-rumination with negative or lonely solitude experiences.   Given that solitude has been associated with a variety of positive and negative affective   8     states and thought patterns, the raises some key questions. Do positive and negative experiences of solitude occur at the same time? Or do they reflect different types of solitude experiences? The research presented in Chapter 2 seeks to answer these questions. A third question that arises is, what conditions or circumstances bring about these different affective experiences and thoughts during solitude? I next examine motivational and situational factors that shape solitude experiences.  1.2.2 Desire for solitude and solitude experiences   Solitude may be tolerated, cherished, or anywhere in between, depending on the motivations at play. On the one hand, time spent in solitude may sometimes be unwanted because it feels lonely or isolating (Hawkley & Cacioppo, 2010; Long & Averill, 2003). However, sometimes solitude is desired. Individuals may actively seek out solitude to, for example, work on personal projects or relax after a long day (Burger, 1995; Long et al., 2003). These different motivations may shape solitude experiences: As long as it is not taken to excess, solitude that happens by choice rather than by circumstance may be experienced more positively (Averill & Sundararajan, 2014; Brown, Silvia, Myin-Germeys, & Kwapil, 2007; Burger, 1995; Long & Averill, 2003). With the exception of research using end-of-day reports (Chua & Koestner, 2008; Nguyen, Ryan, & Deci, 2017), few studies have examined how desire for solitude fluctuates over time within a given individual (Brown et al., 2007; Brown, 1992; McAdams & Constantian, 1983). I expected that moments of solitude that are desired will be experienced as more positive, and less lonely, than moments of solitude that are not desired.   Desire for solitude not only varies from one moment to the next, but may also operate as a stable individual difference factor shaping solitude experiences. Solitude-seeking has been   9     shown to stem not only from social disinterest or social avoidance, but from a genuine enjoyment of time alone (Chua & Koestner, 2008; Kwapil et al., 2009; Leary et al., 2003; Nikitin & Freund, 2010). Burger (1995) developed the idea that some individuals seek solitude more than others, and studies of Preference for Solitude have found that people scoring high on this trait spend more time alone and find this alone time to be more pleasant and positive (Burger, 1995; Wascowic & Cramer, 1999). People high in Preference for Solitude also recall more frequent inner-directed solitude experiences (Long et al., 2000), which speaks to the idea that people seek solitude for self-reflection and to address their personal needs. Moreover, autonomous motivation for solitude-seeking has been linked with lower loneliness and higher wellbeing at a between-person level (Chua & Koestner, 2008; Nguyen et al., 2017). Individuals who believe that solitude is undesirable and to be avoided, on the other hand, may be less likely to derive any such benefit (Long & Averill, 2003). I expected that individuals with a stronger overall desire for solitude would tend to experience solitude as more positive and less lonely.  1.2.3 Other time-varying situational correlates of solitude experiences   People may find themselves in solitude at any given moment due to daily routines, being in a certain situation (such as at home or at work), or a host of other factors that impede social interaction or encourage solitude. The three studies presented here account for the potential role of such time-varying situational factors in shaping solitude experiences.    Solitude may be experienced differently depending on whether one is alone (no one nearby) or experiencing “solitude in company” (others nearby but no social interaction). Long and colleagues (2006) use the example of a group hiking trip in which everyone is immersed in their own thoughts, reaping the benefits of contemplation while maintaining the sense of security   10     and camaraderie that come from being part of a group. This kind of solitude may be experienced more positively compared to being completely alone. On the other hand, being completely alone (and away from others’ eyes and ears) may bring a certain freedom and lack of self-consciousness not possible when others are present, and may thus be conducive to meditation, emotional processing, or creative pursuits (Burger, 1995; Leary et al., 2003; Long & Averill, 2003). Because of these two competing hypotheses, I made no specific predictions regarding whether being alone (compared to having others present) would be associated with more negative versus more positive solitude experiences.   In a daily life context, solitude tends to co-occur with certain times, locations, and activities. Solitude is more common at certain times of the day. In a study of married and unmarried older adults, Larson and colleagues (1985) showed that solitude peaks in the morning, whereas afternoon and evening are the most social part of the day. For unmarried older adults, solitude rates again rise in the evening (Larson et al., 1985). In general, the times of day when solitude is likely to occur are also those times of day when one is more likely to be at home. Previous research with adolescents and university students suggests that most daily life solitude occurs at home (approximately 50-70%; Larson et al., 1982; Long, 2000). Home, a place conducive to privacy and social withdrawal, may also be a particularly common location for solitude among middle-aged and older adults. The second most common location for solitude among university students was outdoor spaces (approximately 25%; Long, 2000). Indeed, adults of all ages may go to parks and other outdoor places for serenity and escape, to the extent that physical health and mobility allow it (Long et al., 2007). Finally, as mentioned earlier, solitude may be conducive to certain types of activities, notably, passive leisure (e.g. reading or relaxing) and working (Long, 2000; Long et al., 2003). Hence, solitude may be most likely to occur, and to   11     be a positive experience, when an individual is engaging in these kinds of activities. Overall, I expected that moments of solitude would be experienced more positively in the morning, when at home or outside, and when engaged in passive leisure or work activities.  1.3 Individual differences in solitude experiences   Aside from situational, time-varying factors, there is also evidence that stable, individual difference factors shape solitude experiences. Are some individuals better able to deal with time spent in solitude, and to derive benefit from it, compared to other individuals? In other words, might some individuals have a greater capacity for solitude (Winnicott, 1958)? I suggest that individuals with more social and personal resources (such as close relationships) and fewer vulnerabilities (such as social anxiety) might experience solitude more positively (Section 1.3.1). Older age and certain cultural contexts may also be associated with having more positive solitude experiences (Sections 1.3.2 and 1.3.3).  1.3.1 Social and personal resources and vulnerabilities shaping solitude experiences   There are a number of individual difference factors (social and personal resources) that may help individuals thrive in solitude. I talk about these first, followed by factors that may be associated with negative solitude experiences (personal vulnerabilities)   It has been argued that benefiting from solitude requires a strong sense of self, high self-esteem, and a secure attachment style (Larson, 1990; Long et al., 2003). To thrive in solitude, it seems to be important that a person feel securely connected to others or to society and be able to maintain a sense of community (Averill & Sundararajan, 2014; de Jong Gierveld, van Tilburg, &   12     Dykstra, 2005; Long & Averill, 2003; Pauly et al., in press). If solitude is experienced within the context of a strong network of social ties, then a person may not truly feel alone, but rather, can think about absent loved ones in ways that foster feelings of intimacy rather than loneliness (Averill & Sundararajan, 2014; Long et al., 2003; Long & Averill, 2003; Pauly et al., in press). In the present work, I examine individual difference in several social resources, with the expectation that such resources may make individuals more prone to positive experiences during solitude, and less prone to loneliness and other negative experiences in solitude. The social resources examined are social network size, social relationship quality, relationship status, perceived social status, and acculturation to one’s mainstream culture.   Certain personal resources may also be associated with thriving in solitude. Individuals high in social self-efficacy, who are confident in their social skills, may feel less threatened by finding themselves in solitude (Di Giunta, Eisenberg, Kupfer, Steca, Tramontano, & Caprara, 2010). Given that thoughts often turn inward during solitude, overall thought styles, or tendencies to engage in certain thought patterns, may also colour solitude experiences. Trait self-reflection, for example, involves a tendency to pay attention to one’s own thoughts in a curious and philosophical way (rather than in a negative, obsessive way; Trapnell & Campbell, 1999). A tendency for self-reflection might make a person more prone to positive solitude experiences because solitude provides space to pursue self-knowledge and self-improvement (Burger, 1995; Dahlberg, 2007; Trapnell & Campbell, 1999).   In addition to differences in available social and personal resources, certain vulnerability factors might make some individuals less likely to thrive in solitude. Individuals who are shy or socially anxious may sometimes avoid social interaction despite having the desire to interact with others, due to anxiety regarding their performance in social situations (Clark & Wells, 1995;   13     Nikitin & Freund, 2010; Spurr & Stopa, 2002). For such individuals, solitude may often be associated with feelings of anxiety or failure, leaving less room for solitude’s potential benefits to take hold. Individuals who are prone to self-rumination (preoccupation with one’s own thoughts, problems, and perceived threats) may also experience solitude more negatively (Trapnell & Campbell, 1999). This thought pattern is generally experienced as unpleasant and uncontrollable, and has been linked with self-sabotaging cycles of anxiety, social avoidance, and loneliness (Cacioppo & Hawkley, 2009; Hartman, 1983; Spurr & Stopa, 2002; Trapnell & Campbell, 1999).   1.3.2 Solitude in the second half of life   Due to age-normative circumstances, including retirement, reduced physical mobility, and shrinking of social networks (Antonucci, 2001; Lang & Carstensen, 1994; Zhang, Yeung, Fung, & Lang, 2011), individuals spend more time alone in old age (Larson, 1990; Larson et al., 1985; Pinquart & Sorensen, 2001; Wethington & Pillenar, 2014). In Western samples, for example, the percentage of time spent alone has been estimated to be 17% in adolescence, 29% in adulthood, 48-59% in young-old adults, and 62-71% in the oldest old (Baltes, Wahl, & Schmid-Furstoss, 1990; Chui et al., 2014; Klumb, 2004; Larson, 1990). However, older adults seem to be faring comparatively well during solitude. Compared to their younger and middle-aged counterparts, older adults show less pronounced decreases in high arousal positive affect and less pronounced increases in low arousal negative affect and loneliness when in solitude (Chui et al., 2014; Larson et al., 1985; Larson, 1990; Lang & Baltes, 1997; Pauly et al., 2017). Some research also suggests that older adults have a greater sense of control and autonomy when in solitude, compared to younger and middle-aged adults (Larson et al., 1985; Larson, 1990;   14     Lang & Baltes, 1997). Hence, there is initial evidence that solitude may be experienced less negatively in old age – at least among older adults in Western societies, where this research has been conducted.   Older adults may be particularly well-equipped to deal with their increased time in solitude, and even to benefit from it. A useful interpretive lens is socioemotional selectivity theory, which holds that as future time becomes more limited, individuals become more motivated to spend time with emotionally meaningful, familiar people, as opposed to novel partners (Carstensen, Gottman, & Carstensen, 1995; Fung, Carstensen, & Lutz, 1999; Fung, Lai, & Ng, 2001). Older adults may, at times, prefer solitude over engaging in meaningless or unpleasant social interactions. To summarize, it seems that older adults may maximize their wellbeing by dealing well with their increased time alone, and even perhaps by seeking out solitude in their daily lives.   Previous time-sampling research comparing time alone (or in solitude) to time with others has often examined solitude experiences across different life phases of the adult lifespan (e.g. Larson, 1990; Pauly et al., 2017) or has focused on older adults in particular (e.g. Baltes et al., 1990; Chui et al., 2014). However, previous research examining autonomous motivation or desire for solitude (using end-of-day reports), and research examining distinct types of solitude experiences (using retrospective reports), has focused almost exclusively on university student samples (e.g. Chua & Koestner, 2008; Long et al., 2003; Nguyen et al., 2017; Wang, 2006). Little is known about solitude desire, and about the possibility of different types of solitude experiences, in the later phases of life, and among individuals of varying socioeconomic and cultural backgrounds. It is well-recognized that the experiences of educated young adults of relatively high socioeconomic status may not always generalize to other populations (Henrich,   15     Heine, & Norenzayan, 2010). Hence, the research presented here focuses on understanding the experiences of community-dwelling adults aged 50 years and above, a relatively understudied population whose experiences may have a lot to teach us about solitude.  1.3.3 Solitude across cultures   The solitude literature is also limited by its Western focus; most research on solitude experiences has been conducted with North American and (to a lesser extent) European samples (Averill & Sundararajan, 2014; Wang, 2006). Notably, solitude and solitude-seeking may be interpreted quite differently across cultures. In individualistic societies (e.g. Canada), where people tend to internalize values of autonomy and independence (Markus & Kitayama, 1991; Triandis, 1988), solitude-seeking may be seen as a healthy indicator of independence, emotional stability, and self-sufficiency (Burger, 1995; Coplan, Prakash, O'Neil, & Armer, 2004; Long & Averill, 2003). Indeed, one study showed that North American students valued the freedom afforded by solitude to a greater extent than Chinese students (Wang, 2006). Members of collectivistic societies (e.g. China), in contrast, tend to internalize values of interdependence and group harmony (Markus & Kitayama, 1991). In these societies, the same solitude-seeking behaviour, by prioritizing individual needs over group needs, may be seen as reflecting social and adjustment problems (Rubin, Cheah, & Menzer, 2009). The greater emphasis on social rules in collectivistic cultures (Suh, Diener, Oishi, & Triandis, 1998) may also breed feelings of guilt or rejection among solitude-seeking members, thereby counteracting solitude’s potential positive effects.   Despite these cultural differences in perceptions of solitude-seeking, however, the story could be completely reversed if we consider other kinds of cultural values, specifically, the value   16     placed on self-reflection, low-arousal activities, and introversion more generally. It has been suggested that individuals who benefit from solitude may be those who more readily engage in self-reflection (Long & Averill, 2003), and self-reflection is a more important value in East Asian cultures compared to North American cultures (Heine, Lehman, Markus, & Kitayama, 1999). Moreover, there is evidence that Chinese students may value solitude more for contemplation and self-cultivation than do American students (Wang, 2006). Furthermore, in comparisons of Chinese and North American individuals, low arousal activities, such as meditation and yoga, were more preferred among Chinese people, whereas high arousal activities, such as parties and social gatherings, were more preferred among North Americans (Tsai, 2007). Finally, introversion is favoured more in collectivistic cultures and extraversion more in individualistic cultures (Markus & Kitayama, 1991; Tsai, Triandis, & Suh, 2002); given that introversion is strongly associated with voluntary social withdrawal (Costa & McCrae, 1992), cultures that prize this personality trait may be more conducive to having positive solitude experiences. Solitude provides an ideal environment for self-reflection, low arousal activities, and general introversion, and hence may be enjoyed by East Asian individuals to a greater extent than North Americans.   Based on the literature, the effects of culture on solitude experiences may be complicated in that there are two relevant forces that could lead to opposing predictions. While North American cultural values of autonomy and individualism might lend themselves to thriving in solitude (Burger, 1995; Coplan et al., 2004; Long & Averill, 2003), the same can be said of East Asian cultural values prizing self-reflection, low arousal activities, and introversion (Heine et al., 1999; Tsai et al., 2002; Tsai, 2007). To test these competing hypotheses regarding cultural differences in propensity to thrive in solitude, I compare solitude experiences in Canadian and   17     Chinese cultural contexts, drawing on data from middle-aged and older adults living in Vancouver and from middle-aged and older adults living in Hong Kong. This study focuses on older adulthood, a time when cultural values might be even more deeply ingrained or internalized (Fung, 2013).    Critics of cross-cultural research have cautioned that studies often compare cultural groups within the same country, and, hence, conflate cultural influences and immigration influences (Chang, 1996; Tsai, 2007). Due to recent historical events (specifically, the handover of Hong Kong to China in 1997), many middle-aged and older adults living in Vancouver are immigrants from China or Hong Kong. This provides a unique opportunity to examine the experiences of middle-aged and older adults who have immigrated from an East Asian culture to a Western society. One quarter of the total population of British Columbia immigrated from Hong Kong or mainland China (Statistics Canada, 2017). At the same time, many middle-aged and older adults who are living in Hong Kong are also immigrants from mainland China. According to the 2011 Hong Kong Census Report, more than 120 mainland Chinese immigrate to Hong Kong every day (Hong Kong Home Affairs Department and Immigration Department, 2011). Hence, comparing these two immigrant groups with local Hongkongese and Canadian middle-aged and older adults offers a way of teasing apart the influences of culture and immigration on individuals’ propensity to thrive during solitude.    In a multicultural context, acculturation is an important factor to consider when examining how culture and immigration might shape solitude experiences. Host culture acculturation (Ryder, Alden, & Paulhus, 2000) speaks to a potential mechanism underlying cultural differences. Specifically, participating in the local culture and having a sense of belonging within mainstream society (Ryder et al., 2000; Tieu & Konnert, 2015) may help   18     individuals smoothly navigate everyday life in a particular cultural context. Previous cross-sectional research has also shown that a sense of belonging to the local community or one’s country of residence is associated with reduced overall levels of loneliness among local and East Asian immigrant older adults living in Canada (de Jong Gierveld, Van der Pas, & Keating, 2015; Syed et al., 2017). Therefore, I expected that host culture acculturation would help explain potential cultural differences in propensity to thrive when in solitude. In any case, examining acculturation (a continuous measure) has the potential to provide more nuanced information than that obtained from dichotomous variables such as cultural heritage and immigration status.  1.4 Overview of research program   In this research program, I aimed to examine the multifacetedness of solitude by asking how solitude may be experienced differently across situations and across people. My focus was on subjective experiences of solitude, specifically, affective states (including loneliness) and thought patterns associated with solitude. This research aimed to identify distinct types of solitude experiences and to identify time-varying motivational and situational correlates of thriving during solitude (e.g. desire for solitude and current location). I also aimed to identify more stable individual difference factors associated with propensity to thrive during solitude. These include person-level resources and vulnerabilities (e.g. trait self-reflection), age, and cultural contextual factors (e.g. acculturation).    In the chapters that follow, I present three studies that examined solitude through different lenses: a cluster analytic approach to identifying distinct types of solitude experiences (Study 1), an age-comparison approach to examining solitude-seeking experiences (Study 2), and a cross-cultural approach to examining solitude-loneliness links (Study 3). Figure 1-1 provides a   19     conceptual overview of this research program’s key constructs and their interrelations.   Figure 1-1: Research program overview    Study 1 (Chapter 2) uses repeated daily life assessments from 100 adults aged 50+ and 50 students living in Vancouver. The study takes a cluster analytic approach to examining solitude experiences, using latent profile analysis to identify distinct types of solitude experiences based on momentary affect/thought patterns. The study investigates the phenomenon of solitude with the expectation that there are at least two different types of solitude experiences. The second aim of this study is to link distinct types of solitude experiences to several motivational and individual difference factors, to examine when and for whom solitude might be experienced positively versus negatively.    Study 2 (Chapter 3) uses data from the 100 adults age 50+ living in Vancouver (the same   20     sample of middle-aged and older adults used in Study 1). The study investigates solitude-seeking, a phenomenon that is common, yet understudied, particularly in the second half of life. The study compares middle-aged and older adults, examining how they feel and where they go when they want time to themselves.    In the Vancouver-based studies (Studies 1 and 2), most of the East Asian participants are also immigrants, so cultural heritage is conflated with immigration status. This motivated the need for a cross-national study (Study 3) to disentangle the roles of culture and immigration on solitude experiences.   Study 3 (Chapter 4) uses data from 96 of the 100 adults aged 50+ living in Vancouver (the same participant sample used in Study 2) and combines these with a new set of data from 56 adults age 50+ living in Hong Kong. Participants include both immigrant and local adults in the two countries. The aim of this study is to examine the distinct roles of cultural context, immigration, acculturation, and solitude desire on solitude experiences, specifically focusing on momentary solitude-loneliness links.   All three studies make use of data from the same sample of 100 middle-aged and older adults living in Vancouver. Study 1 incorporates an additional sample of 50 students living in Vancouver, and Study 3 incorporates an additional sample of 56 middle-aged and older adults living in Hong Kong. Sample sizes ranged from 100 to 150 participants across the studies, with each participant providing approximately 30 momentary assessments. These sample sizes at the person level and at the situation level are in line with sample size recommendations for maximizing power in multilevel study designs (Scherbaum & Ferreter, 2009). Recognizing that power in such designs is limited mainly by the number of clusters (i.e. number of people) rather than the size of each cluster (i.e. number of momentary assessments; Mathieu, Aguinis,   21     Culpepper, & Chen, 2012; Snijders, 2005), the studies were planned to maximize the number of participants sampled. More formal power analyses for the planned study models were conducted using the Optimal Design (Raudenbush et al., 2011) and PINT (Bosker, Snijders, & Guldemond, 1999) programs for multilevel power analysis. Assuming intraclass correlations ranging from 0.2 to 0.6, and a variety of covariance structures for the fixed effects and random effects, our design had power greater than 0.8 to detect effect sizes of approximately 0.2-0.3 (for level 1 and level 2 regression coefficients and cross-level interactions in models with continuous outcomes). The addition of model covariates increased power to the extent that they were correlated with the dependent variable, and decreased power to the extent that they were correlated with predictors of interest (Aberson, 2011). Given the number of parameter effect sizes that must be estimated to conduct power analysis in a multilevel modeling context, the power estimates reported here must be taken as rough approximations.    22     Chapter 2 By myself and liking it? Predictors of distinct types of solitude experiences in daily life (STUDY 1)  Note: A version of this chapter has been accepted for publication as: Lay, J. C., Pauly, T., Graf, P., Biesanz, J. C., & Hoppmann, C. A. (in press). By myself and liking it? Predictors of distinct types of solitude experiences in daily life. Journal of Personality.  2.1 Introduction   Time spent alone has a bad reputation, and perhaps for good reason. Loneliness is linked to poor health and wellbeing, notably, depressive symptoms, cardiovascular disease, and cognitive decline (Hawkley & Cacioppo, 2010). Yet, despite a need for social connection, people across the adult lifespan spend a lot of time alone, and sometimes choose time alone over time with others (Burger, 1995; Chua & Koestner, 2008; Larson, 1990; Lay, Pauly, Graf, Mahmood, & Hoppmann, 2018; Leary et al., 2003; Long & Averill, 2003). Integrating these seemingly contradictory perspectives, this study examines the multifaceted nature of everyday solitude (defined as the absence of social interaction; Burger, 1995) and links different kinds of solitude with time-varying motivational and more stable person-specific factors. To do so, we collected approximately 30 electronic daily life assessments per person over 10 days from 100 adults aged 50+ and 50 students.     Most psychological research on solitude emphasizes the negative correlates and consequences of loneliness (Long & Averill, 2003). Yet, a wealth of philosophical, spiritual, and popular work lauds the benefits of solitude for self-attunement and growth (Burger, 1995; Long et al., 2003). How can solitude be both lonely and nourishing? A factor contributing to this   23     paradox may be that the extant literature does not always conceptually distinguish between solitude, aloneness, and loneliness (Larson, 1990; Lay et al., 2018; Long & Averill, 2003; Pauly et al., 2017). Solitude is most clearly defined by the absence of social interaction, whereas aloneness is defined by the physical absence of other people, at a given moment (Burger, 1995; Larson, 1990). One can be in solitude but not alone when reading a book in a busy coffee shop. Conversely, one can be physically alone but not in solitude when chatting on the phone with a friend. Solitude and aloneness are defined by objective situational characteristics and their definitions do not have any specific emotional connotations (Larson, 1990). Loneliness, in contrast, is a negative emotional experience resulting from a “discrepancy between one’s desired and achieved levels of social relations” (Perlman & Peplau, 1981, p. 32). By this definition, one can feel lonely whether alone or surrounded by other people (de Jong Gierveld et al., 2005).   2.1.1 Different types of solitude experiences in everyday life   Solitude is a double-edged sword. On the one hand, studies of affective experiences have shown that, compared to being with others, being alone (and not interacting with others) is associated with increased negative affect and loneliness, and decreased positive affect and energy (Chui et al., 2014; Larson, et al., 1982, 1985; Nguyen et al., 2017; Pauly et al., 2017). On the other hand, studies also suggest people may seek solitude for escape or relaxation, fostering emotional renewal, greater low arousal positive affect, and lower self-consciousness (Burger, 1995; Larson, 1990; Larson et al., 1982; Long et al., 2003; Pauly et al., 2017). Research on cognitions associated with solitude points to a similar two-sidedness. Solitude may trigger maladaptive thought patterns such as self-doubt and rumination (a preoccupation with negative thoughts and perceived threats; Long & Averill, 2003; Trapnell & Campbell, 1999). Yet, solitude   24     may also bring benefits by fostering creativity, problem-solving, concentration, self-reflection, autonomy, and personal growth (Burger, 1995; Larson et al., 1982, 1985; Long et al., 2003).    These seemingly contradictory findings from the social psychological, adult lifespan developmental, and health literatures illustrate the complex nature of solitude. For the present study, we embraced this complexity by considering the broad spectrum of affective and cognitive correlates of solitude reported in previous research, while adopting a well-circumscribed definition of solitude (the absence of social interaction; Burger, 1995). Specifically, we link everyday solitude with (a) concurrent affective experiences (high and low arousal positive/negative affect; Feldman Barrett & Russell, 1998; Kashdan & Steger, 2006; Russell, 1996; Tsai, Knutson, & Fung, 2006) and (b) concurrent thought patterns (low cognitive effort thoughts, high cognitive effort thoughts; Baer et al., 2006; Trapnell & Campbell, 1999). We make a distinction between high and low cognitive effort thoughts because these thought patterns have distinct neural and affective correlates (e.g. Farb, Anderson, & Segal, 2012). High cognitive effort thoughts are those that involve evaluation, interpretation, and executive control; such thoughts are often more self-relevant and experienced as more difficult (Farb et al., 2012). Low cognitive effort thoughts, in contrast, are those that are non-evaluative, non-interpretative, and attuned to the present moment; such thoughts are often experienced as more pleasant and unconstrained (Farb et al., 2012; Mason et al., 2007). Taken together, we expected that these diverse affective and cognitive correlates of solitude would form at least two separable solitude experience clusters: one reflecting negative experiences such as loneliness and difficult thoughts (negative solitude experiences), and the other reflecting positive experiences such as calm affect and pleasant thoughts (positive solitude experiences).     25     2.1.2 Stable individual difference correlates and time-varying motivational correlates of everyday solitude experiences   Research suggests that social resources may play a key role in how we experience solitude. Individuals with large social networks, high-quality social relationships, and high social status may experience solitude more positively than individuals with fewer social resources (Adler & Stewart, 2007; Antonucci, 1986; de Jong Gierveld et al., 2005; Long & Averill, 2003; Pauly et al., in press; Ryff & Keyes, 1995). With respect to personal resources, individuals high in social self-efficacy may be less prone to self-doubt in the absence of social feedback, and hence be better able to reap solitude’s benefits (Di Giunta et al., 2010). Moreover, individuals high in trait self-reflection might actively seek out and savour solitude because they enjoy having space for contemplation (Burger, 1995; Trapnell & Campbell, 1999). Hence, we expected trait levels of social self-efficacy and self-reflection to be tied to propensity for positive solitude in daily life.   In addition to social and personal resources, certain trait vulnerabilities might make individuals prone to negative solitude experiences. Socially anxious individuals may avoid interaction despite a desire to connect, thereby perpetuating feelings of loneliness and anxiety; hence, we expected they may be more prone to experience solitude negatively (Ernst & Cacioppo, 2000; Spurr & Stopa, 2002). We also expected individuals high in self-rumination, who may engage in maladaptive thought patterns that perpetuate negative affect, to be more prone to experience solitude negatively (Long & Averill, 2003; Trapnell & Campbell, 1999).    Finally, the likelihood of experiencing solitude positively versus negatively may also depend on whether an individual wants to interact with others at a particular moment. Undesired solitude may be difficult to tolerate whereas solitude that is desired may be cherished (Chua &   26     Koestner, 2008; Long et al., 2003). Furthermore, individual differences in overall desire for solitude are thought to shape momentary solitude experiences: We expected individuals with greater desire for solitude to be more prone to experiencing momentary solitude positively, compared to individuals with low desire for solitude (Burger, 1995).   2.1.3 Current study   The purpose of this study was to examine the complexity of solitude as it naturally occurs in daily life, and to determine under what circumstances and for whom solitude may be experienced positively or negatively. We used repeated daily life assessments (‘time sampling’) to capture time-varying emotional and cognitive correlates of everyday solitude (Bolger et al., 2003; Hoppmann & Riediger, 2009). Using latent profile analysis on approximately 30 momentary affect and thought assessments from 150 individuals, we sought to classify solitude episodes into distinct types, with the expectation that there would be at least two separable types of solitude experiences (one negative; one positive). We hypothesized that individuals with large social networks and high-quality social relationships, and those high in perceived social status, social self-efficacy, and self-reflection would have a greater propensity to experience solitude positively, as compared to individuals with fewer of these resources. In contrast, individuals high in social anxiety and self-rumination were expected to be more prone to negative solitude experiences than individuals with fewer such vulnerabilities. Finally, we expected that current desire for solitude and stronger overall (trait level) desire for solitude would be positively associated with the likelihood of experiencing solitude positively.  2.2 Method   27     2.2.1 Participants   One hundred community-dwelling adults aged 50-85 years (M = 67.0, SD = 8.7) and 50 undergraduate students aged 18-28 years (M = 20.0, SD = 1.8) in Metro Vancouver were recruited for a study on social engagement and wellbeing. We combined the two samples to maximize statistical power and to represent individuals across a range of ages and backgrounds. Middle-aged and older adults were recruited through community organizations, posters, referrals, and a database, and students were recruited through a university research subject pool. The middle-aged/older adult sample was 64% female, 56% East Asian, 36% European, and 8% of other/mixed heritage; 72% had at least some post-secondary education. The student sample was 92% female, 42% East Asian, 22% European, and 20% of other/mixed heritage. Fifty-seven percent of the middle-aged/older adults and 28% of the students were in a romantic relationship, and both samples were in good health (M = 3.2 on 5-point subjective health scales). Nine additional participants left the study due to time constraints (4 middle-aged/older adults, 3 students) or difficulties with the electronic assessments (2 middle-aged/older adults), and two middle-aged/older adults were excluded due to technical issues resulting in data loss. Middle-aged and older adults were reimbursed with up to $100 or the iPad mini they had used in the study. Students were reimbursed with 3 course credits and up to $30 (differences in compensation between the two samples reflect that middle-aged and older adults were part of a longitudinal study, whereas students were not). The study was approved by the university behavioural research ethics board.  2.2.2 Procedure   This study consisted of a baseline session, a time-sampling period, and an exit session. In   28     the baseline session, participants completed questionnaires measuring individual differences (e.g. trait self-reflection) and received training in the use of portable electronic devices. Then, for a 10-day time-sampling period beginning the day after the baseline session, participants were beeped three times daily (once in the morning, once in the afternoon, once in the evening). On each occasion, participants completed a brief questionnaire concerning their thoughts, affect, and current and desired social situation using a touch screen interface on an iPod or iPad mini (iDialogPad; G. Mutz, 2011, University of Cologne, Germany). To avoid conflicts with predetermined commitments, beeps were adjusted to participants’ schedules, with at least 4 hours between beeps. Participants completed an average of 30.5 valid questionnaires (SD = 9.6, range = 4-71). In cases when a participant completed two questionnaires within 90 minutes of one another, we deleted both questionnaires. An additional 180 time-sampling questionnaires (3.8%) were thereby excluded from analyses. Nearly two-thirds of participants continued completing time-sampling questionnaires after the end of the 10-day study period, as beeps were not programmed with a specific end date. Hence, the average number of questionnaires completed exceeds the expected number (3 daily questionnaires x 10 days = 30 questionnaires). Excluding these extra questionnaires did not substantively change the reported findings. Hence, we kept all completed questionnaires in the reported analyses.   Within two weeks after the time-sampling period, participants attended an exit session to complete further individual difference measures and a debriefing. Participants reported that the time-sampling period was typical of their everyday lives (M = 3.5 on a 5-point scale) and that the study did not interfere with their daily routines (M = 1.8/5) or change their behaviour (M = 1.7/5). Data were collected year-round (August 2014–May 2016). All materials were translated into Chinese and translations were verified via independent backward-translation. Middle-aged   29     and older adult participants completed the study in English (57% of participants), Mandarin (28%), or Cantonese (15%). Student participants completed the study in English.  2.2.3 Measures and basic descriptive data 2.2.3.1 Time-sampling measures   Current thoughts. At each beep, participants were first asked, “What were you just thinking about?” and they recorded a brief answer using the keyboard or voice recorder. They then responded to eight items concerning their current thoughts (each item used a 100-point scale: 0 = “not at all true”, 100 = “completely true”). These items were adapted from measures of reflection and rumination (Trapnell & Campbell, 1999) and mindfulness (Baer et al., 2006). We grouped the items into two parcels reflecting (a) low cognitive effort thought and (b) high cognitive effort thought. The low cognitive effort parcel consisted of four items assessing present focus (“I was thinking about something that happened in the past” [reverse coded], M = 68.2, SD = 33.4, “I was thinking about something happening in the future” [reverse coded], M = 54.7, SD = 45.3), pleasantness (“My thoughts were pleasant”, M = 54.7, SD = 28.6), and mindfulness (“I was just watching my thoughts go by without getting caught up in them”, M = 43.7, SD = 32.6). The high cognitive effort parcel consisted of four items assessing self-focus (“My thoughts were mainly about myself”, M = 50.7, SD = 35.2), reflection (“I was exploring new or 'deep' ideas”, M = 31.4, SD = 30.5), rumination (“I was having a hard time shutting off negative thoughts”, M = 26.2, SD = 28.5), and lack of clarity (“It was difficult to describe my thoughts just now”, M = 25.3, SD = 26.5).   Current affect. The next twelve items used a 100-point scale (0 = “not at all”, 100 = “very much”) to assess participants’ current affective and cognitive-emotional states. Items were   30     drawn from previous work to capture a spectrum of positive and negative affective states of both high and low arousal (Feldman Barrett & Russell, 1998; Tsai et al., 2006), and to probe feelings of shyness (Kashdan & Steger, 2006; Spurr & Stopa, 2002) and loneliness (Russell, 1996). Items were grouped into four affect parcels representing (a) high arousal positive affect (2 items, “I am happy”, M = 61.1, SD = 25.2, “I am excited”, M = 36.8, SD = 28.3), (b) low arousal positive affect (3 items, “I am calm”, M = 68.7, SD = 24.4; “I am satisfied”, M = 55.5, SD = 27.8, “I feel close to others”, M = 53.6, SD = 29.1), (c) high arousal negative affect (4 items, “I am anxious”, M = 30.6, SD = 29.5; “I am irritated”, M = 23.9, SD = 27.3, “I feel shy”, M = 16.0, SD = 20.9; “I am worried about what other people might think of me”, M = 25.6, SD = 29.3), and (d) low arousal negative affect (3 items, “I am sad”, M = 22.7, SD = 25.9; “I am tired”, M = 46.0, SD = 32.3, “I am lonely”, M = 23.3, SD = 26.8).   Current social situation. To collect information about participants’ social situation at each beep, participants were asked, “What was your situation when you were reminded to do this questionnaire?” They responded by selecting one of the following options, which were adapted from McAdams and Constantian (1983): (a) interacting with someone, (b) others nearby but not interacting, or (c) alone. Instances when participants selected (b) or (c) were categorized as solitude episodes (absence of social interaction). Participants were also asked to indicate the activities they had been engaged in when beeped by selecting one or more of the following activity categories: social activity, physical activity, cognitive activity, volunteering, passive leisure, self-care/health care, work, other. Instances when participants had been engaged in a social activity were removed from the pool of solitude episodes to eliminate times when participants may have been talking on the phone or communicating online. Consistent with the foregoing criteria, of the 4571 valid questionnaires completed by participants, we classified 2944   31     (64%) as solitude episodes and used these in the analyses (M = 19.6 episodes per participant; SD = 9.0, range = 1–69). Instances when participants selected (c), “alone”, were coded as episodes of aloneness. These constituted 64% of solitude episodes (M = 13.6 alone episodes per participant, SD = 9.1, range = 0–53). Analyses controlled for aloneness to disentangle being alone from being in solitude.   Current desire for solitude. We also used the three social situation options from McAdams and Constantian (1983) to collect information about participants’ current ideal or desired social situation. The results showed that 15% of the solitude episodes were times when participants had wanted social interaction (a), 28% were times when they had wanted others nearby but no interaction (b), and 57% were times when they had wanted to be alone (c). Instances when participants chose options (b) or (c) were coded as desire for solitude, and instances when they chose option (a) were coded as desire not to be in solitude.  2.2.3.2 Individual difference measures   Social and personal resources. Social network size was measured in the exit session using the Personal Networks Questionnaire (Antonucci, 1986), which requires participants to list people in their network in three concentric circles according to how close they feel to each person. Social network size was quantified as the total number of individuals listed in all circles (M = 20.3, SD = 11.9). We assessed social relationship quality (perception of having close, supportive relationships with others) in the exit session using the 3-item “positive relations” subscale of the Ryff Scales of psychological wellbeing (short version, 5-point Likert scale; Ryff & Keyes, 1995; M = 3.6, SD = 0.7, α = 0.55). Perceived social status was assessed in the exit session using the MacArthur scale (Adler & Stewart, 2007). For this scale, participants circle a   32     rung on a 10-rung ladder to indicate their social status relative to others in their community (one ladder) and in their country (another ladder), and the average is taken (M = 5.5, SD = 1.4, α = 0.61). Social self-efficacy (self-efficacy as it pertains to social skills) was assessed in the exit session using the 5-item Perceived Social Self-Efficacy scale, on a 5-point Likert scale (Di Giunta et al., 2010; M = 3.6, SD = 0.6, α = 0.74). The 12-item Rumination-Reflection Questionnaire (Trapnell & Campbell, 1999) was completed in the baseline session; it includes a 6-item subscale assessing self-reflection (tendency to enjoy reflecting on one’s inner self) on a 5-point Likert scale (M = 3.4, SD = 0.7, α = 0.79).   Personal vulnerabilities. Social anxiety was assessed in the exit session using the 6-item short version of the Social Interaction Anxiety Scale (Fergus, Valentiner, McGrath, Gier-Lonsway, & Kim, 2012; M = 2.2, SD = 0.8, α = 0.92).  Self-rumination (tendency to ruminate over past mistakes or negative thoughts) was assessed in the baseline session using the corresponding 6 items from the Rumination-Reflection Questionnaire (Trapnell & Campbell, 1999; M = 3.4, SD = 0.8, α = 0.82). Both measures use a 5-point Likert scale.  2.2.4 Statistical analyses   We used a 2-stage procedure to (1) classify solitude episodes into different types based on momentary affect/thought dimensions, and (2) predict the likelihood of experiencing each type of solitude in daily life from a set of time-varying and person-level predictors. This procedure is illustrated in Figure 2-1; details are described below and in Appendices A and B.    33      Figure 2-1: Conceptual model and data analytic stages: Identification of distinct solitude classes based on affect/thought dimensions (latent profile analysis) and prediction of solitude class membership from situation- and person-level characteristics (latent class regression)  2.2.4.1 Stage 1: Multilevel latent profile analysis   For each solitude episode, participants responded to 12 affect and eight thought items. We grouped these items into 4 affect parcels (high arousal positive affect, low arousal positive affect, high arousal negative affect, low arousal negative affect) and 2 thought parcels (low cognitive effort thought, high cognitive effort thought) based on theoretical groupings of affect and thought dimensions. This parceling gives equal weight to affect/thought dimensions reflecting positive/unchallenging experiences and those reflecting negative/effortful experiences. We expected that affect and thought response patterns would reveal at least two distinct types of solitude experiences. Latent profile analysis (LPA; Masyn, 2013) was used to test this hypothesis (see top part of Figure 2-1). LPA fits a set number of latent classes to data by maximizing intra-Solitude class 1Solitude class 2Solitude class 3etc...Level 1 - alone or not?predictors - desiring solitude or not?Level 2 - demographicspredictors - personal/social resources and vulnerabilities- person means of  situation-level predictorsSolitude episodes - 4 affect dimensions- 2 thought dimensionsLEVEL 1 (situation level)LEVEL 2 (person level)STEP 1: Latent Profile Analysis (LPA)STEP 2: Latent Class Regression (LCR)  34     class homogeneity and class separation. We generated several candidate models (with different model specifications and different numbers of classes), and selected a final model based on fit indices, residuals, classification diagnostics, parsimony, and theoretical considerations. Given the nested data structure (momentary affect/thoughts nested within individuals), multilevel modeling was used to account for person-level clustering in solitude class assignment (Henry & Muthén, 2010). Multilevel LPA was conducted in Mplus (Muthén & Muthén, 2007) using the parametric approach described by Vermunt (2003). Details of our LPA model specifications, modeling decisions, and procedure for final model selection are provided in Appendix A.  2.2.4.2 Stage 2: Multilevel latent class regression analysis   After classifying solitude episodes into different types through LPA, we used multilevel latent class regression (LCR) to test hypotheses regarding situational and individual difference factors predicting the likelihood of having each type of solitude experience (Henry & Muthén, 2010; Masyn, 2013). Odds of experiencing a certain class of solitude, relative to a reference class, was regressed on our set of predictors, again using multilevel modeling to account for the nested data structure (see bottom part of Figure 2-1). We used the 3-step approach recommended by Vermunt (2010) to account for uncertainty in solitude class membership by incorporating probabilistic class assignments. Mplus was used for LCR analyses. Details of the multilevel LCR procedure, including model equations, are provided in Appendix B.   To test hypotheses regarding time-varying motivational factors, current solitude desire was added as a dichotomous Level 1 (situation level) predictor. Person-average solitude desire was added at Level 2 (person level). Hypotheses regarding individual difference factors were tested by adding resources (social network size, social relationship quality, perceived social   35     status, social self-efficacy, self-reflection) and vulnerabilities (social anxiety, self-rumination) at Level 2. Several covariates were also added: current aloneness (dichotomous) at Level 1 and person-average aloneness, age, ethnicity, education, and relationship status at Level 2. All variables were grand mean centered. Refer to Appendix B for further details.  2.3 Results 2.3.1 Descriptive findings   Bivariate correlations for person-level variables are shown in Table C-2 (Appendix C). Individuals higher in certain personal/social resources (social relationship quality, perceived social status, and social self-efficacy) reported higher levels of positive affect and lower negative affect, whereas individuals higher in personal vulnerability factors (social anxiety and self-rumination) reported higher negative affect and lower positive affect. Although personal/social resources tended to be negatively correlated with vulnerability factors, one exception was that self-reflection and self-rumination were positively correlated, suggesting a tendency for self-focused thought common to these two traits. Mean time alone, time in solitude, and desire for solitude were all positively correlated, in line with the idea that people seek social situations that match their desires. Correlations among mean affect/thought dimensions suggested that tendency for high- and low-arousal positive affect and low cognitive effort thought go together, whereas tendency for high and low arousal negative affect and high cognitive effort thought go together. Situation-level variable descriptives and inter-correlations are provided in Table 2-1. They are discussed in the context of the latent profile analysis results. 2.3.2 Different types of solitude experiences in everyday life: Latent profile analysis results   36       The first aim of this study was to identify patterns of affect and thoughts characterizing different types of solitude experiences, with the expectation that at least two distinct types of solitude experiences would emerge. Multilevel LPA was used to classify solitude episodes into a number of classes (types) based on the 6 momentary affect/thought parcels, while accounting for person-level clustering of solitude class membership. Two model types were tested; one with and one without an additional indicator-specific random intercept (Appendix A provides further details on these model specifications). For each model type, a 1-class model was fitted, then a 2-class model, a 3-class model, and so on until the model was no longer identifiable. One-, 2-, and 3-class solutions were identified for both model types, but 4-class solutions were not identifiable. Table A-1 (Appendix A) gives further information on all models generated, including class proportions, model fit indices, residuals, and classification indices, and the Appendix A text explains each of these indices in detail. Scree plots for the Akaike Information Criterion (AIC) and Adjusted Bayesian Information Criterion (BIC) suggested that 2- and 3-class solutions for both model types were viable. However, the models with indicator-specific random intercepts were removed from further consideration due to their large residuals for the means, variances, and covariances. Classification indices showed that all models had good class separation and classification accuracy, but one stood out as performing best: the 2-class model with no indicator-specific random intercept (Entropy = 0.89; Average Posterior Probability = 0.98 for Class 1, 0.96 for Class 2; Odds of Correct Classification = 30.04 for Class 1, 33.79 for Class 2). Hence, the residuals and classification indices point to this 2-class model as being the best fit, supporting our hypothesis that at least two solitude classes would be distinguishable. The choice of this 2-class model over the 3-class model was also informed by model parsimony and theoretical considerations. It is generally recommended, in this situation, to pick the model with   37     the smaller number of theoretically meaningful classes (Masyn, 2013). Moreover, as we explain in the Appendix A, the solitude classes in the 3-class solution were not all well-separated or theoretically distinct. In our final model, two qualitatively distinct types of solitude experiences are well-identified, enabling direct tests of our hypotheses regarding positive and negative solitude experiences.    Table 2-1 shows class-specific mean ratings across the six affect and thought dimensions for the final LPA model classes. These and overall sample means are provided in Table 2-1. Class 1, comprising 56.7% of solitude episodes, reflected negative experiences, characterized by elevated levels of high and low arousal negative affect and high cognitive effort thought. Mean ratings on these three dimensions were around 40/100 (9-13 points above the overall sample means), and ratings for low arousal positive affect were 7 points below the sample mean. Hence, this class was labelled “negative solitude experiences”. The second class, comprising 43.3% of solitude episodes, was more positive. It was characterized by elevated levels of low arousal positive affect (69/100, 10 points above the overall sample mean) and slightly elevated levels of high arousal positive affect and of low cognitive effort thought. Most notably, high arousal negative affect ratings were near 0, and low arousal negative affect and high cognitive effort thought ratings were around 20/100, for this solitude class (11-18 points below overall sample means). To capture contrasts between the two solitude experience classes, we labelled Class 2 “positive solitude experiences”.        38     Table 2-1: Overall sample descriptives for momentary affect and thought dimensions and class-specific means, standard deviations, and standardized mean class distances for the final 2-class model from latent profile analysis (n = 2944 solitude episodes)      Note 1. All variable correlations are significant, p < .001. In this final model, indicator variances and covariances vary across classes, with no indicator-specific random intercept. All affect and thought dimensions are on 100-point scales. SMD is standardized mean distance, an adaptation of Cohen’s d indicating degree of class separation, calculated for each class indicator (parcel) separately. Values greater than 2 indicate < 20% overlap in class distributions; values less than 0.85 indicate > 50% overlap in distributions. Situation level SMDs are calculated for each affect/thought dimension across all solitude episodes; person level values are calculated for person-means of each affect/thought dimension. Note 2. To examine the generalizability of the solitude experience classes, we conducted the same multilevel LPA procedure on the full set of 4571 momentary assessments (both solitude and non-solitude). The reported 2-class solution was closely replicated in terms of class contents and class proportions, suggesting that the two solitude clusters also reflect overall clusters of negative and positive experiences.        Correlations CLASS 1  CLASS 2 Situation level  SMD Person  level  SMD  Affect/thought dimension M SD   2   3   4   5   6 M SD M   SD  1.  High arousal positive affect 48.9 22.1  .63  .32 -.26 -.43 -.08 46.0 21.9 52.7 21.7 0.31 0.40  2.  Low arousal positive affect 59.4 21.4   .36 -.48 -.51 -.26 51.9 20.2 69.0 19.0 0.87 1.00  3.  Low cognitive effort thought 53.0 15.6   -.29 -.31 -.24 50.6 15.2 56.2 15.4 0.37 0.75  4.  High arousal negative affect 24.0 20.4     .67  .56 37.6 16.5   6.3   7.0 2.53 3.89  5.  Low arousal negative affect 30.7 21.3      .43 41.7 19.5 16.3 13.6 1.50 2.02  6.  High cognitive effort thought 33.4 19.4      42.3 17.8 21.7 14.5 1.26 1.63   39      Caption: Thicker lines show class-specific means, and thinner lines show standard deviations, for the 6 momentary affect/thought dimensions in the final LPA model (indicator variances/covariances vary across classes, no indicator-specific random intercept).  Figure 2-2: Two types of solitude experiences: Class-specific means of momentary affect and thought dimensions for final 2-class model from latent profile analysis (N = 150 individuals, n = 2944 solitude episodes)  40     The class-specific descriptives in Table 2-1 give further insight into the solitude class structure. For all six affect/thought dimensions, variances within each class were smaller than overall sample variances, indicating that the LPA successfully generated homogeneous classes (Masyn, 2013). A second key indicator of LPA success is the extent to which classes are clearly separable (show little overlap in indicator values). The standardized mean distances in Table 2-1 reveal that, of all the affect/thought dimensions, high arousal negative affect showed the largest separation between the two solitude classes, followed by low arousal negative affect, high cognitive effort thought, and low arousal positive affect. Hence, these four dimensions are the most useful for distinguishing between positive and negative types of solitude experiences. Appendix A gives further details on the assessment of class homogeneity and class separation.    The positive solitude experience class was marked by consistently low levels of negative affect and of high cognitive effort thought, whereas the negative solitude experience class captured the rest of the negative affect/high cognitive effort thought spectrum. Further insight may be gleaned from the distributions of responses on the six affect/thought dimensions; Figure C-2 (Appendix C). The zero-inflated distributions for high and low arousal negative affect and high cognitive effort thought seem to indicate an underlying dichotomy, rather than continuity. That is, having little or no negative affect/high cognitive effort thought at a given moment appears to be a distinctly different experience from having some amount of negative affect/high cognitive effort thought. These different experiences are reflected in qualitatively distinct types of solitude. 2.3.3 Stable individual difference correlates and time-varying motivational correlates of everyday solitude experiences: Latent class regression    Our second aim was to link stable individual difference factors and time-varying   41   motivational factors with the likelihood of having positive and negative solitude experiences at a given moment. Solitude experiences (likelihood of negative vs. positive solitude class membership) varied both between people and within a given person across time, with most of the variability (80%) occurring at the between-person level. This between-person variability is illustrated in Figure A-1 (Appendix A), which shows the distribution of types of solitude experiences across people. Based on our classification of participants’ momentary experiences, most participants experienced only one type of solitude over the course of the study – negative solitude experiences only (~50% of participants) or positive solitude experiences only (~25% of participants) – while the remaining participants experienced a mix of both types.  This preliminary assessment suggests that the experience of different types of solitude might be better predicted by stable individual difference factors rather than time-varying factors. To test our hypotheses regarding predictors of distinct types of solitude experiences, log-odds of having positive over negative solitude experiences were regressed on several situation- and person-level variables using multilevel LCR (see Table 2-2).    Counter to expectations, social network size, social relationship quality, and perceived social status were not significantly associated with propensity for positive solitude experiences. As hypothesized, however, perceived social self-efficacy was linked with greater propensity for positive solitude experiences, b = 0.87, SE = 0.44, p = .048 (variable γ011 in Table 2-2). A 1-point increase in social self-efficacy meant 139% greater odds of positive solitude experiences. The association between trait self-reflection and solitude experiences was the opposite of that expected; self-reflection was linked with greater propensity for negative solitude experiences, b = -0.62, SE = 0.31, p = 0.045 (γ012 in Table 2-2). A 1-point increase in self-reflection meant 86% greater odds of having negative solitude experiences. As expected, trait self-rumination was   42   linked with greater propensity for negative solitude experiences, b = -0.62, SE = 0.31, p = .049 (γ014 in Table 2-2). A 1-point increase on the self-rumination scale meant 86% greater odds of negative solitude experiences. The expected association between social anxiety and negative solitude experience propensity, however, was not found.    We expected solitude desire to be linked with positive solitude experiences at the situation level (current solitude desire) and at the person level (person-mean solitude desire). Only the person-level association was significant, b = 3.99, SE = 1.27, p = .002 (variable γ03 in Table 2-2). A 10% increase in person-mean solitude desire (how often a person in fact wanted to be in solitude, when they were in solitude) meant a 49% increase in their odds of experiencing positive solitude.    Notably, person-mean solitude desire was only associated with positive solitude experiences, and not with having positive experiences in general. As shown in Table C-2 (Appendix C), correlations of person-mean solitude desire with scores on the six affect/thought dimensions were weak or nonsignificant. Moreover, as indicated in the footnotes to Table 2-2, although the reported solitude class structure was replicated when instances of social interaction were included in latent profile analyses, person-mean solitude desire was not predictive of having more positive experiences in general. This specific link between solitude desire and propensity for experiencing positive over negative solitude provides initial evidence of the solitude classes’ construct validity.       43   Table 2-2: Multilevel latent class regression predicting log-odds of experiencing positive solitude over negative solitude (N = 150 individuals, n = 2944 solitude episodes) using maximum likelihood estimation with robust standard errors    Log-odds of positive (Class 2) over negative (Class 1) solitude, logit(Pij)   Parameter Coefficient (unstandardized)     SE Relative odds p value LEVEL 1 β1j  Aloneness 0.09    (0.06) Class 2,  1.09 : 1  .152  β2j  Solitude desire  0.19    (0.11) Class 2,  1.21 : 1  .085 LEVEL 2 γ00  Intercept -0.48    (0.22) Class 1,  1.62 : 1  .025  γ01  Overall time in solitude 0.77    (1.17) Class 2,  2.16 : 1  .506  γ02  Person-mean aloneness  0.57    (0.83) Class 2,  1.77 : 1  .489  γ03  Person-mean solitude desire  3.99    (1.27) Class 2,  54.05 : 1  .002  γ04  Age  0.01    (0.01) Class 2,  1.01 : 1  .639  γ05  Ethnicity -0.30    (0.52) Class 1,  1.35 : 1  .564  γ06  Education  0.91    (0.75) Class 2,  2.48 : 1  .223  γ07  Relationship status  -0.23    (0.60) Class 1,  1.26 : 1  .697  γ08  Social network size -0.01    (0.02) Class 1,  1.01 : 1  .753  γ09  Social relationship quality  0.74    (0.40) Class 2,  2.10 : 1  .065  γ010  Perceived social status  0.12    (0.17) Class 2,  1.13 : 1  .472  γ011  Social self-efficacy  0.87    (0.44) Class 2,  2.39 : 1  .048  γ012  Self-reflection -0.62    (0.31) Class 1,  1.85 : 1  .045  γ013  Social anxiety -0.54    (0.34) Class 1,  1.72 : 1  .115  γ014  Self-rumination -0.62    (0.31) Class 1,  1.86 : 1  .049 Note 1. Overall time in solitude is proportion of all beeps when participant was in solitude. Person-mean aloneness (solitude desire) is proportion of solitude instances when participant was alone (desiring solitude). Age is in years; ethnicity is 1 = European, 0 = not European; education is 1 = some post-secondary, 0 = none; relationship status is 1 = in a relationship, 0 = not in a relationship. Social network size is total number of people listed. Social status is on a 10-point scale. All other variables are on 5-point scales. Bayesian multiple imputation (Muthén & Muthén, 2007) was used to impute missing data for age (N = 5), relationship status (3), social status (5), social self-efficacy (3), and social anxiety (3).  Note 2. Adding participant sample as a person-level predictor (1 = middle-aged/older adults, 0 = students) did not change reported findings, hence, we omitted this variable from the model for parsimony. We also added the following situation-level predictors: time of day (hours since 4:00am), time of day squared, daily precipitation, daily hours of sunlight, current location: outside (1 = outside, 0 = not outside), current location: at home (1 = at home, 0 = not at home), current activity: work (1 = working, 0 = not working), and current activity: passive leisure (1 = engaged in passive leisure, 0 = not engaged in passive leisure). Person-means of each situation-level predictor were added at Level 2. None of these predictors showed associations with solitude experiences, except for person-mean activity: work. Individuals who spent more of their time in solitude engaged in work were more likely to experience   44   solitude negatively. Due to potential power issues (adding current and person-mean work activity to the model reduced several of the reported associations to marginal significance), we omitted these and the other situation-level variables from the final model. Associations between working and solitude experiences are discussed in Chapter 5. Note 3. We also tested the reported LCR model on the 2-class solution derived from the entire set of momentary assessments (both solitude and non-solitude; see Table 2-1 footnote 2). Social self-efficacy was associated with having positive experiences in general, and self-reflection and self-rumination with negative experiences in general, in line with the reported associations with positive and negative solitude experiences. However, person-mean desire for solitude was not predictive of having positive experiences in general (despite its association with positive solitude experiences).  2.4 Discussion   The aim of this study was to embrace the complexity of everyday solitude experiences by examining distinct types of solitude experiences and by asking how and for whom solitude might be experienced positively versus negatively. We captured correlates of solitude (defined as the absence of social interaction) by asking older and younger adults to report their thoughts, affect, and current and desired social situations three times daily over 10 days. Multilevel latent profile analysis identified two types of solitude experiences, one positive and one negative, characterized by distinct patterns of affect and thought. Individuals higher in social self-efficacy and overall desire for solitude were more prone to positive solitude experiences, and those higher in self-rumination and self-reflection were more prone to negative solitude experiences. Findings are discussed in the context of the social psychological and adult lifespan developmental literatures.  2.4.1 How is solitude experienced in daily life? Distinct types of solitude experiences   This study used a valence-neutral definition of solitude (absence of social interaction) to capture qualitatively distinct solitude experience clusters based on affective states and thought patterns that co-occur with solitude. Our findings showed that solitude is indeed a multifaceted   45   construct that is best described by two distinct clusters: a “negative solitude experience” cluster characterized by negative affect and more effortful/complex/self-focused thought and a “positive solitude experience” cluster characterized by positive affect, simpler/pleasant/present-focused thoughts, and the near-absence of negative affect. Importantly, our approach takes into account that a given person may sometimes experience solitude negatively and sometimes positively. To illustrate, consider a person named Nadia, who lives by herself and who, after commuting on a crowded subway, is home alone at day’s end. Whether on the subway or at home, she is in solitude. Sometimes, Nadia may be preoccupied by worries or ruminations, her solitude marred by anxiety, sadness, or loneliness. This negative kind of solitude experience occurred most frequently in our study (about 57% of solitude instances), reflecting the negative contours of solitude (Long & Averill, 2003). On the other hand, Nadia may also experience the kind of solitude that helps her relax after a demanding day; at those times, she might be feeling calm and be enjoying the present moment, free of loneliness, anxiety, or intrusive thoughts. This positive kind of solitude experience represented a little under half (43%) of solitude instances in the present study, reinforcing the idea that solitude can be nourishing (Burger, 1995).   By examining solitude experiences as they occur in older and younger adults’ daily lives, our study complements and extends previous work using retrospective reports of solitude experiences from student samples (Long et al., 2003). Our solitude clusters suggest that, at the moment when it occurs, deep contemplation is a feature of negative solitude experiences. However, this label does not imply that negative solitude experiences are necessarily unhealthy, maladaptive, or indicative of a lonely existence (Cacioppo, Cacioppo, & Boomsma, 2014). Indeed, the challenges of introspection are thought to be among solitude’s key benefits to the extent that they foster problem-solving and self-growth (Burger, 1995; Long & Averill, 2003).   46   For example, Long and colleagues (2003) identified three kinds of solitude (outer-directed, inner-directed, and loneliness), based on reports of the importance of different kinds of solitude experiences, and suggested that inner-directed solitude may be remembered as a difficult process of self-reflection leading to inner peace. Unlike this previous work, our study captured snapshots of solitude as they occurred, before being subject to retrospection and subjective importance ratings, and before individuals might have benefited from working through tough problems during solitude. Moreover, many of our momentary affect/thought measures emphasized self-focused aspects of solitude experience, rather than outer-directed aspects such as spirituality and connectedness to others. Hence, it may not be surprising that the negative and positive solitude experience types we uncovered do not directly map onto those identified in previous research (Long et al., 2003).  2.4.2 For whom and under what circumstances is solitude likely to be a negative or positive experience?   Solitude has been described as a “unique experiential niche” in which some people are more likely to thrive than others (Larson, 1990, p. 156). Indeed, the present study points to systematic individual differences in solitude experiences: one-half of our sample had only negative solitude experiences, another quarter had only positive solitude experiences, and the rest had a mix of negative and positive solitude experiences. Moreover, we identified key individual difference factors underlying propensity to experience solitude negatively versus positively.   As expected, having high social self-efficacy was associated with experiencing solitude positively. This finding adds to the literature linking high self-esteem, communication skills, and secure attachment style with lower loneliness (de Jong Gierveld et al., 2005; Ernst & Cacioppo,   47   2000; Larson, 1990; Long & Averill, 2003). Counter to expectations, social network size, social relationship quality, and perceived social status were not significantly associated with positive solitude experience propensity. Building on previous work showing that strong social ties protect against negative solitude experiences (Pauly et al., in press), we suggest that, accounting for the quality of one’s social relations, having high confidence in one’s own social skills (social self-efficacy) may be a key to experiencing solitude positively.    In contrast to what we hypothesized, trait self-reflection was associated with greater propensity to experience solitude negatively. It is possible that, while self-reflection that is focused on self-attunement may be conducive to positive solitude experiences (Burger, 1995; Long & Averill, 2003; Leary et al., 2003), self-reflection that is focused on self-critical thinking may backfire and contribute to loneliness (Cacioppo et al., 2014). Indeed, engaging in the kind of deep introspection that is conducive to self-growth may in fact be a challenging, unpleasant experience in the moments of solitude when it occurs. The present study’s solitude cluster findings support this interpretation, revealing that high cognitive effort thought is a defining characteristic of negative, rather than positive, momentary solitude experiences.   The present study also examined individual differences in the propensity to experience solitude negatively. Findings showed that self-rumination was associated with greater likelihood of having negative solitude experiences. In solitude, thoughts often turn inward, and if an individual habitually has uncontrollable negative thoughts, these may negatively colour their experience (Long et al., 2003). We also expected highly socially anxious people to be more prone to negative solitude; however, we found no significant association. It may be that for socially anxious individuals, feelings of loneliness and social inadequacy in solitude (Ernst & Cacioppo, 2000) are balanced by feelings of calm and relief from social pressures (Long &   48   Averill, 2003; Spurr & Stopa, 2002), thereby neutralizing any negative effects. It may also be that our sample’s social anxiety scores (M = 2.2 on a 5-point scale) were too low to show an impact on solitude experiences.    Finally, this study embraced the fact that some people desire solitude more than others, and that this desire may ebb and flow in daily life. As expected, people with greater overall desire for solitude were more prone to experience solitude positively. However, fluctuations in solitude desire were not significantly associated with positive solitude experiences at the momentary level. Solitude desire hence seems to operate primarily as an individual difference factor. This finding aligns with previous research linking retrospective reports of overall preference for solitude to solitude enjoyment (Burger, 1995) and extends it to a broader range of affect and thought dimensions accompanying solitude experiences. We also build on research based on retrospective reports of positive and negative solitude experiences (Long et al., 2003) by showing how, when participants are not explicitly asked to think about their solitude experiences, their overall solitude desire still shapes their thoughts and affective states reported in the moment. Overall solitude desire was linked specifically with positive solitude experiences, but was not associated with having more positive experiences in general; this specificity constitutes further evidence for the existence of two distinct types of solitude experiences.   To demonstrate the validity of cluster analysis results, they must be grounded in well-established individual difference factors. This study revealed that high social self-efficacy, overall solitude desire, and low self-ruminative and self-reflective tendencies are particularly key to thriving in solitude. By linking these traits to daily life solitude, we took a first step toward validating the two types of solitude experiences (negative and positive) that emerged from older and young adults’ lived experiences.   49    2.4.3 Limitations and future directions   Our aim was to examine the complexity of solitude as it occurs in everyday life, and findings need to be interpreted in light of certain limitations. We sought to capture snapshots of naturally occurring experiences without interfering with participants’ daily routines, and hence chose a sampling frame that took into account participants’ pre-existing commitments. Doing so led to high compliance: Participants completed an average of 25 out of 30 possible assessments within the 10-day sampling frame. This approach could have resulted in oversampling of solitude instances. However, solitude rates in our study were similar to those in other time-sampling studies using quasi-random (Pauly et al., 2017) and random sampling frames (Larson et al., 1982, 1985), which gives us confidence that we captured naturally-occurring solitude episodes.   This study included older and younger adults across a broad cultural and social spectrum. We specifically aimed to include middle-aged and older adults who are less well-represented in research, such as recent immigrants and individuals of various socioeconomic statuses. As a result, more than half of our middle-aged/older adult sample were East Asian immigrants to Canada, and approximately half had incomes falling below the provincial low-income threshold. Although this limits generalizability, our study provides insight into the experiences of a large and growing population of middle-aged and older adult immigrants often missed in psychological research. Cultural factors may also shape solitude. Individuals of East Asian heritage may experience solitude more positively as it is conducive to self-reflection and low-arousal leisure activities, activities that are valued more in East Asian than in Western cultures (Averill & Sundararajan, 2014; Tsai, 2007). Although we found no cultural differences in solitude experiences in the current set of analyses, more research is needed to compare solitude   50   experiences across cultures. Finally, our middle-aged/older adult sample comprised mostly retired individuals, and our young adult sample comprised undergraduate students, with recognized limits to generalizability (Henrich et al., 2010). Life phase specific goals and social roles may make solitude a particularly common experience for both older adults and students (Larson, 1990; Lay et al., 2018; Pauly et al., 2017). In contrast, working adults with children at home may have less time or freedom to pursue solitary activities (Lay et al., 2018). To account for such life phase factors, our findings need to be replicated in samples representing the full adult lifespan.   2.4.4 Conclusions   Solitude need not be lonely. Our findings demonstrate that solitude is a multifaceted construct that can have positive as well as negative connotations. By combining momentary affect and thought assessments during time in solitude, we identified two types of solitude experiences, one negative and one positive, and linked them with well-established individual difference factors. Key characteristics of people likely to thrive in solitude were being high in social self-efficacy and desire for solitude, and being low in self-rumination and self-reflection. To further understand this emerging and important phenomenon, potential causal mechanisms (such as the role of self-rumination in producing negative solitude) need to be tested experimentally (e.g. Nguyen et al., 2017). Positive and negative solitude may also differentially shape longer term outcomes; this could be tested by examining time-ordered associations between solitude experiences and subsequent changes in wellbeing.    A particularly key finding of this study was the robust link between having a desire for solitude and having positive solitude experiences. This is in line with previous research on   51   university student samples, suggesting that autonomous or desired solitude may be experienced more positively than solitude that is unwanted (Long & Averill, 2003; Chua & Koestner, 2008; Nguyen et al., 2017). Solitude-seeking and wellbeing have been examined extensively among children, adolescents, and emerging adults (e.g. Bowker, Nelson, Markovic, & Luster, 2014; Coplan et al., 2004; Galanaki, 2015; Larson, 1997; Long & Averill, 2003). However, solitude-seeking among middle-aged and older adults is poorly understood. Due to age-normative circumstances, individuals spend a greater proportion of their time in solitude as they age (Larson, 1990). However, compared to their younger and middle-aged counterparts, older adults may in fact experience solitude more positively (Chui et al., 2014; Larson et al., 1985; Larson, 1990; Lang & Baltes, 1997; Pauly et al., 2017). The research presented in the next two chapters focuses on community-dwelling adults age 50 and above, for whom solitude and solitude-seeking may be particularly common.    52   Chapter 3 Choosing solitude: Age differences in situational and affective correlates of solitude-seeking in midlife and older adulthood (STUDY 2)  Note: A version of this chapter has been accepted for publication as: Lay, J. C., Pauly, T., Graf, P., Mahmood, A., & Hoppmann, C. A. (2018). Choosing solitude: Age differences in situational and affective correlates of solitude-seeking in midlife and older adulthood. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences. Advance online publication. doi: 10.1093/geronb/gby044  3.1 Introduction We need social contact to thrive (Cohen, 2004; Hoppmann & Gerstorf, 2016), but this does not mean we need, or want, to interact with others all the time. Individuals sometimes choose to spend time by themselves (Burger, 1995; Leary et al., 2003; Long & Averill, 2003), for example, spending an evening home alone or going on a hike with everyone immersed in their own thoughts. Solitude is defined as the absence of social interaction (Burger, 1995; Larson, 1990): an objectively defined situation without any specific emotional connotations. Loneliness, in contrast, is a negative emotional experience resulting from a perceived lack of social relations (Perlman & Peplau, 1981). Previous research on solitude has largely focused on the negative experiences that go along with loneliness (Ernst & Cacioppo, 2000). Little is known about instances when individuals seek out solitude (Long et al., 2003). This study focused on situational and affective correlates of solitude-seeking in midlife and older age, a phase in life when people spend a significant amount of time alone (Larson, 1990). We used approximately   53   30 electronic daily life assessments collected over a 10-day period from 100 community-dwelling adults aged 50-85 years. Solitude is a ubiquitous experience in midlife, and even more so in older age (Larson, 1990). Percentage of time spent alone ranges from 48% in samples aged 65 years and above to 71% in the oldest old (Chui et al., 2014; Larson et al., 1985). Solitude is a problem only to the extent that it is experienced as lonely or isolating (Cohen, 2004; Hawkley, & Cacioppo, 2010). Across the adult lifespan, individuals actively seek out solitude in daily life (Burger, 1995; Leary et al., 2003; Larson, 1990) to escape social pressures, to work, or to relax (Long et al., 2003; Pauly et al., in press). When solitude occurs by one’s own choosing, it is experienced more positively than when it is undesired (Lay, Pauly, Graf, Biesanz, & Hoppmann, in press; Long & Averill, 2003). Building on this assertion, this study investigated how much of the everyday solitude that middle-aged and older adults experience happens by choice, and how solitude-seeking is linked with time-varying situational characteristics and affective experiences.   3.1.1 Solitude-seeking and situation selection Most everyday solitude has been shown to occur at home (approximately 55-72%), followed by outdoor spaces (approximately 25%; Larson et al., 1982; Long, 2000). The comfort of one’s home may allow individuals to be “off-stage” and recharge, whereas the outdoors may be conducive to spirituality and inner peace (Long, 2000). However, individuals may avoid seeking solitude in public places like cafés because social interaction expectations are stronger in public settings (Goffman, 2008; Long, 2000). We therefore expect that individuals seeking solitude are more likely to be at home or outdoors as compared to other places.     54   3.1.2 Solitude-seeking and affective experiences Solitude, defined by the absence of social interaction, is an objective state that may be linked to specific feelings; likewise, solitude-seeking may be associated with distinct affective states (Burger, 1995; Larson, 1990). Research examining time-varying social context-affect associations across the adult lifespan has shown that moments spent alone are associated with elevated negative affect and loneliness, and lower positive affect, compared to moments spent with others (Chui et al., 2014; Larson, 1990; Pauly et al., 2017). Experimental work has linked solitude with a reduction in high arousal affective states (Nguyen et al., 2017), whereas time-sampling work has shown an association with increased low arousal affect (e.g. Pauly et al., 2017). Hence, both valence and arousal matter, but results are mixed. Previous research has further shown that individuals seem to seek solitude when feeling anxious, sad, or low in energy, potentially as a means of coping (Brown, 1992; Long & Averill, 2003). Importantly, research on the affective correlates of solitude desire is sparse; we address this research gap by examining the affective correlates of everyday solitude-seeking. Specifically, we expect that, like solitude itself, solitude-seeking is associated with lower concurrent positive affect and elevated negative affect, relative to social interaction-seeking. Given the particular relevance of loneliness to time in solitude, we also explore associations between solitude-seeking and momentary loneliness.  3.1.3 Solitude-seeking in midlife and older adulthood Solitude is particularly common in older adulthood (Larson et al., 1985). Interestingly, older adults experience solitude more positively than middle-aged or young adults - as indicated by less pronounced decreases in high arousal positive affect and less pronounced increases in low arousal negative affect and loneliness when alone (Chui et al., 2014; Larson et al., 1985; Larson, 1990; Lang & Baltes, 1997; Pauly et al., 2017). This age difference may be due, at least   55   in part, to older adults’ social preferences and increased emotion-regulation abilities. Prominent aging models like socioemotional selectivity theory posit that older adults strive to optimize their well-being by selectively focusing on emotionally meaningful social interactions (Carstensen, Fung, & Charles, 2003). Consequently, older adults may proactively seek out solitude to ‘escape’ meaningless or unpleasant social situations. Older adults may also have fewer work-related constraints on their time than middle-aged adults, allowing them more freedom to decide when and where to seek solitude. Such life-phase differences might make it more likely that older adults seek solitude for leisure (and hence enjoy this experience), and that middle-aged adults seek solitude to focus on work (which may be less enjoyable). We expect that older adults may be more likely to actively seek out locations that allow them to have a moment to themselves when they desire it, and that they may also experience this solitude-seeking more positively, than middle-aged adults.  3.1.4 Current study This study examined situational (location) and affective correlates of solitude desire in midlife and older adulthood. Extending previous research using retrospective self-reports of solitude-seeking in younger samples (Chua & Koestner, 2008; Long et al., 2003), we used up to 30 repeated daily life assessments over 10 days (‘time-sampling’, Bolger et al., 2003; Hoppmann & Riediger, 2009) to capture everyday solitude-seeking in a sample of middle-aged and older adults. This design enabled us to examine naturally-occurring solitude-seeking as participants went about their daily routines in their own environments and to minimize self-report biases arising from retrospective designs (Bolger et al., 2003). We expected that participants would seek solitude at home or outside more than in other places. We also hypothesized that solitude-seeking would be associated with increased negative affect and decreased positive affect,   56   compared to times when individuals desired social interaction. Given evidence of the importance of affect arousal level as well as valence, models distinguished between high and low arousal forms of positive and negative affect, as well as loneliness, although no arousal-specific hypotheses were developed. We also expected age differences in location and affective correlates of solitude-seeking. Specifically, we expected older adults to be more likely to be at home or outdoors when seeking solitude than middle-aged adults. Similarly, we expected older adults to show weaker solitude desire-affect associations (less pronounced decreases in positive affect, and less pronounced increases in negative affect) than middle-aged adults.   3.2 Method 3.2.1 Participants  The sample consisted of 100 community-dwelling adults aged 50-85 years (M = 67.0, SD = 8.7) from Metro Vancouver: 64% female, 72% post-secondary educated, 76% retired, 57% in a relationship, 56% of East Asian heritage, 36% of European heritage, 8% of other/mixed heritage, and 65% having little or no experience using tablets. Participants reported good health (M = 3.2 on 5-point scales). Six additional participants did not complete the study due to time constraints (4) or device difficulties (2). They were less likely to have post-secondary education (Χ2(1) = 5.58, p = .018). Two participants’ data were lost due to technical issues. Participants received up to CAD $100 or an iPad mini. The study was approved by the University of British Columbia Behavioural Research Ethics Board.  3.2.2 Procedure  This study was part of a larger project on social engagement and well-being in old age.   57   Participants completed a baseline session, a time-sampling phase, an exit session, and a 6-month follow-up. At baseline, participants completed training on using tablets and individual difference measures (detailed below). For the next 10 days (time-sampling phase), participants were prompted three times daily to report their current affect, location, activities, and actual and desired social context using an iPad mini app (iDialogPad; G. Mutz, Cologne, Germany). Everyday surveys were scheduled in the morning, afternoon, and evening (with a minimum of 4 hours between assessments) at times that avoided conflicts with participants’ predetermined commitments (e.g. work, appointments). Participants were beeped for each assessment but were able to open and complete questionnaires at any time. Each participant provided valid data for an average of 32.0 sampling occasions (SD = 10.1, range = 10-71). Several participants provided data beyond the 10-day period, which we included in the reported analyses; retaining only the measurement points completed during the 10-day period (26.3 measurement points per participant, on average) did not change the reported findings. Participants also occasionally provided data between scheduled beeps. In cases when two questionnaires were completed within a 90-minute period (137 questionnaires, 4.1%), we discarded both questionnaires to omit any data that may not reflect momentary experiences (e.g. participants “correcting” their responses or making up for missed questionnaires). At the exit session, participants completed additional measures, including study feedback. Participants considered the 10-day time-sampling phase to be typical of their everyday lives (M = 3.5 on a 5-point scale) and perceived it as neither interfering with their routines (M = 1.8/5) nor changing their behaviour (M = 1.7/5). Six months later, participants attended another session to complete measures including perceived social status. Data was collected year-round (August 2014-December 2015). All materials were translated into Chinese and independently backward-translated for verification. We used   58   previously-validated Chinese measures of affective states (Tsai et al., 2006) to minimize interpretation differences. Fifty-seven percent of participants completed the study in English, 28% in Mandarin, 15% in Cantonese.   3.2.3 Measures Current affect. Nine items assessed current affective states on 100-point scales (0 = “Not at all”, 100 = “Very much”). Items covered positively and negatively valenced high and low arousal states (Tsai et al., 2006): high arousal positive affect (“happy”, “excited”, M = 54.3, SD = 19.6), low arousal positive affect (“calm”, “satisfied”, M = 67.8, SD = 19.8), high arousal negative affect (“anxious”, “irritated”, M = 23.3, SD = 21.5), low arousal negative affect (“sad”, “tired”, M = 29.9, SD = 21.4), and loneliness (“lonely”, M = 20.7, SD = 23.0; Russell, 1996). Current location. Participants were able to choose between six current location options: “outside”, “home”, “public building”, “other person’s home”, “traveling”, “other”. Participants reported being outside 7% of the time and at home 79% of the time. Current activities. Participants reported their current activities by selecting one or more activities from the following eight options: “social activity”, “physical activity”, “passive leisure”, “cognitive activity”, “self-care or health care”, “volunteering”, “work”, “other”. Participants reported working 9% and passive leisure 33% of the time. Passive leisure was defined as leisure that did not involve any physical activity (e.g. reading, relaxing). Passive leisure activities were reported most often, followed by self-care/health care activities (26% of the time), cognitive (23%), physical (21%), and social (20%); volunteering activities were reported 5% of the time.  Current social context. Participants indicated their current social context by selecting one   59   of three options: (a) interacting with someone (29.8% of sampling occasions), (b) others nearby but not interacting (26.3% of occasions), or (c) alone (43.9% of occasions; McAdams & Constantian, 1983). Occasions when participants selected (b) or (c) were coded as solitude (the absence of social interaction), and occasions when participants selected (a) as non-solitude. Solitude occasions when “social activity” had been reported for the current activities measure (231 instances) were then re-coded as non-solitude, to account for times when participants might have been chatting online or on the phone. Current desire for solitude. Participants were again presented with the three social context options above and were asked to indicate, “which of these situations would you most like to be in?” (McAdams & Constantian, 1983). Of the 3195 sampling occasions, 33.1% were times when participants desired social interaction (option a), 25.7% when they wanted others nearby but no interaction (option b), and 41.2% when they wanted to be alone (option c). Occasions when participants chose options (b) or (c) were coded as desire for solitude, and occasions when they chose option (a) were coded as desire not to be in solitude. Covariates. We controlled for demographic variables (age, gender, ethnicity, education, retirement, and relationship status) and perceived social status (Adler & Stewart, 2007). Survey time was included as a time-varying covariate to control for time of day effects.  3.2.4 Statistical analyses   We used multilevel modeling (R lme4 package; Bates, Mächler, Bolker, & Walker, 2015) to account for the hierarchical data structure (momentary assessments nested within people). Logistic models were used for the two dummy-coded location outcomes: currently at home and currently outside. Linear models were used for the five affect outcomes: high arousal   60   positive affect, low arousal positive affect, high arousal negative affect, low arousal negative affect, and loneliness.  Models included current solitude, solitude desire, work activity, passive leisure activity, time, and time squared at level 1 (momentary level), and person-averages of current solitude, solitude desire, work, and passive leisure at level 2 (person level). We included cross-level interactions of current solitude and current solitude desire with age, and level 1 interactions of current solitude desire with current working, passive leisure, time, and time squared. Several level 2 covariates were included. Further model details are provided in Appendix D.  3.3 Results 3.3.1 Descriptive statistics Participants completed 3195 time-sampling assessments, 63.0% of which were solitude occasions (n = 2013; M = 19.5 per participant, SD = 9.7, range = 1-69). Most solitude (85.8%) was desired (wanting to be alone: 55.3%, wanting others nearby without interaction: 30.5%). Hence, solitude was a common phenomenon in this sample, and usually occurred by choice. Solitude and solitude-seeking were most likely to occur in the early morning (before 7am) or evening (after 6pm). When in solitude, participants were less likely to be outside, more likely to be at home, and more likely engaged in passive leisure, and they reported lower levels of high arousal positive and negative affect (Appendix D, Table D-1). These same patterns emerged for solitude desire. Solitude desire was also associated with reduced low arousal negative affect and loneliness. At the person level, average time in solitude and desire for solitude were positively correlated (Appendix D, Table D-2). Individuals who reported more solitude and more solitude desire spent more time at home and less time outside. Solitude desire was negatively associated   61   with loneliness. Age was not significantly associated with solitude or desire for solitude, but older adults spent more time at home and less time working, and they reported lower levels of low arousal negative affect than middle-aged adults. See Appendix D, Table D-3 for age group base rates.    3.3.2 Solitude-seeking and situation selection in midlife and older adulthood  We first examined time-varying solitude desire-situational context associations and potential age differences therein. Models predicted log-odds of being in certain locations (see Table 3-1). Current solitude (as compared to non-solitude) was associated with 5.1 times greater odds of being at home and 3.2 times greater odds of not being outside at that moment.  As expected, currently desiring solitude was also associated with being at home, specifically, a 1.7 times greater likelihood of being at home, compared to times when desiring social interaction. This solitude desire-home association was most pronounced at the beginning and end of each day (as indicated by significant linear and quadratic time effects). It was also moderated by age. Age was treated as a continuous variable in analyses; for illustrative purposes, Figure 3-1 shows simple slopes for older adults (individuals 1 SD above the mean age, 75.8 years) and middle-aged adults (individuals 1 SD below the mean age, 58.4 years). Simple slopes analyses (Figure 3-1a) revealed that older adults were more likely to be at home when desiring solitude (b simple slope = 0.83, SE = 0.24, 95% CI [0.35, 1.31]); this was not the case for middle-aged adults (b simple slope = 0.05, SE = 0.21, 95% CI [-0.36, 0.46]). There was no overall association between solitude desire and being outside. However, there was a cross-level interaction between age and momentary solitude desire, with a significance value of .05. As illustrated in Figure 3-1b, older adults desiring solitude were less likely to be outside (b simple   62   slope = -0.97, SE = 0.44, 95% CI [-1.82, -0.12]); this was not the case for middle-aged adults (b simple slope = 0.21, SE = 0.22, 95% CI [-0.23, 0.65]). To summarize, older adults who desired solitude were more likely to be at home and less likely to be outside. There were no such solitude desire-location associations in middle-aged adults.  We compared model deviances for the full models and the reduced models that excluded current and person-average solitude desire. Solitude desire improved the fit of the “home” model, as indicated by a significant deviance reduction (Table 3-1).  Table 3-1: Logistic multilevel models: Current location by solitude desire. N = 95 individuals, n = 3058 momentary assessments  Log-odds of being home Log-odds of being outside LEVEL 1          Current solitude  1.64 (0.20)*** [1.25, 2.04] -1.16 (0.38)** [-1.90, -0.42]    Current solitude desire  0.52 (0.18)** [0.16, 0.87] -0.54 (0.37) [-1.26, 0.18]    Currently working -0.08 (0.22) [-0.51, 0.36] -0.16 (0.33) [-0.80, 0.49]    Current passive leisure  0.90 (0.15)*** [0.60, 1.20] -0.61 (0.23)** [-1.07, -0.15]    Time -0.70 (0.08)*** [-0.85, -0.54]  0.70 (0.12)*** [0.46, 0.94]    Time squared  0.03 (0.00)*** [0.03, 0.04] -0.03 (0.01)*** [-0.04, -0.02] LEVEL 2           Intercept  2.13 (0.15)*** [1.84, 2.42] -3.78 (0.23)*** [-4.24, -3.33]    Person-average solitude  0.00 (0.01) [-0.02, 0.01]  0.01 (0.01) [-0.01, 0.03]    Person-average solitude desire  0.01 (0.01) [-0.01, 0.02] -0.01 (0.01)† [-0.03, 0.00]    Person-average working  0.00 (0.01) [-0.02, 0.02]  0.01 (0.01) [-0.01, 0.03]    Person-average passive leisure -0.01 (0.01) [-0.02, 0.00]  0.00 (0.01) [-0.01, 0.01]    Age  0.03 (0.02) [-0.01, 0.07] -0.02 (0.02) [-0.07, 0.03]    Ethnicity  -0.27 (0.30) [-0.86, 0.32] -0.22 (0.37) [-0.94, 0.50]    Education   0.71 (0.29)* [0.14, 1.28] -0.80 (0.35)* [-1.49, -0.11]    Gender  0.16 (0.28) [-0.39, 0.71] -0.58 (0.33)† [-1.22, 0.07]    Retirement status  0.44 (0.37) [-0.28, 1.17]  0.33 (0.46) [-0.57, 1.22]    Relationship status  0.58 (0.29)* [0.02, 1.15] -0.49 (0.36) [-1.19, 0.21]    Perceived social status -0.02 (0.10) [-0.21, 0.17]  0.00 (0.12) [-0.23, 0.23] INTERACTIONS        Age x Current solitude  0.01 (0.02) [-0.03, 0.05]  0.00 (0.03) [-0.06, 0.07]   63     Age x Current solitude desire  0.04 (0.02)* [0.00, 0.07] -0.05 (0.03)† [-0.10, 0.00]   Working x Current sol. desire  0.39 (0.38) [-0.34, 1.13]  0.06 (0.53) [-0.98, 1.10]   Pass. leis. x Current sol. desire -0.17 (0.28) [-0.73, 0.38] -0.02 (0.42) [-0.85, 0.81]   Time x Current solitude desire -0.40 (0.14)** [-0.69, -0.12]  0.09 (0.22) [-0.35, 0.53]   Time squared x Current sol. desire  0.02 (0.01)** [0.01, 0.03]  0.00 (0.01) [-0.02, 0.02] DEVIANCE REDUCTION   Χ2(10) = 25.06**     Χ2(10) = 14.09 Note for Table 3-1 and Table 3-2. Current solitude, solitude desire, working, passive leisure coded 1 = reported that situation/activity, 0 = did not report situation/activity. Person-averages are percentages of beeps when participant reported that situation/activity. Time is hours since 4am; age in years; ethnicity coded 1 = European, 0 = not European; education coded 1 = some post-secondary, 0 = no post-secondary; gender coded 1 = F, 0 = M; retirement and relationship status dummy-coded; perceived social status is on a 10-point scale. All variables grand mean centered, coefficients unstandardized. Location models used maximum likelihood estimation, Laplace approximation; affect models used restricted maximum likelihood. Deviance reduction compares full models to models that exclude solitude desire. Missing data for relationship (N = 2) and perceived social status (N = 5) were multiply imputed (predictive mean matching, R mice package; Buuren & Groothuis-Oudshoom, 2011); missing data for age (N = 5) was not imputed, resulting in a final N = 95. Models without control variables (gender, ethnicity, education, retirement, relationship status, perceived social status, time, and time squared) show the same results (age x solitude desire moderations). Hence, we retain all controls in the models to demonstrate the robustness of the reported findings. 95% confidence intervals are shown. ***p < .001, **p < .01, *p < .05, †p < .1 Note for Table 3-1. We tested additional models that included current aloneness and current desire to be alone, and their interactions with age, as predictors of the two location outcomes. The inclusion of these variables accounted for the reported age x solitude desire interaction for location: home, specifically older adults seeking to be alone were more likely to be at home, but this was not the case for middle-aged adults. Individuals desiring to be alone were less likely to be outside, but age did not moderate this association. Hence, desiring aloneness, rather than desiring solitude, may be a key factor in middle-aged and older adults’ current location.    64   Figure caption. Graphs show simple slopes for interactions between current solitude desire and age in predicting location and affect outcomes. Graphs (a) and (b) show log-odds of currently being at home (a) and of currently being outside (b) when solitude is currently desired versus not desired. Positive log-odds values indicate that the participant is more likely to be in that location (at home, outside) than not in that location at that moment. Graph (c) shows momentary high arousal positive affect (on a 100-point scale) reported when solitude is currently desired versus not desired. Solid lines denote participants aged one standard deviation below the mean age (58.4 years; middle-aged adults), and dotted lines denote participants aged one standard deviation above the mean age (75.8 years; older adults). Age was measured on a continuous scale; simple slopes for the two age groups are shown for illustrative purposes only.             Figure 3-1: Associations of momentary solitude desire with current location (home, outside) and high arousal positive affect in midlife and older adulthood.    65   3.3.3 Solitude-seeking and affective experiences in midlife and older adulthood We next examined the affective correlates of momentary solitude desire, and age differences in these associations (Table 3-2). Current solitude (as compared to social interaction) was associated with lower levels of high arousal positive affect and elevated levels of loneliness. There were no main effects of current solitude desire on any affect outcomes. However, cross-level interactions indicated that currently desiring solitude (as compared to desiring social interaction) was associated with lower levels of high arousal positive affect in middle-aged adults (b simple slope = -2.45, SE = 1.11, 95% CI [-4.64, -0.27]); there was no such solitude desire-affect association in older adults (b simple slope = 1.56, SE = 1.16, 95% CI [-0.72, 3.83]; see Figure 3-1c). Findings pertaining to low arousal positive affect showed a similar pattern. For middle-aged adults, there was a marginally significant association between solitude desire and lower levels of low arousal positive affect (b simple slope = -2.00, SE = 1.19, 95% CI [-4.33, 0.34]), but this was not the case for older adults (b simple slope = 1.49, SE = 1.24, 95% CI [-0.94, 3.92]). To summarize, middle-aged adults currently desiring solitude reported lower levels of positive affect, but older adults did not. Counter to expectations, no solitude desire-age interactions emerged for high or low arousal negative affect or loneliness.   Individuals who desired solitude more overall were less lonely, and participants of East Asian heritage reported higher levels of loneliness than those of European heritage. Individuals engaged in passive leisure while solitude-seeking reported higher levels of low arousal positive affect. The inclusion of current and person-average solitude desire improved model fit for low arousal positive affect, high and low arousal negative affect, and loneliness (Table 3-2).   66   Table 3-2: Multilevel models: Current affect outcomes by solitude desire. N = 95 individuals, n = 3058 momentary assessments.   High arousal  positive affect Low arousal  positive affect High arousal  negative affect Low arousal  negative affect Loneliness LEVEL 1                   Current solitude -2.44 (0.77) [-3.95,-0.93] -0.75 (0.72) [-2.17,0.67] 0.32 (0.87) [-1.39,2.02] 0.40 (0.79) [-1.14,1.94] 2.99 (0.95) [1.13,4.86]    Current solitude desire -0.45 (0.79) [-2.01,1.10] -0.25 (0.85) [-1.92,1.41] -1.11 (1.04) [-3.15,0.92] 0.35 (0.93) [-1.48,2.18] -1.75 (1.01) [-3.72,0.22]    Currently working -1.09 (1.10) [-3.26,1.07] -0.33 (1.02) [-2.32,1.67] 3.81 (1.12) [1.61,6.01] 1.61 (1.12) [-0.59,3.81] 1.68 (1.04) [-0.37,3.73]    Current passive leisure 0.06 (0.68) [-1.26,1.39] 0.98 (0.62) [-0.24,2.20] -1.12 (0.69) [-2.46,0.23] -0.01 (0.69) [-1.35,1.34] 0.42 (0.64) [-0.83,1.67]    Time 0.94 (0.30) [0.34,1.53] 0.33 (0.28) [-0.22,0.88] 0.28 (0.31) [-0.32,0.89] -0.10 (0.31) [-0.71,0.50] 0.10 (0.29) [-0.46,0.67]    Time squared -0.05 (0.01) [-0.07,-0.02] -0.01 (0.01) [-0.04,0.01] -0.01 (0.01) [-0.04,0.01] 0.03 (0.01) [0.01,0.06] 0.00 (0.01) [-0.03,0.02] LEVEL 2               Intercept 55.01 (1.36) [52.35,57.66] 67.78 (1.49) [64.85,70.71] 24.31 (1.66) [21.05,27.57] 30.50 (1.56) [27.44,33.56] 21.32 (1.88) [17.63,25.02]    Person-average solitude 0.09 (0.09) [-0.09,0.26] -0.05 (0.10) [-0.25,0.14] 0.04 (0.11) [-0.18,0.26] 0.04 (0.10) [-0.16,0.24] 0.15 (0.13) [-0.09,0.40]    Person-average solitude desire -0.13 (0.07) [-0.27,0.01] 0.00 (0.08) [-0.17,0.16] -0.06 (0.09) [-0.24,0.12] -0.12 (0.09) [-0.29,0.05] -0.27 (0.10) [-0.48,-0.07]    Person-average working 0.15 (0.10) [-0.04,0.34] -0.09 (0.11) [-0.31,0.12] 0.06 (0.12) [-0.18,0.29] 0.01 (0.11) [-0.21,0.23] 0.07 (0.14) [-0.20,0.33]    Person-average passive leisure -0.03 (0.06) [-0.15,0.10] 0.01 (0.07) [-0.12,0.15] -0.03 (0.08) [-0.18,0.12] 0.02 (0.07) [-0.12,0.16] -0.03 (0.09) [-0.20,0.14]     67      Age -0.01 (0.20) [-0.40,0.37] -0.09 (0.22) [-0.52,0.34] -0.10 (0.24) [-0.58,0.38] -0.21 (0.23) [-0.65,0.24] -0.08 (0.28) [-0.62,0.46]    Ethnicity  -3.98 (3.24) [-10.33,2.37] 4.22 (3.60) [-2.84,11.28] -6.36 (4.02) [-14.25,1.53] -1.00 (3.77) [-8.38,6.39] -10.72 (4.55) [-19.63,-1.80]    Education  -1.01 (3.14) [-7.18,5.15] -7.05 (3.50) [-13.92,-0.18] 1.61 (3.91) [-6.07,9.28] 1.38 (3.66) [-5.80,8.55] -1.74 (4.41) [-10.39,6.92]    Gender 2.33 (2.98) [-3.52,8.17] -1.19 (3.33) [-7.72,5.33] 4.13 (3.72) [-3.16,11.41] 6.53 (3.47) [-0.28,13.34] 0.16 (4.18) [-8.05,8.36]    Retirement status 7.48 (4.03) [-0.42,15.37] 7.52 (4.53) [-1.37,16.41] -3.80 (5.03) [-13.67,6.07] -4.91 (4.71) [-14.15,4.33] -6.76 (5.66) [-17.86,4.35]    Relationship status 3.39 (3.10) [-2.68,9.46] 3.99 (3.43) [-2.74,10.72] -3.56 (3.82) [-11.04,3.93] -4.60 (3.57) [-11.60,2.40] -6.79 (4.31) [-15.25,1.67]    Perceived social status 1.59 (1.04) [-0.45,3.63] 2.41 (1.18) [0.09,4.72] -1.63 (1.29) [-4.16,0.90] -1.19 (1.21) [-3.57,1.19] -0.53 (1.52) [-3.51,2.45] INTERACTIONS             Age x Current solitude  -0.03 (0.09) [-0.20,0.14] -0.03 (0.08) [-0.19,0.13] 0.02 (0.10) [-0.18,0.21] 0.07 (0.09) [-0.10,0.24] -0.09 (0.11) [-0.30,0.13] Age x Current solitude desire 0.23 (0.09) [0.05,0.41] 0.20 (0.10) [0.00,0.40] -0.11 (0.12) [-0.34,0.13] -0.07 (0.11) [-0.28,0.15] -0.09 (0.12) [-0.32,0.14] Work x Current solitude desire -0.94 (2.05) [-4.96,3.07] -1.08 (1.94) [-4.88,2.73] 3.86 (2.16) [-0.38,8.10] 1.11 (2.15) [-3.10,5.31] 0.20 (2.03) [-3.77,4.17] Passive leisure x Curr sol. desire 0.69 (1.33) [-1.92,3.29] 2.89 (1.25) [0.45,5.33] -0.32 (1.38) [-3.03,2.40] -1.64 (1.37) [-4.34,1.05] -1.93 (1.30) [-4.48,0.61] Time x Current solitude desire -0.39 (0.65) [-1.66,0.89] 0.04 (0.60) [-1.13,1.22] 0.47 (0.66) [-0.83,1.77] 1.25 (0.66) [-0.05,2.55] -0.09 (0.62) [-1.30,1.12] Time2 x Current solitude desire 0.01 (0.03) [-0.04,0.06] -0.01 (0.03) [-0.06,0.04] -0.02 (0.03) [-0.07,0.04] -0.03 (0.03) [-0.09,0.02] 0.01 (0.03) [-0.04,0.06] DEVIANCE REDUCTION Χ2(10) = 15.26 Χ2(10) = 22.45* Χ2(10) = 21.04* Χ2(10) = 28.98** Χ2(10) = 42.76***    68   Note for Table 3-2. Bolded values are significant at p < .05. We tested additional models that included current aloneness and current desire to be alone, and their interactions with age, as predictors of the five affect outcomes. None of the age interactions were significant for any of the affect outcomes, nor did the inclusion of these variables alter the reported age x current solitude desire interactions. Hence, we report the more parsimonious models that exclude current aloneness and desire to be alone.  69   3.4 Discussion This study examined situational and affective correlates of everyday solitude-seeking in middle-aged and older adults. Counter to expectations, older adults did not spend more time in solitude than middle-aged adults, nor did they desire it more. In line with general expectations, older adults (and not middle-aged adults) were more likely to either be at home or not outdoors when seeking solitude. Furthermore, whereas middle-aged adults experienced a dip in high and low arousal positive affect when seeking solitude, older adults did not. Findings are discussed in the context of the lifespan developmental and emotion regulation literatures.  3.4.1 Solitude-seeking in midlife and older adulthood Solitude (the absence of social interaction) was common in our sample and typically happened by participants’ own choosing. Participants were in solitude at about two-thirds of the beeps, a rate similar to that reported in previous research in a comparable age group (Larson et al., 1985). Notably, solitude occurred primarily by choice, extending previous evidence from younger samples (Chua & Koestner, 2008; McAdams & Constantian, 1983). Unlike previous research (Klumb, 2004; Larson, 1990; Pauly et al., 2017), there was no evidence of age-related differences in solitude or solitude-seeking. This could be due to the sample’s restricted age range (50-85 years); participants may have shared more experiences (e.g. empty nest) than lifespan samples in other studies. Moreover, as participants were required to use tablets, they were relatively healthy, with little physical or cognitive impairment. They may also have had more control over social contexts than lower-functioning older adult samples (who may be at home more due to common age-related conditions or mobility limitations). Finally, our definition of solitude differs slightly from those used in previous time-sampling research   70   (Klumb, 2004; Larson, 1990; Pauly et al., 2017) in that it does not necessitate the physical absence of other people. It may be that time alone increases with age but time in solitude (the absence of social interaction) does not.  3.4.2 Solitude-seeking and situation selection in midlife and older adulthood  Where do individuals go when they want time to themselves? Solitude and solitude-seeking varied by time of day, peaking in the morning and evening, when individuals were typically at home. Middle-aged and older adults were more likely to be at home during solitude, and older adults were also more likely to be at home when desiring solitude. These findings dovetail with previous research indicating that home is a place where individuals seek privacy and social reprieve (Brown, 1992; Long et al., 2003). We expected that the outdoors would also be a prime location for solitude and solitude-seeking, but found the opposite. When in solitude, middle-aged and older adults were less likely to be outdoors. Older adults seeking solitude were also less likely to be outdoors than when not seeking solitude. This divergence between our findings and previous evidence may be due, in part, to age differences. Previous findings on solitude-seeking in nature (e.g. Long et al., 2003) are based exclusively on university student samples. Given that older adults are more likely to live alone than young adults (Statistics Canada, 2012), going outside may be an important avenue for them to connect rather than to be alone. This finding reinforces the need for a lifespan developmental approach to solitude and solitude-seeking, as findings from student samples may not generalize to later life phases. All participants lived in an urban environment close to both quiet (e.g. parks) and congested public spaces (Chaudhury et al., 2011); further research is needed to examine how neighbourhood characteristics (e.g. urban versus rural) shape individuals’ propensity to seek solitude outdoors.   71    Older adults were more likely to be at home or indoors when seeking solitude, but middle-aged adults showed no such location-specificity in solitude-seeking. This may be because older adults spent more time at home and less time working than middle-aged adults, in line with the idea that older adults might have more control over their social contexts due to reduced work obligations. Indeed, previous research suggests that older adults feel more autonomous and in control when in solitude (Larson et al., 1985; Lang & Baltes, 1997). Such age differences in location-seeking may also reflect an underexplored implication of socioemotional selectivity theory (Carstensen et al., 2003): Compared to middle-aged adults, older adults may have less tolerance for social interactions that do not serve their socio-emotional goals. Hence, when solitude seems like a better option than socializing, older adults may be more proactive in seeking solitary spaces.  3.4.3 Solitude-seeking and affective experiences in midlife and older adulthood  Overall, participants reported reduced high arousal positive affect and increased loneliness during solitude, compared to when interacting with others. This aligns with previous research linking solitude with lower levels of happiness, cheerfulness, and high arousal positive affect, and elevated levels of loneliness (Larson, 1990; Nguyen et al., 2017; Pauly et al., 2017).  Desiring solitude (as compared to desiring social interaction) was associated with lower levels of high and low arousal positive affect – but only for middle-aged, not older, adults. Whereas older adults’ positive affect tended to be maintained when desiring solitude, middle-aged adults tended to show a decrease in positive affect. This aligns with previous findings showing that, compared to their younger counterparts, older adults show more emotional stability (Röcke, Li, & Smith, 2009) and more motivation to actively maintain their positive affect in   72   daily life (Riediger, Schmiedek, Wagner, & Lindenberger, 2009). We speculate that positive affect maintenance while solitude-seeking might also reflect older adults’ enhanced emotion regulation skills (Isaacowitz & Blanchard-Fields, 2012). Social situation selection may be a particularly effective emotion regulation strategy for older adults because it capitalizes on cognitive and social resources that remain intact in old age (Urry & Gross, 2010). Life phase might also help explain why only middle-aged adults tended to experience a positive affect dip when solitude-seeking. Due to greater work-related obligations and time spent away from home, middle-aged adults might be less able than older adults to escape to desired solitary locations, making solitude-seeking less pleasant. Notably, both middle-aged and older adults experienced higher levels of low arousal positive affect when solitude-seeking coincided with passive leisure, suggesting that seeking solitude for relaxation may be particularly beneficial.   Solitude desire-age interactions were linked with positive affect only; we found no similar associations with negative affect or loneliness. This may be because older adults pay more attention to positive than to negative information (Mather & Carstensen, 2005) and, hence, may be more attuned to positive than negative aspects of solitude-seeking. Our finding also underscores the importance of disentangling positive from negative affect using unipolar affect scales (Pauly et al., 2017) rather than bipolar scales (such as “happy—sad”), which are common in research on social context and solitude (e.g. Larson et al., 1985). A key strength of this study is its inclusion of participants of diverse backgrounds. The study was offered in both Chinese and English because nearly 30% of the Vancouver population is of East Asian heritage, with ~15% speaking a Chinese dialect as their primary language. We also conducted sessions in the community to reduce participation barriers for individuals who are less well-represented in aging research, such as recent immigrants and those of lower   73   socioeconomic status. Indeed, 68% of participants were born outside Canada and approximately 50% had an annual income below the governmental low-income threshold. Individual difference associations with loneliness also emerged. Individuals who desired solitude more often overall tended to feel less lonely, underscoring the point that solitude-seeking differs from loneliness. Furthermore, individuals of East Asian heritage tended to feel lonelier than those of European heritage. As many East Asian participants were immigrants, their higher loneliness might reflect lower social integration or being unaccustomed to societal norms in individualistic North American culture, having come from more collectivistic cultures (Stewart et al., 2011; Triandis, 1988). However, the present study does not allow us to disentangle cultural influences from immigration influences on loneliness (Chang, 1996; Tsai, 2007). To better understand solitude across cultures, the work presented in Chapter 4 examines solitude experiences of older adults living Canada and China, including those who have immigrated and those who are aging in place.  3.4.4 Limitations Solitude-seeking and affect were assessed concurrently in the present study; our findings therefore do not allow causal inferences. Conceptually, it makes sense that solitude-seeking leads to decreases in positive affect, but it is also conceivable that decreases in positive affect might motivate individuals to seek solitude (Brown, 1992; Lay et al., 2018). In this study, most solitude-seeking (81%) happened when participants were already in solitude, hence, it seems more likely that affective states were a response, rather than an antecedent, to one’s current situation. One way to test this would be to examine lagged effects. Our sampling frame (4+ hours between assessments) was not fine-grained enough to examine such processes; situation   74   selection and affective change in response to solitude-seeking may occur within minutes, and affective responses may hence dissipate before the next assessment.  We chose our sampling frame to capture snapshots of daily life solitude-seeking while minimizing participant burden. This design resulted in high compliance: participants completed 88% of scheduled assessments (a rate similar to those in previous time-sampling research; Green, Rafaeli, Bolger, Shrout, & Reis, 2006), and 75% of these within 90 minutes of being beeped. Participant-based adjustments in the timing of beeps could have resulted in over-sampling instances of solitude-seeking. However, solitude rates were similar to those reported in previous time-sampling studies using random sampling designs with middle-aged and older adults (Larson et al., 1982, 1985), which gives us confidence that we did in fact capture naturally-occurring solitude-seeking. Solitude-seeking may be less appropriate or normative in certain situations (e.g. during a social gathering) or certain phases of life (e.g. when raising a young family), and such norms may reduce participants’ willingness to report desiring solitude in those circumstances. Demand characteristics might also encourage participants to simply report desiring their current social situation (whatever it may be), thereby potentially conflating solitude with solitude-seeking. Future research examining the affective correlates of solitude-seeking might disentangle such self-presentation effects from genuine solitude-seeking by, for example, manipulating social motivations (e.g. social avoidance) in different social contexts (e.g. library versus restaurant; Epley & Schroeder, 2014; Nikitin & Freund, 2010).  3.4.5 Conclusions and future directions   Solitude may not always be negative and can provide space for emotional renewal (Long   75   & Averill, 2003); the present study suggests that older adults may be particularly likely to benefit from solitude-seeking. Unlike middle-aged adults, older adults reported being in locations conducive to solitude, with no decrease in positive affect, when seeking solitude. Findings may reflect enhanced situation selection and emotion regulatory capacities in old age (Urry & Gross, 2010). Further research is needed to unpack underlying causal mechanisms. For example, one could manipulate solitude-seeking by having people pursue social interaction or solitude while riding the bus (Epley & Schroeder, 2014) and measure subsequent affective change. Future research could also examine how personality variables (e.g. Extraversion; Burger, 1995) may shape solitude-seeking experiences. More work is needed to further explore potential benefits of everyday solitude-seeking, particularly in late life.     76   Chapter 4 Solitude in context: On the role of culture, immigration, acculturation, and solitude desire in the experience of time to oneself (STUDY 3)  Note: A version of this chapter will be submitted for publication as: Lay, J. C., Fung, H. H., Jiang, D., Mahmood, A., Graf, P., & Hoppmann, C. A. (in preparation for International Journal of Psychology). Solitude in context: On the role of culture, immigration, acculturation, and solitude desire in the experience of time to oneself.   4.1 Introduction   What does solitude feel like? Time in solitude (the absence of social interaction) is ubiquitous in daily life and it is often associated with feelings of loneliness (Hawkley, & Cacioppo, 2010; Larson, 1990; Jylhä & Saarenheimo, 2010). However, most of the literature examining everyday solitude and loneliness focuses exclusively on Western samples; little is known about the extent to which solitude-loneliness links may be shaped by cultural contextual factors (Averill & Sundararajan, 2014; Jiang et al., in press; Lay et al., 2018; Long et al., 2007). The aim of the present study was to disentangle the roles of cultural heritage, local culture (sample location), immigration, acculturation, and desire for solitude on time-varying associations between solitude and loneliness. This study was based in Vancouver, Canada, and was extended to Hong Kong, SAR. East Asians are the largest ethnic minority in Vancouver, and differences between East Asian and Western cultural values may be particularly informative in understanding solitude experiences (Averill & Sundararajan, 2014; Jiang et al., in press). Hence, we collected data from 65 immigrant and 30 Canadian-born adults aged 50+ living in   77   Vancouver, and 29 immigrant and 27 Hong Kong-born adults aged 50+ living in Hong Kong (M age 68.7 years, range 51-85). Participants completed 30 ecological momentary assessments of current and desired social context and affective experiences over a 10-day period.   Due to normative age-related changes, including retirement and reduced mobility, older adults spend a greater proportion of their time in solitude (Larson, 1990). This solitude brings an increased risk of loneliness, with associated physical and mental health consequences (Chen et al., 2014; Jylhä & Saarenheimo, 2010; Perissinotto et al., 2012). We and others (e.g. Averill & Sundararajan, 2014; Lay et al., 2018; Pauly et al., 2017) have pointed to the need to distinguish between solitude (defined as the absence of social interaction, a situation without any particular emotional connotations; Larson, 1990), and loneliness (defined as a perceived lack of closeness with others, Perlman & Peplau, 1981). Research using repeated daily life assessments as individuals engage in their everyday lives (time-sampling; Bolger et al., 2003; Hoppmann & Riediger, 2009) has shown that time alone (as compared to time with others) is associated with higher levels of loneliness and other negative affective states, across the adult lifespan and into old age (Chui et al., 2014; Larson, 1990; Larson et al., 1982, 1985; Pauly et al., 2017). However, depending on characteristics of the person and of the situation, there are also circumstances in which solitude may be associated with low-arousal positive affect and emotional renewal, demonstrating that solitude does not inevitably lead to loneliness (Larson, 1990; Larson et al., 1982; Nguyen et al., 2017; Pauly et al., 2017).    4.1.1 Culture and solitude-loneliness associations   Solitude needs to be understood within the cultural context in which it occurs (Long et al., 2007). Previous time-sampling work examining everyday solitude-loneliness associations (e.g. Larson, 1990) has exclusively focused on Western samples, and it unclear whether this   78   association generalizes to other cultures (Averill & Sundararajan, 2014; Jiang et al., in press; Lay et al., 2018). For example, in Chinese culture as compared to North American culture, solitude may be more highly valued for providing freedom from social regulation and space for self-cultivation, in line with principles of Taoism and a tradition of eremitism (Averill & Sundararajan, 2014; Jiang et al., in press; Koch, 1989; Long & Averill, 2003; Wang, 2006). Similarly, self-reflection is also a more important value in East Asian as compared to North American cultures (Heine et al., 1999). Furthermore, compared to North American individuals, Chinese individuals score higher in measures of introversion and express a preference for low arousal activities, such as meditation and yoga, whereas North Americans are higher in extraversion and prefer high-arousal activities, such as parties and social gatherings (Mooradian and Swan, 2006; Tsai, 2007). Because solitude provides an ideal environment for self-reflection (Long & Averill, 2003) and low arousal activities, solitude may feel less lonely for Chinese individuals as compared to North Americans (Jiang et al., in press; Wang, 2006). For the same reasons, momentary solitude-loneliness associations may be reduced for individuals living in Hong Kong (who have had greater exposure to Chinese culture), as compared to those living in Vancouver (who have had greater exposure to North American culture; Jiang et al., in press).  4.1.2 Immigration and solitude-loneliness associations   Solitude may feel particularly lonely for middle-aged and older adults who have immigrated to a new country because immigration often involves living away from family. For various socio-political reasons (e.g. the handover of Hong Kong to China, or emigration to seek employment; Chen et al., 2014; Wu & Penning, 2015) one quarter of the total population of British Columbia emigrated from Hong Kong or other East Asian cities, with the vast majority living in Vancouver (Statistics Canada, 2017). At the same time, due in part to recent   79   urbanization trends (Chen et al., 2014), many individuals living in Hong Kong emigrated from mainland China or other parts of East Asia (Hong Kong Census and Statistics Department, 2017). A difference between immigrants living in Vancouver and in Hong Kong is that Vancouver immigrants mainly come from a different (non-North American) culture, whereas Hong Kong immigrants mainly come from another East Asian culture. Notwithstanding such differences, however, moving to another country poses many challenges to any immigrant, including loss of social ties or reduced frequency of social contact, language and cultural barriers, discrimination, social isolation, and reductions in socioeconomic status (Barrio et al., 2008; de Jong Gierveld et al., 2015; Stewart et al., 2011; Wong, Chou, & Chow, 2012). Hence, time spent in solitude may feel lonelier for middle-aged and older adults who had immigrated, as compared to those who had been born locally.  4.1.3 Host culture acculturation and solitude-loneliness associations   Solitude need not feel lonely if an individual has formed strong connections with other people and with the society in which they live; these connections may help sustain a sense of belonging when other people are absent (Averill & Sundararajan, 2014; Koch, 1994; Long & Averill, 2003; Pauly et al., in press). There are recognized benefits to participating in the local culture and having a sense of belonging within local society (host culture acculturation; Ryder et al., 2000), as this makes it easier to navigate daily life. Indeed, previous research using cross-sectional surveys has shown that a sense of belonging to the local community or one’s country of residence is associated with reduced overall levels of loneliness among local-born and East Asian immigrant adults age 55+ living in Canada (de Jong Gierveld et al., 2015; Syed et al., 2017). There is also evidence that being married and having high-quality social relations reduce   80   concurrent associations between solitude and negative affect (Averill & Sundararajan, 2014; Koch, 1994; Larson et al., 1985; Pauly et al., in press). However, it is an open question whether such protective effects are specific to close relationships, or whether they might generalize to other social resources like feeling connected with the broader society in which one lives. We hypothesized that moments of solitude may feel less lonely for individuals who are more acculturated to the local or host culture (Ryder et al., 2000).  4.1.4 Solitude desire and solitude-loneliness associations   Beyond the characteristics of a given person (culture, immigration, acculturation), solitude experiences may also be shaped by time-varying situational factors, specifically, fluctuations in desire for solitude (Long & Averill, 2003). A growing body of research suggests that individuals across the adult lifespan sometimes choose to spend time in solitude for relaxation, contemplation, or simply to work (Burger, 1995; Larson, 1990; Lay et al., 2018; Leary et al., 2003; Long et al., 2003). Solitude that happens by choice may be a source of autonomy and self-growth, and may therefore be experienced as less lonely than involuntary solitude (Averill & Sundararajan, 2014; Chua & Koestner, 2008; Long & Averill, 2003). Desire for solitude may be particularly salient in old age. Previous work suggests that older adults often choose to spend more of their time alone, and feel less negative and more in control when alone, compared to younger and middle-aged adults (Larson, 1990; Larson et al., 1985; Pauly et al., 2017). Although research examining solitude desire has focused on North American samples, as we alluded to earlier, desire for solitude may be even stronger in East Asian cultures (Jiang et al., in press; van Zyl, Dankaert, & Guse, 2018). Overall, we expect that everyday solitude-loneliness associations will be weaker at times when that solitude is desired.    81   4.1.5 Current study   The present study examined concurrent associations between solitude and loneliness in the daily lives of middle-aged and older adults, who spend a larger proportion of their time in solitude with increasing age (Larson, 1990). We did so by collecting repeated daily life assessments of current social context, desired social context, and loneliness over the course of 10 days (~30 assessments for each of 151 participants; Bolger et al., 2003). By comparing immigrant and local-born individuals living in Vancouver and Hong Kong, the present study aimed to tease apart the roles of culture, immigration, and acculturation, while taking into account whether solitude occurred by choice. We expected that compared to moments of social interaction, moments of solitude would be associated with increased concurrent loneliness (Hypothesis 1). We further expected that this solitude-loneliness link would be weaker for individuals of East Asian heritage (Hypothesis 2) and those living in Hong Kong (Hypothesis 3). We also expected weaker solitude-loneliness associations for local-born individuals as compared to immigrants (Hypothesis 4) and for individuals higher in host culture acculturation (Hypothesis 5). Finally, we expected everyday solitude-loneliness links to be reduced at moments when solitude was desired (Hypothesis 6). Analyses controlled for demographic and social contextual factors that have been associated with loneliness in previous research, including age, gender, education, retirement status, relationship status, perceived social status, and social relationship quality (Cohen-Mansfield, Hazan, Lerman, & Shalom, 2016; Pinquart & Sorenson, 2001).  4.2 Method 4.2.1 Participants   Community-dwelling adults aged 51–85 years (M age = 68.7, SD = 7.7) in Metro Vancouver, Canada (N = 95) and in Hong Kong (N = 56) were recruited for a study on social   82   engagement and wellbeing. The Vancouver sample consisted of 30 individuals born in Canada, most of whom were of European heritage (76.7%; 16.7% of East Asian heritage, 6.7% of Indigenous heritage), and 66 individuals not born in Canada, most of whom had immigrated from East or Southeast Asian countries (81.8%; Europe 16.7%, United States 1.5%). Vancouver immigrant participants had spent an average of 30.6 years in Canada (SD = 15.9, range 2–67) ,and most had immigrated as young adults (M age at immigration = 34.0 years, SD = 19.0, range 2–73). The Hong Kong sample consisted of 27 individuals born in Hong Kong (of East Asian heritage) and 29 individuals who had immigrated from mainland China (75.9%) or from other East or Southeast Asian countries (24.1%). Hong Kong immigrant participants had spent an average of 57.6 years in Hong Kong (SD = 10.9, range 35–71), and most had immigrated as adolescents (M age at immigration = 15.6 years, SD = 11.7, range 2–42).    Participants in both samples reported good health (M = 3.1, range ~1.5–5.0, on 5-point scales of subjective physical, mental, and overall wellbeing). Further sociodemographic characteristics of the two samples are provided in Table 4-1. Ten additional participants did not complete the study due to time constraints (7) or difficulties with the electronic assessments (3). They were approximately 7 years older than the final sample (Welch’s t(6.82) = 2.76, p = .03). A further 7 participants were excluded due to missing data for immigration status (4) or acculturation (1), or due to technical difficulties resulting in data loss (2). Vancouver participants received up to CAD 100 or an iPad mini, and Hong Kong participants received up to HKD 1000 (approximately CAD 160). The study was approved by the respective institutional research ethics boards.     Table 4-1: Participant sample characteristics and person-level variable intercorrelations   Van sample (N = 95)  M (SD)  HK sample (N = 56) M (SD)  Sample  difference  (Welch’s t or χ2) 2 3 4 5 6 7 8 9 10 11 12 13 1. Age 67.19 (8.65) 71.02 (5.08) t(143.6) = -3.37,  p < .001 -.13 -.05 .51 .03 .03 -.02 -.07 .04 .01 .02 -.03 -.06 2. Gender (female) 65.26% 60.71% χ2(1) = 0.15, p = .67  -.09 .12 -.26 .01 -.01 -.02 .12 -.05 -.04 .00 .05 3. Some post-secondary education 72.63% 23.21% χ2(1) = 32.71, p < .001   -.11 -.11 .22 .06 -.11 .00 -.40 .21 .17 -.21 4. Retired 76.84% 96.43% χ2(1) = 8.70, p = .00    .01 -.03 -.03 -.11 .19 .09 -.01 -.02 -.14 5. In a relationship 60.22% 80.36% χ2(1) = 5.60, p = .02     .02 .19 .18 .01 .32 -.33 -.25 -.03 6. Perceived social status 5.55 (1.47) 5.10 (1.20) t(133.9) = 2.06,  p = .04      .28 -.07 .16 -.17 .03 -.05 -.16 7. Social relationship quality 3.67 (0.68) 3.67 (0.49) t(142.1) = 0.01,  p = .99       .03 .24 .04 -.14 -.10 -.27 8. Immigrants 68.42% 51.79% χ2(1) = 3.47, p = .06        -.21 .35 -.05 .00 .23 9. Acculturation to host culture 6.15 (1.42) 6.86 (0.82) t(149.0) = -3.87,  p < .001         .01 -.04 -.03 -.19 10. East or Southeast Asian heritage 61.05% 100.00% χ2(1) = 26.82, p < .001          -.24 -.12 .22 11. Solitude (% of beeps) 63.54 (22.57) 46.82 (25.73) t(103.7) = 4.04,  p < .001           .76 .07 12. Solitude desire (% of beeps) 68.59 (25.68) 50.30 (28.22) t(106.9) = 3.98,  p < .001            -.19 13. Mean loneliness  20.03 (18.08) 26.60 (15.89) t(127.5) = -2.33,  p = .02             Notes. Van = Vancouver, HK = Hong Kong. Perceived social status in on a 10-point scale; social relationship quality is on a 5-point scale; acculturation to host culture is on a 9-point scale. Mean loneliness is the person-mean across all momentary assessments, on a 100-point scale. Solitude and desire for solitude are the percentage of all beeps when in solitude and when desiring solitude, respectively. Sample difference tests compare Vancouver sample to Hong Kong sample using Welch’s t (for means) or chi-square difference test (for proportions).       4.2.2 Procedure This study consisted of a baseline session, a 10-day time-sampling period, and an exit session; Vancouver participants also completed a 6-month follow-up session. At baseline, participants completed individual difference measures (including socio-demographic information) and training on the use of tablets for repeated daily life assessments. Participants were beeped three times daily, for a 10-day period beginning the day after the baseline session, to complete brief questionnaires concerning their current affect and actual and desired social context. Questionnaires were administered through an iPad miniTM app (iDialogPad; G. Mutz, Cologne, Germany) and were scheduled to occur once in the morning, once in the afternoon, and once in the evening (with a minimum of 4 hours between assessments) while avoiding conflicts with participants’ predetermined commitments. Questionnaires that were completed within 90 minutes of one another (392 questionnaires, 7.6%) or that contained electronic timestamp errors (impossible dates, 14 questionnaires, 0.3%) were excluded from analyses. This left an average of 30.8 valid questionnaires per participant (SD = 8.7, range 8–71). Some participants completed questionnaires for more than 10 days (excluding this extra data did not change the reported findings, hence, all data were retained in analyses to maximize power). Participants completed additional questionnaires, including study feedback and a measure of acculturation, at the exit session and (for Vancouver participants only) at a 6-month follow-up session. Participants reported that the 10-day time-sampling period was representative of their typical daily lives (M = 3.5 on a 5-point scale) and that the study did not disrupt their daily routines (M = 1.7/5) or cause them to behave differently (M = 1.7/5). Data were collected year-round from August 2014 to January 2017. Materials were developed in English and translated to simplified and traditional Chinese; translations were verified through independent backward-translation. Fifty-seven   85   percent of Vancouver participants completed the study in English, 28% in Mandarin, and 15% in Cantonese. Hong Kong participants completed the study in Cantonese.  4.2.3 Measures 4.2.3.1 Time-sampling measures   Current loneliness. At each beep, participants were asked to report on their current affective states, including loneliness (Russell, 1996; Tsai et al., 2006), on a 100-point scale (0 = “not at all”, 100 = “very much”. The mean loneliness score was 22.2 (SD = 21.8, Mode = 0, within-person scale reliability estimate R1R = 0.62; person-level score reliability estimate RKF = 0.63; Cranford et al., 2006).   Current social context (solitude). Participants indicated their current social context, “what was your situation when you were beeped?”, by selecting one of three options (McAdams & Constantian, 1983): (a) interacting with someone, (b) others nearby but not interacting, or (c) alone (no one nearby). Occasions when participants selected (a) were coded as social interaction, and occasions when they selected (b) or (c) were coded as solitude (defined as the absence of social interaction). In a separate question, participants were asked about their activities at the time of the beep by selecting from a list of activity types. Solitude occasions when participants had selected “social activity” (316 occasions) were recoded as social interaction, to account for times when participants may have been alone but chatting online/over the phone. Of the 4654 momentary assessments used in analyses, 57.2% were coded as solitude.   Current desire for solitude. Using the same three social context options (McAdams & Constantian, 1983), participants were asked, “which of these situations would you most like to be in?” Occasions when participants chose option (a) were coded as desire for social interaction, and occasions when they chose (b) or (c) were coded as desire for solitude. Of the 4654   86   momentary assessments, 61.1% were coded as desire for solitude (85.5% of all solitude occasions and 28.6% of all social interaction occasions were times when participants desired solitude).   4.2.3.2 Individual difference measures Acculturation to host culture. The 10-item host culture acculturation subscale of the Vancouver Index of Acculturation (VIA; Ryder et al., 2000; Tieu & Konnert, 2015) was used to measure identification with the local or host culture (“North American” culture for participants in Vancouver; “Hong Kong” culture for participants in Hong Kong). Participants responded to items, such as “I believe in mainstream [North American/Hong Kong] values” and “I enjoy social activities with typical [North American/Hong Kong] people.” on a 9-point scale (1 = “Strongly disagree”, 9 = “Strongly agree”), M = 6.4, SD = 1.3, Cronbach’s alpha = 0.88. The measure was administered at baseline (Hong Kong participants) or 6-month follow-up (Vancouver participants). Sociodemographics and covariates. Immigration status, cultural heritage, age, gender, education, retirement status, and relationship status were assessed at baseline. Perceived social status was assessed at baseline (Hong Kong participants) or at the 6-month follow-up (Vancouver participants) using the 2-item MacArthur ladder scale (Adler & Stewart, 2007), M = 5.4 on a 10-point scale, SD = 1.4. We assessed social relationship quality at exit using the “positive relations” subscale of the Ryff scales of psychological wellbeing (short version, Ryff & Keyes, 1995), M = 3.7 on a 5-point scale, SD = 0.6.   4.2.4 Statistical analyses   87   Given the hierarchical data structure (momentary assessments nested within people), we used multilevel modeling (Snijders & Bosker, 2002) to test hypotheses regarding momentary solitude-loneliness associations and moderators of these associations. Models predicted momentary loneliness from current solitude (Level 1 predictor) and its interactions with current desire for soltiude (Level 1 predictor) and sample location, cultural heritage, immigration status, and host culture acculturation (Level 2 predictors). Person-average solitude and desire for solitude, and several covariates (age, gender, education, retirement status, relationship status, perceived social status, and social relationship quality), were added at Level 2. Model equations and detailed variable information (variable coding, centering) are provided in Appendix E.  4.3 Results 4.3.1 Descriptive findings   Participants reported being in solitude at 57.3% of their momentary assessments, on average (SD = 25.0, range 0–100%). As shown in Table 4-1, participants in Vancouver reported solitude at more occasions than those in Hong Kong (63.5% of beeps versus 46.8%), and they also reported more occasions desiring solitude than those in Hong Kong (68.6% of beeps versus 50.3%). Across both samples, most of the occasions participants spent in solitude were desired (86.3% of solitude instances in Vancouver, 83.6% in Hong Kong). Notably, participants in Vancouver also tended to report lower levels of loneliness (across all momentary assessments) than those in Hong Kong (M = 20.0 versus 26.6). Participants in Hong Kong reported stronger acculturation to their host culture than participants in Vancouver (6.86 versus 6.15 on a 9-point scale), and people who had immigrated reported weaker host culture acculturation than those who had not immigrated (r = -.21). Participants living in Vancouver were younger, more likely to have post-secondary education, perceived their social status to be higher, and were less likely   88   to be retired or in a relationship. There were no gender, social relationship quality, or immigration status differences between samples.   The person-level variable intercorrelations in Table 4-1 show a positive relationship between solitude and desire for solitude (r = .76), and a negative relationship between desire for solitude and loneliness (r = -.19). Being of East Asian heritage was associated with less occasions spent in solitude (r = -.24) and higher overall loneliness (r = .22), compared to being of European/North American heritage; however, follow-up multilevel models confirmed that differences in overall solitude and loneliness were driven by sample location (Vancouver versus Hong Kong) rather than cultural heritage. Stronger host culture acculturation was associated with lower overall loneliness (r = -.19). Having post-secondary education was associated with a greater proportion of occasions in solitude (r = .21) and desiring solitude (r = .17), and lower loneliness levels (r = -.21).   4.3.2 Momentary solitude-loneliness associations   Momentary loneliness varied both within a given person (within-person variance = 167.63) and between people (between-person variance = 305.33; Intraclass Correlation Coefficient = 0.65). Hence, we used multilevel models to test our hypotheses. Results are presented in Table 4-2. Overall, individuals reported greater loneliness if they were living in Hong Kong, had immigrated, desired solitude less, or had lower quality social relationships.     Table 4-2: Fixed effects for multilevel model predicting current loneliness (N = 151 individuals, n = 4654 momentary assessments)  Current loneliness - Model A Current loneliness - Model B  Estimate SE p 95% CI Estimate SE p 95% CI LEVEL 1         Current solitude status 2.07 (0.73) .00 [0.65, 3.49] 2.16 (0.70) .00 [0.78, 3.54] Current solitude desire -0.52 (0.76) .50 [-2.02, 0.98] -0.84 (0.77) .28 [-2.35, 0.67] LEVEL 2             Intercept 22.43 (1.25) .00 [19.99, 24.88] 22.60 (1.26) .00 [20.13, 25.06] Mean solitude 0.11 (0.08) .16 [-0.04, 0.27] 0.14 (0.08) .08 [-0.02, 0.30] Mean solitude desire -0.18 (0.07) .01 [-0.32, -0.04] -0.20 (0.07) .00 [-0.34, -0.06] Sample location (Hong Kong) 7.54 (3.73) .04 [0.23, 14.84] 8.00 (3.77) .03 [0.61, 15.39] Ethnic heritage (East Asian) 4.19 (3.80) .27 [-3.26, 11.64] 3.86 (3.84) .32 [-3.67, 11.39] Immigration status (immigrant) 7.84 (2.98) .01 [2.00, 13.69] 7.83 (3.03) .01 [1.89, 13.77] Host culture acculturation -1.27 (1.10) .25 [-3.42, 0.87] -1.76 (1.11) .11 [-3.93, 0.41] Age 0.06 (0.20) .77 [-0.33, 0.45] 0.06 (0.20) .75 [-0.33, 0.45] Gender (female) 2.95 (2.80) .29 [-2.55, 8.45] 2.95 (2.79) .29 [-2.53, 8.42] Education (at least some post-secondary) -0.63 (2.98) .83 [-6.47, 5.20] -0.40 (2.96) .89 [-6.21, 5.41] Retirement status (retired) -7.84 (4.16) .06 [-15.98, 0.31] -7.86 (4.14) .06 [-15.97, 0.25] Relationship status (in a relationship) -3.19 (3.08) .30 [-9.23, 2.85] -3.00 (3.07) .33 [-9.01, 3.02] Perceived social status -0.17 (0.98) .86 [-2.10, 1.75] -0.15 (0.98) .88 [-2.06, 1.77] Social relationship quality -6.28 (2.23) .01 [-10.64, -1.91] -6.39 (2.22) .00 [-10.74, -2.05] INTERACTIONS         Current solitude x current solitude desire     -2.51 (1.26) .05 [-4.98, -0.04] Current solitude x mean solitude     0.09 (0.04) .05 [0.00, 0.17] Current solitude x mean solitude desire     -0.05 (0.04) .21 [-0.12, 0.03] Current solitude x sample location     1.23 (1.69) .47 [-2.08, 4.54] Current solitude x immigration status     0.24 (1.48) .87 [-2.66, 3.15] Current solitude x ethnic heritage     -1.26 (1.84) .49 [-4.87, 2.34] Current solitude x host culture acculturation     -1.40 (0.51) .01 [-2.40, -0.40] DEVIANCE REDUCTION χ2(20) = 142.55, p < .001 (compared to model with no predictors) χ2(7) = 17.74, p = .01 (compared to model with no interactions)   90   Note 1. Coefficients are unstandardized. Perceived social status is on a 10-point scale; acculturation to host culture is on a 9-point scale. Mean loneliness is the person-mean across all momentary assessments, on a 100-point scale. Solitude and desire for solitude are the percentage of all beeps when in solitude and when desiring solitude, respectively. Deviance reduction compares full model to empty model without predictors. Missing data for age (N = 5), relationship status (N = 2), and perceived social status (N = 5), were multiply imputed (predictive mean matching, R mice package; Buuren & Groothuis-Oudshoom, 2011); missing data for immigration status (N = 4) and acculturation to host culture (N = 1) were not imputed. Models were estimated using the lme4 package in R (Bates, Mächler, Bolker, & Walker, 2015); restricted maximum likelihood estimation. Note 2. Additional post-hoc models tested for potential boundary conditions of the host culture acculturation moderation effect on solitude-loneliness associations. Specifically, we examined 3-way interactions of host acculturation x current solitude with (1) cultural heritage, (2) sample location, (3) immigration status, and (4) current solitude desire. Although interactions (3) and (4) were significant, the overall moderating effect of host culture acculturation also remained significant. Three-way interactions were excluded from the reported analyses due to concerns about power and lack of theoretical underpinning. Additional models examining acculturation to heritage culture rather than host culture (Ryder et al., 2000) revealed that individuals who were more acculturated to their heritage culture also showed weaker solitude-loneliness associations, as compared to individuals lower in heritage culture acculturation. The inclusion of heritage culture acculturation in the model did not change the reported host culture acculturation moderation finding. Note 3. Additional models included time of day (linear and quadratic effects), current location (outside, at home), current activity (passive leisure, work), and age as moderators of solitude-loneliness associations. Currently being outside and spending more time overall engaged in passive leisure activities were associated with stronger associations between solitude and loneliness. Older age was associated with weaker solitude-loneliness associations. Additional models examining aloneness (no one present) instead of solitude (no social interaction) showed that desire to be alone reduced concurrent alone-loneliness associations, but there were no moderating effects of host culture acculturation or of overall time alone. Hence, the reported moderating effect of host culture acculturation seems to be specific to solitude-loneliness associations. Another set of additional models examined affective outcomes aside from loneliness, specifically, high arousal positive affect (happy, excited), low arousal positive affect (calm, relaxed), high arousal negative affect (anxious, irritated), and low arousal negative affect (sad, tired). None of the reported findings (moderating effects of solitude desire, host culture acculturation, and overall time in solitude) generalized to these other affect outcomes. Interestingly, however, immigrants (as compared to local-born middle-aged/older adults) showed increased associations between current solitude and high arousal positive affect, and decreased associations between current solitude and low arousal negative affect. These additional findings were excluded from the reported model for parsimony but are discussed in Chapter 5.  Note 4. To examine whether the reported findings were specific to solitude (no social interaction) rather than aloneness (no one physically present), we tested additional models that included current aloneness and current desire to be alone as predictors of loneliness. Being alone (as compared to being with others) was associated with greater loneliness, but less so when aloneness was desired (thus paralleling the reported solitude desire moderation finding). Host culture acculturation did not moderate the association between aloneness and loneliness, but still moderated the association between solitude and loneliness when the additional “aloneness” variables were added to the model. Because our theoretical model centered on experiences of solitude rather than on experiences of being alone, aloneness and desire to be alone were excluded from the final model, but are discussed further in Chapter 5.      91     In line with Hypothesis 1, participants reported increased loneliness when in solitude, compared to when engaged in social interaction (Model A). Counter to expectations (Hypotheses 2, 3, and 4), cultural heritage, sample location, and immigration status did not moderate momentary solitude-loneliness associations (Model B). However, as expected (Hypothesis 5), there was a significant interaction between host culture acculturation and momentary solitude-loneliness associations (Model B). Specifically, as shown in Figure 4-1a, for individuals low in acculturation (M – 1 SD), there was a significant positive association between momentary solitude and loneliness (b simple slope = 3.88, SE = 0.97, p < .001, 95% CI [1.98, 5.77]), but for individuals high in acculturation (M + 1 SD), the association between momentary solitude and loneliness was not significant (b simple slope = 0.33, SE = 0.96, p = .730, 95% CI [-1.54, 2.20]).    As expected (Hypothesis 6), currently desiring solitude reduced the association between momentary solitude and loneliness (Model B). As shown in Figure 4-1b, when solitude was undesired, there was a significant positive association between momentary solitude and loneliness (b simple slope = 3.63, SE = 1.03, p < .001, 95% CI [1.62, 5.65]), whereas when solitude was desired at that moment, the association between momentary solitude and loneliness was not significant (b simple slope = 1.13, SE = 0.87, p = .195, 95% CI [-0.58, 2.84]). Individuals who spent more of their occasions in solitude also showed stronger concurrent solitude-loneliness associations than individuals who spent less of their occasions in solitude.    To summarize, solitude was associated with increased momentary loneliness, but this association was diminished for individuals who were highly acculturated to their local (host) culture and who spent less time in solitude overall, and when individuals wanted to be in solitude at that moment. Contrary to expectations, cultural heritage, sample location, and immigration status did not moderate concurrent solitude-loneliness associations.   92     (a)  (b) Figure 4-1. Associations between current solitude and current loneliness as a function of acculturation to host culture (a) and current solitude desire (b)  4.4 Discussion   This study examined the role of culture (heritage culture and sample location), immigration, host culture acculturation, and solitude desire in concurrent associations between solitude and feeling lonely, as middle-aged and older adults living in Vancouver and in Hong Kong went about their daily lives. Solitude was associated with increased concurrent loneliness, but only for individuals who were low in acculturation to their host culture or who did not desire *** ***   93   solitude at that moment, and less so for individuals who spent little time in solitude overall. There were also notable sample location differences in overall time spent in solitude, solitude desire, and loneliness.    4.4.1 Culture and solitude-loneliness associations   Middle-aged and older adults in Vancouver reported solitude more often and desired solitude more often, and yet also felt less lonely overall, than middle-aged/older adults living in Hong Kong. These location differences were independent of participants’ cultural heritage and immigration status. It is possible that individuals living in Vancouver are more accustomed to a solitary lifestyle because, compared to Hong Kong, Vancouver is less densely populated and more individuals live alone (Hong Kong Census and Statistics Department, 2017; Statistics Canada, 2017).    As expected, moments spent in solitude were associated with higher levels of loneliness than moments spent engaged in social interaction. However, counter to expectations, neither cultural heritage nor sample location were significantly associated with momentary solitude-loneliness associations. One previous study has found that exposure to Chinese culture is associated with experiencing moments of solitude more positively (Jiang et al., in press). It is possible that the expected cultural differences did not emerge in the present study because our sample included a larger proportion of emigrants from Hong Kong (with significant exposure to Western values through the British Commonwealth) rather than from mainland China (Chan, 2013). It is also possible that dichotomous variables such as heritage culture may carry different meanings for different individuals (Chan, 2013; Wu & Penning, 2015), only serving as rough proxies for cultural contextual processes that shape solitude experiences; this may also help   94   explain why the present study found no moderating effects of culture on solitude-loneliness associations.  4.4.2 Immigration and solitude-loneliness associations   Contrary to expectations, older immigrants did not feel any lonelier during solitude than local-born middle-aged and older adults. This is in line with previous null findings (Jiang et al., in press) and also suggests that comparing immigrants to locals may be an oversimplification when examining momentary solitude experiences. The effects of immigration are time-and context-dependent, shaped by factors such as the age of the immigrant, how long they have lived in their host country, and how different the mainstream host culture is from their heritage culture (de Jong Gierveld et al., 2015, Wu & Penning, 2015). Indeed, moving from mainland China to Hong Kong may be quite different from moving from China to Canada. Moreover, individuals who immigrated long ago and at a young age may feel particularly connected with their host culture and with others around them, compared to individuals who immigrated more recently or at an older age (Cheung, Chudek, & Heine, 2011; de Jong Gierveld et al., 2015); for such individuals, moments of solitude may feel less oppressive, threatening, or lonely. Future research comparing immigrant subgroups may reveal additional immigration-related factors shaping solitude experiences. However, although their solitude experiences did not feel any more lonely on a moment-to-moment level, immigrants still reported greater overall loneliness than local middle-aged and older adults, in line with the literature on immigration (de Jong Gierveld et al., 2015).  4.4.3 Acculturation and solitude-loneliness associations   95     As expected, host culture acculturation moderated solitude-loneliness associations: Individuals who were low in acculturation to the local (host) culture felt lonelier during moments of solitude compared to moments of social interaction. Individuals who were high in host culture acculturation, however, showed no such momentary solitude-loneliness association. Host culture acculturation captures a sense of identification and belonging with the mainstream culture in which one lives (Ryder et al., 2000), a rich psychological construct that may speak to a process not captured by categorical variables such as cultural heritage, location, and immigration status. Exploratory analyses (reported in the Table 4-2 footnote) showed that host culture acculturation moderated solitude-loneliness associations for both immigrants and local-born individuals, and regardless of whether they were living in Vancouver or in Hong Kong, ruling out explanations having to do with the immigration process or with acculturation to Western versus East Asian culture. To further examine further boundary conditions, heritage culture acculturation (Ryder et al., 2000; Tieu & Konnert, 2015) was added to models; heritage acculturation also moderated solitude-loneliness associations (Table 4-2 footnote). It is possible that heritage culture operates much like a mainstream host culture, for example, for East Asian immigrant middle-aged and older adults living in a “Chinatown” neighbourhood in Vancouver (Syed et al., 2017). We suggest that acculturation is associated with feeling less lonely in moments of solitude because it captures a sense of belonging or closeness with others (Averill & Sundararajan; Pauly et al., in press).  4.4.4 Desire for solitude and solitude-loneliness associations Although it is important to feel connected, individuals may sometimes seek out solitude in their daily lives (Burger, 1995; Larson, 1990; Lay et al., 2018; Long & Averill, 2003; Nguyen   96   et al., 2017). Our findings revealed that, not only do middle-aged and older adults sometimes seek solitude, but this motivation keeps them from feeling lonely during moments of solitude: When solitude is desired, the solitude-loneliness association disappears. We thus build on a small but growing literature suggesting that desire or autonomous motivation is key to thriving in solitude (Averill & Sundararajan, Chua & Koestner, 2008; Nguyen et al., 2017). We speculate that middle-aged and older adults sometimes seek out solitude as a source of strength and personal growth (Averill & Sundararajan, 2014; Larson, 1990; Long & Averill, 2003).   Most of the solitude that was reported was desired at that moment. Individuals who had higher overall desire for solitude also reported lower levels of loneliness overall. Building on previous research based on university student samples (Chua & Koestner, 2008; Larson et al., 1982; Long & Averill, 2003; Nguyen et al., 2017), these findings suggest that middle-aged and older adults also regularly seek out solitude in their daily lives.  4.4.5 Further exploratory findings   Individuals who had higher-quality social relationships reported lower levels of loneliness overall, confirming well-established findings linking social relationship quality with lower overall loneliness among older adults (de Jong Gierveld et al., 2015). Controlling for relationship quality did not change or diminish the central findings of this study. Middle-aged and older adults who spent more time overall in solitude also showed stronger links between solitude and loneliness. This is in line with previous research (e.g. Pauly et al., 2017) and suggests that the benefits of solitude may be undermined if one experiences so much solitude that it is an indicator of social isolation (Chen et al., 2014; Hawkley & Cacioppo, 2010).    97   4.4.6 Limitations and future directions   Reflecting well-established demographic differences between adults aged 50+ living in Canada and in China (Hong Kong Census and Statistics Department, 2017; Statistics Canada, 2017), our Vancouver participants had higher levels of education and perceived social status, and were less likely to be of East Asian heritage or in a relationship, than Hong Kong participants. Compared to Vancouver participants, Hong Kong participants had lived longer in their country of residence and were more acculturated to the host culture. Having post-secondary education was associated with greater overall time in solitude and desiring solitude, and lower overall loneliness, hence, differences in education levels may help explain why individuals living in Vancouver and in Hong Kong also differed on these variables. The samples of adults aged 50+ in this study were non-representative in terms of education level and number of female participants, both of which were higher than population averages. This is common in convenience samples. Further research is needed to examine solitude-loneliness links in more representative national samples (e.g. Almeida, 2005).   In the present research, we sought to disentangle the roles of cultural context and immigration in solitude experiences. Given the study locations (Vancouver and Hong Kong), this study focuses on European/North American and East Asian individuals (who make up 33% of the Vancouver population; Statistics Canada, 2017). However, further research is needed to examine whether our findings generalize to other cultural groups, for example, Hispanic middle-aged and older adult immigrants to the United States (Russell & Taylor, 2009). Moreover, we used cultural heritage and local culture/sample location as dichotomous proxies for what are really very complex cultural influences, and hence, further research is needed to understand the processes underlying cultural differences in solitude, solitude desire, and loneliness. For   98   example, previous research has linked cultural differences in solitude-seeking to specific individualistic and collectivistic cultural values (Averill & Sundararajan; Maes, Wang, van den Noortgate, & Goossens, 2015). It would be interesting to examine how internalization of values related to individualism and collectivism (e.g. independent and interdependent self-construal; Markus & Kitayama, 1991) might specifically shape solitude experiences.  4.4.7 Conclusions   Solitude is a ubiquitous part of daily life (Larson, 1990) that may be linked with greater loneliness for individuals across countries (such as Canada and Hong Kong; Chen et al., 2014; Syed et al., 2017). This study identified high host culture acculturation and solitude desire as key factors that de-couple solitude from loneliness, and that may enable middle-aged and older adults to thrive in moments of solitude (Long & Averill, 2003; Pauly et al., in press). These protective effects held across two cultures (Canadian and Chinese), and for both immigrant and local-born adults aged 50+. By examining concurrent associations between solitude and loneliness, this study identified daily life processes that help us to contextualize the links between culture/immigration and loneliness in old age (de Jong Gierveld et al., 2015). Future research should examine causal mechanisms underlying the host culture acculturation moderation finding, for example, by manipulating belongingness or independent/interdependent self-construal in a controlled setting and examining their influence on solitude experiences (Blackhart, Nelson, Winter, & Rockney, 2011; Brewer & Gardner, 1996; Nguyen et al., 2017). Such future directions are discussed further in Chapter 5.    99   Chapter 5 General discussion  5.1 Synthesis    The aim of this research program was to examine the multifacetedness of solitude experiences. To address a need for definitional precision, solitude (the absence of social interaction) was differentiated from both loneliness and being alone. A series of three studies examined solitude in a daily life context to understand how individuals experience solitude, and how solitude experiences vary depending on motivational and situational factors that vary within a given person. These studies also examined more stable individual difference factors that may shape solitude experiences, including social and personal resources and vulnerabilities, age, and cultural factors. The overall aim was to understand what solitude looks like (the affective states and thoughts characterizing different types of solitude experiences), and under what circumstances and for whom solitude may be experienced negatively versus positively.   The three studies presented here examined daily life solitude experiences from different angles: (1) a cluster analytic approach to identify different types of solitude experiences and link them with situation- and person-specific factors, (2) an age-comparison approach to examine middle-aged and older adults’ solitude-seeking experiences, and (3) a cross-cultural approach to examine solitude and loneliness among local-born and immigrant middle-aged and older adults living in two countries. Table 5-1 summarizes the characteristics and key findings of each of the three studies.    100   Table 5-1: Summary of study characteristics and findings  STUDY 1  STUDY 2 STUDY 3  Sample(s) 100 adults age 50+ and 50 students in Vancouver 100 adults age 50+ in Vancouver 96 adults age 50+ in Vancouver and 56 adults age 50+ in Hong Kong Statistical approach Multilevel latent profile analysis and latent class regression Multilevel models (logistic and continuous outcomes) Multilevel models  Primary outcomes (situation level) Distinct types of solitude experiences Location and affect when seeking solitude Solitude-loneliness associations Primary predictors (situation-level and person-level) Desire for solitude, social and personal resources (e.g. social self-efficacy) and vulnerabilities (e.g. self-rumination) Age, current activities, time of day Cultural heritage, sample location, immigration, acculturation, solitude desire Role of loneliness Indicator used to classify solitude episodes Dependent variable in regression model Dependent variable in regression model Role of solitude desire Situation-level and person-level predictor of distinct types of solitude experiences Primary variable of interest (situation-level); person-level control variable Situation-level predictor variable; person-level control variable Key findings - Two distinct types of solitude experiences from affect-thought patterns - Person-level solitude desire, social self-efficacy associated with positive solitude experiences - Person-level self-rumination, self-reflection associated with negative solitude experiences - Most solitude was desired - Older (but not middle-aged) adults more likely at home or outside when seeking solitude - Middle-aged (but not older) adults showed reduced positive affect when seeking solitude - Middle-aged/older adults in Hong Kong and immigrants more lonely - Host culture acculturation and situation-level solitude desire buffered concurrent solitude-loneliness associations    Taken together, these studies show that solitude is not always lonely, and that desire for solitude plays a key role in differentiating solitude from loneliness and from other negative experiences. These studies specify the characteristics (thought and affect) associated with different kinds of solitude experiences, and also identify situational and individual difference factors that may be conducive to thriving when in solitude. Below, I discuss key findings from   101   this research program. These findings provide initial empirical evidence for key aspects of a new theoretical model of solitude experiences, illustrated in Figure 5-1. This figure is an updated version of Figure 1-1 (Chapter 1), highlighting the key findings.   Figure 5-1: Conceptual model of solitude experiences: Key findings from the present research program  5.1.1 Momentary affect and thoughts during solitude   This research program linked solitude with specific affective states and thought patterns. In Study 1, two distinct types of solitude experiences were identified, one characterized by the presence of effortful thought and negative affective states including loneliness (“negative” solitude), and the other characterized by calm and the absence of negative affect and effortful   102   thought (“positive” solitude). Importantly, Study 1 showed that loneliness and other negative affective states are not an inevitable correlate of solitude, and that a more positive kind of solitude is also common in daily life. Interestingly, despite common perceptions that solitude is lonely (Hawkley & Cacioppo, 2010; Jylhä & Saarenheimo, 2010), Study 2 showed that the vast majority of solitude is in fact desired. Study 3 further demonstrated that the specific link between solitude and loneliness can be diminished: Under certain conditions (e.g. when solitude is desired) and for certain people (e.g. those acculturated to the mainstream culture), solitude is in fact not associated with loneliness.    Based on previous research suggesting individuals may seek solitude for contemplation and self-reflection (Burger, 1995; Long et al., 2003; Long & Averill, 2003; Wang, 2006), I expected that inner-directed thought would feature prominently during daily life solitude. In Study 1, momentary thought dimensions were split into two parcels (high-cognitive-effort thought and low-cognitive-effort thought). Interestingly, this study showed that the presence of high-cognitive-effort thought specifically (e.g. having new ‘deep’ thoughts and uncontrollable negative thoughts), was associated with negative solitude experiences. Low-cognitive-effort thought (e.g. pleasant, present-focused thoughts), in contrast, did not differentiate between different kinds of solitude experiences.   To summarize, high-cognitive-effort thought was associated with negative solitude experiences in the present research. This finding seems, at first glance, to conflict with previous research linking solitude with pleasant self-reflection and productive problem-solving (inner-directed solitude; Long et al., 2003). However, what looks like a paradox may be a reflection of differences in study designs: the present study captured momentary solitude experiences, whereas Long and colleagues (2000, 2003) use retrospective reports of solitude experiences   103   (specifically, ratings of the subjective importance of various types of solitude experiences). As the authors elaborate, if an individual uses their solitude for self-reflection or problem-solving, the resulting self-understanding or growth may lead to feelings of calm and satisfaction at a later time (Long et al., 2003). Hence, while our study’s negative solitude experience class captured less-pleasant solitude experiences as they occurred in the moment, the solitude experience types that that Long and colleagues identified reflected participants’ retrospection on past solitude experiences and what came out of them. This is consistent with the idea that difficult momentary experiences can pave the way for calm over the course of a solitude episode or in the longer term. Importantly, the present research program demonstrates how solitude is experienced at the moment when it occurs, before being subject to retrospection or to the interpretive lens of time.  5.1.2 Desire for solitude and solitude experiences   Solitude desire emerged as a key factor shaping solitude experiences, confirming suggestions in the literature (e.g. Averill & Sundararajan, 2014; Long & Averill, 2003; van Zyl, Dankaert, & Guse, 2018). In Study 1, individuals with higher overall desire for solitude were more likely to have positive solitude experiences. Desiring solitude at a given moment also diminished concurrent solitude-loneliness associations in Study 3. Moreover, Studies 2 and 3 both showed that individuals who desired solitude more overall also felt less lonely at any given moment. Taken together, these studies suggest that solitude desire is negatively correlated with loneliness.    These studies further suggest that, even though desire for solitude varies more within people than it does between people (Study 2), solitude desire operates on both the between-person level and on the within-person (momentary) level in shaping experiences of solitude and   104   loneliness. At the between-person level, solitude desire had main effects on loneliness and on solitude experience types: Individuals with stronger overall desire for solitude felt less lonely (Studies 2 and 3) and were more likely to have negative solitude experiences (Study 1) at any given moment. However, at the within-person level, fluctuations in current desire for solitude also shaped whether solitude felt lonelier than social interaction at any given moment (Study 3). Hence, solitude desire should be examined at both the between person level and at the within person level to get a fuller picture of its role in solitude experiences and loneliness. The present research underscores the utility of a time-sampling design and multilevel modeling for understanding these processes (e.g. Chui et al, 2014; Hoppmann & Riediger, 2009).  5.1.3 Other time-varying situational correlates of solitude experiences   Certain situations and contexts may make solitude more normative, or may make solitude-seeking feel more “right”. In line with previous research on married and unmarried older adults (Larson et al., 1985), Study 2 showed that individuals were more likely to be in solitude or to be seeking solitude in the early morning and in the evening, whereas the late-morning and afternoon hours were more social. Morning and evening are also times when individuals are more likely to be at home. Indeed, solitude was more likely to occur at home, and less likely to occur outdoors, among adults age 50+ in Study 2. Being outdoors at a given moment was also associated with stronger solitude-loneliness links in this age group (Study 3, Table 4-2 footnote). These findings suggest that, unlike for younger adults (Long, 2000; Long et al., 2003), the outdoors is not an ideal place for middle-aged and older adults to seek solitude. Differences between the present research findings and those derived from university student samples on locations associated with solitude-seeking (e.g. Long et al., 2003) underscore the need to look at solitude experiences across the adult lifespan (Coplan, Ooi, & Baldwin, 2018; Larson, 1990).    105     Certain activities may also be more likely pursued during solitude. For example, individuals age 50+ were more likely to engage in passive leisure activities, such as reading, relaxing, or watching TV, when in solitude than when interacting with others (Study 2). Study 2 also showed that seeking solitude for passive leisure (or during passive leisure) was associated with higher levels of low-arousal positive affect than seeking solitude during other types of activities. Hence, solitude seems to be conducive to restful leisure activities (Burger, 1995; Larson, 1990; Long et al., 2003). Interestingly, however, spending more time overall engaged in passive leisure activities was also associated with stronger concurrent solitude-loneliness associations (Study 3, Table 4-2 footnote). This suggests a potential boundary condition: Seeking solitude to relax may be effective, but spending too much time engaged in solitary passive leisure may dilute its beneficial effects. Individuals may also sometimes seek solitude to focus or concentrate on work activities (Larson, 1990; Long et al., 2003). Although the present studies found no association between current work activity and being in solitude, there was some indication that individuals who spent more of their overall time in solitude engaged in work activities were more likely to experience solitude negatively (Study 1, Table 2-2 footnote). Overall, it seems that the activities one typically engages in during solitude (passive leisure, work) can shape how solitude is experienced.   It has been further suggested that solitude may only be beneficial in moderation, that is, when not prolonged (Larson, 1990; Long & Averill, 2003; Pauly et al., in press). Indeed, sustained solitude or social isolation (as occurs when, for example, individuals live alone or have little contact with close others) has been clearly linked with poor health and wellbeing outcomes, including depression, cardiovascular disease, and mortality (de Jong Gierveld et al., 2005; Seeman, 1996). Although the present research focuses on momentary experiences of solitude   106   rather than longer-term consequences of sustained solitude, findings from Study 3 suggest that prolonged solitude negatively colours momentary solitude experiences. Specifically, individuals who spent more time overall in solitude reported greater loneliness in moments of solitude compared to individuals who spent less time in solitude overall (Table 4-2). Further research is needed to determine how much solitude is too much, that is, the point at which solitude may be best described as social isolation.   Finally, when it comes to momentary solitude experiences, does it matter whether an individual is “alone in a crowd” versus completely alone (Rokach, 2004)? The findings reported in this research program are specific to solitude (the absence of social interaction) and solitude-seeking, and do not generalize to being alone and seeking aloneness. Aloneness and desire to be alone were added to the study models (as described in Study 1 Table 2-2, Study 2 footnote to Table 3-2, and Study 3 footnote to Table 4-2); these variables neither changed nor replicated the vast majority of reported findings pertaining to solitude and desire for solitude. These findings confirm that being in solitude is not experienced the same way as being alone, and that it is important to differentiate between these two social situations in research on solitude.  5.1.4 Social and personal resources and vulnerabilities shaping solitude experiences   This set of studies also examined more stable individual difference characteristics that may be associated with overall propensity to experience solitude positively versus negatively. Study 1 showed that individuals who felt confident in their social selves (high social self-efficacy; Di Giunta et al., 2010) were more likely to have positive solitude experiences. This finding extends previous work showing that solitude that occurs within a context of strong social relationships is experienced more positively (Averill & Sundararajan, 2014; Long et al., 2007;   107   Pauly et al., in press). Further research is needed to examine whether factors such as social self-efficacy might mediate the association between social relationship quality and positive solitude experiences (Pauly et al., in press).   With respect to individual differences in thought styles, I had expected that individuals who tend to enjoy self-reflection would be more likely to have positive solitude experiences, and that individuals with a tendency for self-rumination would be more likely to have negative solitude experiences (Trapnell & Campbell, 1999). These hypotheses were based on the idea that solitude is particularly conducive to self-reflection but that it poses risks to affective wellbeing if individuals are mired in repetitive negative thought or rumination (Long & Averill, 2003). Interestingly, both trait self-reflection and trait self-rumination were associated with propensity for negative solitude experiences in Study 1, suggesting that a general tendency to engage in introspection is associated with loneliness and other negative affective experiences during solitude. This does not mean that deep or intense thoughts always co-occur with loneliness when in solitude (Larson, 1990). However, high cognitive-effort thought and loneliness do seem to go together when in solitude, on both a moment-to-moment level (as explained above, Section 5.1.1) and on a between-person level.  5.1.5 Solitude in the second half of life   Life phase is a key individual difference factor shaping solitude experiences (Larson, 1990). Study 2 showed that older adults (aged 68-85 years) in particular were more likely to be in locations conducive to solitude (at home, not outdoors) when they were in fact seeking solitude, and that, unlike middle-aged adults (aged 50-67), they did not experience reduced levels of positive affect when seeking solitude. Hence, the present research not only extends the   108   solitude-seeking literature to later life phases, but it also suggests ways in which older adults may have an enhanced capacity to thrive in solitude (Larson, 1990; Larson et al., 1985; Pauly et al., 2017; Rowe & Kahn, 1997).   More broadly, this set of studies focuses on solitude among adults aged 50+, who may be at particular risk of social isolation and loneliness (Jylhä & Saarenheimo, 2010). Populations in industrialized countries are rapidly aging, and the number of people aged 60 years and above is at a record high across nations (United Nations, 2015). It is estimated that by 2050, nearly one third of Canadians will be aged 60 and above, and in China, the proportion of people in this age group is expected to more than double to reach 37 percent over the same period (United Nations, 2015). Given this historically unprecedented situation, there is a pressing need to foster successful aging and wellbeing in late life: Social isolation and loneliness in old age are important concerns for researchers and health professionals across the globe (Chen et al., 2014; Jo Cox Commission on Loneliness, 2017; Kobayashi et al., 2009). However, although there are strong links between old age and risk of loneliness (Heylen, 2010; Jylhä & Saarenheimo, 2010), it is important not to paint old age (or solitude) with too broad of a brush. The present research might perhaps serve as encouragement to treat older adults not as victims of certain circumstances (e.g. reduced social networks), but rather as active agents who may have certain abilities and resources that may help them to thrive during solitude (Larson, 1990; Pauly et al., 2017; Rowe & Kahn, 1997). Put another way, optimal aging does not have to mean “escaping” loneliness; rather, there is a need to place more emphasis on capacity to thrive during solitude in later life (Mikkelsen, 2016).  5.1.6 Solitude across cultures   Most literature investigating everyday solitude experiences focuses exclusively on   109   samples in Western cultural contexts (Averill & Sundararajan, 2014). There is some evidence of cultural differences in solitude preferences and experiences of loneliness (Burger, 1995; Long & Averill, 2003; Tsai et al., 2002; Tsai, 2007). These differences may shape solitude experiences in different ways in North American and East Asian cultures, and especially in old age, when cultural differences become even more ingrained (Fung, 2013; Jiang et al., in press; Wang, 2006). Study 3, a cross-national study of Canadian and Chinese middle-aged and older adults, showed those who felt more connected with their local culture (high host culture acculturation; Ryder et al., 2000; Tieu & Konnert, 2015) did not feel any lonelier in moments of solitude than in moments of social interaction. This acculturation moderation effect applied to individuals of both European/North American and East/Southeast Asian ethnicity, and to individuals in both Vancouver and Hong Kong, hence, it seems to be about more than just culture. I interpret host culture acculturation as a broader construct involving feeling connected to the mainstream society in which one lives. Building on the idea that solitude that occurs against a backdrop of strong social relations feels less lonely (Averill & Sundararajan, 2014; Long & Averill, 2003; Pauly et al., in press), I suggest that a sense of belonging to the culture or society in which one lives also makes solitude feel less lonely. Notably, host acculturation moderated solitude-loneliness associations above and beyond any role of social relationship quality (Study 3), suggesting that acculturation taps a resource not captured by the presence of strong social relationships (Pauly et al., in press). Further research is needed to explain the role of host culture acculturation in shaping solitude experiences.   Although host culture acculturation buffered concurrent solitude-loneliness associations, the categorical variables of heritage culture, sample location, and immigration showed no such buffering effects. Interestingly, however, for immigrant middle-aged/older adults (as compared   110   to local individuals), solitude was more strongly associated with high arousal positive affect, and was more weakly associated with low arousal negative affect (Study 3, footnote to Table 4-2). Middle-aged and older adults who immigrated to a new country often face social network losses, reduced frequency of social contact, language and cultural barriers, discrimination, social isolation, and reduction in socioeconomic status (Barrio et al., 2008; de Jong Gierveld et al., 2015; Emami, Torres, Lipson, & Ekman., 2000; Stewart et al., 2011; Wong et al., 2012). Hence, older immigrants may be particularly used to solitude, and also particularly at risk of loneliness. Further research is needed to examine how factors related to immigration (e.g. years in host country, differences between host and heritage culture) might shape solitude experiences (Barrio et al., 2008; de Jong Gierveld et al., 2015; Emami et al., 2000; Ryder et al., 2000). More generally, investigating the difficulties faced by immigrants can guide the development of preventive measures and interventions – to help Chinese and Canadian adults who are at particular risk of loneliness and its health consequences (Hawkley & Cacioppo, 2010; Jylhä & Saarenheimo, 2010).   5.2 Contributions 5.2.1 Disentangling solitude from related constructs   Solitude experiences were examined using a specific and theoretically driven definition of solitude (the absence of social interaction) that differentiates it from other related constructs: loneliness and being alone. The main contributions of this research were to show that solitude is not always lonely, and that desire for solitude is a key to making solitude a positive (or at least non-lonely) experience. The present studies distinguished solitude from its affective correlates, and identified situational factors that shape how individuals feel when in solitude. Hence, this   111   research program underscores the importance of studying the objective characteristics of an event (e.g. solitude) in order to disentangle the event itself from one’s affective responses to the event (e.g. loneliness, calm; Lazarus, 1999).    Solitude was measured without telling participants that solitude was the construct of interest: At each momentary assessment, participants reported their thoughts and affective experiences, and then specified whether they had been “interacting with others”, “alone”, or in the company of “others nearby but not interacting”. The word “solitude” may have certain emotional connotations for research participants: For example, some of the scientific literature, and much popular writing, treats solitude as a distinctly positive kind of experience (e.g. “positive aloneness”; Galanaki, 2015). Therefore, by avoiding the word “solitude” and instead using a more objective definition of solitude (the absence of interaction), this study avoided the potential confound of capturing solitude experiences that had mostly positive connotations for participants. Moreover, individuals of different ages and cultural backgrounds might differ in how they think about solitude, or in what solitude means to them (Larson, 1990; Wang, 2006). Importantly for the present cross-cultural studies, solitude may have a somewhat different meaning when it is translated from English to Mandarin or Cantonese (Averill & Sundararajan, 2014; Wang, 2006). Hence, by not using the word solitude in the time-sampling measures, this set of studies avoided potential confounds stemming from how people interpret the word across cultures and languages.   Solitude also needs to be differentiated from being physically alone. Henry David Thoreau wrote that “A man thinking or working, is always alone, let him be where he will. Solitude is not measured by the miles of space that intervene between a man and his fellows.” (1854). Turning inward can enable a person to escape, relax, work, or create, even in a crowded   112   place such as a park, library, or restaurant. Still, however, there are key differences between “solitude in company” (for example, when sitting in silence among strangers or family members) and being completely alone. Exploratory analyses conducted for Study 2 (Table 3-1 and Table 3-2 footnotes) and Study 3 (Table 4-2 footnote) suggest that, while solitude and aloneness are both associated with increased loneliness, they are qualitatively distinct experiences. Specifically, solitude and solitude-seeking are specifically linked with reductions in high arousal positive affect (e.g. excitement), whereas aloneness and aloneness-seeking are specifically linked with increases in low arousal negative affect (e.g. tiredness) and with being at home. These findings suggest that, compared to being in solitude, being physically alone (and without social stimulation in the form of others’ presence) may have a stronger deactivating effect, and be more strongly linked with a desire to rest (Burger, 1995; Leary et al., 2003; Long & Averill, 2003). Importantly, however, when examining individual difference factors shaping affective experiences across the three studies, individual differences moderated affective experiences associated with solitude and solitude-seeking specifically. Older age (Study 2) and higher host culture acculturation (Study 3) were associated with more favorable affective experiences when in solitude (compared to when interacting with others), but did not moderate affective experiences when alone (compared to when with others). Moreover, the distinction between being alone and being with others was not predictive of different types of solitude experiences in Study 1. Hence, findings from this research program suggest that solitude and solitude-seeking may be the key constructs to employ when examining how everyday social situations, contexts, and individual differences shape affective experiences, but that aloneness and desire to be alone should also be examined to enable fine-grained distinctions among different kinds of affective experiences.    113     A key advantage of this program’s operationalization of solitude (the absence of social interaction) is that it encompasses both “solitude in company” and being alone, thus enabling these distinct forms of solitude to be disentangled. Both forms of solitude may confer unique benefits. Specifically, the presence of others may provide a sense of comfort and security, and reduce feelings of being isolated, while still allowing an individual mental space to reflect or relax (Long et al., 2003, 2006). Moreover, solitude in company need not be silent in order to be beneficial; for example, strangers’ conversations in a coffee shop may provide a welcome aural backdrop for under-stimulating work tasks. However, in other circumstances, complete aloneness and silence may be necessary, for example, when engaging in tasks requiring sustained concentration or when processing difficult emotions (Burger, 1995; Leary et al., 2003; Long & Averill, 2003). Hence, the benefits of being in solitude “with company” versus being alone may depend on situational needs and individual preferences. This research program’s operationalization of solitude does not, however, distinguish in-person communication from other forms of social interaction (e.g. texting or chatting online), leaving the situation of being alone but engaged in electronic communication unexplored. Further research is needed to examine how individuals navigate the boundary between solitude and social interaction, and their experiences of solitude and aloneness, in contexts when social interaction occurs without others being physically present (Harley, Morgan, & Frith, 2018).  5.2.2 Time-sampling approach to examine lived experiences of solitude   This research program contributes to the solitude literature by using time-sampling methods to examine different facets of solitude as it occurs in a daily life context. Previous time-sampling studies have linked solitude to specific affective states, including loneliness and low arousal positive affect (e.g. Larson, 1990; Pauly et al., 2017), but has treated these affective   114   states as being independent from one another. It is not clear based on this previous work whether, for example, loneliness and calm may occur at the same time during solitude, or whether they reflect distinct types of solitude experiences (Long et al, 2003). The present research program strongly supports this second interpretation, that there exist distinct kinds of solitude experiences, and that these are brought about by distinct motivational factors (e.g. desire for solitude), situational factors (e.g. engaging in passive leisure activities), and individual difference factors (e.g. tendency for self-rumination, host culture acculturation).   By using time-sampling methods, the present research captured thoughts and affective experiences as they occurred, reducing the risk that memory biases, self-schemas, and other influences might change self-reported experiences of solitude over time (Lay et al., 2017; Robinson & Clore, 2002). The present work builds on previous research using retrospective self-reports, which has identified specific types or dimensions of solitude (outer-directed, inner-directed, and loneliness; Long, 2000; Long et al., 2003). This work also builds on previous research using end-of-day reports, which has linked autonomous or desired solitude (as compared to undesired solitude) with enhanced wellbeing (e.g. Chua & Koestner, 2008; Nguyen et al., 2017).  5.3 Limitations and future directions 5.3.1 Other approaches to measuring solitude and solitude-seeking    The present research prioritized ecological validity, capturing solitude experiences as they occurred in participants’ daily lives. Indeed, rates of solitude captured in the present research match rates reported in previous research using adult lifespan samples (e.g. Larson, 1990). However, this correlational (time-sampling) design cannot establish causal mechanisms. For example, what is it exactly about acculturation to one’s local (host) culture that makes   115   solitude feel less lonely (Study 3)? Does acculturation operate as a measure of belongingness to broader society (de Jong Gierveld et al., 2005), or perhaps as a measure of interdependent self-construal (Markus & Kitayama, 1991)? A laboratory-based solitude induction paradigm (e.g. Nguyen et al., 2017) could be used to answer this question by manipulating potential causal mechanisms underlying host culture acculturation, including belongingness (Blackhart, Nelson, Winter, & Rockney, 2011) or interdependent self-construal (Brewer & Gardner, 1996; Trafimow et al., 1991). Such a study could also speak to the mechanisms underlying affective experiences and thought patterns during solitude (e.g. Nguyen et al., 2017; Wilson et al., 2014).   This research program is based on self-report data (subjective experiences in solitude). Participants reported on their affective experiences and thoughts as they went about their daily lives (Bolger et al., 2003; Hoppmann & Riediger, 2014), including moments of solitude and moments of social interaction. Indeed, the only way to find out what people are thinking and feeling is to ask them. Hence, this research relies on introspection – on participants being willing and able to report on their subjective experiences at any given moment. To reduce potential reactivity effects (Bolger et al., 2003) due to participants habituating to the time-sampling study procedures over the course of the 10-day period, variables for “day in study” (an integer starting from 1) and “time since study start” (in minutes) were also added to the reported multilevel models for Studies 1, 2, and 3. Neither of these variables was predictive of core outcomes nor did their inclusion change reported findings (hence they were excluded from the reported models), demonstrating the consistency of participants’ self-reports over the course of the study. Still, future research examining how solitude shapes affective wellbeing could supplement self-reports with autonomic nervous system (ANS) indices to examine, for example, whether negative solitude experiences might also be associated with physiological arousal (Kamarck, Peterman, &   116   Raynor, 1998; van der Ploeg, Brosschot, Thayer, & Verkuil, 2016). By combining self-reported affect with ANS indices, future research can work towards a fuller picture of how solitude affects people.   Future research should also build upon the measure of current solitude desire used here. The dichotomous measure used in the present studies is not able to distinguish between, for example, times when individuals are seeking solitude out of a genuine enjoyment of solitude (low social approach motivation or high solitropy) and times when individuals are seeking solitude out of desire to avoid interacting with others (high social avoidance motivation; Brown et al., 2007; Kwapil et al., 2009; Nikitin & Freund, 2010). Continuous (scale) measures of state social approach and social avoidance motivation may be particularly useful in solitude research (Brown, 1992; Brown et al., 2007). Importantly, different motivations for seeking solitude (including social anhedonia and social anxiety) may have different consequences for affective wellbeing when in solitude (Brown et al., 2007; Burger, 1995; Kwapil et al., 2009; Long & Averill, 2003). Another important reason for considering motivations underlying solitude-seeking is that these may change over the life course. Solitude-seeking may be more normative and adaptive at certain developmental stages, namely adolescence, when periods of solitude may be conducive to self-identity formation, and potentially in old age, when solitude may be associated with autonomy and self-sufficiency (Coplan et al., 2018; Larson, 1990, 1997; Larson et al., 1984). Further research is needed to compare motivations for seeking solitude at different phases of the lifespan, and to link these with solitude-seeking behaviour (for example, the length and frequency of solitude-seeking bouts).  5.3.2 Participant sample limitations, generalizability, and replicability   117     The participant samples were limited in size for practical reasons. As a result, power to detect individual differences (person-level relationships) was lower than power to detect time-varying factors (situation-level relationships) in these studies (Mathieu et al., 2012; Scherbaum & Ferreter, 2009). Participants also came from select samples, and the extent to which findings may generalize needs to be demonstrated. Specifically, the younger adult sample (Study 1) comprised university students (a WEIRD population, Henrich et al., 2010), and the Vancouver sample of adults aged 50+ had higher rates of post-secondary education than population averages (Statistics Canada, 2017). Both samples also included many more women than men. Moreover, the samples of adults aged 50+ were relatively healthy and high-functioning, and were willing and able to use tablet technology in order to participate in the study. Overall, it seems that the younger and “age 50+” participants in this set of studies are select samples, and more similar to one another than typical younger and “age 50+” adults in Canada and in China (Hong Kong Census and Statistics Department, 2017; Statistics Canada, 2017). Future research on solitude could solve problems of sample size and sample selectivity by comparing findings to large, nationally-representative samples (e.g. National Study of Daily Experiences, MIDUS study; Almeida, 2005; Almeida, Neupert, Banks, & Serido, 2005).   This research program was conceived in Vancouver, Canada, whose middle-aged and older adult population consists of a large proportion of individuals who emigrated from East or Southeast Asian countries (Statistics Canada, 2017). Hence, this research focused on differences between North American/European individuals and East/Southeast Asian individuals. However, Vancouver may be an unusual case because of its large East Asian middle-aged and older adult population (Statistics Canada, 2017); many individuals emigrating from East Asian countries moved to large communities within which they were able to carry on customary daily activities   118   in their own languages (Syed et al., 2017). It would be important to examine whether findings pertaining to solitude and loneliness generalize to other settings, for example, when comparing Hispanic immigrants and local-born individuals living in the United States (Russell & Taylor, 2009). Historical and current immigration patterns may be shaping how middle-aged and older adults experience solitude.   Finally, this set of studies took place in urban areas. Anthropological researchers have suggested that solitude-seeking may be part of a distinctly urban way of life (Coleman, 2009, 2014). Due to factors such as differences in population density and culture, it may be that experiences of solitude and loneliness in urban and suburban communities in Vancouver and Hong Kong may not translate to surrounding rural communities in British Columbia or South China (Havens, Hall, Sylvestre, & Jivan, 2004). For example, it is possible that in rural communities with lower population density and greater availability of outdoor locations for solitude-seeking, solitude may be a more common and normative experience.   The core findings reported in this dissertation (including affective experiences in solitude, age differences, and associations with desire for solitude) were tested in multiple ways to demonstrate their robustness and likelihood of being replicated. Given that my analytical approaches allowed many researcher degrees of freedom (for example, in multilevel model specification, selection and coding of variables used to test hypotheses, and inclusion/exclusion of data points and covariates; Simmons, Nelson, & Simonsohn, 2011), I examined several versions of the study models to determine under what conditions my hypotheses were and were not supported. This approach is similar to the “multiverse analysis” approach (Steegen, Tuerlinckx, Gelman, & Vanpaemel, 2016) but applied to both data cleaning decisions and data analytic decisions (Simonsohn, Simmons, & Nelson, 2015). In discussing findings and overall   119   conclusions drawn from this research program, I focus on those findings that held under an overwhelming majority of conditions. Moreover, when possible, I attempted to replicate findings derived from one sample (e.g. adults aged 50+ in Vancouver) in another sample (e.g. adults aged 50+ in Hong Kong). Although certain findings were not replicated across samples (e.g. age differences in location correlates of solitude-seeking), core findings pertaining to loneliness and experiences of solitude and solitude-seeking were robust across samples, providing initial evidence for their generalizability.  5.4 Final remarks   Solitude is a part of daily life, sometimes avoided and sometimes cherished, and experienced differently depending on the situation and on the person. Whether beneficial or harmful, motivation is a key factor shaping solitude experiences; individuals need to balance their personal needs and desires with situational demands at a given time. This research program provides initial evidence of solitude’s multifacetedness and identifies factors that may help make the best of it – a starting point for future work on this ubiquitous phenomenon.   120   References  Adler, N., & Stewart, J. (2007). The MacArthur scale of subjective social status. John D. and Catherine T. MacArthur Research Network on Socioeconomic Status and Health. Retrieved from http://www.macses.ucsf.edu/Research/Psychosocial/subjective.php . Almeida, D. M. (2005). Resilience and vulnerability to daily stressors assessed via diary methods. Current Directions in Psychological Science, 14, 64-68. doi: 10.1111/j.0963-7214.2005.00336.x Almeida, D. M., Neupert, S. D., Banks, S. R., & Serido, J. (2005). Do daily stress processes account for socioeconomic health disparities?. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 60(Special_Issue_2), S34-S39. doi: 10.1093/geronb/60.special_issue_2.s34 Antonucci, T. C. (1986). Social support networks: A hierarchical mapping technique. Generations: Journal of the American Society on Aging, 10, 10-12. Antonucci, T. C. (2001). Social relations: An examination of social networks, social support, and sense of control. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging (5th ed.) (pp. 427–453). San Diego, CA: Academic Press. Averill, J. R., & Sundararajan, L. (2014). Experiences of solitude. In R. J. Coplan & J. C. Bowker (Eds.), Handbook of solitude: Psychological perspectives on social isolation, social withdrawal, and being alone (pp. 90-108). Malden, MA: Wiley/Blackwell. Baer, R. A., Smith, G. T., Hopkins, J., Krietemeyer, J., & Toney, L. (2006). Using self-report assessment methods to explore facets of mindfulness. Assessment, 13(1), 27-45. doi: 10.1177/1073191105283504   121   Baltes, M. M., Wahl, H. W., & Schmid-Furstoss, U. (1990). The daily life of elderly Germans: Activity patterns, personal control, and functional health. Journal of Gerontology, 45(4), 173-179. doi: 10.1093/geronj/45.4.P173. Barrio, C., Palinkas, L. A., Yamada, A.-M., Fuentes, D., Criado, V., Garcia, P., & Jeste, D. V. (2008). Unmet needs for mental health services for Latino older adults: Perspectives from consumers, family members, advocates, and service providers. Community Mental Health Journal, 44(1), 57-74. doi: 10.1007/s10597-007-9112-9 Bates, M. M., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67, 1-48. doi: 10.18637/jss.v067.i01 Blackhart, G. C., Nelson, B. C., Winter, A., & Rockney, A. (2011). Self-control in relation to feelings of belonging and acceptance. Self and Identity, 10(2), 152-165. doi: 10.1080/15298861003696410 Brewer, M.B., & Gardner, W.L. (1996). Who is this ‘we’? Levels of collective identity and self representations. Journal of Personality and Social Psychology, 71, 83-93. doi: 10.1037/0022-3514.71.1.83 Bolger, N., Davis, A. & Rafaeli, E. (2003). Diary methods: Capturing life as it is lived. Annual Review of Psychology, 54(1), 579-616. doi: 10.1146/annurev.psych.54.101601.145030 Bosker, R. J., Snijders, T. A. B., & Guldemond, H. (1999). PINT,(Power IN Two-level Designs): Estimating standard errors of regression coefficients in hierarchial linear models for power calculations: User's manual. Snijders. Bowker, J. C., Nelson, L. J., Markovic, A., & Luster, S. (2014). Social withdrawal during adolescence and emerging adulthood. In R. J. Coplan & J. C. Bowker (Eds.), Handbook of   122   solitude: Psychological perspectives on social isolation, social withdrawal, and being alone (pp. 167-183). Malden, MA: Wiley/Blackwell. Brown, B. B. (1992). The ecology of privacy and mood in a shared living group. Journal of Environmental Psychology, 12(1), 5-20. doi: 10.1016/s0272-4944(05)80293-2 Brown, K. W., & Ryan, R. M. (2003). The benefits of being present: Mindfulness and its role in psychological well-being. Journal of Personality and Social Psychology, 84(4), 822. doi: 10.1037/0022-3514.84.4.822 Brown, L. H., Silvia, P. J., Myin-Germeys, I., & Kwapil, T. R. (2007). When the need to belong goes wrong: The expression of social anhedonia and social anxiety in daily life. Psychological Science, 18(9), 778-782. doi: 10.1111/j.1467-9280.2007.01978.x Burger, J. M. (1995). Individual differences in preference for solitude. Journal of Research in Personality, 29(1), 85-108. doi: 10.1006/jrpe.1995.1005   Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45(3), 1-67. doi: 10.18637/jss.v045.i03 Cacioppo, J. T., & Hawkley, L. C. (2009). Perceived social isolation and cognition. Trends in Cognitive Sciences, 13(10), 447-454. doi: 10.1016/j.tics.2009.06.005 Cacioppo, J. T., Cacioppo, S., & Boomsma, D. (2014). Evolutionary mechanisms for loneliness. Cognition and Emotion, 28, 3-21. doi: 10.1080/02699931.2013.837379 Carstensen, L. L., Fung, H. H., & Charles, S. T. (2003). Socioemotional selectivity theory and the regulation of emotion in the second half of life. Motivation and Emotion, 27(2), 103-123. doi: 10.1023/a:1024569803230   123   Carstensen, L. L., Gottman, J. M., Carstensen, L. L. (1995). Evidence for a life-span theory of socioemotional selectivity. Current Directions in Psychological Science, 4(5), 151-156. doi: 10.1111/1467-8721.ep11512261 Chan, A. B. (2013). Chinese canadians. In The Canadian Encyclopedia. Retrieved from http://www.thecanadianencyclopedia.ca/en/article/chinese-canadians/ . Chang, E. C. (1996). Cultural differences in optimism, pessimism, and coping: Predictors of subsequent adjustment in Asian American and Caucasian American college students. Journal of Counseling Psychology, 43(1), 113. doi: 10.1037/0022-0167.43.1.113 Chaudhury, H., Sarte, A. F., Michael, Y. L., Mahmood, A., Keast, E. M., Dogaru, C., & Wister, A. (2011). Use of a systematic observational measure to assess and compare walkability for older adults in Vancouver, British Columbia and Portland, Oregon neighbourhoods. Journal of Urban Design, 16, 433-454. doi: 10.1080/13574809.2011.585847 Chen, Y., Hicks, A., & While, A. E. (2014). Loneliness and social support of older people in China: A systematic literature review. Health & Social Care in the Community, 22(2), 113-123. doi: 10.1111/hsc.12051 Cheung, B. Y., Chudek, M., & Heine, S. J. (2011). Evidence for a sensitive period for acculturation: Younger immigrants report acculturating at a faster rate. Psychological Science, 22(2), 147-152. doi: 10.1177/0956797610394661 Chua, S. N., & Koestner, R. (2008). A self-determination theory perspective on the role of autonomy in solitary behavior. The Journal of Social Psychology, 148, 645-648. doi: 10.3200/SOCP.148.5.645-648 Chui, H., Hoppmann, C. A., Gerstorf, D., Walker, R., & Luszcz, M. A. (2014). Social partners and momentary affect in the oldest-old: The presence of others benefits affect depending on   124   who we are and who we are with. Developmental Psychology, 50(3), 728-740. doi:10.1037/a0033896   Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In M. Liebowitz & R. G. Heimberg (Eds.), Social phobia: Diagnosis, assessment, and treatment (pp. 69-93). New York: Guilford Press. Cohen, S. (2004). Social relationships and health. American Psychologist, 59(8), 676. doi: 10.1037/0003-066x.59.8.676 Cohen-Mansfield, J., Hazan, H., Lerman, Y., & Shalom, V. (2016). Correlates and predictors of loneliness in older-adults: A review of quantitative results informed by qualitative insights. International Psychogeriatrics, 28(4), 557-576. doi:10.1017/S1041610215001532 Coleman, L. (2014). A view from anthropology: Anomie and urban solitude. The handbook of solitude. Psychological perspectives on social isolation, social withdrawal, and being alone (pp. 483-498). Chichester, West-Sussex: Wiley Blackwell. Coleman, L. (2009). Being alone together: From solidarity to solitude in urban anthropology. Anthropological Quarterly, 82(3), 755-777. doi: 10.1353/anq.0.0075 Coplan, R. J., Ooi, L. L., & Baldwin, D. (2018). Does it matter when we want to be alone? Exploring developmental timing effects in the implications of unsociability. New Ideas in Psychology. Advance online version. doi: 10.1016/j.newideapsych.2018.01.001 Coplan, R. J., Prakash, K., O'Neil, K., & Armer, M. (2004). Do you" want" to play? Distinguishing between conflicted shyness and social disinterest in early childhood. Developmental Psychology, 40(2), 244. doi: 10.1037/0012-1649.40.2.244   125   Cornwell, E. Y., & Waite, L. J. (2009). Social disconnectedness, perceived isolation, and health among older adults. Journal of Health and Social Behavior, 50(1), 31-48. doi: 10.1177/002214650905000103 Costa, P. T., & McCrae, R. R. (1992). Revised NEO personality inventory (NEO-PI-R) and NEO five-factor inventory (NEO-FFI). Psychological Assessment Resources. doi: 10.1037/t03907-000 Cranford, J. A., Shrout, P. E., Iida, M., Rafaeli, E., Yip, T., & Bolger, N. (2006). A procedure for evaluating sensitivity to within-person change: Can mood measures in diary studies detect change reliably?. Personality and Social Psychology Bulletin, 32(7), 917-929. doi: 10.1177/0146167206287721 Dahlberg, K. (2007). The enigmatic phenomenon of loneliness. International Journal of Qualitative Studies on Health and Well-being, 2(4), 195-207. doi: 10.3402/qhw.v2i4.4960 de Jong Gierveld, J., Van der Pas, S., & Keating, N. (2015). Loneliness of older immigrant groups in Canada: Effects of ethnic-cultural background. Journal of Cross-Cultural Gerontology, 30(3), 251-268. doi: 10.1007/s10823-015-9265-x de Jong Gierveld, J., van Tilburg, T., & Dykstra, P. A. (2005). Loneliness and social isolation. In A. Vangelisti & D. Perlman (Eds.), The Cambridge handbook of personal relationships (pp. 485-500). New York, NY: Cambridge University Press.  Di Giunta, L., Eisenberg, N., Kupfer, A., Steca, P., Tramontano, C. & Caprara, G. V. (2010). Assessing perceived empathic and social self-efficacy across countries. European Journal of Psychological Assessment, 26, 77-86. doi: 10.1027/1015-5759/a000012   126   Emami, A., Torres, S., Lipson, J. G., & Ekman, S. L. (2000). An ethnographic study of a day care center for Iranian immigrant seniors. Western Journal of Nursing Research, 22(2), 169-188. doi: 10.1177/01939450022044340 Ernst, J. M., & Cacioppo J. T. (2000). Lonely hearts: Psychological perspectives on loneliness. Applied and Preventive Psychology, 8, 1-22. doi: 10.1016/s0962-1849(99)80008-0 Epley, N., & Schroeder, J. (2014). Mistakenly seeking solitude. Journal of Experimental Psychology: General, 143, 1980-1999. doi: 10.1037/a0037323 Farb, N. A., Anderson, A. K., & Segal, Z. V. (2012). The mindful brain and emotion regulation in mood disorders. The Canadian Journal of Psychiatry, 57(2), 70-77. doi: 10.1177/070674371205700203 Feldman Barrett, L., & Russell, J. A. (1998). Independence and bipolarity in the structure of current affect. Journal of Personality and Social Psychology, 74, 967. doi: 10.1037/0022-3514.74.4.967 Fergus, T. A., Valentiner, D. P., McGrath, P. B., Gier-Lonsway, S. L. & Kim, H.-S. (2012). Short forms of the social interaction anxiety scale and the social phobia scale. Journal of Personality Assessment, 94(3), 310-320. doi: 10.1080/00223891.2012.660291 Fung, H. H. (2013). Aging in culture. The Gerontologist, 53(3), 369-377. doi: 10.1093/geront/gnt024 Fung, H. H., Carstensen, L. L., & Lutz, A. M. (1999). Influence of time on social preferences: Implications for life-span development. Psychology and Aging, 14(4), 595. doi: 10.1037/0882-7974.14.4.595   127   Fung, H. H., Lai, P., Ng, R. (2001). Age differences in social preferences among Taiwanese and Mainland Chinese: The role of perceived time. Psychology and Aging, 16, 351-356. doi: 10.1037/0882-7974.16.2.351 Galanaki, E. P. (2015). Solitude as a state of positive aloneness in childhood and adolescence. In C. J. Kowalski, J. P. Cangemi, & A. Rokach (Eds.), Loneliness in life: Education, business, and society (pp. 168-190). Dubuque, IA: McGraw-Hill.  Goffman, E. (2008). Behavior in public places. New York, NY: Simon and Schuster. Green, A. S., Rafaeli, E., Bolger, N., Shrout, P. E., & Reis, H. T. (2006). Paper or plastic? Data equivalence in paper and electronic diaries. Psychological Methods, 11, 87. doi: 10.1037/1082-989x.11.1.87 Harley, D., Morgan, J., & Frith, H. (2018). Being alone. In Cyberpsychology as everyday digital experience across the lifespan (pp. 153-173). London: Palgrave Macmillan. Harrison, T. (2018, January 15). Alone together: Who’s lonely and how do we measure it? [Blog post]. Retrieved from https://www.thersa.org/discover/publications-and-articles/rsa-blogs/2018/01/alone-together-whos-lonely-and-how-do-we-measure-it Hartman, L. M. (1983). A metacognitive model of social anxiety: Implications for treatment. Clinical Psychology Review, 3(4), 435-456. doi: 10.1016/0272-7358(83)90023-5 Havens, B., Hall, M., Sylvestre, G., & Jivan, T. (2004). Social isolation and loneliness: Differences between older rural and urban Manitobans. Canadian Journal on Aging/la revue canadienne du vieillissement, 23(2), 129-140. doi: 10.1353/cja.2004.0022 Hawkley, L. C., & Cacioppo, J. T. (2010). Loneliness matters: A theoretical and empirical review of consequences and mechanisms. Annals of Behavioral Medicine, 40(2), 218-227. doi: 10.1007/s12160-010-9210-8   128   Hays, J. C. (2002). Living arrangements and health status in later life: A review of recent literature. Public Health Nursing, 19(2), 136-151. doi: 10.1046/j.1525-1446.2002.00209.x Heine, S. J., Lehman, D. R., Markus, H. R., & Kitayama, S. (1999). Is there a universal need for positive self-regard?. Psychological Review, 106(4), 766. doi: 10.1037/0033-295x.106.4.766 Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33, 61-83. doi: 10.2139/ssrn.1601785 Henry, K. L., & Muthén, B. (2010). Multilevel latent class analysis: An application of adolescent smoking typologies with individual and contextual predictors. Structural Equation Modeling, 17, 193-215. doi: 10.1080/10705511003659342 Heylen L. (2010). The older, the lonelier? Risk factors for social loneliness in old age. Ageing & Society 30, 1177-1196. doi: 10.1017/s0144686x10000292 Home Affairs Department and Immigration Department. (2011). Home Affairs Department and Immigration Department Statistics on new arrivals from the Mainland (Third Quarter of 2011). Retrieved from http://www.legco.gov.hk/yr08-09/chinese/hc/sub_com/hs52/papers/hs520614cb2-766-1-ec.pdf  Hong Kong Census and Statistics Department. (2017). 2016 Population Census: Hong Kong Government Special Administrative Region. Retrieved from https://www.statistics.gov.hk/pub/B11200942016XXXXB0100.pdf Hoppmann, C. A., & Gerstorf, D. (2016). Social interrelations in aging: The sample case of married couples. In K. W. Schaie & S. L. Willis (Eds.), Handbook of the psychology of aging (8th Edition) (pp. 263-277). San Diego: Elsevier. doi: 10.1016/B978-0-12-411469-2.00014-5In  Hoppmann, C. A., & Riediger, M. (2009). Ambulatory assessment in lifespan psychology: An   129   overview of current status and new trends. European Psychologist, 14, 98-108. doi: 10.1027/1016-9040.14.2.98 Huffziger, S., Ebner-Priemer, U., Eisenbach, C., Koudela, S., Reinhard, I., Zamoscik, V., ... & Kuehner, C. (2013). Induced ruminative and mindful attention in everyday life: An experimental ambulatory assessment study. Journal of Behavior Therapy and Experimental Psychiatry, 44(3), 322-328. doi: 10.1016/j.jbtep.2013.01.007 Isaacowitz, D. M. & Blanchard-Fields, F. (2012). Linking process and outcome in the study of emotion and aging. Perspectives on Psychological Science, 7, 3-17. doi: 10.1177/1745691611424750 Klumb, P. L. (2004). Benefits from productive and consumptive activities: Results from the Berlin Aging Study. Social Indicators Research, 67, 107-127. doi: 10.1023/b:soci.0000007336.64239.a6 Jiang, D., Fung, H. H., Lay, J. C., Ashe, M. C., Graf, P., & Hoppmann, C. A. (in press). Everyday solitude and well-being in old age: The role of culture versus immigration. Aging and Mental Health. doi: 10.1080/13607863.2018.1479836 Jo Cox Commission on Loneliness (2017). Combatting loneliness one conversation at a time: A call to action. Retrieved from https://www.jocoxloneliness.org Jylhä, M., & Saarenheimo, M. (2010). Loneliness and aging: Comparative perspectives. In C. R. Phillipson & D. Dannefer (Eds.), The SAGE handbook of social gerontology (pp. 317–328). Los Angeles/London: SAGE. doi: 10.4135/9781446200933.n24 Kamarck, T. W., Peterman, A. H., & Raynor, D. A. (1998). The effects of the social environment on stress-related cardiovascular activation: Current findings, prospects, and implications. Annals of Behavioral Medicine, 20(4), 247-256. doi: 10.1007/bf02886374   130   Kashdan, T. B., & Steger, M. F. (2006). Expanding the topography of social anxiety: An experience-sampling assessment of positive emotions, positive events, and emotion suppression. Psychological Science, 17, 120–128. doi: 10.1111/j.1467-9280.2006.01674.x Koch, P. J. (1994). Solitude: A philosophical encounter. Chicago: Open Court. Koch, P. J. (1989). Solitude in Ancient Taoism. Diogenes, 37(148), 78-91. doi: 10.1177/039219218903714805 Kwapil, T. R., Silvia, P. J., Myin-Germeys, I., Anderson, A. J., Coates, S. A., & Brown, L. H. (2009). The social world of the socially anhedonic: Exploring the daily ecology of asociality. Journal of Research in Personality, 43(1), 103-106. doi: 10.1016/j.jrp.2008.10.008 Lang, F. R., & Baltes, M. M. (1997). Being with people and being alone in late life: Costs and benefits for everyday functioning. International Journal of Behavior Development, 21, 729-746. doi: 10.1080/016502597384640 Lang, F. R., & Carstensen, L. L. (1994). Close emotional relationships in late life: Further support for proactive aging in the social domain. Psychology and Aging, 9(2), 315. doi: 10.1037/0882-7974.9.2.315 Larson, R. W. (1997). The emergence of solitude as a constructive domain of experience in early adolescence. Child development, 68(1), 80-93. doi: 10.2307/1131927 Larson, R. W. (1990). The solitary side of life: An examination of the time people spend alone from childhood to old age. Developmental Review, 10, 155-183. doi: 10.1016/0273-2297(90)90008-R Larson, R., & Csikszentmihalyi, M. (1978). Experiential correlates of time alone in adolescence. Journal of Personality, 46(4), 677-693. doi: 10.1007/978-94-017-9094-9_13   131   Larson, R., Csikszentmihalyi, M., & Graef, R. (1982). Time alone in daily experience: Loneliness or renewal? In L. A. Peplau & D. Perlman (Eds.), Loneliness: A sourcebook of current theory, research, and therapy (pp. 40–53). New York: Wiley-Interscience. Larson, R., Zuzanek, J., & Mannell, R. (1985). Being alone versus being with people: Disengagement in the daily experience of older adults. Journal of Gerontology, 40, 375-381. doi: 10.1093/geronj/40.3.375 Lay, J. C., Gerstorf, D., Scott, S. B., Pauly, T., & Hoppmann, C. A. (2017). Neuroticism and extraversion magnify discrepancies between retrospective and concurrent affect reports. Journal of Personality, 85(6), 817-829. doi: 10.1111/jopy.12290 Lay, J. C., Pauly, T., Graf, P., Biesanz, J. C., & Hoppmann, C. A. (in press). By myself and liking it? Predictors of distinct types of solitude experiences in daily life. Journal of Personality. Lay, J. C., Pauly, T., Graf, P., Mahmood, A., & Hoppmann, C. A. (2018). Choosing solitude: Age differences in situational and affective correlates of solitude-seeking in midlife and older adulthood. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences. Advance online publication. doi: 10.1093/geronb/gby044 Leary, M. R., Herbst, K. C., & McCrary, F. (2003). Finding pleasure in solitary activities: Desire for aloneness or disinterest in social contact?. Personality and Individual Differences, 35, 59-68. doi: 10.1016/S0191-8869(02)00141-1 Liu, L. J., & Guo, Q. (2007). Loneliness and health-related quality of life for the empty nest elderly in the rural area of a mountainous county in China. Quality of Life Research, 16(8), 1275-1280. doi: 10.1007/s11136-007-9250-0 Long, C.R. (2000). A comparison of positive and negative episodes of solitude. Unpublished   132   master's thesis. University of Massachusetts, Amherst. Long, C. R. & Averill, J. R. (2003). Solitude: An exploration of benefits of being alone. Journal for the Theory of Social Behaviour, 33, 21-44. doi: 10.1111/1468-5914.00204 Long, C., More, T. A., & Averill, J. R. (2007). The subjective experience of solitude. In Proceedings of the 2006 Northeastern Recreation Research Symposium. Newtown Square, PA. Long, C. R., Seburn, M., Averill, J. R. & More, T. A. (2003). Solitude experiences: Varieties, settings, and individual differences. Personality and Social Psychology Bulletin, 29, 578-583. doi: 10.1177/0146167203029005003 Maes, M., Wang, J. M., van den Noortgate, W., & Goossens, L. (2015). Loneliness and attitudes toward being alone in Belgian and Chinese adolescents: Examining measurement invariance. Journal of Child and Family Studies, 25, 1408-1415. doi: 10.1007/s10826-015-0336-y Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98(2), 224. doi: 10.1037/0033-295x.98.2.224 Mason, M. F., Norton, M. I., Van Horn, J. D., Wegner, D. M., Grafton, S. T., & Macrae, C. N. (2007). Wandering minds: The default network and stimulus-independent thought. Science, 315, 393-395. doi: 10.1126/science.1131295 Masyn, K. E. (2013). Latent class analysis and finite mixture modeling. In T. D. Little (Ed.), The Oxford handbook of quantitative methods, volume 2: Statistical analysis (pp. 551-611). New York, NY: Oxford University Press. Mather, M., & Carstensen, L. L. (2005). Aging and motivated cognition: The positivity effect in attention and memory. Trends in Cognitive Sciences, 9, 496-502. doi: 10.1016/j.tics.2005.08.005   133   Mathieu, J. E., Aguinis, H., Culpepper, S. A., & Chen, G. (2012). Understanding and estimating the power to detect cross-level interaction effects in multilevel modeling. Journal of Applied Psychology, 97(5), 951. doi: 10.1037/a0029358 Matias, G. P., Nicolson, N. A., & Freire, T. (2011). Solitude and cortisol: Associations with state and trait affect in daily life. Biological Psychology, 86(3), 314-319. doi: 10.1016/j.biopsycho.2010.12.011 McAdams, D. P. & Constantian, C. A. (1983). Intimacy and affiliation motives in daily living: An experience sampling analysis. Journal of Personality and Social Psychology, 45, 851. doi: 10.1037/0022-3514.45.4.851 Mikkelsen, H. H. (2016). Unthinkable solitude: Successful aging in Denmark through the Lacanian Real. Ethos, 44(4), 448-463. doi: 10.1111/etho.12144 Mooradian, T. A., & Swan, K. S. (2006). Personality-and-culture: The case of national extraversion and word-of-mouth. Journal of Business Research, 59(6), 778-785. doi: 10.1016/j.jbusres.2006.01.015 Muthén, L. K. and Muthén, B. (2007). Mplus User’s Guide. Los Angeles: Muthén & Muthén. Nagin, D. S. (2005). Group-based modeling of development. Cambridge, MA: Harvard University Press. Nikitin, J., & Freund, A. M. (2010). When wanting and fearing go together: The effect of co‐occurring social approach and avoidance motivation on behavior, affect, and cognition. European Journal of Social Psychology, 40(5), 783-804. doi: 10.1002/ejsp.650 Nguyen, T. T., Ryan, R. M., & Deci, E. L. (2017). Solitude as an approach to affective self-regulation. Personality and Social Psychology Bulletin, 44, 92-106. doi: 10.1177/0146167217733073   134   Pauly, T., Lay, J. C., Nater, U. M., Scott, S. B., & Hoppmann, C. A. (2017). How we experience being alone: Age differences in affective and biological correlates of momentary solitude. Gerontology, 63(1), 55-66. doi: 10.1159/000450608 Pauly, T., Lay, J. C., Scott, S. B., & Hoppmann, C. A. (in press). Social relationship quality buffers negative affective correlates of everyday solitude in an adult lifespan and an older adult sample. Psychology and Aging. doi: 10.1037/pag0000278 Pedersen, D. M. (1999). Model for types of privacy by privacy functions. Journal of Environmental Psychology, 19(4), 397-405. doi: 10.1006/jevp.1999.0140 Pedersen, D. M. (1997). Psychological functions of privacy. Journal of Environmental Psychology, 17(2), 147-156. doi: /10.1006/jevp.1997.0049 Perlman, D., & Peplau, L. A. (1981). Toward a social psychology of loneliness. In R. Gilmour & S. Duck (Eds.), Personal relationships: 3. Relationships in disorder (pp. 31-56). London: Academic Press. Perissinotto, C. M., Cenzer, I. S., & Covinsky, K. E. (2012). Loneliness in older persons: A predictor of functional decline and death. Archives of Internal Medicine, 172(14), 1078-1084. doi: 10.1001/archinternmed.2012.1993 Pinquart, M., & Sorensen, S. (2001). Influences on loneliness in older adults: A meta-analysis. Basic and Applied Social Psychology, 23(4), 245-266. doi: 10.1207/153248301753225702 Riediger, M., Schmiedek, F., Wagner, G. G., & Lindenberger, U. (2009). Seeking pleasure and seeking pain: Differences in prohedonic and contra-hedonic motivation from adolescence to old age. Psychological Science, 20, 1529-1535. doi: 10.1111/j.1467-9280.2009.02473.x   135   Robinson, M. D., & Clore, G. L. (2002). Belief and feeling: Evidence for an accessibility model of emotional self-report. Psychological Bulletin, 128(6), 934. doi: 10.1037/0033-2909.128.6.934 Röcke, C., Li, S. C., & Smith, J. (2009). Intraindividual variability in positive and negative affect over 45 days: Do older adults fluctuate less than young adults?. Psychology and Aging, 24, 863-878. doi: 10.1037/a0016276 Rokach, A. (2004). Loneliness then and now: Reflections on social and emotional alienation in everyday life. Current Psychology, 23(1), 24-40. doi: 10.1007/s12144-004-1006-1 Rowe, J. W., & Kahn, R. L. (1997). Successful aging. The Gerontologist, 37(4), 433-440. doi: 10.1093/geront/37.4.433 Rubin, K. H., Cheah, C. S. L. & Menzer, M. (2009). Peer relationships. In M. Bornstein (Ed.), Handbook of cross-cultural developmental science (pp. 223-238). New York: Psychology Press. Russell, D. W. (1996). UCLA loneliness scale (Version 3): Reliability, validity, and factor structure. Journal of Personality Assessment, 66, 20-40. doi: 10.1207/s15327752jpa6601_2 Russell, D., & Taylor, J. (2009). Living alone and depressive symptoms: The influence of gender, physical disability, and social support among Hispanic and non-Hispanic older adults. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 64(1), 95-104. doi: 10.1093/geronb/gbn002 Ryder, A. G., Alden, L. E., & Paulhus, D. L. (2000). Is acculturation unidimensional or bidimensional? A head-to-head comparison in the prediction of personality, self-identity, and adjustment. Journal of Personality and Social Psychology, 79, 49-65. doi: 10.1037/0022-3514.79.1.49   136   Ryff, C. D. & Keyes, C. L. M. (1995). The structure of psychological well-being revisited. Journal of Personality and Social Psychology, 69, 719. doi: 10.1037/0022-3514.69.4.719 Scherbaum, C. A., & Ferreter, J. M. (2009). Estimating statistical power and required sample sizes for organizational research using multilevel modeling. Organizational Research Methods, 12(2), 347-367. doi: 10.1177/1094428107308906 Seeman, T. E. (1996). Social ties and health: The benefits of social integration. Annals of Epidemiology, 6, 442-451. doi: 10.1016/s1047-2797(96)00095-6 Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22(11), 1359-1366. doi: 10.1037/e519702015-014 Simonsohn, U., Simmons, J. P. & Nelson, L. D. (2015). Specification curve: Descriptive and inferential statistics on all reasonable specifications. SSRN Electronic Journal. doi: 10.2139/ssrn.2694998 Snijders, T. A. B. (2005). Power and sample size in multilevel linear models. In B. S. Everitt & D. C. Howell (Eds.), Encyclopedia of statistics in behavioral science, volume 3 (pp. 1570-1573). Chicester: Wiley. Snijders, T. A. B. & Bosker, R. J. (2002). Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage Publications, Inc. Spurr, J. M., & Stopa, L. (2002). Self-focused attention in social phobia and social anxiety. Clinical Psychology Review, 22, 947-975. doi: 10.1016/S0272-7358(02)00107-1 Statistics Canada. (2017). Vancouver [Population centre], British Columbia and British Columbia [Province] (table). Census Profile. 2016 Census. Statistics Canada Catalogue no.   137   98-316-X2016001. Ottawa, Ontario. Retrieved from http://www12.statcan.gc.ca/census-recensement/2016/dp-pd/prof/index.cfm Statistics Canada. (2012). Living Arrangements of Seniors. Statistics Canada Catalogue no. 98-312-X2011003. Ottawa, Ontario. Retrieved from http://www12.statcan.gc.ca/census-recensement/2011/as-sa/98-312-x/98-312-x2011003_4-eng.cfm  Steegen, S., Tuerlinckx, F., Gelman, A., & Vanpaemel, W. (2016). Increasing transparency through a multiverse analysis. Perspectives on Psychological Science, 11(5), 702-712. doi: 10.1177/1745691616658637 Stewart, M., Shizha, E., Makwarimba, E., Spitzer, D., Khalema, E. N., & Nsaliwa, C. D. (2011). Challenges and barriers to services for immigrant seniors in Canada: “you are among others but you feel alone.” International Journal of Migration, Health and Social Care, 7, 16-32. doi: 10.1108/17479891111176278 Suh, E., Diener, E., Oishi, S., & Triandis, H. C. (1998). The shifting basis of life satisfaction judgments across cultures: Emotions versus norms. Journal of Personality and Social Psychology, 74(2), 482. doi: 10.1037//0022-3514.74.2.482 Syed, M. A., McDonald, L., Smirle, C., Lau, K., Mirza, R. M., & Hitzig, S. L. (2017). Social isolation in Chinese older adults: Scoping review for age-friendly community planning. Canadian Journal on Aging/La Revue canadienne du vieillissement, 36(2), 223-245. doi: 10.1017/s0714980817000101 Thoreau, H. D. (1854). Walden; or, Life in the Woods. Boston, MA: Ticknor and Fields. Tieu, Y., & Konnert, C. (2015). Measuring acculturation and enculturation among Chinese Canadian older adults. Canadian Journal on Aging/La Revue Canadienne du Vieillissement, 34(01), 36-46. doi: 10.1017/s071498081400049x   138   Trafimow, D., Triandis, H.C., & Goto, S.G. (1991). Some tests of the distinction between the private self and the collective self. Journal of Personality and Social Psychology, 60, 649-655. doi: 10.1037/0022-3514.60.5.649 Trapnell, P. D., & Campbell, J. D. (1999). Private self-consciousness and the five-factor model of personality: Distinguishing rumination from reflection. Journal of Personality and Social Psychology, 76, 284. doi: 10.1037/0022-3514.76.2.284 Triandis, H. C. (1988). Collectivism vs. individualism: A reconceptualization of a basic concept in cross-cultural social psychology. In G. K. Verma & C. Bagley (Eds.), Cross-cultural studies of personality, attitudes and cognition (pp. 60–95). New York: St. Martin Press. Tsai, J. L. (2007). Ideal affect: Cultural causes and behavioral consequences. Perspectives on Psychological Science, 2, 242-259. doi: 10.1111/j.1745-6916.2007.00043.x Tsai, J. L., Knutson, B., & Fung, H. H. (2006). Cultural variation in affect valuation. Journal of Personality and Social Psychology, 90, 288. doi: 10.1037/0022-3514.90.2.288 Tsai, J. L., Triandis, H. C., & Suh, E. M. (2002). Cultural influences on personality. Annual Review of Psychology, 53(1), 133-160. doi: 10.1146/annurev.psych.53.100901.135200 United Nations, Department of Economic and Social Affairs, Population Division (2015). World Population Prospects: The 2015 Revision, Volume II: Demographic Profiles. ST/ESA/SER.A/380. Urry, H. L., & Gross, J. J. (2010). Emotion regulation in older age. Current Directions in Psychological Science, 19, 352-357. doi: 10.1177/0963721410388395 van der Ploeg, M. M., Brosschot, J. F., Thayer, J. F., & Verkuil, B. (2016). The implicit positive and negative affect test: Validity and relationship with cardiovascular stress-responses. Frontiers in Psychology, 7. doi: 10.3389/fpsyg.2016.00425   139   van Zyl, C. J., Dankaert, E., & Guse, T. (2018). Motivation for solitude: A cross-cultural examination of adolescents from collectivist and individualist cultures in South Africa. Journal of Child and Family Studies, 27(3), 697-706. doi: 10.1007/s10826-017-0916-0 Vermunt, J. K. (2003). Multilevel latent class models. Sociological Methodology, 33, 213-239. doi: 10.1111/j.0081-1750.2003.t01-1-00131.x Vermunt, J. K. (2010). Latent class modeling with covariates: Two improved three-step approaches. Political Analysis, 450-469. doi: 10.1093/pan/mpq025 Vermunt, J. K., & Magidson, J. (2002). Latent class cluster analysis. Applied Latent Class Analysis, 11, 89-106. doi: 10.1017/cbo9780511499531.004 Wang, Y. (2006). Culture and solitude: Meaning and significance of being alone. Unpublished master's thesis, University of Massachusetts, Amherst. Wethington, E., & Pillemer, K. (2014). Social isolation among older people. In R. J. Coplan & J. C. Bowker (Eds.), Handbook of solitude: Psychological perspectives on social isolation, social withdrawal, and being alone (pp. 242-259). Malden, MA: Wiley/Blackwell. Winnicott, D. W. (1958). The capacity to be alone. The International Journal of Psycho-analysis, 39(5), 416-420. doi: 10.1093/med:psych/9780190271374.003.0060 Wong, W. K. F., Chou, K.-L., & Chow, N. W. S. (2012). Correlates of quality of life in new migrants to Hong Kong from Mainland China. Social Indicators Research, 107(2), 373-391. doi: 10.1007/s11205-011-9853-2 Wu, Z., & Penning, M. (2015). Immigration and loneliness in later life. Ageing & Society, 35(1), 64-95. doi: 10.1017/s0144686x13000470   140   Zhang, X., Yeung, D. Y., Fung, H. H., & Lang, F. R. (2011). Changes in peripheral social partners and loneliness over time: The moderating role of interdependence. Psychology and Aging, 26(4), 823. doi: 10.1037/a0023674     141   Appendices Appendix A: Study 1 multilevel latent profile analysis procedure and results      Of the 4571 momentary assessments collected from our 150-participant sample, 2944 were classified as solitude episodes. We used multilevel latent profile analysis (LPA) to classify these solitude episodes into a set of latent classes or solitude experience types, based on their momentary affect-thought profiles (4 affect dimensions and 2 thought dimensions). A final model (set of solitude classes) was selected based on model fit indices, residuals, classification diagnostics, parsimony, and theoretical considerations. We used the following procedure, based on recommendations for LPA model selection and class enumeration provided by Masyn (2013). Step 1: Identify a set of model types to test  In a multilevel LPA context, choosing which model types to test involves making decisions about (1) the covariance structure of the class indicators (momentary affect/thought dimensions), and (2) how to account for the nested structure of the data (solitude episodes and momentary affect/thought indicators nested within people).   For all reported models, indicator variances and covariances were allowed to vary within and between classes (Masyn, 2013). We had no reason to believe that (a) different indicators would have the same variance within a solitude class, that (b) the same indicator would have the same variance across solitude classes, or that (c) indicators would all have the same covariances within or across classes. Hence, we did not impose any such constraints on the indicator covariance structure.   Multilevel LPA also accounts for the nested data structure by allowing the latent class means (defining Level 1 solitude class membership) to vary across Level 2 units (people). In   142   other words, each of the K-1 latent class variables for a K-class model has its own random intercept (Henry & Muthén, 2010). This allows us to model person-level influences on solitude class membership, that is, to account for the possibility that people may vary in their propensity to experience one type of solitude over another. All of our reported models account for the nested data structure in this way. In addition, multilevel LPA allows the option of adding another random intercept for the thought/affect indicators themselves. This indicator-specific random intercept enables us to model person-level clustering of momentary affect/thought dimensions, independently of solitude class membership (Henry & Muthén, 2010). In other words, people might differ in their mean levels of high arousal positive affect, high cognitive effort thought, etc. in ways that are not accounted for by classifying their momentary affect/thought profiles into distinct types of solitude experiences. We found it reasonable to suppose that, in addition to solitude class propensities varying between people, momentary affect and thoughts might also vary between people, and that accounting for this between-person variability in indicator means might improve model fit. Hence, we tested two model types: one without an indicator-specific random intercept and one with an indicator random intercept (in the form of one common random factor for the 6 affect/thought dimensions; Henry & Muthén, 2010). We report 1-, 2-, and 3-class solutions for these two model types.  A final consideration in multilevel LPA is whether to use a parametric or a non-parametric approach to model the random latent class means. In the parametric approach (the approach we use here), the latent class means at Level 1 are assumed to be normally distributed across Level 2 units (Henry & Muthén, 2010; Vermunt, 2003). In other words, this assumes individuals are normally distributed in terms of their mean propensity to experience one type of solitude over another. The non-parametric approach, in contrast, does not make the assumption   143   of normally distributed random class means. Instead, the K-1 random means from the K Level 1 classes are used as indicators of a second set of latent classes at Level 2 (Henry & Muthén, 2010; Vermunt, 2003). This means that, in addition to having distinct solitude classes at Level 1 (situation level), we would also have distinct solitude-propensity classes at Level 2 (person level), e.g. “people who always have positive solitude experiences” and “people who have a mix of positive and negative solitude experiences”. We had no a priori hypotheses regarding different types or latent classes of people (above and beyond different types of solitude experiences). Hence, fitting and comparing non-parametric models is not necessary to answer our research questions, and is beyond the scope of this manuscript. For all reported models, we used the parametric approach; this allows us to test our hypotheses by identifying solitude classes and examining situation-level and person-level predictors of having these different types of solitude experiences. Step 2: Generate 1-class, 2-class, 3-class, etc. models for all model types   For each of the two model types, we generated a series of models, starting with a 1-class model, and increasing the number of classes until the model was no longer well-identified. The results are summarized in Appendix A, Table A-1. For model type 1 (variances/covariances vary across classes, no indicator random intercept), we generated a 1-class, a 2-class, and a 3-class model, and the 4-class model was not identifiable. We did the same for model type 2 (variances/covariances vary across classes, with indicator-specific random intercept): 1-class, 2-class, and 3-class solutions were generated, and the 4-class solution was not identifiable. Notably, for both model types, the 3-class solutions presented convergence issues, whereas the 2-class solutions did not. Step 3: Compare fit indices (Scree plots)   144     We examined fit indices for each of the six models generated (1-, 2-, and 3-class solutions for model type 1, and 1-, 2-, and 3-class solutions for model type 2). The Aikike Information Criterion (AIC) and Adjusted Bayesian Information Criterion (BIC) for each model are presented in Appendix A, Table A-1; these allow us to compare models of the same type, with differing numbers of classes (i.e. they do not allow us to compare solutions for model type 1 vs. model type 2). Smaller values indicate better model fit. Because only 1-, 2-, and 3-class solutions were identifiable, the AIC and BIC Scree plots for the two model types were not able to show a clear “elbow” indicating the optimal number of classes for that model type. Hence, at this stage, all the 2- and 3-class models seemed to be viable candidate models for identifying different types of solitude (4 candidate models in total). Step 4: Compare model residuals and classification indices  Model residuals and classification indices were examined for all models generated and are summarized in Supplementary Materials A, Table A-1. a. Residuals for the indicator means, variances, covariances, univariate skewness, and univariate kurtosis are indicators of absolute model fit. As shown in the table, introducing an indicator-specific random intercept resulted in very large residuals for the means and variances/covariances, and particularly so for the 2-class model. This model specification (the inclusion of an indicator random intercept) seems to be a poor fit to the data. b. Model entropy is an overall summary of latent class assignment error. Values range from 0 to 1, and values near 0 may indicate model misfit (Masyn, 2013). Entropy for all four candidate models is acceptable, and is highest for model type 1, 2-class model. c. Average posterior class probability (AvePP) was computed for each class. Values above 0.7 indicate good class separation and classification accuracy (Nagin, 2005). All   145   candidate model classes were well above this threshold, and AvePP values were highest for model type 1, 2-class model. d. Odds of correct classification (OCC) was computed for each class based on modal class assignments. Values above 5 indicate good class separation and assignment accuracy (Nagin, 2005), and again, all candidate model classes were well above threshold.  Step 5: Examine class homogeneity and class separation indices Class homogeneity and class separation indices are provided in Table 1 of the main manuscript for the final selected model (model type 1, 2-class).  Smaller within-class variances and covariances, as compared to overall sample values, indicate greater class homogeneity. As shown in Table 2-1, all Class 1 and Class 2 indicator variances are smaller than their respective overall sample variances, and this is especially true for the negative affect and high cognitive effort thought dimensions. Overall, Class 2 (“positive” solitude) is more homogeneous than Class 1 (“negative” solitude). Class separation was assessed by calculating the standardized mean distances between classes; this is a class indicator-specific adaptation of Cohen’s d. Values greater than 2 indicate less than 20% overlap in class distributions, i.e. high separation between classes on that particular indicator. Values less than 0.85 indicate greater than 50% overlap in class distributions, i.e. low class separation on that indicator. Table 2-1 values indicate a particularly high degree of separation between solitude Class 1 and Class 2 on the high arousal negative affect dimension, and good class separation on the low arousal negative affect, high cognitive effort thought, and low arousal positive affect dimensions. Class separation is poor on the high arousal positive affect and low cognitive effort thought dimensions. Step 6: Examine class contents to select a final model   146    For each of the top candidate models (identified based on model fit, residuals, and classification indices) the final step in model selection is to examine the class contents, taking into consideration parsimony and theoretical meaningfulness: “to what extent do these classes reflect qualitatively distinct types of solitude experiences?” Based on their high residuals, models with indicator-specific random effects were removed from the pool of candidate models (Step 4 above). The remaining top-candidate models were the 2-class model and 3-class models for model type 1 (no indicator-specific random effects). In general, parsimony considerations would suggest we pick the model with the lower number of classes, i.e., the 2-class model (Masyn, 2013). Moreover, inspection of class contents and class separation indices suggested that in the 3-class model, two of the solitude classes were in fact very similar to one another, indicating they were capturing redundant information. The existence of these two redundant classes also made little theoretical sense because one class was characterized by higher means on all six affect/thought dimensions than the other class. Therefore, we selected the 2-class model as our final model. Table 2-1 in the manuscript provides this final model’s class-specific means and standard deviations for the 6 affect/thought dimensions, and Figure 2 plots the means for the two classes. Further elaboration on the solitude classes’ theoretical meaning is provided in the main manuscript.  147   Table A-1: Identifying a set of distinct types of solitude experiences using latent profile analysis: Class proportions, model fit indices, residuals, and classification indices for candidate models Model type Number  of classes Number  of free  para-meters Class  pro-portions Model fit indices  Largest residuals  Classification indices AIC Adjusted BIC  Mixed means Mixed variances Mixed co-variances Mixed univariate skewness Mixed univariate kurtosis  Entropy AvePP OCC Model  Type 1: Variances/ covariances vary across classes 1 27 1  1.00 68268.43 68344.31  n/a n/a n/a n/a n/a  n/a 1  1.00 n/a 2 56 1  0.57 2  0.43 65251.48 65408.85  0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.24  0.23  0.18  0.65 -0.29  0.20  0.891 1  0.98 2  0.96 1  30.04:1 2  33.79:1 3 86 1  0.20 2  0.36 3  0.44 63528.73 63770.41  0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00  0.28  0.26  0.24 -0.67 -0.46  0.35  0.886 1  0.94 2  0.96 3  0.95 1  62.74:1 2  42.72:1 3  22.90:1 4 117 Model not identified           Model Type 2: Variances/ covariances vary across classes, indicator-specific random intercept 1 32 1  1.00 65094.12 65184.05  n/a n/a n/a n/a n/a  n/a 1  1.00 n/a 2 62  1  0.60  2  0.40 62762.37 62936.60  -3.60 -3.38 -2.46 -413.60 -404.80 -184.30 -409.70 -276.10  274.50  0.54  0.42  0.31 -0.57 -0.47 -0.39  0.845 1  0.97 2  0.95 1  19.44:1 2  25.11:1  3 93 1  0.39 2  0.19 3  0.43 61384.73 61646.08  -2.96 -1.93 -0.47   -70.90   -15.40     -5.80   -34.90    31.70    30.20  0.24  0.23  0.15 -0.59  0.33 -0.28  0.854 1  0.96 2  0.92 3  0.92 1  34.23:1 2  52.49:1 3  15.67:1    4 125 Model not identified           Note: AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; Adjusted BIC accounts for sample size. Lower AIC and Adjusted BIC values indicate better model fit when comparing models of the same type with different numbers of classes. Entropy is a measure of posterior classification uncertainty; values near 0 may indicate poor class separation. AvePP = average posterior class probability and OCC = odds of correction classification; higher values indicate better class separation and classification accuracy. The bolded model was selected as the final model, based on model fit indices, residuals, classification indices, parsimony, and class contents.  148    Caption: Distribution of person-level solitude class membership, i.e. each individual’s proportion of solitude episodes categorized as solitude type 1 (“negative solitude experience”) versus type 2 (“positive solitude experience”). Most individuals had either exclusively negative solitude experiences (N = 73) or exclusively positive solitude experiences (N = 37).  Figure A-1: Person-level distribution of solitude experience classes (N = 150 individuals)   149   Appendix B: Study 1 multilevel latent class regression procedure  Latent Class Regression Model Equations and Variables In the latent profile analysis procedure (Supplementary Materials B), solitude episodes were classified into two types (Class 1: “negative solitude experiences” and Class 2: “positive solitude experiences”) based on momentary affect/thought profiles. Our next aim was to predict for whom and under what circumstances solitude would be experienced negatively or positively. To do this, we built on the final 2-solitude-class model by adding a set of Level 1 (situation-level) and Level 2 (person-level) predictors of solitude class membership, using multilevel latent class regression (LCR; Henry & Muthén, 2010; Masyn, 2013). Multilevel LCR allows us to model random effects, i.e. person-level clustering of situation-level solitude class membership, using a logistic model. This is the equivalent of adding covariates to the latent profile analysis model, but is done after establishing the final solitude class structure. Model equations and variable interpretations are provided below. All predictors were grand mean centered. Level 1: logit(Pij) = β0j + β1j ALONEij + β2j DES_SOLij Level 2: β0j = γ00 + γ01 PROP_SOLj + γ02 ALONE_Mj + γ03 DES_SOL_Mj   + γ04 AGEj  + γ05 EUROj  + γ06 UNIVj + γ07 RELATj   + γ08 NET_SIZEj + γ09 REL_QUALj + γ010 SOC_STATj   + γ011 SOC_EFFj + γ012 REFLj + γ013 SOC_ANXj + γ014 RUMIj + U0j  β1j = γ10 β2j = γ20 – Subscript i indicates level 1 units (solitude episodes), and j indicates level 2 units (persons) – Pij is the probability of experiencing positive (Class 2) rather than negative (Class 1) solitude at a given moment. logit(Pij) is the log-odds; values greater than 0 indicate greater odds of   150   experiencing positive solitude and values less than 0 indicate greater odds of negative solitude. – β0j is the random intercept of the logit outcome (the log-odds of experiencing positive over negative solitude is allowed to vary randomly across people). – γ00 is the average log-odds of having positive over negative solitude experiences (when all model predictors are at their grand means). – γ10 is the average change in log-odds when currently alone (ALONE = 1) versus not alone (ALONE = 0), when all other predictors are at their grand means. – γ20 is the average change in log-odds when currently desiring solitude (DES_SOL = 1 vs. 0), when all other predictors are at their grand means. – γ01 through γ14 are the average changes in log-odds for a one unit change in the respective person-level variable, when all other predictors are at their grand means. The predictors are: proportion of instances of solitude across all momentary assessments (PROP_SOL); proportion of solitude instances when participant was alone (ALONE_M, person-average of ALONE); proportion of solitude instances when participant desired solitude (DES_SOL_M, person-average of DES_SOL); age in years (AGE); being European (EURO = 1 vs. 0); having at least some post-secondary education (UNIV = 1 vs. 0); being in a relationship (RELAT = 1 vs. 0); number of individuals in social network (NET_SIZE); social relationship quality on a 5-point scale (REL_QUAL); perceived social status on a 10-point scale (SOC_STAT); social self-efficacy on a 5-point scale (SOC_EFF); self-reflection on a 5-point scale (REFL); social anxiety on a 5-point scale (SOC_ANX); and self-rumination on a 5-point scale (RUMI). – U0j is the residual influence of Level 2 units (people) after accounting for all model predictors; assumed to be normally distributed. This intercept random effect was fixed to zero for model convergence.   151   Latent Class Regression Modeling Procedure  In estimating this latent class regression (LCR) model, we used the 3-step approach recommended by Vermunt (2010) to account for uncertainty in solitude class assignment. Modal class assignments are weighted by probabilistic class assignments when determining the influence of the predictors on class membership. The steps are as follows: 1.  After picking a final LPA model (see Appendix A), look at the solitude class assignment results. Each solitude episode is assigned to a single class (modal class assignments, with values of either 0 or 1 for each class). The LPA results also give us information about classification uncertainty for each solitude episode (probabilistic class assignments, values ranging from 0 to 1 for each class), including logits for the classification probabilities for the most likely class membership (one logit per class). All this information is part of the Mplus LPA output. 2. Create a nominal “most likely class” variable to use in the LCR analysis, based on the modal class assignments. Then, using the logits for the classification probabilities for the most likely latent class memberships, pre-fix the class-specific measurement error rates for this “most likely class” variable to match the misclassification rates from the LPA analysis. This process is analogous to gathering reliability information for a particular measure, and then using this reliability information to specify error variance when using the measure in a subsequent model.  3. Run the LCR model, including the “most likely class” variable from the previous step as a nominal indicator of solitude class membership. Add Level 1 and Level 2 covariates, as appropriate, to test hypotheses regarding situation-level and person-level predictors of solitude class membership.     152   Appendix C: Study 1 variable descriptive information     153   Table C-2: Intercorrelations of person-level variables and person-averaged momentary variables (N = 150 individuals)   Correlations     2   3   4   5   6   7   8   9   10   11   12   13   14   15   16   17   18   19   20  1. Age  .09 -.30  .26 -.10  .09  .10  .05 -.33 -.44 -.58  .01 -.03  .01  .40  .47  .43 -.31 -.38 -.07  2. Ethnicity   .17 -.14  .25  .16  .23  .02 -.10 -.18 -.17  .10  .06 -.15 -.01  .12 -.07 -.10  .00 -.24  3. Education   -.11  .25  .00  .09  .02  .22  .03  .20  .01  .10  .09 -.19 -.27 -.19  .07  .05 -.10  4. Relationship status     -.06  .27  .15  .14 -.18 -.28 -.25 -.45 -.27 -.16  .24  .25  .10 -.17 -.21  .07  5. Social network size       .12  .15  .14  .13  .09  .10 -.10 -.12 -.19 -.02 -.05 -.01  .08  .06 -.03  6. Social relationship quality          .31  .42 -.01 -.30 -.19 -.17 -.12 -.12  .32  .38  .11 -.34 -.32 -.12  7. Perceived social status           .25  .01 -.27 -.12 -.06 -.07 -.06  .17  .24  .05 -.25 -.24 -.19  8. Social self-efficacy            .17 -.19 -.02 -.20 -.23 -.25  .38  .28  .10 -.24 -.24 -.12  9. Self-reflection            .15  .48  .04 -.10 -.14 -.02 -.09 -.17  .25  .22  .18 10. Social anxiety             .47  .09  .13  .09 -.30 -.37 -.29  .44  .37  .19 11. Self-rumination             .10  .11  .05 -.30 -.43 -.30  .43  .42  .28 12.   Mean time alone             .58  .40 -.03 -.05  .00  .02 -.02 -.12 13. Mean time in solitude              .60 -.10 -.14 -.05  .01  .00 -.09 14. Mean desire for solitude                 -.12 -.09  .11 -.10 -.20 -.14 15. Mean high arousal positive affect                .69  .43 -.29 -.46 -.07 16. Mean low arousal positive affect                 .43 -.53 -.58 -.32 17. Mean low cognitive effort thoughts                 -.38 -.48 -.26 18. Mean high arousal negative affect                   .81  .68 19. Mean low arousal negative affect                    .57 20. Mean high cognitive effort thoughts                   Note. Bolded values are significant at α = .05. Age is in years; ethnicity is 1 = European, 0 = non-European; education is 1 = at least some post-secondary, 0 = no post-secondary; relationship status is 1 = in a relationship, 0 = not in a relationship. Social network size is total number of individuals listed. Perceived social status is on a 10-point scale. Social relationship quality, social self-efficacy, self-reflection, social anxiety, and self-rumination are on 5-point scales. Mean time alone is proportion of all (4571) momentary assessments when participant was alone; mean time in solitude is proportion of assessments when participant was in solitude; mean desire for solitude is proportion of assessments when participant desired solitude. Mean affect and thought dimensions are person-averages of all momentary assessments, on 100-point scales.   154    Figure C-2: Distributions of momentary affect and thought dimensions (n = 2944 solitude episodes)    155   Appendix D: Study 2 data analytic approach and descriptive findings  Study 2 models - Models 1-2: Currently at home (1 = at home, 0 = not at home) and currently outside (1 = outside, 0 = not outside); logistic multilevel models - Models 3-7: Current high arousal positive affect, low arousal positive affect, high arousal negative affect, low arousal negative affect, and loneliness (100-point scales); linear multilevel models  All models were of the following form: Level 1 (momentary level) OUTCOMEij = b0j + b1jCURR_SOLij + b2jCURR_SOL_DESij + b3jCURR_WORKij + b4jCURR_LEISUREij + b5jTIMEij + b6jTIME_SQij +   b7j(CURR_SOL_DESij x CURR_WORKij) + b8j(CURR_SOL_DESij x CURR_LEISUREij) +  b9j(CURR_SOL_DESij x TIMEij) + b10j(CURR_SOL_DESij x TIME_SQij) + eij Level 2 (person level) b0j  =  γ00 + γ01AVG_SOLj + γ02AVG_SOL_DESIREj + γ03AVG_WORKj + γ04AVG_LEISUREj + γ05AGEj + γ06ETHNICITYj + γ07EDUCATIONj + γ08GENDERj +   γ09RETIREMENTj + γ010RELATIONSHIPj + γ011SOCIAL_STATUSj + u0j b1j  = γ10 + γ11AGEj + u1j          b2j  =  γ20 + γ21AGEj + u2j          b3j = γ30          b4j = γ40           b5j = γ50          b6j = γ60          b7j = γ70          b8j = γ80          b9j = γ90          b10j = γ100     156   Key variables (all variables grand mean centered) - CURR_SOL: current solitude (1 = in solitude, 0 = not in solitude) - CURR_SOL_DES: current solitude desire (1 = solitude desired, 0 = solitude not desired) - CURR_WORK: current working activity (1 = working, 0 = not working), grand mean centered - CURR_LEISURE: current passive leisure activity (1 = passive leisure, 0 = not passive leisure) - TIME: current time (hours since 4am)        - TIME_SQ: current time squared        - AGE (years)           - AVG_SOL: person-average solitude (percentage of occasions in solitude) - AVG_SOL_DESIRE: person-average solitude desire (percentage of occasions desiring solitude) - AVG_WORK: person-average working (percentage of occasions working) - AVG_LEISURE: person-average passive leisure (percentage of occasions in passive leisure)  Hypothesis testing Hypothesis 1: Solitude-seeking will be associated with greater likelihood of being at home and lesser likelihood of being outside: b2j coefficient for Models 1-2 Hypothesis 2: Solitude-seeking will be associated with decreased positive affect and increased negative affect and loneliness: b2j coefficient for Models 3-7 Hypothesis 3: Compared to middle-aged adults, older adults will be more likely to be either at home or outdoors when seeking solitude: γ21 coefficient for Models 1-2 Hypothesis 4: Compared to middle-aged adults, older adults will show lesser decreases in positive affect, and lesser increases in negative affect and loneliness, when seeking solitude: γ21 coefficient for Models 3-7   157   Table D-1: Situation-level (within-person) variable descriptives by solitude situation and by solitude desire (n = 3195 momentary assessments)   In solitude  (n = 2013) Not in solitude  (n = 1182) Difference test (In solitude vs. not in solitude) Desiring solitude  (n = 2139) Not desiring solitude  (n = 1056) Difference test (Desiring vs. not desiring solitude) Percent occasions in solitude    80.7% 27.1% Χ2(1) = 870.89*** Percent occasions desiring solitude 85.8% 34.9% Χ2(1) = 870.89***    Percent occasions at home 88.6% 62.9% Χ2(1) = 299.17*** 85.0% 66.8% Χ2(1) = 141.27*** Percent occasions outside 4.3% 11.3% Χ2(1) = 55.83*** 4.6% 11.6% Χ2(1) = 51.58*** Percent occasions working 8.9% 9.4% Χ2(1) = 0.19 8.9% 9.5% Χ2(1) = 0.25 Percent occasions passive leisure 37.5% 24.1% Χ2(1) = 60.57*** 34.2% 29.4% Χ2(1) = 7.21** Mean high arousal positive affect (SD) 52.6 (19.9) 57.2 (18.7) t(2599) = -6.55*** 53.0 (19.6) 57.0 (19.2) t(2146) = -5.59*** Mean low arousal positive affect (SD) 67.3 (20.1) 68.5 (19.4) t(2546) = -1.65 67.5 (19.8) 68.4 (19.9) t(2097) = -1.28 Mean high arousal negative affect (SD) 22.4 (21.6) 24.9 (21.5) t(2516) = -3.13** 22.2 (21.5) 25.6 (21.2) t(2124) = -4.33*** Mean low arousal negative affect (SD) 29.4 (21.7) 30.6 (20.9) t(2550) = -1.49 28.7 (21.4) 32.2 (21.3) t(2115) = -4.41*** Mean loneliness (SD) 20.4 (23.6) 21.1 (22.0) t(2616) = -0.86 18.5 (22.1) 25.0 (24.2) t(1937) = -7.33***  Note. t tests use Welch’s t for unequal variances.   ***p < .001, **p < .01, *p < .05, †p < .1      158   Table D-2: Intercorrelations of person-level variables and of person-averaged situation-level variables (N = 90-100 individuals)      2   3   4   5   6   7   8   9   10   11   12   13   14   15   16 17 18 1. Age  .10  .12 -.14  .54  .00  .10  .10  .07 -.04  .21  .08  .10 -.17 -.25 -.17 -.24 -.13 2. Ethnicity   .24  .04 -.07 -.30  .19  .14 -.09 -.07 -.05 -.14  .07 -.11  .06 -.11 -.04  .07 3. Education    -.00 -.04 -.03  .15  .09  .11 -.29  .13 -.10 -.16 -.02 -.01 -.14 -.12  .12 4. Gender     .12 -.19  .03 -.06 -.03 -.17  .01  .09 -.03  .10  .19 -.03 -.16 -.04 5. Retirement status     -.01  .06  .10  .08 -.06  .28  .22  .19 -.12 -.19 -.18 -.33 -.09 6. Relationship status       .06 -.29 -.15 -.01  .08  .21  .16 -.08 -.14 -.08  .09  .01 7. Perceived social status       -.04 -.07 -.05 -.02  .15  .27 -.25 -.18 -.18 -.17 -.02 8. Mean time in solitude         .69 -.26  .38 -.18 -.12 -.03  .00 -.06 -.15  .17 9. Mean desire for solitude          -.33  .35 -.19 -.11 -.04 -.10 -.22 -.08  .09 10. Mean time outside          -.55  .03  .08 -.10 -.09  .16  .16 -.16 11. Mean time at home             .05  .01 -.05 -.19 -.16 -.15  .04 12. Mean high arousal positive affect              .62 -.26 -.38 -.14  .04 -.07 13. Mean low arousal positive affect              -.68 -.58 -.46 -.15 -.04 14. Mean high arousal negative affect                .77  .72  .15 -.03 15.   Mean low arousal negative affect                 .59  .10  .03 16. Mean loneliness                 .16  .02 17. Mean time working                  .05 18. Mean time in passive leisure                  Note. Age is in years; gender coded 1 = female, 0 = male; ethnicity coded 1 = European, 0 = non-European; education coded 1 = some post-secondary, 0 = no post-secondary, retirement status coded 1 = retired, 0 = not retired, relationship status coded 1 = in a relationship, 0 = not in a relationship. Perceived social status is a score on a 10-point scale. Mean time in solitude, desire for solitude, time outside, time at home, time working, and time in passive leisure are the percentage of occasions when the individual was in the respective situation/location. Affect dimensions are person-averages of momentary assessments (100-point scale).  N ranges from 90 to 100 individuals due to missing data for age (N = 5), relationship status (N = 2), and perceived social status (N = 5). Bolded values are significant at α = .05.      159   Table D-3: Situation-level variable descriptives by age group (n = 3195 momentary assessments, N = 95 individuals)   Middle-aged adults (age 50-67 years,  N = 47) Older adults  (age 68-85 years,  N = 48) Difference test (middle-aged vs. older adults) Percentage of occasions in solitude 60.5% 67.4% t(90) = -1.51 Percentage of occasions desiring solitude 66.9% 69.7% t(92) = -0.51 Percentage of occasions at home 76.2% 82.7% t(90) = -1.83† Percentage of occasions outside 7.5% 5.8% t(93) = 1.00 Mean high arousal positive affect 54.0 (13.2) 55.8 (14.5) t(92) = -0.66 Mean low arousal positive affect 66.4 (14.2) 69.1 (16.0) t(92) = -0.88 Mean high arousal negative affect 26.4 (15.4) 22.2 (16.9) t(93) = 1.28† Mean low arousal negative affect 33.3 (14.4) 28.0 (16.8) t(91) = 1.66† Mean loneliness 23.0 (18.3) 19.8 (20.5) t(92) = 0.80 Percentage of occasions working 12.6% 5.5% t(73) = 2.34* Percentage of occasions passive leisure 33.6% 30.3% t(90) = 0.68  Note. t tests use Welch’s t for unequal variances. All variables are person-means for each age group. Affect variables are on a scale from 0 to 100. ***p < .001, **p < .01, *p < .05, †p < .1        160   Appendix E: Study 3 multilevel model equations and variables      Multilevel modeling was used to test hypotheses pertaining to momentary solitude-loneliness associations and moderators of these associations.  Model equations and variable interpretations are provided below; subscript i indicates Level 1 units (momentary assessments), and j indicates Level 2 units (persons). All variables were grand mean centered. Level 1: CURRENT_LONELINESSij = β0j + β1j CURRENT_SOLITUDEij  + β2j CURRENT_SOLITUDE_DESIREij  + β3j CURRENT_SOLITUDE X CURRENT_SOLITUDE_DESIREij + eij Level 2: β0j = γ00 + γ01 OVERALL_SOLITUDEj  + γ02 OVERALL_SOLITUDE_DESIREj   + γ03 HONGKONGj  + γ04 ASIANj  + γ05 IMMIGRANTj  + γ06 HOST_ACCULTURATIONj + γ07 AGEj + γ08 FEMALEj  + γ09 UNIVERSITYj + γ010 RETIREDj  + γ011 IN_RELATIONSHIPj  + γ012 SOCIAL_STATUSj + γ013 SOCIAL_RELATIONSHIP_QUALITYj  + U0j    β1j = γ10 + γ11 CURRENT_SOLITUDE X OVERALL_SOLITUDE   + γ12 CURRENT_SOLITUDE X OVERALL_SOLITUDE_DESIRE  + γ13 CURRENT_SOLITUDE X HONGKONG   + γ14 CURRENT_SOLITUDE X ASIAN   + γ15 CURRENT_SOLITUDE X IMMIGRANT   + γ16 CURRENT_SOLITUDE X HOST_ACCULTURATION + U1j β2j = γ20 + U2j  β3j = γ30 + U3j   161    Level 1 variables. The situation-level outcome, current loneliness, was predicted from current solitude (1 = in solitude, 0 = not in solitude) to test our Hypothesis 1 linking current solitude to current loneliness. Current solitude desire (1 = solitude desired, 0 = solitude not desired) was also included at Level 1.  Level 2 variables. The person-means of current solitude and current solitude desire (“overall solitude” and “overall solitude desire”, respectively) were added at Level 2, along with several other person-level predictors: sample location (1 = Hong Kong, 0 = Vancouver), cultural heritage (1 = East or Southeast Asian, 0 = European or North American), immigration status (1 = immigrant, 0 = non-immigrant), and host culture acculturation (9-point scale). The following control variables were added at Level 2: age, gender (1 = female, 0 = male), education (1 = at least some post-secondary, 0 = no post-secondary), retirement status (1 = retired, 0 = not retired), relationship status (1 = in a romantic relationship, 0 = not in a relationship), perceived social status (10-point scale), and social relationship quality (5-point scale).  Interactions. At Level 1, we added the 2-way interaction of current solitude with current solitude desire to test whether currently desiring solitude weakens momentary solitude-loneliness associations (Hypothesis 6). We added cross-level interactions to test hypotheses regarding person-level moderators of solitude-loneliness associations, specifically, current solitude x ethnicity (Hypothesis 2, East Asian participants will show weaker solitude-loneliness associations), current solitude x sample (Hypothesis 3, participants living in Hong Kong will show weaker solitude-loneliness associations), current solitude x immigration status (Hypothesis 4, immigrated participants will show stronger solitude-loneliness associations), and current solitude x host culture acculturation (Hypothesis 5, participants more acculturated to the host culture will show weaker solitude-loneliness associations). We also included cross-level   162   interactions of current solitude with overall solitude and with overall solitude desire. Random effects. Random effects for the model intercept (U0j), current solitude slope (U1j), current desire for solitude slope (U2j), and current solitude x solitude desire slope (U3j) were estimated. That is, models allowed for the residual influence of Level 2 units (people) on current loneliness, current solitude-loneliness associations, and current solitude desire-loneliness associations. The Level 1 residuals (eij) were assumed normally distributed.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0368723/manifest

Comment

Related Items