How relational is ‐evaluation?

Self‐esteem and social relationships across

life span and family situations

Inaugural‐Dissertation

in der Philosophischen Fakultät und Fachbereich Theologie

der Friedrich‐Alexander‐Universität

Erlangen‐Nürnberg

vorgelegt von

Jenny Wagner

aus

Leipzig

D 29

Tag der mündlichen Prüfung: 28.05.2009

Dekan: Universitätsprofessor Dr. Jens Kulenkampff

Erstgutachter: Universitätsprofessor Dr. Frieder R. Lang

Zweitgutachter: Universitätsprofessor Dr. Karen L. Fingerman

Acknowledgement

First of all I would like to thank my supervisor Prof. Dr. Frieder R. Lang for his support and his inspiring scientific expertise that helped greatly to extend my . His advice was always challenging and improving this dissertation.

Secondly, I am very grateful to Prof. Karen L. Fingerman for the opportunity to work in her lab, where she took much time to exchange ideas and discuss my dissertation, opening up new perspectives.

This dissertation is part of the project “ and Relationship Regulation across Adulthood”. My special thanks go to the German Research Foundation for funding and to Prof. Franz J. Neyer and Dr. Cornelia Wrzus, who never were short of helpful in‐ sights. I deeply appreciate how much effort and Cornelia dedicated to this collabora‐ tion. She became a fundamental part of my scientific and .

A great many more people have contributed in important ways to this dissertation, and I am very grateful to all of them: my colleagues and the lab of Prof. Fingerman whose suggestions and criticism improved the work’s quality, the students in Halle and Erlangen that helped to collect the data, and Lindsay Pitzer and Florence W. Jones for proofreading the dissertation. Many special thanks go to Margund Rohr, who has always valuable com‐ ments, helps out at all times, and manages to be a scientific sparring‐partner and a friend.

Very warm thanks go to my parents and family for their in what I do and for sharing their always highly regarded opinions with me; and to Toralf for everything.

Table of Contents iv

TABLE OF CONTENTS

1 INTRODUCTION ...... 1

2 ACTIVE SELF: THE SIGNIFICANCE OF SELF‐ESTEEM ...... 7

2.1 Self‐Evaluation: Maintaining a Positive Appraisal of the Self ...... 7

2.2 Self‐Esteem – The Construct ...... 8

2.3 Relational Self‐Esteem: Relationship Specific Evaluations of the Self ...... 15

2.4 Summary ...... 18

3 SOCIAL TIES AND RELATIONSHIP REGULATION: PERSONAL NETWORKS IN

DIFFERENT CONTEXTS ...... 19

3.1 Two Perspectives on Social Relationships ...... 21

3.2 Integrative Framework of Lifelong Relationship Regulation: Contextual and

Relationship Effects ...... 24

3.3 ‐Parent Ties: Relationship Regulation and Self‐Evaluation in the

Family of Origin ...... 35

3.3.1 Closeness and Reciprocity in Adult Child‐Parent Relationships...... 35

3.3.2 Self‐Esteem as Regulatory Relationship Outcome ...... 37

3.3.3 Effects of Contextual Characteristics: Family Constancy ...... 38

3.4 Childless Couples in Middle Adulthood: Relationship Regulation and Self‐

Evaluation in Different Family Situations ...... 40

3.4.1 Predictors and Differentiation of Fertility ...... 43

3.4.2 Motivated and Involuntary Childlessness: The Two Observed

Conditions ...... 45

3.5 Summary ...... 51

4 SELF‐EVALUATION ACROSS RELATIONAL CONTEXTS: GENERAL SUMMARY AND

RESEARCH QUESTIONS ...... 52 Table of Contents v

5 METHOD ...... 57

5.1 Online Study: Relational Self‐esteem across the Life Span ...... 57

5.1.1 Sample Descriptions of Online Study ...... 58

5.1.2 Measures of Online Study ...... 59

5.2 Family Study: Personality and Social Relationships in the Family Context ...... 69

5.2.1 Target Sample: Description of and Couples ...... 71

5.2.2 Measures of Target Sample ...... 76

5.2.3 Intergenerational Relations: Adult Children and Their Parents ...... 81

5.2.4 Measures on Intergenerational Relationships ...... 84

5.3 Procedures and Methods of Statistical Analyses ...... 86

5.3.1 Data Structures ...... 87

5.3.2 Statistical Software Packages ...... 92

6 RESULTS ...... 93

6.1 Self‐Esteem across Life Span and Family Situations ...... 93

6.1.1 Online Study: Effects of Age on Self‐Esteem ...... 93

6.1.2 Family Study: Effects of Family Situations on Self‐Esteem ...... 103

6.1.3 Summary: Self‐Esteem across Life Span and Family Situations ...... 107

6.2 Social Networks and Relationship Regulation across Life Span and Family

Situations ...... 107

6.2.1 Online Study: Effects of Age on Social Networks and Relationship

Regulation ...... 108

6.2.2 Family Study: Effects of Family Situation on Social Relationships ...... 115

6.2.3 Summary: Social Networks and Relationship Regulation across Life

Span and Family Situations ...... 126

6.3 Functionality of Relationships – Relational and Self‐Evaluation ...... 127

6.3.1 Online Study: Relational Self‐Esteem in Social Networks ...... 127

6.3.2 Family Study: Self‐Evaluation in Adult Child‐Parent Ties ...... 141

6.3.3 Family Study: Self‐Evaluation in Romantic Partnerships ...... 154

6.3.4 Summary: Functionality of Relationships – Relational and Individual

Self‐Evaluation ...... 159

6.4 Summary of Findings in Relation to Hypotheses ...... 161 Table of Contents vi

7 DISCUSSION ...... 163

7.1 Different Facets of Self‐Evaluation across Life Span and Family Situations ...... 163

7.2 Social Networks and Relationship Regulation Characteristics: Differential

Patterns Pertaining to Age Group and Family Situation ...... 169

7.3 Social Relationships Differentially Affect Self‐Evaluation across the Life Span

and in Specific Relationship Dyads ...... 175

7.3.1 Calibration of Relational Self‐Esteem ...... 175

7.3.2 Relationship Quality and Relational Self‐Esteem...... 176

7.3.3 Effects of Level of Data: Relational and Individual Self‐Esteem...... 177

7.3.4 Self‐Evaluation in Specific Dyadic Relationship Contexts ...... 179

7.4 There Is a Relational Nature to Self‐Evaluation – But That Is Not the Whole

Story ...... 182

7.5 Motivated and Involuntary Childlessness – Differentiation versus Convergence

in Reproduction and Post‐Reproduction Times ...... 184

7.6 Limitations of the Study ...... 187

7.7 Outlook ...... 190

8 REFERENCES ...... 193

9 APPENDIX ...... 9‐1

Index of Tables vii

INDEX OF TABLES

Table 3.1 Research Summary on Motivated and Involuntary Childlessness ...... 48

Table 5.1 Applied Instruments of the Online Study ...... 60

Table 5.2 Descriptive Statistics of Ego‐centered Social Networks Regarding the Entire Network and in Ten Relationship Categories ...... 62

Table 5.3 Variables of the Relationship Level – Item Characteristics of the Online Study ...... 66

Table 5.4 Covariates – Scale Characteristics and Intercorrelations of Online Study (N = 557) ...... 68

Table 5.5 Samples of Family Study ...... 71

Table 5.6 Demographic Characteristics by Family Type ...... 74

Table 5.7 Applied Instruments of the Family Study...... 76

Table 5.8 Descriptive Statistics of Entire Ego‐centered Social Networks and Composition in Eleven Different Relationship Types ...... 78

Table 5.9 Variables of the Relationship Level – Item Characteristics of the Family Study (N = 4561) ...... 79

Table 5.10 Covariates – Scale Characteristics and Intercorrelations of the Family Study (N = 342) ...... 80

Table 5.11 Descriptive statistics, Internal Consistencies, Inter‐Correlations, and Dyadic Similarity in Key Variables ...... 85

Table 6.1 Means, Standard Deviations, and Effect Sizes (ES) of Global Trait and Domain‐Specific State Self‐Esteem Measures by Gender, Partnership Status, Parental Status, and across Age Groups ...... 95

Table 6.2 Age Specific Intercorrelations between Global Trait and Domain‐Specific State Self‐Esteem Measures ...... 98

Table 6.3 Regression of Age and Two Indicators of Well‐Being on Trait and State Self‐Esteem Measures ...... 101

Table 6.4 Means, Standard Deviations, and Effect Sizes (ES) of Global Trait and Domain‐Specific State Self‐esteem Measures by Gender, Partnership Status, Parental Status, Family Constancy and Family Situations in the Follow‐Up Sample (N = 175) ...... 105

Table 6.5 Descriptive Statistics of Ego‐Centered Social Networks by Age Groups ...... 109

Table 6.6 Parameter Estimates (PE) of Relationship Measurement Two‐Level Regression Model on Perceived Closeness to Social Network Partners in Five Different Age Groups ...... 114 Index of Tables viii

Table 6.7 Differences between Familial Situations in Average Number of Overall Size of Social Networks as well as in Specific Relationship Types ...... 117

Table 6.8 Differences between Family Types in Characteristics of Social Network Partners ...... 119

Table 6.9 Parameter Estimates (PE) of Relationship Measurement Two‐Level Regression Model on Perceived Closeness to Social Network Partners in Four Different Familial Situations ...... 124

Table 6.10 Parameter Estimates (PE) of Multilevel Model on Perceived Communal‐ Esteem to Social Network Partners in Three Relationship Systems ...... 131

Table 6.11 Parameter Estimates (PE) of Multilevel Model on Perceived Communal‐ Esteem to Social Network Partners in Different Age groups in Kin and Non‐Kin Relationships ...... 133

Table 6.12 Parameter Estimates (PE) and Odds Ratios (OR) of HGLM on Perceived Mutual‐esteem to Social Network Partners in Three Relationship Systems ...... 135

Table 6.13 Parameter Estimates (PE) of Multilevel Model on Perceived Communal‐ Esteem to Social Network Partners in Three Relationship Systems with Cross Level Effects with Trait and State Self‐Esteem, Life Satisfaction and Subjective ...... 138

Table 6.14 Descriptive Statistics of Relationship Specific Qualities to Parent and Aggregated Grand Means of Closeness and Reciprocity for Adult Child ...... 143

Table 6.15 Descriptive Statistics of Relationship Specific Qualities to Adult Child and Aggregated Grand Means of Closeness and Reciprocity for Old Parent ...... 144

Table 6.16 Standardized Regression Coefficients Indicating the Effect of Generation, Family Constancy, Grand‐Mean Variable of Respective Outcome, and Gender on Perceived Closeness and Reciprocity in Intergenerational Relations ...... 147

Table 6.17 Standardized Parameter Estimates of Actor‐Partner Interdependence Models of Closeness and Reciprocity Regressed on Global Self‐Esteem in Adult Child ‐ Parent Dyads ...... 150

Table 6.18 Descriptive Statistics of Relationship Perceptions in the Couple Sub‐ Sample at T1 and T2 (N = 52 couples) ...... 155

Table 6.19 Descriptive Statistics of Self‐Esteem Measures in the Couple Sub‐Sample at T1 and T2 (N = 52 couples) ...... 156

Table 6.20 Summary of Results...... 161 Index of Figures ix

INDEX OF FIGURES

Figure 1. Mean level trajectory of self‐esteem, for the total sample as well as separately for males and females (pictured results extracted from Robins et al., 2002, p. 429) ...... 10

Figure 3. Proportion of Childlessness in Women in Germany (Table is based on Data of the Statistisches Bundesamt [Federal Agency of Statistics], 2007b) ...... 42

Figure 4. The Actor‐Partner Interdependence Model (APIM, Based on Cook & Kenny, 2005, p. 102). Child = A; Parent = Person B; UC = Residual (Unexplained) Proportion of Child; UP = Residual (Unexplained) Proportion of Parent ...... 88

Figure 5. Trajectories of Cross‐Sectional Data on Global Trait Self‐Esteem as well as Performance and Social State Self‐Esteem across Seven Age Groups ...... 96

Figure 6. Interaction Plotted From Life Satisfaction and Age Regressed on Global Self‐Esteem (Regarding Both Predictors ʹHighʹ and ʹLowʹ Is Based on +/‐ 1 SD) ...... 100

Figure 7. Differences in Social Self‐Esteem by Family Constancy and Gender ...... 106

Figure 8. Mean Perceived Emotional Closeness as a Function of Genetic Relatedness and Perceived Under‐ and Over‐Benefit in Social Networks ...... 123

Figure 9. Mean Perceived Emotional Closeness To Non‐Kin (left) and Kin (right) as a Function of Perceived Under‐ and Over‐Benefit in Ego‐Centered Social Networks among Four Family Situations ...... 125

Figure 10. Group Differences in Relationship Perceptions by Generation of Adult Children and Old Parents ...... 146

Figure 11. Standardized Parameter Estimates of Path Analysis Predicting Closeness, Reciprocity, and Self‐esteem at Follow‐up in High and Low Family Constancy ...... 153

Figure 12. Standardized Parameter Estimates of Path Analysis Predicting Closeness, Reciprocity, and Global Self‐Esteem at Follow‐up in Romantic Partnerships ...... 158

Abstract x

ABSTRACT

The present research aims at three issues: (1) exploration of different self‐esteem measures with respect to age related patterns and contextual sensitivity, (2) consid‐ eration of structural and procedural issues of social networks and regulative mecha‐ nisms in the context of age and family situation, and (3) examination of relational self‐evaluation across networks, in intergenerational ties and romantic partnerships.

The research is based on two studies with independent samples. Study 1 includes a cross‐sectional online sample of 562 people (age 17 to 88 years), who completed a network‐ and self‐esteem questionnaire. Study 2 is based on a quasi‐experimental study of 342 (age 30 to 45 years) with different family forms (e.g., motivated childless, traditional family). This study explores effects of family situation and pa‐ renthood status on social embeddedness and self‐evaluation. Investigations focus on ego‐centered social networks, global and domain‐specific self‐esteem as well as a new construct of relationship functioning, that is, relational self‐esteem. Findings suggest differential age‐related pathways and patterns of associations between global and domain‐specific self‐esteem. Social self‐esteem indicates a particular sensitivity to reflect contextual differences of family forms. Considerations of networks and re‐ gulative strategies support differential structures of kinship orientation across age groups and family situations. Regulation of over‐ and under‐benefit indicates lack of distinctiveness. Results on relational self‐esteem support a differential esteem of rela‐ tionship partners, with perceptions of closeness and reciprocity serving as predictive variables. Patterns of dyadic analyses show substantial effects of closeness on self‐ esteem in intergenerational ties and of reciprocity in spousal relations. Contextual effects were less distinctive than expected. In accordance with relational self‐esteem, findings point to a relational nature of self‐evaluation. The differential and relation‐ ship specific perspective opens up a new perspective of personal functioning across the life span.

Words count: 286 Zusammenfassung xi

ZUSAMMENFASSUNG

Die vorliegende Dissertationsschrift beschäftigt sich mit der Frage, inwieweit Selbst‐ bewertungsprozesse auf beziehungsspezifischen Charakteristiken basieren. In einem ersten Schritt konzentriert sich die Arbeit auf die Auseinandersetzung mit bisherigen

Erhebungsinstrumenten zum globalen und bereichsspezifischen Selbstwert, betrach‐ tet altersspezifische Verläufe und kontextuelle Effekte unterschiedlicher Familien‐ formen. In einem zweiten Schritt werden strukturelle und regulative Komponenten des sozialen Netzwerkes analysiert, da diese als grundlegende Struktur der Selbst‐ bewertung verstanden werden. Im letzten Schritt wird anhand eines neuentwickel‐ ten Konstruktes, dem relationalen Selbstwert, das Zusammenspiel von Beziehungs‐ charakteristiken und der Selbstbewertung untersucht. Zusätzlich zur Netzwerkbe‐ trachtung werden hier zwei spezifische Beziehungen herausgegriffen, um anhand dyadischer Analysen die Interdependenz von Beziehungscharakteristiken und

Selbstbewertung darzustellen.

Die empirische Umsetzung der Studie beruht auf zwei unabhängigen Stich‐ proben. Studie 1 umfasst 562 Personen (zwischen 17 bis 88 Jahren) einer querschnitt‐ lichen Onlineerhebung, aus welcher sowohl Netzwerkdaten als auch unterschiedli‐ che Selbstwertskalen vorliegen. In Studie 2 wurden 171 heterosexuelle Paare (zwi‐ schen 30 und 45 Jahren) aus unterschiedlichen Familienformen (z.B. Motiviert Kin‐ derlos, Traditionelle Familie) befragt. Mit Hilfe dieser Studie konnten familienspezi‐ fische Analysen zur sozialen Einbindung und Selbstbewertung durchgeführt wer‐ den. Die empirische Erhebung umfasste einerseits strukturelle und qualitative As‐ pekte ego‐zentrierter sozialer Netzwerke und zum zweiten unterschiedliche Skalen zur Erfassung von Selbstwert. Besonderes Augenmerk wurde auf ein neuentwickel‐ tes Konstrukt zum beziehungsspezifischen Selbstwert gelegt.

Die Ergebnisse der Studien legen dar, dass der soziale und globale Selbstwert

über die Lebensspanne differentiell verlaufen, wobei der soziale Selbstwert eine stär‐ kere Sensitivität für kontextuelle Unterschiede zwischen einzelnen Familienformen verdeutlicht. Soziale Netzwerkstrukturen und Regulationsmechanismen illustrieren Zusammenfassung xii deutliche Altersunterschiede und Besonderheiten der Familienformen. Im späteren

Erwachsenenalter zeigt sich eine verstärkte Konzentration auf verwandtschaftliche

Beziehungen und deren Abgrenzung von Nichtverwandten. Dagegen unterstreichen motiviert, jedoch nicht ungewollt kinderlose Personen ein geringeres Interesse an familialen Beziehungen. Dieses Ergebnis konnte auf unterschiedlichen Ebenen bestä‐ tigt werden. Die Resultate zur Aushandlung von Reziprozität zeigen kein klares

Muster. Der dritte Teil der Ergebnisanalysen konnte verdeutlichen, dass die Wert‐ schätzung und Wichtigkeit von Personen für die eigene Selbstbewertung bezie‐ hungsspezifisch variiert und durch Beziehungscharakteristiken der genetischen

Verwandtschaft sowie der wahrgenommenen Nähe und Reziprozität beeinflusst wird. Dyadische Analysen konnten zusätzlich veranschaulichen, dass in Abhängig‐ keit der betrachteten Beziehung Nähe und Reziprozität unterschiedlich stark mit dem Selbstwert assoziiert sind.

Insgesamt unterstreichen die Ergebnisse, dass Selbstbewertungsprozesse durch beziehungsspezifische Charakteristiken beeinflusst werden. Die differentielle und beziehungsspezifische Sichtweise eröffnet neue Perspektiven im Verständnis von lebenslanger Plastizität und Variabilität.

Words count: 372

Introduction 1

1 INTRODUCTION

“The self is relational – or even entangled – with sig‐ nificant others (…) this has implications for self‐ definition, self‐evaluation, self‐regulation, and, most broadly, for the personality functioning, expressed in relation to others.” Andersen & Chen, 2002, p. 619

The need to belong is a well known and frequently described basic desire of our spe‐ cies (e.g., Andersen & Chen, 2002; Baumeister & Leary, 1995). are social creatures, longing for social connectedness and (Dyer, 2000). At the same time, individuals are equipped with a strong sense of the self, which seeks to maxi‐ mize the capabilities and executive functions of the individual (Baumeister & Vohs,

2003; Carver & Scheier, 1998; Tesser, 2003). In this context, some researchers describe closeness to and dependency on others as a risk (Murray, 2005). Others assume and interconnectedness in close relationships to be a prerequisite of independence and (Feeney, 2007). These partially contradictory and desires indicate a need for continual self‐regulation of behaviors, feelings, and ex‐ pressions. The regulation process must address a particular condition that requires either stabilization or change. To identify this status, self‐evaluation has been pro‐ posed as a major process to initiate self‐regulative mechanisms (Crocker & Park,

2004). One frequently identified self‐evaluative indicator is self‐esteem.

Self‐esteem is the evaluation and appraisal of the self (Asendorpf, 1999; Bau‐ meister et al., 2003; Leary & Baumeister, 2000; Schütz, 2000a, 2000b). Research studies support and assess trait and state aspects of self‐esteem (Marsh & O’Neal, 1984; Hea‐ therton & Polivy, 1991). Meta‐analyses (cf., Trzesniewski, Donnellan, & Robins, 2003) and cross‐sectional investigations (cf., Robins, Trzesniewski, Tracy, Gosling, & Potter,

2002) emphasize the life‐long development of the construct. However, investigations are primarily based on global self‐esteem. In order to explore possible different Introduction 2 pathways across the life span, considerations need to widen and include a broader range of domain specific facets of self‐esteem.

The human ability to evaluate and regulate the self has also been shown to af‐ fect people’s social functioning and versa (cf. Rawn, Vohs & Finkel, 2006; Tang‐ ney, Baumeister, & Boone, 2004). In this context, theory suggests that peo‐ ple do not strive for self‐esteem for its own sake (Leary & Baumeister, 2000; Leary,

Tambor, Terdal, & Downs, 1995); but rather to serve as an individual indicator of relational belonging (Kirkpatrick & Ellis, 2001, 2006, Kirkpatrick, Waugh, Valencia, &

Webster, 2002; Leary, 2003a, 2003b; Leary et al, 1995). To date, theoretical outlines and empirical research seem to remain at a fairly global and broad level of explana‐ tion. Even when empirical investigations suggest the importance of specific relation‐ ship types (Denissen, Penke, Schmitt, & van Aken, 2008), theoretical elaborations do not follow.

With this in , current social relationship research is considered. It is well known that within given biological and societal borders individuals actively partici‐ pate in shaping their contexts, developmental surroundings, and social relationships

(Lang, 2005). As a result the social network composition often mirrors recent life tasks or of a person (Lang & Heckhausen, 2006). Such thinking builds on as‐ sumptions about a continuous and dynamic interaction between the individual and context (Asendorpf, 2004). Whether it is the child starting school, the young person leaving home, or the older adult retiring, individuals vast changes in so‐ cial environments across the life span and must continually adjust social relation‐ ships to contexts. Self‐regulation and self‐evaluation a key role in these processes. Adaptive and flexible self‐evaluation is possible only if it is based on a flexible and dynamic mechanism. Relational self‐esteem is proposed to fulfill these re‐ quirements.

Relational self‐esteem describes a relationship specific monitoring process of the value or relevance a certain partner has for the individual’s self‐evaluation. With‐ in a complex and dynamic environment, individuals can differentially weight feed‐ Introduction 3 back from social interactions and rate themselves based on this information. Result‐ ing perceptions of conditions and stages may then be compared to the expected or desired status to initiate self‐regulative behaviors or to break up relationships.

Based on this theoretical outline, the current project offers a psychological con‐ tribution to research on childlessness that involves perspectives from life span perso‐ nality psychology and relationship research. Empirical research on the topic of chil‐ dlessness describes a multiple‐cause structure for dropping birth rates, e.g. the politi‐ cal and social situation, increasing individualization in modern times with the ac‐ companying of self‐realization, as well as the growing occupational orientation and independence of woman (Carl, Bengel & Strauß, 2000; Lutz, 2006). Anglo‐

American research generally considers childlessness as a phenomenon of white, ur‐ ban, non‐religious, and well‐educated women (Carl et al., 2000). However, German statistics suggest that low fertility rates apply to different kinds of social back‐ grounds (Statistisches Bundesamt, 2006). Hence, simple structural and economic de‐ terminants seem to offer insufficient explanations.

Applications of social relationship and personal network research to childless‐ ness are scarce. In one of the few examples, Bernardi (2003) describes processes of social selection and segregation among couples with different family planning ten‐ dencies. Other findings show that nearly half of involuntary childless couples do not talk about their fertility problems (Sabatelli, Meth, & Gavazzi, 1988; Voß, Soeffner,

Krämer, & Weber, 1994), possibly resulting in even greater segregation. Examination of relationship regulation characteristics suggests that childless individuals report lower levels of kinship orientation (Neyer & Lang, 2003). But this study fails to diffe‐ rentiate between types of childlessness (postponed, involuntary, motivated, etc.), which may invalidate results. One aim of the current study is to distinguish among types of childless couples (and to compare them with parents) with respect to their social network structures and relationship regulation characteristics. The framework of lifelong relationship regulation is used as theoretical background in this context

(Lang & Neyer, 2008). Introduction 4

Besides showing the flexibility and adaptability of network and regulation characteristics, the current research addresses the functionality of various network and relationship structures. Utilizing different indicators of self‐esteem, diverse pat‐ terns of self‐evaluation are explored. Relational self‐esteem is used to emphasize rela‐ tionship‐specific values and to investigate in age‐related patterns of self‐evaluation.

In order to explore the dynamics of childlessness, two specific relationship types are selected. Patterns of relationship perceptions, mutual interdependency, and self‐ evaluation are examined in adult child‐parent ties and romantic partnerships.

Intergenerational relationships between adult children and their parents are addressed, knowing that the parent‐child relationship remains important throughout lifetime (cf., Bianchi, 2006; Fingerman, 2003; Schwarz & Trommsdorff, 2005a;

Schwarz, Trommsdorff, Albert, & Mayer, 2005). Also, a mutual influence of dyadic partners is expected. However, structural aspects of the family may affect regulative characteristics and interdependence of the relationship. Instability in childhood has been shown to accelerate biological development and destabilize relationship strate‐ gies (e.g., Belsky, 1997; Belsky, Jaffee, Caspi, Moffitt, & Silva, 2003; Ellis, 2004; Nebe‐ rich, Lehnart, Penke, & Asendorpf, in press). Extrapolating from these findings, ex‐ periences of low family constancy are suggested to be associated with lower interde‐ pendencies between children and parents and with specific self‐regulative relation‐ ship strategies. For example, children are expected to rely on themselves more often.

In addition to the intergenerational tie, couple relationships among motivated and involuntary childless individuals are discussed. Romantic partnerships are known to be the closest intimate relationship (Doherty & Feeney, 2004), to represent a key component of the support system (Neyer & Lang, 2003), and to have positive consequences for the individual’s personality and psychological well being (Lemme,

2006). Prior research on motivated childlessness for example accentuates that moti‐ vated childless women report greater marital cohesion, more intellectual exchange, and higher levels of consensus with their romantic partner compared to involuntary childless women or mothers (Callan, 1987; Houseknecht, 1979). The focus of analyses Introduction 5 is the examination of interdependent structures between men and women from dif‐ ferent family types.

To approach and answer the questions posed here, the current research ex‐ amines two samples: (1) an online study of 557 individuals aged 17 to 86, and (2) a family study of 171 heterosexual couples in middle adulthood and their old parents.

Regarding the online study, questions of cross‐sectional mean‐level differences on self‐esteem and relationship characteristics will be explored. With respect to the family study, three different sub‐samples are utilized. First, age‐homogeneous partic‐ ipants are sampled to address the specific contextual characteristics of childlessness in middle adulthood. This sample distinguishes four family situations: (1) motivated childlessness (n = 82), (2) involuntary childlessness (n = 70), (3) blended families (n =

94), and (4) traditional families (n = 96). Second, a dyadic sample of older parents and adult children is used to analyze dyadic connectedness in intergenerational relation‐ ships (n = 92 dyads). Third, 52 of the middle‐aged couples participated in a follow‐up study. This sample is utilized to explore the association between relationship quali‐ ties and self‐evaluation characteristics across time. Data structures on child‐parent dyads and heterosexual couples not only facilitates consideration from both sides of the relationship, but also allows dyadic analyses in the context of different familial situations (past and current).

This dissertation is organized as follows. The next two chapters present core theoretical considerations. The second chapter briefly illustrates self‐regulative processes, highlighting the importance of the evaluation of the self – specifically cap‐ tured with the indicator self‐esteem. Furthermore, it offers a new perspective on this construct, termed relational self‐esteem. The focus of the third chapter is social rela‐ tionship research. After the definition of important constructs, the integrative framework of relationship regulation is presented and linked with theoretical as‐ sumptions of relational self‐esteem. Finally two specific relationship contexts are ex‐ amined in detail: first, the crucial role of intergenerational relationships (i.e., family of origin); and second, romantic partnerships and their effects on the family situa‐ Introduction 6 tion. Chapter four summarizes theoretical considerations and presents research hy‐ potheses. In chapter five the empirical implementation focuses on the description of the two conducted studies and utilized instruments. The explanation of existing data structures and statistical processes of data analyses conclude this chapter of the dis‐ sertation.

The results section consists of three major parts, its structure being derived from the research questions and hypotheses. The first part focuses on global and do‐ main‐specific self‐esteem indicators, their developmental trajectories, and contextual sensitivity. The second part examines social networks and relationship regulation strategies in different age groups and among motivated and involuntary childless individuals. Finally, relational self‐esteem and relationship specific analyses of self‐ esteem indicators examine the functionality of various relationship patterns. In the last chapter, results are discussed and integrated into the existing theoretical and empirical literature. Limitations of the investigations are noted and final conclusions are drawn.

Theoretical Considerations 7

2 ACTIVE SELF: THE SIGNIFICANCE OF SELF‐ESTEEM

“No matter how much I read, I always had the sense that I did not quite understand what self‐esteem really did and I was never sure whether this reflected my own ignorance or widespread confusion in the field.“ Leary, 2003b, p. 270

How do people perceive, describe, judge, and express themselves? What does it mean to ʺbe oneselfʺ? Do relationships matter in processes of self perception, descrip‐ tion, estimation, and expression? These are fundamental issues for laypersons and professionals alike, leading to widespread considerations of selfhood in modern times. The term self is often used simultaneously for diverse constructs, with self‐ regulative processes being at the core of all of them. Yet, self‐regulation has to be stimulated. One significant condition in this context is the evaluation of the self, often termed and assessed as self‐esteem. Prior to more detailed theoretical examinations of the significance of self‐esteem in social relationship regulation, the differential use of the term self demands an outline.

2.1 Self‐Evaluation: Maintaining a Positive Appraisal of the Self

Within the „Psychologie des Selbst”1, a book on the debate on the nature of the self, Greve (2000) defined the self as a complex and dynamic system that consists of descriptive and regulative structures. The former include internal representations of characteristics such as attitudes, behaviors and interests – or knowledge about the self – which distinguish between the self and others. The latter structures are regula‐ tive mechanisms of activation and functionality, describing the self as acting on itself

(Baumeister & Vohs, 2003). Operating with the descriptive structures, they establish a sense of individual continuity and consistency (see also Freund, 1995; Markus &

Wurf, 1987). Three aspects are particularly important in this context: 1. People are

1 In English the title would be „Psychology of the self“. Theoretical Considerations 8 driven to establish certain goals or desired stages; 2. they engage in appropriate be‐ haviors to obtain these goals; and 3. they monitor the progress that they make to‐ wards goals. In that self‐regulation encompasses processes that guide the individ‐ ual’s decisions and behaviors to maximize its capability, mastery, and satisfaction

(Baumeister & Vohs, 2003; Carver & Scheier, 1998; Tesser, 2003). Self‐evaluation is considered as one necessary condition to stimulate self‐regulative processes, since one major goal is the maintenance of a positive evaluation of the self. The construct of self‐evaluation is frequently termed and assessed as self‐esteem (Leary, 2003b;

Schütz, 2000a).

Use of self‐regulative and evaluative as well as resources may even be necessary for interpersonal functioning (Vohs & Ciarocco, 2004). Vohs and Finkel

(2006) point out that in the last few decades self and relationship have become hot top‐ ics in psychological research as evidenced by many articles and books that address the interplay or possible causal relationship between the two subjects (e.g. Cross,

Gore, & Morris, 2003; Hannover, 2000; Markus & Wurf, 1987; Vohs & Ciarocco, 2004).

Upcoming pages will, first, focus on the current of self‐esteem, pre‐ ceded by a brief definition and outline of the construct is given. Second, Sociometer

Theory is introduced as one theoretical approach to self‐esteem as an interpersonal monitor. The final part of the section states a new way to view the connection be‐ tween self‐esteem and social relations, Relational Self‐Esteem.

2.2 Self‐Esteem – The Construct

Self‐esteem has been studied for decades and is perhaps one of the most extensively considered constructs in behavioral research (Leary & Baumeister, 2000). The concept has been examined in many contexts, such as demographics (e.g. McMullin & Cair‐ ney, 2004), performance (e.g. Baumeister, Campbell, Krueger, & Vohs, 2003), or be‐ havior (e.g. Baumeister, Smart & Boden, 1996). Yet not only Leary (2003b) admits an uncertainty regarding many other aspects of self‐esteem (see also Kernis, 2003; Theoretical Considerations 9

Pyszczynski, Greenberg, Solomon, Arndt, & Schimel, 2004). Still the questions re‐ main: What is self‐esteem? And in particular what is its purpose?

Outline and Definition of Self‐Esteem. In the following self‐esteem is defined as evaluation or appraisal of the self (Asendorpf, 1999; Baumeister et al., 2003; Leary &

Baumeister, 2000; Schütz, 2000a, 2000b). Hence, self‐esteem as an evaluative compo‐ nent must be clearly differentiated from the descriptive, denoted self concept. Be‐ sides the individual level of the construct (high vs. low, e.g. Baumeister et al., 2003;

Baumeister et al., 1996, Neiss, Sedikides, & Stevenson, 2002; Zeigler‐Hill & Showers,

2007), researchers have focused on several other structural aspects of self‐esteem.

They distinguish between facets of stability or situational interference (trait vs. state, e.g. Asendorpf, 1999; Asendorpf & van Aken, 2003; Marsh, Trautwein, Lüdtke,

Köller, & Baumert, 2006; Neyer & Asendorpf, 2001; Robins, Hendin, & Trzesniewski,

2001; Robins & Trzesniewski, 2005; Robins et al., 2002; Trzesniewski et al., 2003), be‐ tween multidimensional contents (global vs. domain‐specific, e.g. Asendorpf & van

Aken, 1993, in Asendorpf, 1999; Heatherton & Polivy, 1991; Marsh et al., 2006; Ro‐ senberg, Schooler, Schoenbach, & Rosenberg, 1995), between components of social cognition (implicit vs. explicit, e.g. Rudolph, Schröder‐Abé, Schütz, Gregg, & Sedi‐ kides, 2008; Schröder‐Abé, Rudolph, & Schütz, 2007; Zeigler‐Hill, 2006), and discuss possible compositions of the ʹoptimalʹ self‐esteem (Kernis, 2003). Recently, several au‐ thors have taken a dynamic and process‐oriented approach to self‐esteem and sug‐ gest assessing optimal self‐esteem through the interplay of intra‐ and interpersonal dynamics (La Guardia & Ryff, 2003). The consideration of additional and wider theo‐ retical approaches is expected to further enlighten the development and functioning of the construct.

Self‐esteem across the Life Span. In this context of process‐orientation and deve‐ lopmental change, the trajectory of self‐esteem should also be displayed across the life span. Recent articles by Trzesniewski and colleagues (Robins et al., 2001; Robins

& Trzesniewski, 2005; Robins et al., 2002; Trzesniewski et al., 2003) consider self‐ esteem across the life span by examining rank‐order stability as well as cross‐ Theoretical Considerations 10 sectional differences. Rank‐order stability of self‐esteem is analyzed with a meta‐ analytic approach considering 50 peer reviewed studies between 1986 and 2003 of global self‐esteem across the life span (Trzesniewski et al., 2003). Test‐retest correla‐ tions illustrate that global self‐esteem is a trait‐like, stable construct, similar to major dimensions of personality such as extraversion or . However, more de‐ tailed examination suggests some variability with less stability in early childhood, a consolidation throughout early and middle adulthood, and a recurring decline dur‐ ing . These results proof against sex differences, usage of different self‐esteem scales, and time of data selection. The authors of this review annotated that research on self‐esteem stability is markedly unequal in its distribution across ages. About 80 percent of studies include samples under the age of 19 and only 5 percent past age

40. Estimates on stability in adulthood and old age are of limited reliability as they refer only to one sample.

4,1

3,9

3,7 (Mean) 3,5

Esteem 3,3

Self 3,1 Total Male s 2,9 Females 2,7 9 ‐ 12 13 ‐ 17 18 ‐ 22 23 ‐ 29 30 ‐ 39 40 ‐ 49 50 ‐ 59 60 ‐ 69 70 ‐ 90

Age (in years)

Figure 1. Mean level trajectory of self‐esteem, for the total sample as well as separately for males and females (pictured results extracted from Robins et al., 2002, p. 429)

Self‐esteem across the life span is pictured by Robins et al. (2001) using a cross‐ sectional online sample of 326,641 people (figure 1). The trajectory shows the highest Theoretical Considerations 11 level in self‐esteem during childhood. In the following decade, between ages 12 to 22, self‐esteem declines strongly. During or just after college a gradual increase takes place and lasts throughout adulthood. Beginning around age 60, self‐esteem marked‐ ly drops again down to adolescent levels. Separate trajectory lines for males and fe‐ males show gender differences in addition to the total sample mean. In childhood the sexes demonstrate no differences in levels of self‐esteem; thereafter the trajectories are parallel with females below males caused by a big difference in the change in , with mean‐levels of girls dropping almost twice as much as boys. This gap is to be fairly stable throughout the life span, narrowing and then disappearing only in older age.

Trzesniewski and colleagues (2003; Robins & Trzesniewski, 2005; Robins et al.,

2002) suggest that variations in self‐esteem occur because of changes in social context or as part of maturation. In that line of reasoning, declines in adolescence and old age appear consistent, since both age periods are characterized by a number of develop‐ mental changes. In old age, the of spouse and friends, decline in physical health from age‐related diseases, and/or a sense of uselessness after retirement are possible explanations for less positive self‐evaluation. On the other hand, the authors pick up theoretical approaches such as those of Erikson (1968; in Robins et al., 2002) or of Baltes and Mayer (1999; in Robins et al., 2002) which suggest that older people are wiser and have „a more modest and balanced view of the self” (Robins et al.,

2002, p. 431). One explanation is that this causes self‐evaluation to become more criti‐ cal and less self‐enhancing in old age.

Several researchers discussed the impact of ‐ or interplay between – evaluations of the self and social relationships (La Guardina & Ryff, 2003; Robins &

Trzesniewski, 2005; Robins et al., 2002; Trzesniewski et al., 2003). One line of re‐ search, namely attachment theory, suggests that processes of self appraisal may be based on different contingencies, and which in turn may be linked to the individualʹs attachment style (Brennan & Bosson, 1998; Park, Crocker, & Mickelson, 2004). Empir‐ ical findings connect secure individuals with strategies of self‐appraisal that are Theoretical Considerations 12 based on a positive relationship to others, particularly to family members. The fol‐ lowing theoretical approach, however, does not endorse attachment, but fosters the role of evolutionary propositions by considering the importance of the connection between social relationships and self‐esteem to be based on the social function of self‐ esteem.

Self‐Esteem and Social Belonging ‐ Sociometer Theory. There is robust evidence for the human desire for social belonging that is somehow accompanied by perceptions and evaluations of the self in social contexts (Bishop & Inderbitzen, 1995; Hendrick,

Hendrick & Adler, 1988; Hill & Buss, 2006; Murray, Holmes, & Griffin, 2000; Neyer,

& Asendorpf, 2001; Schütz, 2000b; Srivasta & Beer, 2005). Sociometer Theory is one theoretical approach capturing this line of reasoning (Leary, 2003a; Leary & Bau‐ meister, 2000; Leary et al., 1995). The outline views self‐esteem to be of concern not for its own sake, but rather as an internal monitor of the individualʹs level of social belongingness (Leary & Baumeister, 2000). Based on the fundamental need to belong

(Baumeister & Leary, 1995), sociometer theory proposes that people protect their self‐ esteem because it measures the individualsʹ “relational value in the eyes of other people” (Leary, 2004a, p. 375). Hence, people are always sensitive, on a preattentive level, to cues that give relevant feedback on their relational value. Early versions of the theory suggested that the sociometer monitors the ʹacceptanceʹ (vs. ʹrejectionʹ) into groups and relationships (Leary & Baumeister, 2000; Leary, 1999, 2003a). More recently, Leary (2007) has shifted to a preference for the terms relational ʹvalueʹ or

ʹimportanceʹ because they allow more precise calibrations (not only acceptance vs. rejection) and offer easier conceptualization in specific situations (feelings of rejection in situations of objective acceptance).

Sociometer theory conceptually distinguishes trait and state self‐esteem

(Leary, 2004a; Leary, Cottrell, & Phillips, 2001; Leary & Baumeister, 2000). The former is the general or chronic level of a personʹs relational value or sense of social accep‐ tance in the absence of new interpersonal feedback. The monitor rests or operates at a relatively high level – or in other words high trait self‐esteem – if a person feels gen‐ Theoretical Considerations 13 erally valued by others, depending particularly on past (Leary, 2004a). In contrast, state self‐esteem is the current or present feeling of relational value within a certain evaluative situation. Changes in state self‐esteem imply changes in perceived likelihood of social inclusion or exclusion. State self‐esteem is particularly sensitive to situations observed by other individuals such as performance on a test or inclusion

(vs. exclusion) on a sports team (Leary, 1999; Leary et al., 1995).

To analyze the precise relationship between social acceptance and state self‐ esteem, Leary and his colleagues (Leary, Haupt, Strausser, & Chokel, 1998) con‐ ducted several experimental studies. Results indicate that increased subjective self‐ esteem accompanies higher relational value and perceived acceptance. However, the association is found to be rather curvilinear, or ogive, not linear. Specifically, the in‐ dividual’s sociometer monitor reacts strongly to evaluations connoting moderate va‐ lence but does not do so with regard to strong negative or positive evaluations

(Leary, 2004a; Leary et al., 1998). Just slightly negative feedback leads to the same pattern of reactions as does strong negative feedback. The same outline appears re‐ garding positive responses. Additional findings point toward a particular sensitivity of the monitoring system to negative feedback. Social exclusion and denial lead to immediate decrease of self‐esteem (Leary, 1999).

Overall, sociometer theory proposes a clear linkage between self‐esteem and the social context. But the theoretical outline has a number of limitations: first, it re‐ mains at a fairly global level. Second, empirical research was conducted only in ar‐ tificial experimental settings. Third, investigations in real relational contexts and on more specific contexts are still very limited.

Kirkpatrick and Ellis (2001, 2006) expand the theoretical model, postulating the existence of not only one but multiple sociometers. This hypothesized differentia‐ tion is based on evolutionary psychology, which suggests a wide range of social do‐ mains to serve different functions. In this respect, humans, as a primarily social spe‐ cies, should be equipped with specific mechanisms and abilities to survive and mas‐ ter challenges in diverse social contexts. Different types of problems, such as finding Theoretical Considerations 14 a mate or conflicts within a group, require specific solution strategies. Natural selec‐ tion is expected to form and sort out this system of specific monitoring processes

(Hill & Buss, 2006; Kirkpatrick & Ellis, 2006). The status of inclusion would then only be one qualitative aspect to be monitored because other relational characteristics af‐ fect the evaluation process as well (Kirkpatrick & Ellis, 2001). In order to fully cap‐ ture the complexity of self‐esteem, examination of descriptive aspects of self‐esteem, such as in particular domains, must be complemented by different functional aspects.

For example, Kirkpatrick and colleagues (Kirkpatrick et al., 2002) addressed the pre‐ dictive value of global and specific functional self‐esteem measures (e.g., social inclu‐ sion or superiority) for aggression. Results support the notion that specific domains of self‐esteem operate in different social contexts. In contrast, global measures of self‐ esteem did not contribute to the prediction of aggression in either study, sustaining the need for specific assessments.

Another empirical investigation of Denissen and his colleagues (Denissen,

Penke, Schmitt, & van Aken, 2008) revealed that there is dynamic interaction be‐ tween state self‐esteem and quantity and quality of social interactions across time.

Multilevel analysis confirmed that across a 25 day period people showed more posi‐ tive self‐evaluations when they spent more time with high‐quality interaction part‐ ners such as a best friend. Moreover, the relationship with the romantic partner emerged as the strongest and most consistent predictor of positive self‐evaluations.

Based on this finding the authors hypothesized that the specificity of romantic rela‐ tionships emphasizes the existence of a specific sociometer.

These empirical findings, applied to the theoretical model of sociometer theory as well as the expansion of Kirkpatrick and Ellis (2001), emphasize the exis‐ tence of multiple (i.e., domain‐specific) sociometers in functionally distinct social in‐ teractions. The following discussion of relationship specific evaluations builds on these theoretical postulations but highlights two additional aspects: (1) theoretical implications of social relationship research, (2) relationship specific estimates of rejec‐ tion or acceptance. Theoretical Considerations 15

2.3 Relational Self‐Esteem: Relationship Specific Evaluations of the Self

Based on the perception of relational value in any given interaction, the sociometer as a psychological mechanism should facilitate self‐regulation in interpersonal

(Leary, 2004b). Thus the perception will affect the individual’s self‐esteem (Denissen et al., 2008; Kirpatrick et al., 2002). In other words, sensitivity to interpersonal cues concerning the perceived relational value in social relationships or interactions influ‐ ences personality characteristics, which in turn interact with interpersonal relations.

The interdependence of personality with social relationships has been addressed by several researchers (e.g. Asendorpf & Wilpers, 1998; Neyer & Asendorpf, 2001; Neyer

& Lehnart, 2005). Most empirical results show effects of personality on relationships to be more dominant than vice versa. In this context, a four year follow up study on young adults illustrates that higher global self‐esteem leads to less insecurity in con‐ tact with friends and colleagues across time (Neyer & Asendorpf, 2001). Hence, the process of self‐evaluation may be closely related to the quality of relationships and thus to the composition of social networks.

Since various types of social affiliations are known to have different functions

(e.g. Clark & Mills, 1979; Fiske, 1992; Hollstein, 2001), people will have specific expec‐ tations, needs, or requirements for various social interaction partners. For this , self‐evaluation is hypothesized to depend on relational partners within associated prospects. For instance, depending on the individualʹs status in a group, the duration of a social affiliation, or the degree of genetic relatedness to the interaction partner, the meaning of ties and, consequently, the importance of perceived relational value may vary. Being judged by unimportant people in one’s life (e.g., a sales person) or with regard to irrelevant personal characteristics (e.g., drawing skills for an athlete), should have less impact on an individualʹs self‐evaluation than judgments of rele‐ vant relationship partners or in meaningful areas of life. In addition to variations in the processes of self‐evaluation with diverse relationship partners, adaptations are expected across time and environmental circumstances (Antonucci, Akiyama, & Ta‐ kahashi, 2004; Lang, 2001; Lang, Neyer, & Asendorpf, 2005). According to life span Theoretical Considerations 16 psychology, relationship characteristics and functions vary across the life span. Deve‐ lopmental changes, the engagement with new goals, or choice of a particular life style will reshape the social network (Lang & Heckhausen, 2006) and alter adapta‐ tions in the process of self‐evaluation. Thus, the sociometer, as an internal gauge of perceived esteem, value, or rejection, is hypothesized to deal with these relation and context specific demands.

An illustration will allow to clarify in what ways the theory is also speaking to mechanisms of relationship regulation: Take two young women, Laura and Anna, who have been best friends since childhood. They went to school together, still live close to each other and regularly exchange their personal problems, confidences and . The opinion of the respectively other is highly important for each of them.

Hence, the self‐esteem monitor is extremely sensitive to the other person’s validation or opinion. At about the same time, both women and their husbands decide to be‐ come parents. The women start to exchange about pregnancy and mother‐ hood. Half a year later, Anna conceives. Although Laura has not; she is happy for her friend and expects the best for herself. But a year later, the women face totally differ‐ ent situations: Anna is the mother of a three month old daughter. Laura in contrast, is sitting in a hospital, waiting for her first fertilization treatment. The interaction and exchange between the two has altered dramatically. Laura protects herself from An‐ na’s joy in motherhood. Anna is totally involved with her young child and all the resultant responsibilities. With less frequent get‐togethers and less to discuss, the relationship declines and the validation that Anna and Laura receive from each other also declines. This is an extreme (yet possible) transformation.

The decreased importance of a former important relationship partner through changes in life circumstances can have extensive effects on the processes of self‐ evaluation and in that on relational self‐esteem. The adaptation to relationship struc‐ tures by means of changed parental and partnership status, geographical mobility, or educational and occupational transformations are of critical importance (Bernardi,

2003; Bidart & Lavenu, 2005; Schneewind et al., 1994). The functionality of change in Theoretical Considerations 17 relational structures is suggested to be captured in the relationship specific monitor‐ ing mechanism of relational self‐esteem. This mechanism is hypothesized to respond directly to feedback or other interpersonal cues of relationship partners and the rela‐ tionship itself. Thus, characteristics of relationship quality are important facets. The expected variation may occur as a function of the relationship specific value of this person for the individual’s self‐evaluation.

Even within a particular relationship, such as a long‐term friendship, envi‐ ronmental or personal changes may alter perceptions of this relationship partner or the relationship itself, guiding changes in perceived relational value. The conceptual connection of the two theoretical approaches, the Sociometer Theory and the Framework of Lifelong Relationship Regulation, assumes the existence of relationship specific moni‐ toring systems that assess the particular function of interpersonal ties. Often changes in environmental or living conditions affect several relationships. Adaptations to var‐ ious social interactions are necessary. Thus, Anna would be expected to focus more on family relations or friends who share the joy of parenthood. In particular, the rela‐ tion with her mother may be more relevant to her instrumental and emotional sup‐ port. Therefore the relational value of this relationship partner becomes particularly important in her self‐evaluation. In the strained situation of infertility treatment,

Laura may instead, avoid gatherings of family and friends with children, concentrat‐ ing solely on more supportive relationships. In the long run, the adaptation process will change the composition and role of her social relationship network and regula‐ tive strategies. This evaluation of the importance or relevance of a specific relation‐ ship partner in changed life situations has to be carried out with a relationship specif‐ ic measurement.

The relational self‐esteem is proposed as psychological mechanisms that moni‐ tors relational value across network partners. In connection with the Framework of

Lifelong Relationship Regulation this psychological mechanism is hypothesized to sort out the functionality of single relationship partners in different age groups and con‐ ditions. Theoretical Considerations 18

2.4 Summary

Theoretical considerations on the significance of self‐esteem tap into an array of top‐ ics and contexts. Even so, a gap still exists, first with regard to research investigations of self‐esteem characteristics in middle and later adulthood, and second with respect to specific aspects or functions in the process of self‐evaluation. By connecting self‐ esteem and social affiliations sociometer theory proposes already the importance of the egoʹs relational value mirrored by the social surrounding (Leary, 2007). Yet, in‐ vestigations have been comparatively broad and empirical investigations remained at the experimental level in artificial lab situations. More defined propositions by

Kirkpatrick and Ellis (2001, 2006) still remained at a reasonably global level, unable to describe and capture specifics of relational value. Thus, the functionality of partic‐ ular social relationships in different age groups or circumstances was still not cap‐ tured.

An additional point to consider is the limitations in the methodological tech‐ niques employed. To date research commonly compares global estimates of the self with such situation or relationship specific features as interaction quality with a ro‐ mantic partner (Denissen et al., 2008) or reactions to specific empirical manipulations

(Kirkpatrick et al., 2002). These assessments occur to be at different levels of repre‐ sentation and/or processing. The following empirical study shifts the focus of as‐ sessment to parallel levels of consideration: Intra‐individual variations of relational relevance will be measured at the specific relational level. In other words, the as‐ sessment of relational value will take place regarding each social relationship. This implementation should lead to a better understanding of the psychological function of the relational self‐esteem in particular social contexts. Theoretical Considerations 19

3 SOCIAL TIES AND RELATIONSHIP REGULATION:

PERSONAL NETWORKS IN DIFFERENT CONTEXTS

Relationship phenomena are (…) interperson‐ al dynamics that are more than the sum of the interactants’ characteristics. Vohs & Finkel, 2006, xi

Social relationships constitute the intersection between person and environment. Act‐ ing in different circumstances and striving for different aspirations, such as being a parent vs. being childless, lead to specific behavioral patterns that eventually become explicit in structures and regulation characteristics of social affiliations (Lang &

Heckhausen, 2006). Perceptions of relational belonging, or feedback in more general, happen to take place right at this junction. Thus, relational self‐evaluation should be associated with characteristics of social relationships, personal networks, and rela‐ tionship regulation strategies. Within this process, individuals are perceived as active participants. They co‐regulate social environments by selecting, maintaining, or re‐ fusing social affiliations with kin and non‐kin. Network structures and relationship characteristics should reflect the differential value of relationship partners in the in‐ dividual’s self‐evaluation. Selecting the phenomenon of childlessness, the present research concentrates on characteristics of social relationships and self‐evaluation in motivated and involuntary childless individuals. The comparison with two other familial structures, blended family and traditional parents, further accentuates simi‐ larities and discrepancies between different contextual circumstances.

Social relationships are defined here as dyadic interactions that are stable over time and across contexts, and framed by the relationship’s history and anticipated future (Antonucci & Akiyama, 1995; Asendorpf, 2004; Hinde, 1993). Social relation‐ ships differ according to structural aspects, such as contact frequency or geographical proximity (Björklund & Pellegrini, 2002), as well as qualitative characteristics, such as perceived closeness, conflict frequency or amount of support. Relationship characte‐ Theoretical Considerations 20 ristics depend on the behavior of involved actors as well as their mutual emotions, cognitions, and circumstances (Asendorpf, 2004; Lang, 2003). Structures of relation‐ ships are dynamic, because actors and contexts of interaction vary across time. Indi‐ viduals respond to changing circumstances and manage opportunities. The regula‐ tion of personal interactions is framed by those conditions but additionally driven by individual goals, attitudes, and motivational processes. The concept of active rela‐ tionship regulation captures those behavioral and cognitive processes that start, ad‐ just, rearrange, or end relationships and thereby alter a social network. Those processes may focus on the self, the relationship partner, or the context (Lang, 2001,

2007a; Lang & Heckhausen, 2006). More generally stated, of the numerous social ties formed across the life some intertwine and strengthen while others disband, become distant and fade away. This process of forming, structuring and dissolving relation‐ ships is actively led by the involved person (Lang, 2003). At a given moment in time the whole of a person’s social relations consists of ties from different stages of the individual’s history, with diverse relationship durations and qualities. This structure mirrors former experiences and achievements as well as recent life tasks, develop‐ mental stages, and future goals (Lang & Heckhausen, 2006).

The social network approach provides a broader view of personal relation‐ ships in context. Social networks are a system of diverse actors and their linking ties

(Hollstein, 2001; van Duijn & Vermunt, 2006). The linking tie might be a simple inte‐ raction or stable relationship between two actors (Hinde, 1993). distin‐ guishes several types of network data. The current thesis focuses on ego‐centered, personal networks, which treat one actor and all (a certain percentage or selected type) of his social relationship partners (Vermunt & Kalmijn, 2006). Ego‐centered so‐ cial networks are defined as the whole extent of relationships of one individual in‐ cluding his/her kind of kin and non‐kin ties (Hollstein, 2003). The assessment of structural aspects like network size or geographical proximity is frequently com‐ bined with functional or qualitative information, such as satisfaction with or support from relationship partners (Hollstein, 2001). Hence, it provides researchers with dis‐ Theoretical Considerations 21 tinguishing information on overall network characteristics as well as features of each single network partner (Antonucci & Akiyama, 1995; Kogovsek & Ferligoj, 2005). The investigation of ego‐centered relationship networks considers social relationships as within and between phenomenon. In consequence, it provides researchers with information on the variation of relational phenomena in addition to individual differences (Gable & Reis, 1999). Empirical research suggests both consistency and variation in the composition of social networks over time, as well as both beneficial and disadvantageous effects on the individual’s development and well‐being (Anto‐ nucci, Akiyama, & Takahashi, 2004).

The following section briefly introduces two of social relationships widely acknowledged in psychological research: the perception of attachment and the striving for equity and balance within personal ties.

3.1 Two Perspectives on Social Relationships

(1) Attachment. Regardless of earlier theoretical considerations, fundamental studies of social relationships did not appear until the middle of the 20th century. Bowlby

(1969) focused on the relationship between and their primary caregivers, usually their mothers or other kin. He coined the term attachment to describe the bond between the two. In his understanding infants are predisposed to maintain proximity with their primary caregiver, particularly in times of danger or threat.

More recent formulations refer to attachment as „the dimension of the caregiv‐ er relationship that is concerned with the seeking of security by the infant and its provision by the caregiver” (Nash, 1988, p. 123). Emerging working models of at‐ tachment are becoming more sophisticated, with the growing child attaching to a broadening social world and to non‐kin relations. In attachment theory the quality of first attachment relationships is prototypical for later relationships outside the family

(Bowlby, 1969). The infant’s first relationships play a pivotal role for well‐being, warm relationships with friends in adolescence, and more satisfying romantic rela‐ Theoretical Considerations 22 tionships in early adulthood (Posada et al., 1999; Schneider, Atkinson & Tardif, 2001;

Simpson, Collins, Tran, & Haydon, 2007; Slade, 1987). More recent theoretical and empirical investigations have adopted and advanced attachment theory to relation‐ ship phenomena across the entire life span (Ainsworth, 1989; Bartholomew & Hor‐ rowitz, 1991; Doherty & Feeney, 2004; Griffin & Bartholomew, 1994; Mikulincer &

Shaver, 2001; Mikulincer, Shaver, Gillath, & Nitzberg, 2005). Nevertheless, attach‐ ment theoretical considerations remain at a descriptive level. They barely address the regulative pattern or relationship between effects and systems.

(2) Equity and Cooperation. Individuals strive for a mutual give‐and‐take within social interactions (e.g., Heider, 1958; Homans, 1961; Thibaut & Kelley, 1959). Percep‐ tions of reciprocity are achieved by different strategies or investigations, such as the evaluation of costs and benefits in interdependent interactions (Thibaut & Kelley,

1959), or the judgment of similarity of exchanged resources and estimations of com‐ petence and neediness of the exchange partner (Foa & Foa, 1976). One constitutive characteristic of reciprocal behavior is seen in self‐interest maximization. However, Pe‐ rugini and colleagues (Gallucci & Perugini, 2000; Perugini, Gallucci, Presaghi, & Er‐ colani, 2003) propose that reciprocity is an internalized, universal social norm or mo‐ tivation (c.f. also Gouldner, 1960; Sahlins, 1972). It should be understood as a perso‐ nality construct with individual variation particularly dependent to “the type of re‐ ciprocators (positive vs. negative), … differences in sensitivity to previous events

(positive vs. negative), sanction opportunities (reward vs. punishment), and sanction consequences (fair vs. unfair)” (Perugini et al., 2003, p. 272). In certain interpersonal circumstances behavioral strategies appear to be driven by a norm or by a to balance out costs and benefits between interaction partners.

Yet, there are situations where people help in the context of non‐ interdependence or anonymity: altruistic behavior. considers the welfare of the other, instead of the self (e.g., Batson & Shaw, 1991; Krebs & Van Hesteren, 1994).

Altruism may vary across a spectrum of devoted benefit and enhancement (Krebs &

Van Hesteren, 1994), or it may vary in its underlying motives and ultimate goals Theoretical Considerations 23

(Batson & Shaw, 1991). The existence of purely altruistic behavior is questioned

(Cialdini, Brown, Lewis, Luce, & Neuberg, 1997; Krebs, 1991). At least it may be li‐ mited to particular forms such as mutual or universal altruism (Krebs, & Van Heste‐ ren, 1994). Therefore altruistic behavior is often addressed in specific contextual cir‐ cumstances such as in kin relations (Hamilton’s theory of inclusive fitness, 1964), and in reciprocal altruism between non‐kin (Trivers theory of reciprocal altruism, 1971; c.f. Fetchenhauer & Bierhoff, 2004, for a review). Alternatively, altruism is seen as promoted behavior in the area of sexual selection because of a preference of altruistic sexual partners (Miller, 2001, in Fetchenhauer & Bierhoff, 2004). An additional aspect in the light of altruistic behaviors is the role of punishment (Fehr & Gächter, 2002).

Altruistic punishment occurs despite being costly and unprofitable for the ego. In experimental setups cooperation increased if punishment was permitted, but de‐ creased if punishment was prohibited. Thus, punishment could be considered as al‐ truistic act, particularly in social interactions between genetically unrelated strangers or in situations with little or no gain.

There is robust evidence that relationship partner prefer balanced social inte‐ ractions to under‐ and to over‐benefiting relations (e.g., Befu, 1980; Gouldner, 1960;

Ikkink & van Tilburg, 1998, 1999; Jung, 1990; Krackhardt & Kilduff, 1999; Lewinter,

2003; Sahlins, 1972; Schulz, 1996). The perception of mutuality in social relationships depends on several factors: (1) such relationship attributes as genetic relatedness, perceived closeness, or interdependence (e.g., Rusbult, 1983; Sahlins, 1972; Schulz,

1996); (2) such partner attributes as personality, social status, or available resources

(e.g., Thibaut & Rieken, 1955); and (3) attributes of exchanged items such as their value, social meaning, or availability (e.g., Foa & Foa, 1976). Nevertheless, reciprocity and its achievement are still an aspect of relationships that is not very well unders‐ tood or satisfactorily explained. One major shortcoming is the often highly descrip‐ tive approach. It neglects underlying mechanisms that are necessary to regulate and adapt in complex social contexts. Theoretical Considerations 24

Taking together, attachment and equity theoretical approaches are major ex‐ planations for personal relationships in social sciences, with each capturing different characteristics of social ties. Several attempts have been made to integrate the two perspectives. Clark and colleagues (Clark & Grote, 1998; Clark & Mills, 1979, 1993;

Clark & Taraban, 1991) introduced the model of communal and exchange relation‐ ships. Fiske and colleagues (1992; Fiske & Haslam, 1996; Fiske, Haslam & Fiske, 1991) suggested the relational‐models theory. Although give‐and‐take and closeness are considered within the theoretical frameworks of both models, a major shortcoming is the descriptive nature of the approaches. Underlying regulatory mechanisms are dis‐ regarded. Combining theoretical assumptions of human evolution and empirical evidences from relationship and personality research, Land and Neyer (2005, 2008) recently introduced a new framework of lifelong relationship regulation that sug‐ gests the operation of two fundamental mechanisms of social relationship regulation.

3.2 Integrative Framework of Lifelong Relationship Regulation: Contextual and

Relationship Effects

The theoretical framework of lifelong relationship regulation (Lang & Neyer, 2005) originates in assumptions of evolutionary psychology: humans have evolved with the psychological mechanisms to succeed in numerous adaptive problems and thus, directly or indirectly, can enhance their reproductive fitness (Buss, 2004). Lang and

Neyer (2005, 2008) attempt to integrate principles of evolutionary and lifespan psy‐ chology based on three proposition for the regulation of behavior in social contexts.

First, human behavior is partly rooted in ultimate principles of sexual selection, kin‐ ship preference, and cooperation, all connected to the basic goals of survival and re‐ production; second, social relationships and personal networks are dynamic and flexible, expected to adapt constantly within the person‐environment transaction; and third, interdependence between actors of dyads and groups leads to goal‐driven Theoretical Considerations 25 actions, directed to alter the self or the environment in order to master developmen‐ tal challenges (Lang, Wagner, & Neyer, in subm.).

The ultimate mechanisms are assumed to be translated or manifested in prox‐ imate psychological processes. The two identified processes, involved in monitoring and differentiating social interactions, are regulation of closeness and reciprocity moni‐ toring (Lang & Neyer, 2005, 2008). The current framework considers these two to be the fundamental mechanisms for understanding interpersonal relations, for the diffe‐ rentiation of relationship partners, and for adaptation to developmental circums‐ tances across the life span.

Regulation of Closeness. ʺBlood is thicker than waterʺ is a well known saying that captures a widely acknowledged aspect of human : kin are pre‐ ferred over non‐kin, particularly in times of crises or resource scarcity. A preference for kin implies that humans know, or have an ability to detect, biological kin. But humans have no obvious ability to distinguish degrees of genetic similarity between the self and interaction partners. Different empirical investigations of Neyer and

Lang (2003) point to a proximate cue for genetic relatedness: perceptions of subjec‐ tive closeness. Statistical analyses found a stable intraindividual association between perceived closeness and genetic relatedness which proved robust across different samples and age groups.

The mechanism of closeness regulation is arguably related to the human at‐ tachment system because it is based in social experiences of early childhood and op‐ erates particularly in the regulation of relations with close and important others, such as kin. However, humans desire to be broadly affiliated (Baumeister & Leary, 1995).

They feel close to non‐kin and seek stable, reliable relationship partners in this group, as well (Ackerman, Kenrich, & Schaller, 2007; Antonucci, Akiyama, & Takahashi,

2004; Doherty & Feeney, 2004). Across time the attachment system undergoes norma‐ tive developments and broadens to non‐kin relationship partners, such as romantic partners (Zeifman & Hazan, 1997). The mechanism of closeness regulation allows to create a sense of kinship, and thereby enables people to develop stable, reliable, and Theoretical Considerations 26 supportive relationships even with non‐genetically related others, e.g., parents with an adopted child (Lang & Neyer, 2008; Lang et al., in subm.). Thus, the mechanism occurs also without the evidence of genetic relatedness, without regular contact, or in relationships of brief acquaintanceship (Lang, 2007a). Regulation of closeness cap‐ tures a human flexibility that facilitates the formation of close ties to kin and to non‐ kin.

Reciprocity Monitoring. Monitoring of reciprocity is a second mechanism that differentiates among social relationship partners (Lang & Neyer, 2005). Within each dyadic interaction a minimum of mutuality is necessary for relational satisfaction.

Social exchange, and in this context the formation of a stable dyadic interaction, de‐ pends largely upon the principles of cost minimization and reward maximization

(Thibaut & Kelley, 1959). However, both actors in a relational dyad strive for effec‐ tiveness and situational control. For that reason, relationships are assumed to rely on the obligatory norm of reciprocity (Befu, 1980; Gouldner, 1960). Establishing a mutual relationship implies a balance or equilibrium in the perceived cost‐reward matrix of both involved individuals. The perception of imbalance will lead to more monitoring or demand (in terms of under‐benefit) or investment (in terms of over‐benefit) to achieve or stabilize a certain degree of reciprocity.

Cooperative relationship partners, such as colleagues or neighbors are particu‐ larly based upon the evolutionary mechanism of cooperative altruism. Perceived im‐ balance in these kinds of relationships frequently leads to their dissolution. Less ri‐ gorous expectations for reciprocity occur in the context of long term or highly inter‐ dependent relations (Lang, 2007a). Clark and Mills (1979) showed that imbalance in non‐kin communal relations, such as friendships, is associated with higher degrees of liking. Kin relations, however, entail an exceptional role in reciprocity monitoring.

For example, Ikkink and Van Tilburg (1998, 1999) emphasize the stability of kin ties even if relationships are perceived as one‐sided or imbalanced. Intergenerational in‐ terdependency assumes all generations to be active providers and receivers at a cer‐ Theoretical Considerations 27 tain time across the life span. Thus, reciprocity is assumed to be more generalized in the contexts of kinship (Sahlins, 1972).

In addition to the type of relationship, relational features such as contact fre‐ quency (e.g., Ikkink & Van Tilburg, 1999), or individual characteristics such as status, money and age (e.g., Grundy, 2005) have an influence on the strength of reciprocity monitoring. Lang (2007b) anticipates that the perception of resource similarity in‐ creases peoplesʹ tolerance of imbalance in exchange relationships. However, the evaluation of resource similarity occurs in a relatively open space of personal inter‐ pretation, likely the process of reciprocity monitoring is highly dynamic. This attribute constitutes not only the diversity in cooperative relationships, but also the difficulty in the assessment of perceived reciprocity and reciprocity monitoring.

Evolutionary Psychological Taxonomy of Social Mechanims Mechanims Relationships

1. Kin Selection Regulation of 1. Biological Kin Psychological Closeness 2. Sexual 2. Elective Kin & Selection Partnership Reciprocity 3. Cooperative Monitoring 3. Colleagues, Altruism (Cost‐Benefit‐Equation) Cooperation

Figure 2 Integrative Framework of Lifelong Relationship Regulation (Lang & Neyer, 2008)

Both mechanisms are proposed to be adaptive in the context of active and self‐ adjustive relationship regulation. They establish and control emotional closeness and perceived reciprocity depending on context and relationship specific opportunities.

Both processes vary intra‐relationally (within one relationship), intra‐individually

(within one person) and inter‐individually (between persons). The interaction be‐ tween the two mechanisms will differ according to relationship types and partners

(see figure 2). For the most part, both mechanisms participate in the process of rela‐ Theoretical Considerations 28 tionship regulation. The variation in intensity serves as differential parameter among types of relationships. This differential variation results in a multiplex relationship system, wherein the two mechanisms exist in an ongoing reciprocal dependency.

As shown in figure 2, the process of closeness regulation in particular charac‐ terizes kin ties, while reciprocity monitoring is more prominent in cooperative rela‐ tionships with neighbors or colleagues. In romantic partnerships both mechanisms operate in high concentration, and a stable partnership relies on both a high level of closeness and the perception of a balanced give‐and‐take. Preliminary discriminative analyses based on empirical network data support the differential pattern of this so‐ cial relationship taxonomy (Neyer, Wrzus, Wagner & Lang, in subm.).

Contextual Effects – Age and Family Situation. Existing examinations regarding age related changes in relationship regulation indicate a stable association between closeness and genetic relatedness that strengthens in old age (Neyer & Lang, 2003;

Lang, Wagner & Neyer, in subm.). Genetic kin retain robust and intimate network partners throughout the life span. Only a little information exists on changes in the observation and regulation of reciprocity. As people age, high levels of independence and mutuality are harder to maintain. Physical and cognitive impairments decrease autonomy and thus threaten perceptions of reciprocity. To avoid this monitoring process, fewer cooperative relationships may be kept in the personal network, since imbalance is more likely to be accepted in close kin ties. Socioemotional selectivity theory (SST) addresses the role of life time perception as motivational explanation for differential priorities in types of goals across the life span (Carstensen, 1992; Car‐ stensen, Isaacowits, & Charles, 1999). Older people tend to focus on emotionally meaningful goals and relationships (e.g., Lang & Carstensen, 1994, 2002). The prima‐ cy of emotional relevant relationship partners may also increase their importance for them for people’s self‐evaluation in general. Thus, regardless of a decline in perfor‐ mance related abilities, people are capable to maintain emotionally meaningful social relationships that may help to sustain a positive view of the self. Social networks in Theoretical Considerations 29 old age are likely to be characterized by kin and elective kin ties, by a high level of perceived closeness, and by high kinship orientation.

The life span perspective emphasizes the simultaneous effect of individual and situational or contextual factors in forming personal networks. Initial analyses by Neyer and Lang (2003) indicated medium sized gender effects in closeness regula‐ tion. Women showed higher kinship orientation than men, but this was only true only in middle and old age (45 and up). Older people with no romantic partner and parents of all age groups demonstrated higher associations between closeness and genetic relatedness. Analysis based on parental status did not specifically consider motivational or situational conditions of the existing situation (cf., section 3.4), and thus needs further confirmation. Substantially lower kinship orientation for non par‐ ents, compared to parents, is expected only in the context of motivated childlessness.

This group may be oriented less towards kin and more towards cooperative, non‐kin relationships. People striving for children, the involuntarily childless individuals, are likely to be very similar to parents in the relationship partners they select and in how they differentiate among them.

Relationship Effects – Taxonomy of Social Relationships. Relationship regulation is an active and self‐adaptive process, important to function on a daily basis and to realize specific goals. Individuals have – within a certain realm of possibility ‐ an ability to adjust their relationship patterns by choosing, forwarding, fostering, or ending particular social interactions. Different relationship types and partners serve various functions and support different developmental periods, contexts, and re‐ quirements. Here, the three major relationship groups of the frameworkʹs taxonomy are distinguished: 1) biological kin, 2) elective kin, and 3) non‐kin ties.

Belonging Together – Biological Kin. Kin networks consist of relationship part‐ ners who share a certain genetic similarity. The line of descent is expressed in the coefficient r ranging from ʹ1ʹ to ʹ0ʹ, with the degree of genetic relatedness moving from 100% in identical twins to 0% in all non‐kin relations, including romantic part‐ ners and in‐ family members. Kinship systems are culturally normative structures Theoretical Considerations 30 or networks of concrete social ties, characterized by specific conditions of solidarity and cooperation (Nave‐Herz, 2004). Structural characteristics of kin ties are compara‐ tively stable. In addition, family relations are affected by kinship norms or obliga‐ tions that may be directly involved in concrete behaviors (e.g., Perrig‐Chiello &

Höpflinger, 2005; Schwarz et al., 2005; Shmotkin, 1999).

Intergenerational ties are bonds of exchange, dependence, obligation, and care giving (Bianchi, 2006). They are characterized by high degrees of intimacy, familiari‐ ty, and harmony (Doherty & Feeney, 2004; Neyer & Lang, 2003). Most people remain closely attached to their family throughout life (Fooken, 1997; Nave‐Herz, 2002). Re‐ search additionally shows that family relationships are not solely close but some‐ times ambivalent (Fingerman, 2003; Fingerman, Hay, & Birditt, 2004; Lüscher &

Pillemer, 1998; Schwarz, et al., 2005). Psychological ambivalence is especially likely in very close (family) relations (Fingerman et al., 2004), possibly because of the interde‐ pendence among family members (Lüscher, 2005). Older adults2 tend to report more solely close and fewer ambivalent relations, as well as less tension with relationship partners, in particular with children (Fingerman et al., 2004; Birditt, Fingerman, &

Almeida, 2005). This finding addresses another important factor of family ties. Since kin relationships are often intergenerational, age dependent changes have to be con‐ sidered. With respect to kin relationships, these functional adaptations of socioemo‐ tional selectivity in older age may increase the reproductive success of one’s genetic family since improved socioemotional regulation seems to be associated with a stronger concentration on close (most notably kin) ties (cf., also Carstensen &

Löckenhoff, 2003).

Coming Together – Elective Kin3. Elective kin relationships are of four different types. First, romantic partners or spouses are not biological family members, but in becoming parents they produce a new lineage of genetic relatedness. The second

2 Old age in this context generally refers to the life period above the age of 65 years.

3 The term “elective” is based on the fact that if listed in ego‐centered social networks, people freely choose to include these non‐kin family members. Theoretical Considerations 31 type includes all in‐law relationships in a stable romantic partnership or .

Third, in a remarriage situation step relationships occur; one may become a step‐ child, step‐sibling, or step‐parent. Fourth, long‐term friends may be regarded as qua‐ si‐kin.

The tie to the romantic partner is the closest, most intimate, social relationship throughout adulthood (e.g., Doherty & Feeney, 2004; Fingerman & Hay, 2002). Satis‐ faction levels and overall commitment are related to costs, rewards, and investments made within partnerships (Rusbult, 1983). Thus mutuality, interdependency, and dyadic cohesion affect perceptions of intimacy and overall perceived relationship quality. Empirical findings further support the interdependent association of com‐ mitment with trust (Wieselquist, Rusbult, Foster, & Agnew, 1999), with willingness to sacrifice (Van Lange, De Bruin, Otten, & Joireman, 1997), and with accommodative behaviors (Rusbult, Verette, Whitney, Slovik, & Lipkus, 1991).

Murray and colleagues (2005; Murray, Holmes, & Griffin, 2000) emphasized the importance of a sense of security to feeling committed to a romantic partner and in regulating and maintaining closeness. The sense of security may be particularly supported through positive regard by the partner. This in return is a fundamental aspect that leads to self‐affirmation and relationship‐promoting behaviors, as they are found in stable dating and spouse partnerships. Empirical studies on depend‐ ency‐regulation processes confirmed the essential role of self‐esteem. People with low self‐esteem were less likely to use their partner’s love and acceptance for self‐ affirmation (Murray, Holmes, MacDonald, & Ellsworth, 1998; Murray et al., 2000).

Low self‐esteem increased the probability of reading too much into problems, lead‐ ing to a decrease of perceived closeness with a romantic partner (Murray, Rose, Bel‐ lavia, Holmes, & Garrett Kusche, 2002). Besides the level of self‐esteem, its source, i.e., relational versus competitive, were shown to impact the relationship between attachment and global self‐esteem in spousal relationships (Brennan & Bosson, 1998).

Longitudinal studies on childhood experiences revealed a long‐term effect on emotional perceptions in romantic relationships (Simpson et al., 2007). A secure at‐ Theoretical Considerations 32 tachment style in childhood was associated with positive social development and led to more positive emotional experiences and less negative effect. The interplay be‐ tween personality and romantic relationship quality were considered in a longitu‐ dinal study by Lehnart and Neyer (2006). Results showed that reciprocal interdepen‐ dencies are higher in comparatively stable relationship structures. A cross‐sectional study on young adult couples revealed an association between the quality of part‐ nerships and current social network structures (Neyer & Voigt, 2004). This confirms the effect on partnerships of the individual’s social embeddedness.

Couple relationships do not exist in a social cocoon, but are surrounded and embedded into a complex social world. One specific consequence of a steady part‐ nership is that two social networks intertwine. New in‐law relationships have to be included in the pre‐existing social network. In‐law relationships differ in one impor‐ tant aspect: relationship partners are not genetically related. One implication is that in‐law ties are complex. This can be seen in numerous fairy tales, but there is also evidence in recent research.

Voland and Beise (2005) analyzed historical data of the Krummhörn commu‐ nity and found three major risk factors for stillbirth mortality: first, the mother‐in‐law is alive; second, the duration of marriage between son and daughter‐in‐law is either very short or very long; and third, high geographical proximity. While striving to reduce paternity uncertainty and protecting her own lineage, the husband’s mother becomes the ʺdevil in the houseʺ (p. 1). Using more recent data, Santos and Levitt

(2007) found support and relationship quality to be important predictors that en‐ inclusion of in‐law family members in social networks. Bucx and colleagues

(Bucx, van Wel, Knijn, & Hagendoorn, 2008) showed that the quality of the in‐law relationship also affected the child’s contact frequency with his/her own parents.

Steady partnerships broaden people’s social network. Thus they may also lead to more integrative and regulative activities that affect relationships both in general and with parents specifically. Theoretical Considerations 33

Taking up the matter of step‐family members, Schmeeckle and colleagues

(Schmeeckle, Giarrusso, Feng, & Bengtson, 2006) asked “What makes someone fami‐ ly?”. Empirical results based on the solidarity‐conflict model showed that several structural variables affect the answer to this question. Adult child’s answers patterns were influenced by such facts as whether the step‐parent was still married to the bio‐ logical parent or whether the child ever lived with the step‐parent. The majority of adult children did not regard stepparents as family. Love and Murdock (2004) ad‐ dressed the interplay between well‐being and intact vs. step‐family structures. Em‐ pirical findings showed that step‐children were less often securely attached, and that attachment served as significant predictor of well‐being. The great diversity of blended‐families is an important aspect to take into account in empirical investiga‐ tions (Dunn, 2002).

One special type of non‐kin ties is friendship. Friendships are close relation‐ ships but also based on support and reciprocal interactions (cf., Deci, La Guardia,

Moller, Scheiner, & Ryan, 2006). They were often defined as selected relationships which are heavily based on similarity in values and interests (cf., Blieszner & Adams,

1992). Furthermore, having friends is positively connected to mastering transitional phases (Hartup, Hertzig, & Faber, 1998; Hartup & Stevens, 1997) and they remain important throughout the life span (Antonucci, Lansford, & Akiyama, 2001). Wom‐ en’s treatment of friends often seems very similar to the way they treat kin (Acker‐ man et al., 2007) and apes some aspects of their spousal relationships (Voss, Markie‐ wicz, & Doyle, 1999). Friendship quality is highly related to self‐evaluation (Voss,

Markiewicz, & Doyle, 1999). Yet the protection of one’s own self‐esteem can have both positive and negative effects in the realm of friendships (cf., also Bagwell, Bend‐ er, Andreassi, Kinoshita, Montarello, & Muller, 2005).

Non‐kin Ties – Reciprocal Cooperation. The range of personal, non‐kin relation‐ ships includes social affiliations with neighbors, colleagues, or people known only through a third party. They obviously represent a heterogeneous group, but all lack a genetic connectedness. A major aspect of this type of relationships is a mutual give‐ Theoretical Considerations 34 and‐take between relationship actors. With neighbors this mutuality may be based on helping out occasionally as needed. Among colleagues it may be the sharing of important information or support in new tasks. Even though it is not necessarily the degree of closeness that distinguishes exchange and communal relationships, ex‐ change relationships are heavily based on “give‐and‐take” structures that are needed for the functioning of modern societies (Clark & Mills, 1993; Clark, Mills, Reis, &

Rusbult, 2004; Clark, Oullette, Powell, & Milberg, 1987).

Linking Relationship Regulation and Relational Self‐Esteem. The discussion of rela‐ tionship taxonomy emphasizes the complexity and diversity of social relationships.

The framework of lifelong relationship regulation is taken as theoretical outline for further explorations of the complex human social environment. The assessment and examination of ego‐centered networks and social relationships constitute a basis for intertwined social ties and self‐evaluation characteristics. Different relationship types appear to be more or less likely to affect self‐valuation. Relevance and esteem of rela‐ tionships seems to change by relationship characteristics. Relational self‐evaluation is expected to be a function of relationship type, degree of genetic relatedness and rela‐ tionship quality. Based on the theoretical framework of life‐long relationship regula‐ tion, the focus is on the two aspects of perceived closeness and reciprocity. Close and reciprocal relationships increase the relational self‐esteem. In contrast, imbalance and particularly perceptions of under‐benefit indicate a threat on people’s positive self‐ evaluation. Thus relationship specific examinations of relational self‐esteem may il‐ luminate the differentiation across network partners and conditions. Differential ef‐ fects of relationship type are expected as a function of age, since the meaning of rela‐ tionship partners is known to change throughout the life span (Antonucci & Akiya‐ ma, 1995; Carstensen, 1992). For example, characteristics of kinship orientation in old age should be also visible in high relational relevance and esteem for kin.

In order to deepen the understanding of working mechanisms, specific envi‐ ronmental and relationship contexts are selected and addressed more closely. Percep‐ tions of emotional closeness and reciprocity, as major relational mechanisms, will Theoretical Considerations 35 serve as predictors for the differential relational value. Two distinctive relationship types have been chosen as specific contextual conditions: adult child‐parent ties and spousal relationships. The following sections discuss existing theoretical outlines and sort out relevant empirical knowledge on effects of the family of origin, as well as the particular circumstance of motivated and involuntary childlessness in adulthood.

3.3 Adult Child‐Parent Ties: Relationship Regulation and Self‐Evaluation in the

Family of Origin

The relationship between children and parents is an intrinsic part of for the devel‐ opment of personal abilities and the self (Anderson & Hayes, 1996; Lüscher, 2005;

Roberts & Bengtson, 1993). Evolutionary psychology underpins the importance of children to their parents as vehicles for genetic survival of the lineage (Buss, 2004).

Not confronted with the problem of paternity uncertainty of their children, mothers in particular show better memory of kin names (Buss, 2004), and higher levels of kin‐ ship orientation (Neyer & Lang, 2003). The quality of the mother‐daughter tie is noted as particularly important for the wider kin network (Fingerman, 2003). For that reason, women are also called ʺkin keepersʺ. Based on theoretical attachment con‐ cepts, also highlights the importance of the mother‐child relationship. Current research on adult children and their parents documents the im‐ portance of the relationship and high intergenerational contact frequency, regardless of gender (Bucx, van Wel, Knijn, & Hagendoorn, 2008).

3.3.1 Closeness and Reciprocity in Adult Child‐Parent Relationships

High‐level intimacy and closeness are major characteristics of the parent‐child relationship (e.g., Bengtson, 2001; Doherty & Feeney, 2004; Fingerman, 2003; Schwarz et al., 2005). Fingerman (2003) for instance reports that relationships between elderly mothers and adult daughters ʺtend to be tighter, the intimacy greater, the interaction more frequent and of a more emotional qualityʺ (p. 37; cf., also Rossi & Rossi, 1990). Theoretical Considerations 36

However, evaluations of relationships differ between parents and children. Parents perceive relationships to their children as special and significant throughout the life span. In contrast, adolescent and young adult children replace the intimate bond by new social relationships, such as romantic partners or close friends (Doherty & Fee‐ ney, 2004). Nevertheless, parents remain significant and serve as a secure base, with attachment to mothers described as primary for adult children’s well‐being (Doherty

& Feeney, 2004; Umberson, 1992).

Theoretical postulations about attachment suggest that the relationship cha‐ racteristics of stability and intimacy will serve as the foundation for , ca‐ regiving, and altruism between parents and children (Mikulincer & Shaver, 2005;

Mikulincer et al., 2005). Translated to the framework of lifelong relationship regula‐ tion, closeness may serve as a kind of protection against the violation of the reciproc‐ ity rule. More precisely, people in all ages prefer balanced relationships. Imbalance usually leads to a relationship break‐up, with the exception of kinship ties (Ikkink & van Tilburg, 1998, 1999). Considerations of give‐and‐take tend to shift in direction between adult children and elderly parents. Even though the major flow of support in families is expected to be downward throughout (Buss, 2004), elderly parents tend to need support and help from their children at some point in their later life (Rossi &

Rossi, 1990). This is particularly important in the context of modern demographic change and greater longevity, or a longer shared life time. Not only do these factors change the overall composition of families, but they also alter structures of mutual interdependence and support (Bengtson, 2001). In the lifetime perspective, all gener‐ ations are providers and receivers at some point in their lives (Lewinter, 2003; Mar‐ coen, 2005). It is important to be not only (or mainly) a recipient of help, but to perce‐ ive oneself as able to actively provide support in the exchange relation with the adult child at all times (Lowenstein, Katz, & Gur‐Yaish, 2007).

From the adult childʹs perspective on the parental relationship, a history of high affection is associated with more support received from parents, but does not affect the support given to the older generation (Parrot & Bengtson, 1999). On the Theoretical Considerations 37 other hand, unbalanced exchange patterns in later life are reported to be more likely if the past parent‐child relationship was lower in affection. In general, high quality relationships seem to enhance positive reporting (Mandemakers & Dykstra,

2008), although the authors found a general tendency in both generations to over‐ report the support given and under‐report the support received. Analysing data from three generations, Lin (2004) explores the interconnectedness between giving and receiving among generations. Although support provided to elderly parents and support to young children are not usually directly competing demands, the amount of help that adult children receive from their parents is a valid predictor of what they give to their own children (Lin, 2008). Ackerman and Kenrick (2008) raise an addi‐ tional aspect by highlighting the possibility of and dynamics behind help refusal.

Despite the anticipation of mutual help and prosocial benefits for kin, growing inde‐ pendence of young adult children and perceptions of conflicting motivations with regard to familial care versus new life challenges may result in a tendency toward greater distance and increasing help refusal.

3.3.2 Self‐Esteem as Regulatory Relationship Outcome

Affectionate and reciprocal relationships between children and parents are linked to a wide array of positive or protective effects for both generations across the life span (e.g., Datta, Marcoen, & Poortinga, 2005; Lowenstein et al., 2007; Roberts &

Bengtson, 1993; Schwarz & Trommsdorff, 2005a; Schwarz et al., 2005; Umberson,

1992). For example, the high importance and steady maintenance of the adult daugh‐ ter‐mother relationship leads to a perceived need to regulate oneʹs own feelings and to shape a relationship context of positive emotions (Fingerman, 1998). In that con‐ text interdependence between relationship partners may support overall tolerance and acceptance. Mature adult child‐parent relations should be characterized by a balance of closeness and criticism, accepting the other person and accompanying feelings (Shulman, Cohen, Feldman, & Mahler, 2006; Fingerman, 1998). Nevertheless, Theoretical Considerations 38 negative tensions can not be neglected if existent, since they are a salient feature for the psychological functioning of both relationship partners (Umberson, 1992).

Relationship partners mutually affect each other’s self‐evaluation, given the degree of closeness and interdependent structure of the tie. Fingerman (2003) points out that mothers influence the way their daughters think about themselves at all ages

(cf., also Anderson & Hayes, 1996). Empirical results correlate secure attachment with more positive self‐evaluations and higher degrees of self‐esteem (Bartholomew

& Horowitz, 1991; Luke, Maio, & Carnelley, 2004). In addition to attachment charac‐ teristics, support structures are positively associated with the childʹs later self‐esteem

(Luke et al., 2004).

Longitudinal data has compared self‐esteem baseline effects with a 14‐year fol‐ low up measure (Roberts & Bengtson, 1993). Results showed a decrease over time in the influence of positive parent‐child relations on self esteem. Nevertheless, the child’s self‐evaluation was still dependent on the parent‐child affection. Adding another wave 20 years after baseline, Roberts and Bengtson (1996) reexamined the effect of parent‐child affection on self‐esteem in late adolescence and early adult‐ hood. The anticipated effect of marital or parental status of children was found only to a small extend (Roberts & Bengtson, 1993). In adulthood similar effects were ob‐ served. Self‐esteem of the daughter has been found to be associated with the quality of the parent‐child relationship (Welsh & Stewart, 1995). Higher self‐esteem is linked to higher relationship quality, which is associated with the daughterʹs well‐being.

Findings indicate a robust and continuous effect of parents on their adult childrenʹs self‐evaluation. Yet, such effects appear to depend in part on contextual variables such as the family background.

3.3.3 Effects of Contextual Characteristics: Family Constancy

The stability of family structure was shown to affect childhood development

(Belsky, Steinberg, & Draper, 1991a, 1991b; Ellis, 2004; Ellis & Garber, 2000; Grund‐ mann, 1996, Neberich et al., in press). Two aspects have a particular impact: early Theoretical Considerations 39 relationship instability and absence of the biological father. They not only affect pu‐ bertal timing in daughters, but also shape the basic understanding of the constancy of relationships, resources, and contexts. Taking advantage of a twin design,

D’Onofrio and colleagues (D’Onofrio et al., 2007) provide data that support this long‐ term effect of instability in the family of origin by showing an association between divorce of parents during childhood and the child’s marital or partnership instability later in life. Thus low family constancy is expected to affect social network and rela‐ tionship structures alike. It may shape the quality (i.e. perception of closeness and reciprocity) of the child‐parent tie in adulthood. Moreover, experiences of instability may also lead to less stable or less traditional family structures (i.e., childlessness, divorce) in adulthood.

Family constancy appears to affect the contact frequency of child‐parent rela‐ tionships. Contact occurs less often if parents were divorced (Bucx et al., 2008). For example, when fathers are living alone or with a new woman there is less contact with their children. Umberson (1992) finds differences as well. In his study divorced parents report less contact and higher parental dissatisfaction. Thus perceptions of closeness and satisfaction also show a correlation with family constancy. Widow‐ hood is often associated with closer relationships between daughters and mothers

(Fingerman, 2003). Effects of divorce are, however, less consistent. This may be due in particular to differences in timing of the parental split up. The comparison of di‐ vorced and non‐divorced mothers showed no differences in perceived relationship quality with their adult children in Shulman’s study (Shulman et al., 2006). In con‐ trast, Fingerman (2003) reports closer relationships between elderly divorced moth‐ ers and their daughters. Lin (2008) addressed the association between support given and divorce in the family of origin. She found that the current partnership status of fathers but not the timing of divorce are related to support given. Adult children provide less support to divorced compared to widowed fathers. This difference does not hold true for the support given to divorced or widowed mothers. Thus, in vari‐ ous ways low family constancy appears to influence perceptions of closeness and Theoretical Considerations 40

(implicit) norms of reciprocity in relationships between adult children and their par‐ ents.

Like the background of family constancy, the current family situation of the adult child influences broader family relations. For example, young adult children reduce contact with parents when entering into a partnership or marriage (Bucx, et al., 2008). Contact shows higher frequency again when pre‐school offspring are present. This difference is particularly obvious if compared to adult children with no offspring. Umberson (1992) found no contact differences based on different marital situations of children. However, divorced children perceived less reciprocity in their parental relationships, particularly to fathers. They were also less affected by rela‐ tionship quality with parents, indicating higher levels of independence.

To summarize, research on adult children and their parents reveals an inti‐ mate and supportive relationship, reflecting the specific role of the tie across the life span. Characteristics emphasize that the relationship is continuously associated with the self‐evaluation of both actors. Patterns of perceived reciprocity and closeness, however, are moderated by such contextual structures as family constancy or current family situation. Findings suggest that unstable family structures and less traditional family characteristics of adult children negatively affect relationship perceptions.

Moreover, interaction may be substantially lower between relationship partners with a background of low family constancy. In this case, children may be less dependent on their parental relationship.

3.4 Childless Couples in Middle Adulthood: Relationship Regulation and Self‐

Evaluation in Different Family Situations

Fertility is a core subject of humankind – considered and praised in religion, art, and literature of virtually all cultures for all times(Nave‐Herz, 1988; Shamana & DeCher‐ ny, 1995). Moreover, motherhood has been a strong feminine gender de‐ scribed as a fulfilling and desirable social role for all women (Gillespie, 2001, 2003; Theoretical Considerations 41

Hird & Abshoff, 2000; Houghton & Houghton, 1987; Letherby, 2002a). In this context, current demographic transformations have alerted public opinions and politicians to the issue of a growing number of childless individuals.

Cultural or religious spaces, such as convents and monasteries, have always existed for childless women and men. However, in the past, marriage meant a com‐ mitment to parenthood. Childlessness was involuntary and led to divorce, or even worse, to ostracism. In general, this was true only for women, who were typically blamed for childlessness (e.g. Nave‐Herz, 2004). Within the last few decades though, a significant transition in reproductive behavior has taken place at least in most in‐ dustrialized countries (Townsend, 2003). More women stay childless or have fewer children later in life. Through the introduction of modern contraceptive methods in the 1960s and 1970s women obtained more control over the decision, timing, and cir‐ cumstances of motherhood (Gillespie, 2001; Hakim, 2003; Townsend, 2003). For ex‐ ample, the use of contraceptives allows heterosexual couples to postpone parent‐ hood. This pathway is frequently followed in order to achieve occupational success and financial security before starting a family (von der Lippe & Bernardi, 2006). Po‐ tentially, these individuals or couples may remain permanently childless. This either happens by choice if they become accustomed to or prefer a childfree way of living.

Alternatively people may positively decide to have a child but experience infertility.

Official statistics provide evidence of a considerable increase in childlessness among German women (Carl et al., 2000; Statistisches Bundesamt, 2007a, 2007b). This increase differs in eastern and western parts of Germany (Figure 3). Childlessness is less frequent in East German women. The statistics of younger birth cohorts are still expected to change because of the modern delay in first motherhood, particularly for highly educated women. There are two major concerns about the extent to which this data represents the full picture. First, due to official research methods, German statis‐ tics on rates of fertility and childlessness are exclusively based on the motherhood data, that is, on women only (Schmitt & Winkelmann, 2005; Statistisches Bundesamt,

2007a). Rates of childlessness in men are virtually disregarded. Schmitt and Winkel‐ Theoretical Considerations 42 mann (2005) showed childlessness to be even more prevalent in men (25%) than in women (20%) with respect to birth cohorts of 1950 to 1960. Second, most reported statistics fail to differentiate among various types of childlessness. Thus, it is un

Proportion of Childlessness in Women, Germany 2006 60 Total 50 Western parts 40 Easter n parts Percent

in 30

20 Proportion 10

0 65‐75 55‐64 45‐54 35‐44 25‐34 (1931‐1941) (1942‐1951) (1952‐1961) (1962‐1971) (1972‐1981)

Age (Year of Birth)

Figure 3. Proportion of Childlessness in Women in Germany (Table is based on Data of the Statistisches Bundesamt [Federal Agency of Statistics], 2007b) known what proportions of these statistics refer to postponement of parenthood, to voluntary childlessness, or to involuntary childlessness (Duschek & Wirth, 2005; Sta‐ tistisches Bundesamt, 2006). Statistics on infertility assume that in Germany about 3 to 9 percent of couples experience infertility problems. This pertains to couples, who did not conceive within twelve months despite a manifest desire to have a child and regular unprotected (Stöbel‐Richter & Brähler, 2006). Despite con‐ tinuous medical improvements, there is an increase of infertility problems. One ex‐ planation is the postponement of parenthood. Gnoth and his colleagues (2003) showed that the probability of conception decreases in all couples with advancing female age. Additional suggestions include characteristics of the environment and life style, such as junk food or drug consumption, growing numbers of sexual part‐ ners, and increasing numbers of genital infections or diseases. Theoretical Considerations 43

Most public and scientific discussions focus on childless women. The sketched portrayal contributes to the continued stigmatization of childless women (Letherby,

2002a, 2002b; Somers, 1993) and disregards additional characteristics of the situation such as the role of men or background. Childlessness should be recognized as a mul‐ ti‐faceted phenomenon that refers to a variety of life situations, circumstances, and social contexts. The next section presents a selected number of predictors or explana‐ tory frameworks to underlay the theoretical point of view on the phenomenon.

3.4.1 Predictors and Differentiation of Fertility Behaviors

Human reproduction is a subject in various scientific disciplines. The majority of ex‐ planations offered for changes in fertility behavior, particularly the decreasing birth rates, result from the relationship between changing societal values and fertility rate.

Such explanations often neglect the psychological relationship mechanisms that in‐ fluence fertility decisions.

Gloger‐Tippelt and her colleagues (Gloger‐Tippelt, Grimmig, & Gomille, 1993) reviewed the psychological literature on the desire for a child in the early 1990s. The evaluation revealed the existence of a major distinction between structural and psy‐ chological factors. Psychological features maybe subdivided into short‐ versus long‐ term aspects (Gloger‐Tippelt et al., 1993; also Mitchell & Gray, 2007). Long‐term as‐ pects relate to relatively stable factors, such as personality, individual , the value placed on children, or the family of origin. These factors combine with those that fluctuate more readily, such as current financial, occupational, and overall life circumstances. Gloger‐Tippelt and her colleagues (Gloger‐Tippelt et al., 1993) also found partnership and dyadic development to be important considerations in the research of fertility behavior. In sum, the authors view behavioral and cognitive va‐ riables to be of higher significance in the explanation of fertility behavior than demo‐ graphic and social‐structural factors. They suggest additional empirical investiga‐ tions, particularly on the relative levels of explanatory power of the two categories. Theoretical Considerations 44

Motivational Perspective. Using the argument for individual agency, most psy‐ chological approaches consider fertility to be rational choice. It is a conscious deci‐ sion for or against children (Rosenstiel, Spiess, Stengel, & Nerdinger, 1984; Hakim,

2003) driven by individual values and motives. Consequently, the decline in fertility has been correlated with changing values and motives, particularly the significance of children for their parents. One frequently stated theoretical consideration is the value‐of‐children approach (Hoffman & Hoffman, 1973). This approach uses the val‐ ue of children versus alternative sources of value, plus the perception of costs, bar‐ riers, and facilitators for cost/benefit analysis of fertility motivation. Hoffman and

Hoffman (1973) consider the value of children to be a pivotal moderator in fertility decisions and behaviors. The nine proposed concrete values that children are ex‐ pected to provide for their parents are empirically reduced to three main dimensions of economic benefits, psychologic‐emotional values, and social‐normative values

(Borchardt & Stöbel‐Richter, 2004). This combination of individual components and structural circumstances in the assessment of fertility motivations has been widely used to explore changes and differences in the valuation of children (e.g., Brähler,

Stöbel‐Richter, & Schumacher, 2001; Callan, 1982; Dyer, 2007; Sam, Peltzer, & Mayer,

2005; Schneewind et al., 1994). In addition, it is utilized to extend and elaborate the theory to a rational choice model of intergenerational relations (Nauck, 1989, 2001).

Value‐of‐children approaches often focus on descriptive analyses of individual and structural characteristics. This may be helpful in the differentiation of specific types of childlessness. For example, there should be substantial differences in values of mo‐ tivated and involuntary childless individuals. Yet, they ignore the relational contexts that affect individuals as well.

Developmental Perspective. Recent developmental theories across adulthood have emphasized the contribution of self‐regulative processes to illustrate the course, direction and outcomes of developmental change (Brandtstädter, 1999, 2001). While intentional self‐development is conditional on cultural and historical contexts, indi‐ viduals are reflecting, evaluating and investing their own actual and intended devel‐ Theoretical Considerations 45 opment. Embedding developmental regulation into an action‐phase model, Heck‐ hausen and colleagues (Heckhausen, Wrosch & Fleeson, 2001) focus on the develop‐ mental and regulative aspects of childlessness across time. Since the introduction of contraceptives, individuals expect to decide the circumstances and the timing of pa‐ renthood as part of their individual life plans (Hakim, 2003; Kowalcek, 1998). Yet, biological limitations on fertility, particularly in women, lead to a clear psychological regulative issue: the biological deadline in childbearing (Heckhausen et al., 2001).

Regulative strategies of striving for primary and secondary control are shown to vary by age and motherhood status of women. Empirical research points to differences in strategies of developmental regulation as a function of the circumstances of timing parenthood (Lamprecht, Wagner & Lang, 2008). Childless couples who felt high ur‐ gency in becoming parents showed higher levels of goal engagement and spent more time and effort in goal achievement behaviors than did childless couples who were postponing parenthood. Further to the regulative strategies of childless men and possible dyadic interactions of the couple relationship is needed.

To sum up, the literature reveals a high complexity concerning fertility moti‐ vations and decisions. Childlessness is habitually perceived as maladaptive, because fertility and reproduction are basic human behaviors, necessary for the survival of the species. Thus researchers have made an effort, and continue to do so, to under‐ stand the developmental trajectories of the phenomenon. One consequence is that there is a greater need to distinguish between two forms of childlessness, that is, mo‐ tivated and involuntary childlessness.

3.4.2 Motivated and Involuntary Childlessness: The Two Observed Conditions

Despite an existing body of research, the situation of childless individuals is not well understood. The connection between relational contexts and self‐evaluative characteristics offers a new perspective on the growing phenomenon. The following section has two goals: first, to present the current understanding of motivated and involuntary childlessness and second, to reveal and discuss the existing gap in know‐ Theoretical Considerations 46 ledge. It should be noted that for most characteristics conflicting results exist. Major causes for this are the frequently small samples and the predominance of studies with exclusively female participants. Therefore, the following results indicate the female side of childlessness.

Many research studies have failed to differentiate between two essential types of childlessness: motivated and involuntary. To draw a precise line between the moti‐ vated childless and others, such as the involuntary or temporary childless, House‐ knecht (1979) summarized six criteria. Preconditions for motivated childlessness are

(1) no existing biological child, (2) no child expected in the future, (3) finality and (4) certainty in the decision, (5) no medical , and (6) not living with non‐ biological children. In short, it refers to the deliberate and voluntary decision to not have a child, even though it is biologically possible.

Research on motivated childlessness concentrates on white, well educated women with high income, frequently good professional occupations, and limited or insignificant religious affiliations (Auferkorte‐Michaelis, Metz‐Göckel, Wergen, &

Klein, 2006; Burman & de Anda, 1985; Callan, 1983; Carl et al., 2000; Gillespie, 2003;

Park, 2005; Somers, 1993). German studies show that childlessness is prevalent with‐ in all demographic structures for women, but for men show a higher frequency of occurrence in those with low income (Carl, 2002; Carl et al., 2000; Tölke & Diewald,

2003). Additional assessments give specific attention to values, goals, and structural characteristics of family of origin (e.g. Carl et al, 2000; Gillespie, 2003; Miller, 1992;

Park, 2005). Table 3.1 summarizes some major findings of the last decades. Three as‐ pects seem particularly important: first, motivated childless individuals are less tra‐ ditional. They invest in different (more material) life domains than parenthood and family formation. Second, structure and context of the family of origin are frequently of low stability. Values shared with parents emphasize freedom and independence.

Third, motivated childless partnerships are of high intimacy and affection. Addition‐ ally, mutual exchange and economic attributes play a pivotal role in spousal ties. Theoretical Considerations 47

Facets of social network or relationship regulation characteristics have not yet been well researched. Only a few studies have addressed these aspects; they suggest tendencies of stigmatization or segregation. This line of argument suggests that mo‐ tivated childless individuals are generally less oriented towards reproduction and family formation and more focused on material or other rewards, which may also be visible in their social networks and relationship regulation strategies. The re‐ lationship to older parents may be one tie that reflects differential characteristics of lower intimacy and less interdependency. In contrast, romantic partnerships of moti vated childless individuals are expected to be close and interdependent. Perceptions of reciprocity are likely to play a pivotal role in dyadic regulations.

Although the external situation appears to be the same, circumstances of invo‐ luntary childlessness differ from those of motivated childlessness in many ways. Invo‐ luntary childlessness is defined as strong desire for a child and a failure to conceive, despite regular and unprotected sexual intercourse for at least two years (Voß et al.,

1994). Additional criteria in this context are, again, that people do not have a biologi‐ cal child at that time or live with a non biological (step or adoptive) child. Empirical investigations only partially meet these criteria.

Studies on involuntarily childlessness have relied almost exclusively on Cau‐ casian, heterosexual, married, upper‐middle class women who participated in fertili‐ ty treatments (Lorber & Bandlamudi, 1993). In a German study, using a sample of about 1200 women, Helfferich and Küppers‐Chinnow (1996) described a subgroup that underwent medical treatment (N = 136). The authors found that the length of time the women waited before seeking medical support was age dependent. The older the woman, and the higher her educational level, the earlier she decided to do so. Moreover, results suggested that women in urban environments more readily seek assistance. It should be mentioned that approximately half of couples with fertil‐ ity problems seek no medical assistance (Henning & Strauß, 2000). There is practi‐ cally no research about this group but it is believed to consist of younger and less educated women (Helfferich & Küppers‐Chinnow, 1996). Theoretical Considerations 48

Table 3.1 Research Summary on Motivated and Involuntary Childlessness

Motivated Childless Involuntary Childless

Attitudes & Goals  Less traditional sex roles (Lamprecht et al.,  More traditional sex roles (Lamprecht et al., 2008) 2008)  Strong materialistic motivation; desire to  Motherhood as life longing, childlessness “leave something behind” (Borgenhoff associated with sense of incompleteness Mulder, 1998; Luttbeg et al., 2000; Aarsen & (Greil et al., 1988; Kotter‐Grün, 2008; Ulrich Tzipporah Altman, 2006) & Weatherall, 2000)  Investment in other life domains (Carl,  Investment into goal of parenthood (Lam‐ 2002) precht et al., 2008)  Lower anticipation of emotional stability  High expectation for children to provide through children (Carl, 2002) , social acceptance, etc. (Ul‐  Importance of freedom, independence, self‐ rich & Weatherall, 2000; Voß et al., 1994) realization (Callan, 1987; Carl, 2002; Gilles‐ pie, 2003; Park, 2005; Somers, 1993)

Family of Origin  High frequency of single parent family (Schneewind et al., 1996)  Experience of low family constancy (Belsky et al., 1991a; Chasiotis, 1999; Ellis, 2004; Ellis et al., 1999; Vigil & Geary, 2006)  Absence of role model in family (Carl, 2002; Carl et al., 2000; Park, 2005)  Parents value independence and success (Carl et al., 2000)

Partner &  High satisfaction, but lower than involunta‐  More satisfied than motivated childless ry childless (Calan, 1987), sometimes higher women and mothers (Callan, 1987) Partnership than mothers (Burman & De Anda, 1986)  More loving relationships (Callan, 1987) and  High level of intimacy, affection, consensus, higher dyadic satisfaction (Sabatelli et al., & cohesion (Burman & De Anda, 1986; Gil‐ 1988) lespie, 2003; Schneewind et al., 1996)  Frustration and lack of  More time spent with husbands, more (Greil et al., 1988) exchange (Callan, 1987)  Decline in sexual relationship satisfaction  High importance of educational and eco‐ (Sabatelli et al., 1988; Houghton & Hough‐ nomic attributes in partner (Callan, 1983) ton, 1987)

Well‐Being  High satisfaction with amount of freedom  High satisfaction with amount of freedom and flexibility, compared to mothers (Cal‐ and flexibility, compared to mothers (Cal‐ lan, 1987) lan, 1987)  High sense of well‐being (equal to mothers)  Lower sense of well‐being, less satisfied (Callan, 1987) with life (Callan, 1987; Greil et al., 1988)

Social Context  Social context without effect on decision to  Possibility 1: Infertility as neglected topic in stay childfree (Carl, 2002; Lippe, 2002), with friendships (Bernardi, 2003; Sabatelli et al., effect in low‐income women (Wilson & 1988; Voß et al., 1994) Koo, 2006)  Possibility 2: Strong support by family and  Less positive reports about support from friends in difficult situation (Sabatelli et al., friends and family (Callan, 1987) 1988; Voß et al., 1994)  Perception of stigmatization by friends and  Feelings of alienation and stigmatization family (Somers, 1993) from friends & family (Houghton & Hough‐ ton, 1987; Letherby, 2002a)  Explicit avoidance of certain social contexts (Exley & Letherby, 2001; Greil et al., 1988; Sabatelli et al., 1988) Theoretical Considerations 49

Table 3.1 summarizes research that goes beyond the description of the medical reasons (for a review, see Gnoth, 2004; Gnoth et al., 2003; Wischmann, 2006), the emotional stressfulness, and applied coping strategies in the situation of infertility

(Greil et al., 1988; Henning & Strauß, 2000; Houghton & Houghton, 1987; Jordan &

Revenson, 1999; Lorber & Bandlamudi, 1993; Matthews & Matthews, 1986). First, this research describes involuntarily childless individuals as traditional in values and goals. They have a strong desire for and invest heavily in the goal of parenthood.

Second, despite the occurrence of communication problems partnerships are de‐ scribed as particularly loving and intimate. Third, childless women are less satisfied with their lives and feel less fulfilled. With regard to effects on their social participa‐ tion two possible coping strategies are described. Involuntarily childless individuals either conceal the fact of infertility, or if talking about it they receive strong support from family and friends. Nevertheless, the childless often felt alienated from other parents. They explicitly described situations they avoided, such as female‐ or family‐ centered gatherings. Results in general underpin both the importance of the social context and the difficult situation between burden on the one side and stigma, myths, or taboos on the other (Letherby, 2002a). Structural and regulative characteristics of social networks will likely mirror this complex situation, indicating similarities to motivated childless individuals and parents alike. In general, family ties are likely to be very important. Spousal relationships may be intimate and interdependent.

Implications for the current empirical study that arise out of these earlier con‐ siderations of childlessness in adulthood are fourfold. First, there is a need to diffe‐ rentiate among several types of childlessness. Available research on motivated and involuntarily childless individuals validates the diversity of those two types and em‐ phasizes the importance of distinguishing between them. Second, empirical studies should include both partners in a childless couple. This implication focuses on the frequent concentration on women – or neglect of men – in the context of empirical considerations. There is a need to overcome this one‐sided point of view to get an idea of the whole picture, i.e. the dyadic interdependencies between romantic part‐ Theoretical Considerations 50 ners. Third, examinations of social relationship characteristics and structural contexts are required. Attention to the social context that underlies personal relationships is essential in the study of fertility behaviors. Social relationships frame and shape our lives. Across the entire life span, long‐lasting (e.g. mothers and fathers) as well as intermittent (e.g. colleagues, neighbors or peers) social affiliations affect our thoughts and emotions, decisions and behaviors. Therefore a closer look at social relationship regulation characteristics of childless people is important for a better understanding of the phenomenon. Fourth, implications of documented relational patterns have to be analyzed. In order to illustrate the functionality of social networks and regulative mechanisms self‐evaluative outcomes, including global, domain‐specific, and rela‐ tional self‐esteem, will be taken into account.

Existing research generally suggests that motivated childlessness is associated with more self‐centeredness and higher self‐esteem than parenthood (Burman & de

Anda, 1985). Satisfaction with ones current life situation, however, rather suggests a general positive evaluation of the self. Thus, no differences are expected between global and domain‐specific self‐esteem measures in a given family type. The theoret‐ ical outline of relational self‐esteem, in contrast, proposes stronger effects of impor‐ tant relationship partners on patterns of self‐evaluation. Thus, high self‐esteem in motivated childless individuals may suggest two particular relational characteristics.

First, their social networks should be less likely to contain family members and rela‐ tionship partners that are parents themselves. Second, self‐evaluations should be af‐ fected to a lesser degree by relational feedback of kin and parts of elective kin. In con‐ trast, the situation of involuntary childlessness may increase the sensitivity to social feedback of the family. Depending on the openness about their infertility relational feedback may be more or less difficult to handle. A positive adjustment to the situa‐ tion however has been found to increase self‐esteem (Daniluk & Tench, 2007). Theoretical Considerations 51

3.5 Summary

The framework of lifelong relationship regulation suggests the occurrence of differential network structures and individual characteristics of closeness regulation and reciprocity monitoring. Given patterns are expected to be efficient for specific environments and demands. The empirical investigation emphasizes the active na‐ ture of individuals in forming and regulating social bonds. Thus, social network size and structures, patterns of kinship orientation and effects of (non‐) reciprocity on relationship perceptions may vary among individuals of different age groups and family contexts. It is further suggested that relational self‐esteem contribute to an understanding of the functionality of relationship characteristics.

Dyadic analyses are conducted to complement intra‐ and inter‐individual cha‐ racteristics. Two meaningful relationship contexts of middle adulthood are selected: adult child‐parent ties and romantic partnerships. Previous findings and the signific‐ ance of relationship partners suggest effects on global characteristics of self‐esteem.

Differential patterns are expected for relationship interdependencies based on con‐ textual variables. In intergenerational relations facets of low family constancy, such as absence of father or divorce of parents, might lead to higher aspirations of inde‐ pendence with regard to social affiliation partners and to a tendency toward self‐ reliance in relationship regulation. High degrees of closeness and interdependence are perceived as a risk for people (cf., Bowlby, 1983; Holmes, 2002; Murray et al.,

2000), one they may take only when trusting the other’s basic responsiveness to their needs. It is suggested that low constancy experiences in the family of origin will de‐ crease that trust. In addition, Feeney (2007) proposed that accepting interdependence is necessary to perceive autonomy in a relationship. Trust will lead to interdepen‐ dence which is a necessary condition to perceive oneself as independent individual within relationships. Here, previous experiences of low family constancy may affect the relationship not only to parents but also to other partners such as the spouse.

High trust may also lead to higher interdependency between romantic partners. Summary and Research Questions 52

4 SELF‐EVALUATION ACROSS RELATIONAL CONTEXTS: GENERAL SUMMARY AND

RESEARCH QUESTIONS

“An individual’s self‐evaluation is not just an isolated, introspective construction; rather, self‐evaluations are feelings that are intricately embedded in social rela‐ tionships and social experience.” Srivasta & Beer, 2005, p. 966

The present research addresses three questions: First, what is the shape of the deve‐ lopmental trajectory of self‐esteem across the life span? Prior findings on differential change in global self‐esteem are extended to domain‐specific measures of perfor‐ mance and social self‐esteem. Analyses are used to examine age related and family situation related mean‐level differences. Second, what are the differences between motivated childless and involuntary childless individuals with respect to personal networks and relationship regulation strategies? Empirical analyses are applied to this question with the intent of replicating prior findings regarding the adaptation of social networks on age and context specific challenges. Then a detailed examination of motivated and involuntary childless individuals elaborates on this very specific contextual structure in middle adulthood. Third, do self‐evaluative mechanisms un‐ derlie the regulation of relationships? Different self‐esteem measures are used to ex‐ plore self‐evaluative patterns across age groups and specific relationship contexts.

Investigations focus on global and domain‐specific measures as well as a new con‐ struct of relationship functioning, that is, relational self‐esteem.

Self‐evaluation is a necessary condition for stimulating self‐regulative processes of any kind. Based on the work of Baumeister and Leary (1995), several research projects have addressed the topic of social belonging with regard to the process of self‐evaluation. Indeed, empirical support exists for a link between self‐ esteem and social interaction, in light of social exclusion and positive and/or stable interaction partners (e.g., Denissen et al., 2008; Leary et al., 1998; Leary et al., 1995). In this context, self‐esteem has been described as the product of a monitoring system Summary and Research Questions 53 that judges ones relational value or relational importance (Leary, 2007). Despite the supposition that interpersonal contexts require specific monitoring mechanisms to efficiently judge different relational situations (e.g., Hill & Buss, 2006; Kirkpatrick &

Ellis, 2006), theoretical outlines and empirical research remained at global, general levels and focused on experimental, often artificial (relationship) situations. However a human being acts within dynamic interpersonal contexts and with constant envi‐ ronmental challenges, driven by individual goals and feelings, and strives simulta‐ neously for relational belonging and personal importance. The complexity of this evaluation process emphasizes the inadequacy of a solely global or system specific gauge for monitoring relational value. Instead, it is hypothesized that a flexible and adaptive mechanism, termed relational self‐esteem, is required to monitor the func‐ tionality of each relationship partner in each interpersonal context.

The current investigation addresses the functionality of social relationships in different age groups and in the situation of motivated and involuntary childlessness.

Specific emphasis is put on the self‐evaluative outcome of self‐esteem. Composition and regulative strategies of ego‐centered social networks are evaluated regarding intra‐individual as well as inter‐individual variations. Focus is placed on age‐specific variations and different family settings, which are both likely to provide formative contextual structures across the life span. A first sample differentiates relational and self‐evaluational structures between five age groups. Investigations focus on mean level variations across the life span with respect to different self‐esteem measures, social network structures, and relationship regulation strategies.

A second sample selected middle aged motivated and involuntary childless individuals. Their self‐esteem and network characteristics are compared with two types of parents that differ with respect to stability and traditionalism of structures: traditional and blended families. Beside network structures two specific types of so‐ cial affiliations are examined: the adult child‐parent tie and the romantic partner rela‐ tionship. With respect to the first dyad, the constancy or nature of the family of birth will be especially considered. Descriptions of intergenerational relationships empha‐ Summary and Research Questions 54 size a stable perception of closeness and intimacy between adult child and parent throughout the life time (e.g., Fingerman, 2003). Interdependency patterns may be expected to differ as family background situations vary from high to low stability.

Dynamics in romantic partnerships are expected to differ with respect to parental structures (parent vs. childless) and the extent of traditional characteristics of the family situation. For example, motivated childless women may strive for indepen‐ dence (Callan, 1987; Carl, 2002; Gillespie, 2003) but are also in need of closeness, con‐ sensus, and cohesion in their partnerships (Burman & de Anda, 1986; Gillespie, 2003,

Schneewind et al., 1996; Somers, 1993). Thus both high and low interdependency are possible structures, which will be explored by dyadic analyses on a sub‐sample of romantic partnerships.

The following research hypotheses are guided by three major aspects: First, different measures of individual self‐esteem are explored with respect to their deve‐ lopmental trajectories and contextual sensibility. Second, structural and procedural issues of social networks and regulative mechanisms are considered, and effects of such contextual variables as age or family situation on regulative patterns are pre‐ sented. Third, functionality of social relationships is addressed, indicating effects of relational self‐esteem across networks and of global self‐esteem in intergenerational and romantic partnerships. Using these major ideas, the following hypotheses will be tested:

Self‐Esteem across Life Span and Family Situations

(1) Effects of Age. Age‐associated patterns of global and domain‐specific self‐esteem measures vary across the life span. The differentiation between global and social in‐ dicators increases in late adulthood and old age. Indicators of positive functioning, such as life satisfaction and subjective health, differentially predict global and do‐ main‐specific self‐esteem indicators.

(2) Effects of Family Situation. Domain‐specific self‐esteem indicators, particularly so‐ cial self‐esteem, show higher sensitivity to differential family characteristics than global self‐esteem. Summary and Research Questions 55

Social Networks and Relationship Regulation across Life Span and Family Situations

(3) Effects of Age. Personal network structures and relationship regulation characteris‐ tics differ across age groups, showing higher proportion of kin network partners and kinship preference in older age.

(4) Effects of Family Situation. Social networks of motivated childless individuals are less oriented toward kin relationships and more focused on cooperative relation‐ ships. In contrast, involuntarily childless individuals will show structures similar to parents, being oriented more toward kin relationships and less toward non‐kin rela‐ tionships. Motivated childless individuals will demonstrate lower levels of kinship preference, compared to the involuntarily childless and parents. The differentiation in reciprocity perception between kin and non‐kin is highest in involuntary child‐ lessness and traditional families and lowest in motivated childlessness and blended families.

Functionality of Relationships – Relational and Individual Self‐Evaluation

(5 a) Effect of Relationship Quality. Self‐evaluations of perceived relational esteem in relationships are relationship and age specific. Mechanisms of closeness and reci‐ procity perceptions in social networks differentially predict the relevance of relation‐ ship partners.

(5 b) Effect of Level of Data. Relational self‐esteem as a relationship specific indicator of self‐evaluation differs from individual self‐esteem measures. Only small interde‐ pendencies occur between the two levels. Strongest effects are expected with social self‐esteem.

(6 a) Effect of Family of Origin. Relationship quality for adult children with low consis‐ tency in the family of origin is characterized by less closeness and more imbalance than that for adult children with high consistency. This difference will be less notice‐ able in the older generation. Older parents perceive intergenerational relations as more positive than adult children, regardless of family constancy. Summary and Research Questions 56

(6 b) High constancy in the family of origin is associated with stronger relational in‐ terdependence. In contrast, low family constancy increases self‐regulative strategies.

(6 c) Relationship perceptions of adult children are characterized by stability and change across a one year period. In high family constancy change is associated with relationship perceptions. In low family constancy, adult children rely on self‐ regulative strategies.

(7 a) Effect of Family Type. In general, the romantic partner is highly important across time and regardless of family type or parental status. The degree of dyadic interde‐ pendence between relationship qualities and self‐esteem will depend more on recip‐ rocity perceptions for motivated childless and blended family couples, but more on emotional closeness in spousal ties of involuntarily childlessness and traditional families. Effects differ with respect to global and domain‐specific self‐esteem. Most effects occur regarding social self‐esteem.

(7 b) Relationship perceptions and self‐esteem of romantic partners are characterized by stability and change across time. In motivated childless individuals and blended families, change is related to perceptions of reciprocity, and in involuntary childless individuals and traditional parents it varies as a function of emotional closeness.

Method 57

5 METHOD

This section will begin with a description of the two empirical investigations. The first part refers to the online study ʹRelational Self‐esteem across the Life Spanʹ4. This study was conducted at the two online portals www.gerotest.de and www.psytest.de. This section depicts the sample, the procedure, and key variables.

Second, an explanation of the design of the family study ʹPersonality and Social Rela‐ tionships across the Life Spanʹ (RELATE) is presented including depictions of the samples, procedures and key variables. Finally, the chapter aims to describe existing key statistical packages, data structures, and strategies of data analysis that have been used throughout the dissertation.

5.1 Online Study: Relational Self‐esteem across the Life Span

The first empirical investigation assessed an online sample of participants covering the entire range between early adulthood up to old age. The online study ʹMirror of the Selfʹ was conducted between June 2007 and January 2008 at online portals of two

German universities. Main goal of the study was the assessment of relationship spe‐ cific ratings of the relational relevance or valence of different types of social relation‐ ships across the lifespan.

Clicking on the study’s button, participants first read a short introductory text that not only illustrated the goal of the study but further described the procedure and specified the average duration needed to complete the questionnaire. There were no particular restrictions of participation with regard to demographic variables. Indeed, the selection of the two different online portals, www.psytest.de and www.gerotest.de, was driven by the ambition to include not only the younger but

4 The original title in the internet was “Mirror of the self“(Spiegel des Selbst). Method 58 also the older generation of internet users into the study.5 At the end of the question‐ naire, all participants received a personal feedback concerning their social and per‐ formance state self‐esteem. In average, the study took participants approximately 45 to 55 minutes, depending on the number of social network partners. Even that a higher number of individuals clicked on the online study, the following empirical results will be based only on those participants that (a) finished the study, and (b) named at least three social relationship partners.

5.1.1 Sample Descriptions of Online Study

Altogether, 557 individuals participated in the online study. As in most online sam‐ ples, the bigger part of participants were female (n = 439, 79%). Age ranges between

17 and 86 years, with an average age of 47 years (M = 47.1, SD = 19.0). With regard to the educational status, the majority of participants reported to have gained at least an

Abitur (n = 164, 29%) or even a university diploma (n = 170, 32%). Yet, only 34 per‐ cent were working full‐time and 13 percent part time. The high number of unem‐ ployment might be due to the age characteristic of the sample, since younger partici‐ pants could contain a high proportion of students and participants older than 65 years are forced to retire from workforce in Germany.

For the diversity of the sample with regard to age, the family status of partici‐ pants varied substantially, as well. About one quarter of participants reported to be single (n = 134, 24%), and nearly half of the participants were married (n = 179, 32%) or lived at least with a romantic partner (n = 113, 20 %). A much smaller proportion of people was divorced (n = 83, 15%) or widowed (n = 48, 9%). The average duration of romantic relationships, for those currently having a partner, was nearly 20 years

(M = 18.5, SD = 16.1), yet again the variety showed to be very high (range 1 to 61 years). Rather surprising numbers became obvious with regard to biological child‐

5 Comparisons between paper‐pencil and web‐based records have been shown to produce equally data regarding quality and effects, even in older age (e.g., Gosling, Vazire, Srivastava, & John, 2004) Method 59 ren. Even that the average number was rather (above) average (M = 1.6, SD = .73, vs. mean number of children per women in Germany 1.3; Statistisches Bundesamt, 2006), actually, 77 percent of participant (n = 429) reported not to have any biological child‐ ren. On one side, this may be due to the proportion of younger participants in the sample. On the other, it would be a possibility that older adults did not include their adult children, since they do not live within their households anymore. For these rea‐ sons, analyses on differences by parental status will be conducted only in several analyses, but mainly the variable will be controlled for. Result interpretation on ef‐ fects of childlessness will focus on findings of the family study.

As reported, gender distributions of the sample were imbalanced. Yet, existing demographic discrepancies may be reported. In average, men appeared to be older

(Male: M = 53.6, SD = 20.1; female: M = 45.4, SD = 18.3; t (551) = 4.22, p < .001, d = .43), and they reported to be higher educated than women (University degree: male =

42%, female = 28%, χ2 (2) = 10.29, p < .01). Furthermore, males were more likely to live with a spouse or romantic partner (Partnership: male = 62%, female = 50%, χ2 (1) =

5.35, p < .05), and existing partnerships had longer durations (Male: M = 26.5, SD =

18.9; female: M = 16.4, SD = 14.5; t (338) = 4.96, p < .001, d = .60).

5.1.2 Measures of Online Study

Centerpiece of the empirical investigation was the assessment of the ego‐centered social network. Relationship specific evaluations of closeness and reciprocity percep‐ tions were complemented by the measurement of relational self‐esteem, since the exploration and development of a more precise measure of relationship functioning was a pivotal goal of the current study. Furthermore, individual measures of self‐ esteem and a number of covariates were added to complete the online question‐ naire.Table 5.1 gives an overview on all used instruments.

Method 60

Table 5.1 Applied Instruments of the Online Study

Dimension Instrument Description Comment

Ego‐centered Name Generator “Please name all individuals you already know for Naming between 1 and 35 Network (Lang, Neyer, Wrzus & a long time and/or have frequent contact with.“ social relationship partners Wagner, 2005)

Genetic Relationship Index Differentiation between biological kin, elective 4 types Relatedness kin, and non‐kin.

Closeness Inclusion of Other in the 7‐point scale Ich Anderer Ich Anderer Ich Anderer Ich Anderer Self Scale, IOS (Aron, Aron & Smollan, Ich An‐ Ich An‐ Ich An‐ 1992) Graphic Closeness Scale, Indicating perceived close‐ GCS ness between 0 and 100. (Lang, Neyer, Wrzus & Wagner, 2005)

Reciprocity Graphic Interdepend‐ 7‐point scale ence Scale, GIS (Lang, et al., 2005) Graphic Balance Scale, 7‐point scale GBS (Lang, et al., 2005)

Relational Communal‐esteem 7‐point scale Ich Anderer Ich Anderer Ich Anderer Ich Anderer Self‐Esteem (based on IOS, Aron, Aron & Smollan, 1992) Ich An‐ Ich An‐ Ich An‐

Mutual‐esteem 7‐point scale (based on GIS, Lang, et al., 2005)

Self‐esteem Global Self‐esteem 6 Items, (Marsh & O’Neal, 1984) 7‐point scale State Self‐Esteem 14 Items, 2 Subscales (Per‐ (Heatherton & Polivy, formance, Social) 1991, Schütz, 19) 7‐point scale

Well‐Being Life Satisfaction Combination of two existing scales: 10 Items 6 items of Ferring, Filipp and Schmidt (1996) 7‐point scale 4 items of Diener, et al. (1985) Subjective Health, 12 Items SCL‐90‐R 7‐point scale (Hessel, et al., 2001)

5.1.2.1 Personal Network Assessment: The Relationship Level

The centrepiece of the current study was the assessment of ego‐centered social net‐ works. Previous studies (e.g., Antonucci, Akiyama, & Takahashi, 2004; Asendorpf &

Wilpers, 1998; Fingerman, Hay, & Birditt, 2004; Neyer & Asendorpf, 2001; Neyer &

Lang, 2003) oftentimes assessed ego‐centered social networks by using the concentric circle measure or verbal name generators (e.g., asking for all ʹcurrently importantʹ Method 61 relationships). In this context, Hollstein (2001) pointed out that the number and ver‐ balization of name generators are particularly important, because they have an influ‐ ence on the resulting network structure and size.

Name Generator. In this current research project one single name generator has been used. First, each participant generated a list of his/her social relationship part‐ ners. Participants were asked to list social relationships that include “all individuals you already know for a long time and/or have frequent contact with“. The above statement implied that the conceptualization of social relationships includes all dif‐ ferent types of social affiliations. Not only are family members and friends included, but also colleagues or neighbors. Furthermore, interaction between partners may be connoted as positive, as well as negative. With this definition, individuals (targets) were enabled to choose all others that the research team qualified as network mem‐ bers. Participants were asked to generate a minimum of 3 and a maximum of 35 rela‐ tionship partners. In sum, the 557 participants named 6399 relationship partners with an average of 11.49 social affiliations per person (SD = 7.78).

To obtain a sufficient number and a heterogeneous structure among network members, participants were initially asked “Where did you get to know this person from?” A drop down menu from the computer assisted assessment presented a selec‐ tion of possible kin and non‐kin relationship types such as mother/father, other kin, or from neighborhood. These served as a primer or recognition technique (Neyer,

1997; cf. Appendix D). Participants had to select one relationship type and subse‐ quently had to specify several characteristics of the chosen relationship partner and the relationship itself, before picking the next social affiliate.

Key information on each relationship partner included: the first name and last initial, age, sex, marital status, and whether or not the individual was a parent. Re‐ garding characteristics of the social relationship, the assessment included questions on the duration of knowing this person (7‐point scale from ʹless than 1 yr.ʹ to ʹsince birthʹ), geographical distance (6‐point scale from ʹin the same householdʹ to ʹmore than

200km awayʹ), contact frequency (5‐point scale from ʹdailyʹ to ʹonce a month or lessʹ, re‐ Method 62 verse coded), as well as perceived conflict (5‐point scale from ʹvery rarely/neverʹ to

ʹvery oftenʹ). Additionally participants were asked directly about the degree of genetic relatedness to relationship partners, differentiating between ʹbiological familyʹ, ʹstep‐ familyʹ, ʹfamily in‐lawʹ, and ʹno family memberʹ. By that, analyses considering variations with regard to relationship type and genetic relatedness were possible. Descriptive statistics of the ego‐centered network structure are shown in Table 5.2.

Table 5.2 Descriptive Statistics of Ego‐centered Social Networks Regarding the Entire Network and in Ten Relationship Categories

M SD Min Max

Overall Network 11.49 7.78 3 35 10 Relationship Types r = .5 Parents 0.82 0.85 0 2 Siblings 0.84 0.98 0 7 Children 0.17 0.61 0 4 r = .25 Close Kin 1.50 1.74 0 11 r ≤ .125 Distant Kin 0.39 0.96 0 9 r = 0 Partner 0.46 0.54 0 3 In‐/Step Family 1.28 1.95 0 13 Colleagues 1.27 1.84 0 13 Acquaintances 2.18 3.03 0 18 Old Friends 2.57 3.01 0 20

5.1.2.2 Indicators of Relationship Quality at the Relationship Level

There are two possible ways to gather information on relationship characteristics in ego‐centered social networks: first, question‐ or column‐wise, and second, relation‐ ship‐partner or row‐wise. The current study used the first strategy – question‐wise.

Thus, each relationship‐specific evaluation, i.e. perception of closeness or of reciproc‐ ity, was accomplished for all listed relationship partners at ones, before going to the next item. Prior research on ego‐centered social network data that were collected question‐wise via web showed high amounts of reliability and validity (Coromina &

Coenders, 2006). In this dissertation, the focus is on three dimensions of the ques‐ Method 63 tionnaire: emotional closeness, perceived reciprocity, and the relational self‐esteem in social relationships, which are represented by two graphical items each.

Perceived Closeness. This dimension was first assessed with the Graphic Close‐ ness Scale (Lang et al., 2005). Participants were instructed to place the grey bar at a point on the line that best indicated the perceived relationship quality to each listed interaction partner. The graphic depiction showed a line with two endpoints: ʹmeʹ

(Ich) and ʹstrangerʹ (Fremder). By moving a slidable grey bar, participants were able to illustrate the degree of perceived closeness with regard to each network member.

That is, the closer the bar was moved to the ʹmeʹ endpoint the participant perceived the relationship as ʹcloseʹ, while closer to the ʹstrangerʹ endpoint indicated a more distant relationship.

Closeness was also assessed by the Inclusion of Other in the Self Scale (Aron,

Aron, & Smollan, 1992). This item included seven pictures of two differentially over‐ lapping circles. Each circle represented each relationship partner, and the degree which the circles overlapped represented the level of perceived intimacy. Therefore, the more overlap between the two circles, the closer one felt to the relationship part‐ ner. On this 7‐point scale, participants were asked to choose the picture that best rep‐ resented each perceived relationship. The two items were z‐standardized and then summed to represent a single index of closeness. Item and index descriptives are displayed in Table 5.9. Reliability analyses indicated that the two measures were an acceptable to good indicator of perceived closeness in personal relationships (α = .76).

We used an unconditional means model via multilevel modelling to obtain the intra‐ class correlation coefficient (ICC). The ICC suggested that 21 percent of the variabil‐ ity in closeness perception was due to differences between study participants, whereas 85 percent was due to differences within the participants. Therefore, much more variation was present at the social relationship partners’ level.

Perceived Reciprocity. Two graphical items assessed the perception of reciproc‐ ity, or balance in social interactions. At first, the Graphic Interdependence Scale (Lang et al., 2005) demonstrated possible variations in perceived reciprocity or balance. Two Method 64 arrows, pointing in opposite directions onto ʹmeʹ (Ich) or ʹotherʹ (Anderer), differed regarding the type of line, e.g. dotted vs. dashed line (cf. figure presented at Table

5.7). The variation in line style specified whether the participant or the other person does more for the relationship. Therefore, the more faded the dashed line becomes, the less that partner does for the relationship. The first three graphics indicated that the participant does more, the fourth pointed out mutuality and picture five to seven demonstrated that the other does more. Participants were instructed to think in terms of help, favours, support, advice, etc., before choosing the best fitting picture to indi‐ cate perceived reciprocity for each relationship.

The second measure of reciprocity was the Graphic Balance Scale (Lang et al.,

2005). Each graphic included seven images of balanced scales and each were used to depict the possible facets of mutual give‐and‐take in social interactions (see Table 5.1 for an illustration). When a person is perceived to invest more in the relationship, the scale is ʹtippedʹ more in their direction. Again, the first three pictures indicated that the participant does more for the relationship, the fourth one illustrated a reciprocal dyadic interaction, and the last three pictures pointed out that the other person in‐ vested more.

To distinguish between different forms of perceived reciprocity, statistical analysis splits up the reciprocity measures into two dummy coded variables, where one is an indicator of perceived over‐benefit and the other variable is an indicator of perceived under‐benefit in social relationships. Balanced relationships serve as refer‐ ence category. To obtain these variables, at first the two reciprocity items were summed to a single index of reciprocity. Subsequently, indices between 1 and 3.5 were recoded into a new dummy variable to indicate under‐benefited relationships, with all other relationships (i.e., reciprocal and over‐benefited) as reference category.

In the same fashion, indices between 4.5 and 7 were recoded into a second dummy coded variable to indicate over‐benefited relationships. The ICC showed that the proportion of variability between interviewed persons in under‐benefited relation‐ Method 65 ships were 16 percent and in over‐benefited relationships were slightly smaller with

11 percent.

With regard to the relationship measurement model, in addition to the intro‐ duced key variables of perceived closeness, perceived under‐benefit, perceived over‐ benefit, and degree of genetic relatedness, two interaction variables are included in investigations of closeness prediction. The first interaction term captures the relation‐ ship between the dummy‐coded variable of under‐benefit x degree of genetic rela‐ tedness. This is an indicator for perceived under‐beneficial kin relationships, only

(i.e., if under‐benefit dummy is ʹ1ʹ and genetic relatedness is ʹ.25ʹ, the interaction term is ʹ.25ʹ, instead it is ʹ0ʹ if the relationship is perceived to be reciprocal). Similarly, the second interaction term between the dummy variable of over‐benefit x degree of ge‐ netic relatedness is an indicator for all relationships to biological kin that participants perceived to be over‐beneficial for themselves. Descriptive statistics of key variables at the relationship level are displayed in Table 5.3.

Communal‐Esteem. At first, the Inclusion of Other into the Self Scale (Aron,

Aron, & Smollan, 1992; cf., Table 5.7) graphic was shown a second time, but a differ‐ ent instruction was used this time. Participants were asked to consider the perceived or assigned esteem in the particular social relationship. In more detail, the instruction described that the circles in the graph now symbolized “how relevant or important

[is] the perceived esteem within the relationship to X”6 for the participant. The higher the degree of overlap between the two circles, the more important one described the communal‐esteem of this particular social interaction partner to be for one’s self‐ evaluation. Again, the scale contained seven pictures and participants had to choose the best fitting picture with regard to each named relationship partner. For upcoming analysis the scale has been transformed into z‐values. The ICC indicated that 28 per‐ cent of variability was due to between participant differences and 72 percent was due to within network variation.

6 The original wording in German was “wie relevant bzw. wichtig für Sie die entgegengebrachte

Wertschätzung innerhalb Ihrer Beziehung zu Person X ist“. Method 66

Table 5.3 Variables of the Relationship Level – Item Characteristics of the Online Study

Descriptive Statistics N M SD Range ICC

Relationship Level Perceived closeness 6397 0.57 0.25 0‐1 .21 Degree of genetic relatedness 6399 0.12 0.19 0‐.5 .05 Perceived under‐benefit 6397 0.39 0.49 0/1 .16 Perceived over‐benefit 6397 0.20 0.40 0/1 .11 Under‐benefit x genetic relatedness 6397 0.05 0.13 0‐.5 .06 Over‐benefit x genetic relatedness 6397 0.03 0.10 0‐.5 .06

Communal‐esteem 5853 0.00 1.00 ‐2.05‐1.34 .28 Mutual‐Esteem, higher‐esteem 5853 0.16 0.37 0/1 .18 Mutual‐Esteem, lower‐esteem 5853 0.22 0.41 0/1 .20

Note. M = Mean, SD = Standard deviation, ICC = Intraclass correlation. Perceived closeness was not z‐ standardized, but both graphical indicators were linear transformed to a range between 0 and 1 before being subsumed to the index of closeness.

Mutual‐Esteem. In addition to the relevance of the communal‐esteem, a second item focused on the perceived mutuality of esteem between relationship partners. In this context, the Graphic Interdependence Scale (Lang, et al., 2005; cf., Table 5.7) was utilized for a second time, but, as in the previous item, was framed by a different in‐ struction. The first picture, showing only one arrow with a solid line pointing from

ʹmeʹ to ʹotherʹ, now indicated that the participant regarded his/her own esteem con‐ cerning the relationship partner to be substantially higher than vice versa. In con‐ trast, the fourth picture, with two solid arrows pointing diametrical, suggested the mutual or equivalent degree of relevance of esteem with regard to both relationship partners. Again, ratings were conducted on a 7‐point graph, where the first three pic‐ tures stood for higher‐esteem of the participant, the fourth represented mutuality, and the fifth to seventh picture indicated lower‐esteem of the participant. Data han‐ dling was similar to the treatment of the reciprocity measures, leading to two dum‐ my coded variables: higher‐esteem, indicator of the first three pictures, and lower‐ esteem, indicator for the last three pictures. The mutuality picture served as contrast Method 67 category. ICCs demonstrated 18 and 20 percent of variability between participants, respectively. Descriptive statistics of relational indicators are captured in Table 5.3.

5.1.2.3 Personal Characteristics: The Individual Level

In addition to relational variables, a number of individual level covariates were in‐ cluded into the online study. First, two measures of self‐esteem were assessed, a global trait and a domain‐specific state self‐esteem, to explore different characteris‐ tics of self‐evaluative developmental trajectories. Second, two indicators of general well‐being were integrated, to examine age differential functions of the developmen‐ tal trajectories. Measures are subsequently introduced and sample specific descrip‐ tive statistics and intercorrelations are shown in Table 5.4. As previously done, all scales were linear transformed to an equal range between 0 (low) and 1 (high).

Self‐Esteem. The six‐item General Self‐Esteem scale of the Self‐Description

Questionnaire III (SDQ III; Marsh & O’Neil, 1984; example item: In general, I accept myself the way I am.) assessed global tendencies in self‐evaluation. A 7‐point agree‐ ment format rating, ranging from 1 (not at all) to 7 (very much) was utilized. The in‐ ternal consistency of the scale was satisfactory (α = .87).

State Self‐Esteem. Based on the scale of Heatherton and Polivy (1991), a German version of the state self‐esteem (Schütz, 1996) was utilized. The original scale con‐ sisted of 3 subscales, whereas the current empirical investigation included only two, the performance and the social subscale. Recent reliability analyses suggested a re‐ vised version of the original factor structure (Rudolph, Schütz, & Schröder‐Abé,

2008a; Rudolph, Schütz, & Schröder‐Abé, 2008b), which will be utilized in the follow‐ ing. In that, the performance subscale consisted of five items (example item: ʹI feel confident about my abilities.ʹ), just as the social subscale did (example item: ʹI feel self‐conscious.ʹ). Evaluations of momentary or state ratings were based on a 7‐point rating scale with a range between 1 (applies not at all) to 7 (applies totally). Reliabili‐ ty measures for both sub‐scales were sufficient to good (performance: α = .71; social:

α = .88), and Rudolph et al. (2008a; 2008b) validated the state sensitivity of the re‐ Method 68 vised scale regarding experimental and naturally occurring events. All social self‐ esteem items were recoded; thus higher social self‐esteem indicates less concern about impressions and more self‐ in social contexts.

Table 5.4 Covariates – Scale Characteristics and Intercorrelations of Online Study (N = 557)

Descriptive Statistics Intercorrelations

Global Performance Social Self‐ Life Scale k M SD  Self‐esteem Self‐esteem esteem Satisfaction

Trait Self‐esteem Global 6 0.71 0.22 .87 State Self‐esteem Performance 5 0.75 0.18 .71 .67 Social 5 0.56 0.27 .88 .58 .51

General Well‐being Life Satisfaction 10 0.64 0.21 .87 .74 .57 .44 Subjective Health 12 0.82 0.16 .83 .35 .28 .26 .35

Note. k = number of items, M = Mean, SD = Standard deviation,  = Cronbach’s Alpha. Boldfaced in‐ tercorrelations are significant at p < .05.

As indicators of General Well‐Being two facets were chosen: Life satisfaction and perceptions of subjective health. The assessment of life satisfaction is based on two already existing scales: (a) the scale of life evaluation (Ferring, Filipp & Schmidt,

1996) and (b) the life satisfaction scale (Diener, Emmons, Larsen & Griffin, 1985). Re‐ garding the first scale, six items were selected, and concerning the second scale, four items were included. A 7‐point agreement format rating, ranging from 1 (not at all) to

7 (very much) was utilized. The ten item scale showed an internal consistency of  =

.87 (example items: ʹI am satisfied with my life.ʹ or ʹThere are a lot of things in my life that make me sad.ʹ). All five negative formulated items were recoded. Hence, in the final index higher ratings specify higher levels of satisfaction with life.

Subjective Health. Based on the revised version of the Symptom Checklist (SCL‐

90‐R; Hessel, et al., 2001), the incidence of certain health symptoms was assessed to measure physical wellbeing of participants. The scale consisted of twelve different symptoms, such as ʺheadacheʺ or ʺback painʺ, and individual health problems were Method 69 evaluated regarding the last seven days by participants themselves. Ratings of inten‐ sity of each symptom were based on a 7‐point scale with a range from 1 (not at all) to

7 (very much). Each single item was reverse coded and averaged to specify a coeffi‐ cient of subjective health. In that, higher scores indicated lower frequency of symp‐ toms of disease or better subjective health status. Internal consistency of the scale was good ( = .83).

5.2 Family Study: Personality and Social Relationships in the Family Context

The second empirical investigation7 was conducted between fall 2005 and spring

2006. The target sample consisted of young and middle aged heterosexual couples

(Age range 25 to 54 yrs.; N = 171) living in particular contextual conditions, like mo‐ tivated childlessness or blended families. Recruitment methods included advertise‐ ments in local newspapers and postings in several public places throughout Berlin and the Halle/Leipzig8 metropolitan area. First, a telephone screening questionnaire was used to select couples. Participants who were 30 to 45‐years‐old, were in a rela‐ tionship for at least three years, and fit in one of the parental status categories (e.g., involuntary childless, traditional family) were included in this study. Participants were then placed into one of the four differentiated family type categories. Individu‐ al appointments for each couple were then scheduled. To remain in the study, both partners had to participate. The couple was questioned at the same time but in sepa‐

7 The RELATE project, which began in 2005, is conducted by Prof. Dr. F. R. Lang, head of the Institute of Psychogerontology at Friedrich‐Alexander University Erlangen‐Nuremberg, and Prof. Dr. F. J.

Neyer, professor of at the Department of Psychology, University of Potsdam.

The project is funded by the German Research Community (DFG).

8 The first empirical investigation has been conducted in Berlin and Halle/Leipzig because at that time

Prof. F. R. Lang was professor of Developmental Psychology at the Department of Psychology, Mar‐ tin‐Luther University Halle‐Wittenberg, and Prof. F. J. Neyer held the position of a PD at the Depart‐ ment of Personality Psychology at the Humboldt‐University Berlin. Method 70 rate rooms to avoid mutual influence. Each time, the research assistant started the assessment with a short warm up to introduce the main goals of the study and to answer possible questions. Afterwards, each participant completed the personal‐ computer‐assisted questionnaire on social network information and individual cha‐ racteristics on their own. During this process, a research assistant was always availa‐ ble to answer questions. The centerpiece of the research study was the assessment of ego‐centered social networks and corresponding relationship qualities. Finally, a de‐ cision task on prosocial orientation concluded the whole procedure. Overall, the in‐ terview session took almost two hours (M = 112 min, SD = 32 min) and each couple received a compensation fee of 30 Euro, approximately $40 (USD).

To further elicit interactions of relationship characteristics between genera‐ tions, all targets were asked to disclose the name and address of their living old par‐ ents, so they could also be contracted to be included in the study. The assessment of older parents, which took place between winter 2004 and spring 2005, was conducted via paper‐pencil questionnaires mailed to the home. The interview packet included an information sheet on the study in general, an instruction form on how to complete the questionnaire, and a stamped, addressed envelope. In addition, the information sheet pointed out which child of the parent took part in our main study. The parent questionnaire did not focus on an entire personal network but assessed six specific relationship types (current partner, child, child‐in‐law, grandchild, friend, and col‐ league). In the end, 199 older parents completed a questionnaire. The following ana‐ lyses however will be based on a sub‐sample of n = 92 adult child–parent dyads.9

Furthermore, young and middle aged adults were contacted again after one year, and in total, 175 individuals (51% of first assessment) were consent to take part at this follow‐up study. This follow‐up sample included some more specific meas‐

9 Of the N = 199 old parents that participated in the study the selection of the 92 parents is based on three aspects: (1) the child named the parent within the personal network, (2) no child is included twice to avoid an inflated sample, and (3) child and parent have the same sex. Selection effects are reported in 5.2.3. Method 71 ures on self‐esteem. It is utilized in analyses on stability and change in intergenera‐ tional relationships. And a sub‐sample of couples that participated in both assess‐ ments is used to examine dyadic dynamics more closely. See Table 5.5 for a descrip‐ tion of the different samples of the family study.

Table 5.5 Samples of Family Study

A B

Target Sample Intergenerational Relations (A1) (A2) (A3) (B1) (B2) Family Types Follow‐ T1/2 Dyadic Sample Follow‐up

up Couples Sex of Adult Old MCL ICL BF TF Total Targets Targets Total Adult Child Participant Child Parent Male 41 35 46 48 171 84 52 36 36 72 64

Female 41 35 46 48 171 91 52 56 56 112 80

Individuals 82 70 94 96 342 175 104 92 92 184 144

Dyads 41 35 46 48 171 77 52 / / 92 /

Note. MCL = Motivated Childless, ICL = Involuntary Childless, BF = Blended Family, TF = Traditional Family

5.2.1 Target Sample: Description of Individuals and Couples

The Family study consisted of assessments across time and in different generations.

This section focuses on investigations on the target sample, namely the middle‐aged individuals that lived in stable partnerships. The first part of this section describes the entire sample of the first data assessment (A1). The second part introduces the follow‐up sample (A2) and distinguishes a sub‐sample of 52 couples (A3) that parti‐ cipated in both studies and named their spouse in both occasions. This sample is uti‐ lized to enlighten relationship interdependencies within couples across time. The third part summarizes the used research methods and scales.

5.2.1.1 Description of Target Sample

The interviewed persons were on average 38 years old (M = 37.7, SD = 5.0). Overall, men were slightly older than women (Men: M = 38.6, SD = 5.3, Women: M = 36.9, SD

= 4.6; t (340) = 3.209, p < .01, d = .34). In addition, the majority of participants did not Method 72 consider themselves to be religious (65.5%), were highly educated (15.6 years), and were working full time (61%). The average partnership duration of participating couples was about ten years (M = 10.5, SD = 5.5) and more than half of them were married (64%).

Demographic Characteristics in Four Familial Situations. Table 5.6 indicates se‐ lected demographic features with regard to the differentiated family types. Overall, a number of characteristics were contradictory to prior findings regarding motivated childlessness (e.g., Carl, 2002, Carl et al., 2000). In contrast to all other three family types, motivated childless individuals were not less religious, did not report higher incomes, and were not distinguishable in the expected direction concerning educa‐ tion and status. Furthermore, the group of motivated childless individ‐ uals did not differ from all other three groups regarding religious affiliation, X2 (9) =

9.758, n.s. This result could be a consequence of the location of the research project.

The area of former East Germany, including Halle/Leipzig and Berlin, is known to be very low in religious affiliation in general, compared to the former western part (e.g.,

Meulemann, 2006). In consequence, for all four family types ‘no religious affiliation’ was chosen as the most frequent response option.

The four family types showed no variation in the average number of years of , F (3, 338) = .362, n.s., but a closer look on highest school education indi‐ cated a few significant discrepancies, X2 (9) = 17.306, p < .05. Examination of adjusted standardized residuals pointed out that more individuals being motivated childless received a university entrance diploma (i.e., German Abitur, 34%) and involuntary childless individuals were less likely to continue school beyond the tenth grade (i.e.,

German Hauptschule, 16%). With regard to reported household income10, differences

10 Income has been questioned as monthly income of the household (1 ʹno informationʹ, 2 ʹless than

1000 Euroʹ to 7 ʹmore than 5000 Euroʹ, 8 ʹunknownʹ). At first, this variable was transformed into a met‐ ric scale to enable the adjustment to the number of household members. The metric variable always indicated the midpoint of the raw scale (e.g. ʹ1500ʹ for ʹ1000 to 1999 Euroʹ). The metric measure has been adjusted by an individual factor, based on the scale of equivalence of the OECD (Lenuweit, Method 73 emerged only between involuntary childless individuals, who reported highest earn‐ ings with an average of 1625 Euro per month (SD = 847), and all other family types, F

(3, 319) = 9.233, p < .001. Motivated childless couples, indicating a monthly income of

M = 1231 Euro (SD = 779), did not differ from the two other family types with chil‐ dren. Regardless of differences in income and level of education, group comparisons regarding current occupational status failed to reach significance, X2 (6) = 12.126, n.s.

However, closer examinations of adjusted standardized residuals indicated that mo‐ tivated childless individuals were unemployed more often than expected (28%) and, in contrast, involuntary childless individuals reported as being employed full‐time

(73%) rather than part‐time (9%).

Taking together all information on educational and occupational status, re‐ sults demonstrated a fairly clear pattern. That is, involuntary childless persons were more educated, were employed full‐time, and earned more money in comparison to all other family types. In contrast, motivated childless individuals indicated that they had high levels of education, but they did not seem to have an occupational advan‐ tage, where they earned less than the involuntary childless and reported higher le‐ vels of unemployment. Indeed, there are a number of possible explanations for these unexpected patterns. First, direct comparisons in terms of education with other sam‐ ples of motivated childless individuals (e.g., the German sample of Carl, 2002), did not indicate differences between the two samples. Second, our study included only individuals living in stable partnerships. Often, partnerships of motivated childless individuals have been reported to be less stable or investigations included single per‐ sons in the sample of motivated childless people. However, this investigation did not. Therefore, the sample could picture a somehow special group of motivated indi‐

2007). This factor is taking into account number and age (older and younger than 14 years) of house‐ hold members. Inter‐correlation for the income variable between male and female was fairly high (r =

.76, p < .001), and the average difference value was 26 Euro (SD = 840). The high standard deviation was caused by 12 couples that differed for more than 1000 Euro, but were equally distributed across family types. Method 74 viduals. Third, recruitment of involuntary childless couples often occurs through specialized infertility clinics, which are known to be highly selective with their

Table 5.6 Demographic Characteristics by Family Type

Motivated Involuntary Blended Traditional

Childless Childless Family Family

(n = 82) (n = 70) (n = 94) (n = 96) Differences

M (SD) M (SD) M (SD) M (SD) F

Age 36.7 (5.4) 37.4 (5.8) 38.4 (4.9) 38.1 (4.1) n. s.

Education (in years) (SD) 15.6 (4.8) 15.6 (4.1) 15.2 (3.8) 15.8 (3.4) n. s.

2 Income 1231a (779) 1625b (847) 1072a (447) 1273a (529) (3, 319) = 9.23

Partnership duration 9.5a (5.6) 9.3a (5.2) 7.8a (4.0) 14.7b (4.3) (3, 338) = 38.30

n % n % n % n % X2

Religion

Evangelic 17 21 16 23 19 20 20 21 n. s. Catholic 4 5 2 3 12 13 11 12

Other 3 4 5 7 6 6 3 3

None 58 71 47 67 57 61 62 65

School Qualification1

Hauptschule 8 10 9 13 8 9 5 5 (9) = 17.306 Realschule 17 21 11(‐) 16 27 29 31 32

Abitur 28(+) 34 23 33 17 18 18 19

University Diploma 29 35 27 39 42 45 42 44

Employment

Full‐time 46 56 51(+) 73 56 60 55 57 n. s. Part‐time 13 16 6(‐) 9 21 22 24 25

None 23(+) 28 13 19 17 18 17 18

Current Family Status

Partnership 46(+) 56 20 29 42(+) 45 15(‐) 16 (3) = 36.44 Marriage 36(‐) 44 50 71 52(‐) 55 81(+) 84

Family Constancy

High 34(‐) 42 51(+) 73 52 55 53 55 (3) = 15.09 Low 48(+) 58 19(‐) 27 42 45 43 45 Note. Means in one row with different notations differ significantly from other means according to Scheffe post‐hoc tests. (‐/+) Indicator that the value contributes significantly (positive or negative) to the overall chi‐square value with (absolute value of) adjusted standardized residuals being greater than 2.00. 1ʹHautschuleʹ equiv. to 9th Grade; ʹRealschuleʹ equiv. to 10th Grade; ʹAbiturʹ equiv. to a high school diploma; 2Adjusted income according to OECD scale of equivalence, assigning 1 for first person of household, 0,5 for all other persons of household with an age equal or above 14 years, and 0,3 for all children younger than 14 years (Lenuweit, 2007). 15 individuals did not specify and 4 indicated to have no knowledge about information on household income. Method 75 clientele (Henning & Strauß, 2000; Helferich & Küppers‐Chinnow, 1996; Lorber &

Bandlamudi, 1993). The same type of involuntary childless individuals was included in this study as well. Therefore, the characteristics akin to the upper‐middle class were not surprising. We will take these characteristics into account when interpreting the results.

Partnership duration of parents in traditional families was substantially longer than in all other types (Traditional Family: M = 14.7, SD = 4.3; Motivated Childless: M

= 9.5, SD = 5.6; Involuntary Childless: M = 9.3, SD = 5.2; Blended Family: M = 7.8, SD =

4.0; F (3, 338) = 38.298, p < .001, η2 = 0.25). Couples in blended families and both moti‐ vated and involuntary childless couples did not differ on the duration of the partner‐ ship. However, upon examination of partnership status, motivated childless couples

(44%) and blended parents (52%) are less likely to be married, compared to involun‐ tary childless (71%) and traditional (84%) couples, X2 (3) = 36.44, p < .001.

Another focus of the current study is the analysis of relationship characteris‐ tics based on the constancy of the family of origin. Family constancy captures the degree of stability and for instability in the structure of the born family of the partici‐ pant. In this context, high family constancy indicated that biological parents were still married, whereas low family constancy referred indicated any instability in the family since the birth of the child (e.g. changes such as divorce or of parent).

Family constancy differed significantly between the four familial types, X2 (3) = 15.09, p < .01. Motivated childless individuals were more likely to report low family con‐ stancy (58%), in contrast involuntary childless individuals were less likely to belong to this group (27%). Both blended family and traditional parents showed equal dis‐ tributions regarding high (55%) and low (45%) levels of family constancy.

5.2.1.2 Relationship Perceptions across Time: The Follow‐up Sample

To conduct analyses of retest stability and to assess domain‐specific indicators of self‐ esteem, targets of the family study were contacted again after one year. In total, 175 individuals (77 couples and 21 individuals; 52% female) were consent to take part at Method 76 this follow‐up study (51% of first assessment). Participants that did not participate in the second assessment did not differ from other persons at the first assessment re‐ garding demographic characteristics such as sex, age, education, or marital and pa‐ rental status. However, those taking part at both assessments reported higher general closeness perception across personal networks. There were no differences with re‐ spect to reciprocity indicators.

A sub‐sample of 52 couples (N = 104) was selected, including only those couples where both partners participated at both occasions and both named their spouse in the first and in the second assessment. Taking demographic statistics of the first assessment, males were slightly older than women (Males: M = 38.4, SD = 4.5; females: M = 36.8, SD = 4.5, d = .36). Couples reported a mean relationship duration of

10.4 years (SD = 5.5) and about 61% of them were married. With respect to all these variables, the sub‐sample did not differ from the target sample of the family study.

Furthermore, the sub‐sample showed to be fairly balanced regarding parental status

(40% parents) and family situation (54% traditional, i.e. involuntary childless, tradi‐ tional family).

5.2.2 Measures of Target Sample

This dissertation focuses on the assessment of social networks and corresponding relationship descriptors, a number of outcome and background variables. Table 5.7 gives an overview of instruments used in the current study. Not included are vari‐ ables of demographic information that have been already reported in section 5.2.1.

Table 5.7 Applied Instruments of the Family Study

Dimension Instrument Description Comment

Ego‐centered Name Generator “Please name all individuals you already know for Naming between 1 and 35 Network (Lang, Neyer, Wrzus & a long time and/or have frequent contact with.“ social relationship partners Wagner, 2005)

Genetic Relationship Index Differentiated assessment of degree of genetic 85 types of social relation‐ Relatedness (Lang, Neyer, Wrzus & relatedness and origin of relationship ships Wagner, 2005) (Table continues) Method 77

Dimension Instrument Description Comment

Closeness Inclusion of Other in the 7‐point scale Ich Anderer Ich Anderer Ich Anderer Ich Anderer Self Scale (Aron, Aron & Smollan, Ich An‐ Ich An‐ Ich An‐ 1992) Graphic Closeness Scale Indicating perceived close‐ (Lang, Neyer, Wrzus & ness between 0 and 100. Wagner, 2005)

Reciprocity Graphic Interdepend‐ 7‐point scale ence Scale (Lang, Neyer, Wrzus & Wagner, 2005) Graphic Balance Scale 7‐point scale (Lang, Neyer, Wrzus & Wagner, 2005)

Desire for Leipziger Fragebogen zu Four sub‐scales, 12 Items, Children Kinderwunschmotiven Selection of 3 items per sub‐scale 7‐point scale (Brähler, Stöbel‐Richter, & Schumacher, 2001)

Life Goals Self‐development, based Differentiation between agency and communion 13 Items, on Lang, Diewald & life goals 10‐point scale Wendt (2004) (cf., Appendix E)

Self‐esteem Global Self‐esteem 6 Items, (Marsh & O’Neal, 1984) 7‐point scale

5.2.2.1 Ego‐centered Social Networks: The Relationship Level

Most of the network and individual measures were identical to those described for the online study. Again, the centerpiece of the study was the assessment of ego‐ centered social networks. Assessment was based on the described name generator and information on relationship partners was similar as well.

Relationship Categories. After generating a list of network members, all named relationship partners were presented at the next page and each one had to be classi‐ fied regarding the relationship index. This index, represented at a supplementary sheet (cf. Appendix E), distinguished among 85 possible relationship types. One ma‐ jor aspect has been the differentiation with regard to the degree of genetic related‐ ness (coefficient r with a range between 1 ʹidentical twinsʹ to 0 ʹnon‐kinʹ). In addition, the index indicated for kin whether the relationship was maternal or paternal, an in‐ law, or step. The index separated non‐kin relations by the origin of relationships, such as ʹfrom workʹ, ʹfrom neighborhoodʹ, or ʹknowing each other through another Method 78 personʹ. Upcoming analyses will work with varying variables regarding relationship differentiation.

Table 5.8 Descriptive Statistics of Entire Ego‐centered Social Networks and Composition in Eleven Different Relationship Types

Sub‐Categories M SD Min Max Overall Network Size / 13.34 8.24 1.00 35.00 11 Relationship Types r = .5 Parents 1.29 0.76 0 2 Siblings 0.92 0.90 0 5 Children 0.88 1.17 0 5 r = .25 Close Kin 1.08 1.43 0 9 r = .125 Distant Kin 0.52 1.04 0 7 r = 0 Partner 0.78 0.44 0 3 Step Family 0.21 0.55 0 3 In‐Laws 1.80 2.39 0 13 Colleagues 1.64 2.04 0 12 Acquaintances 2.13 2.69 0 16 Old/Long‐time Friends 2.10 2.67 0 28 Note. M = Mean of named relationship type, SD = Standard deviation

Based on evolutionary assumptions, a number of analyses will focus on the degree of genetic relatedness. The current sample varied between r = .5 (parents, sibl‐ ings, etc.), r = .25 (grandparents, uncles, etc.), r = .125 (cousins, etc.), r = .0625 (distant, other genetic kin), and r = 0 (partner, in‐laws, coworkers, etc.). Another level of anal‐ ysis will examine effects across eleven specific relationship types: parents, siblings, children, close kin, distant kin, partner, step‐kin, in‐law family, colleagues, acquain‐ tances, and old/long term friends (Table 5.8). Both categories will be used to get a more differential view on the composition of ego‐centered networks and relevant regulative mechanisms.

5.2.2.2 Indicators of Relationship Quality at the Relationship Level

Again, relationship perceptions of closeness and reciprocity were assessed by the identical graphical items as in the online study (5.1.2.2). Descriptive statistics of the measurement model are displayed in Table 5.9. Method 79

Table 5.9 Variables of the Relationship Level – Item Characteristics of the Family Study (N = 4561)

Descriptive Statistics Intercorrelations N M SD Range ICC 1 2 3 4 5

Perceived closeness 4554 0.00 0.90 ‐2.1‐1.4 .15 Degree of genetic relatedness 4561 0.14 0.21 0‐.5 .04 .23 Perceived under‐benefit 4554 0.32 0.47 0/1 .13 ‐.07 .13 Perceived over‐benefit 4554 0.20 0.40 0/1 .08 .01 .03 ‐.34 Under‐benefit x genetic relatedness 4554 0.06 0.15 0‐.5 .11 .13 .58 .59 ‐.20 Over‐benefit x genetic relatedness 4554 0.03 0.12 0‐.5 .05 .06 .41 ‐.20 .57 ‐.10

Note. M = Mean, SD = Standard deviation, ICC = Intraclass correlation. Bold intercorrelations were significant at p < .05.

The sub‐sample on couple relationships is based on the evaluation of the rela‐ tionship to the spouse only. In that 104 people reported perceptions of emotional closeness and reciprocity with respect to their spouse at both investigations. Assess‐ ment of closeness and reciprocity is identical as described in the online study. De‐ scriptive statistics are displayed and discussed in section 6.3.3.

5.2.2.3 Background and Outcome Variables: The Individual Level

The entire RELATE project questionnaire consisted of several scales and other items, that assessed family specific characteristics. For these analyses, selected scales will serve as background and some as outcome variables. To simplify data interpretation and comparison, all following scales were linear adjusted to represent equal ranges, rated from 0 (low) to 1 (high). Means, standard deviations, skewness, internal consis‐ tencies, and intercorrelations of considered scales are shown in Table 5.10. With re‐ spect to the sub‐sample of couple relations, only the global self‐esteem indicator of

Marsh and O’Neil (1984) and the state self‐esteem measure of Heatherton and Polivy

(1991) were utilized. Descriptive statistics are reported in the result section (6.3.3).

Desire for Children. To evaluate precise information on motives regarding the

Desire for Children, a shortened version of the Leipziger Fragebogen zu Kinderwun‐ schmotiven (LKM; Brähler, Stöbel‐Richter, & Schumacher, 2001) was included. This Method 80 questionnaire included four sub‐scales, with four items each. In this research study, only three items per subscale were integrated. Selection of items was based on item statistics, assessed with a representative sample and presented by Brähler, Stöbel‐

Richter and Schumacher (2001). The four subscales were “Emotional Stability” (ex‐ ample item: A child gives true meaning to my life.), “Personal Constraints” (example item: Having a child, I would not be able to maintain all my friendships.), “Social

Acceptance” (example item: Having a child is essential to me to be acknowledged as an adult.), and “Insufficient Support” (example item: There is not enough daycare for children.). Participants had to evaluate influence of each statement using a 7‐point scale, ranging from 1 (has no influence) to 7 (has a strong influence). Internal consis‐ tency was sufficient to good (α = .62 ‐ .84).

Table 5.10 Covariates – Scale Characteristics and Intercorrelations of the Family Study (N = 342)

Scale Descriptive Statistics Intercorrelations Life Goals Desire for Children k M SD SK  CLG ALG ES PC SA IS

Life Goals Communion 6 0.81 0.13 ‐1.21 .58 Agency 5 0.78 0.14 ‐0.91 .67 .43 Desire for Children Emotional stability 3 0.59 0.33 ‐0.44 .84 .54 ‐.02 Personal constraints 3 0.25 0.23 1.02 .62 ‐.22 .08 ‐.23 Social acceptance 3 0.12 0.18 1.91 .67 .16 .06 .28 .13 Insufficient support 3 0.37 0.30 0.46 .73 .02 .11 .03 .48 .19

Self‐esteem 6 0.76 0.16 ‐0.89 .75 .23 .17 .10 ‐.14 ‐.11 ‐.05

Note. k = number of items, M = Mean, SD = Standard deviation, SK = Skewness,  = Cronbach’s Alpha. CLG = Communion life goals, ALG = Agency life goals, ES = Emotional stability, PC = Personal con‐ straints, SA = Social acceptance, IS = Insufficient support. Bold intercorrelations were significant at p < .05.

Life Goals. A scale was developed for this study to assess life goals. Items in the scale were based on a list of life goals by Lang, Diewald and Wendt (2004). All to‐ gether, 13 different goals had to be judged in terms of personal importance from 1 Method 81

(not important) to 10 (very important). Subsequently, eleven11 of the goals were placed among the two main dimensions of agency and communion (Abele & Wo‐ jciszke, 2007; Bakan, 1966). Assignment to subscales was guided by theoretical as‐ sumptions. The sub‐scale of communion enclosed six items, such as “experiencing love” or “harmonious family life” with an internal consistency of  = .58. Agency ori‐ ented life goals were represented by items such as “independence and self‐ determination” or “occupational success”. The included five items led to an internal consistency of  = .67.

Self‐Esteem. With regard to global self‐esteem the identical scale and agree‐ ment format as in the online study was used. The internal consistency of the Marsh and O’Neil (1984) scale was satisfactory (α = .75).

5.2.3 Intergenerational Relations: Adult Children and Their Parents

To enhance the work on relationship dynamics and the interplay with family con‐ stancy, two sub‐samples addressed the specific role of intergenerational ties. At first, adult children and their parents were examined by more differential dyadic analyses.

Second, adult children relationship perceptions were followed across a one year time period to illustrate characteristics of stability and change.

5.2.3.1 A Sample of Adult Child‐Parent Dyads (B1)

This sub‐sample consisted of 92 dyads of adult children and their parents.

Composition of dyads divided into 56 (61 %) adult daughter–mother and 36 (39%) adult son–father dyads. Family constancy was reported to be high in 60 (65%) of the

92 dyads. Considering the adult child’s current family, 16 (17%) were motivated childless, 23 (17%) reported to be involuntary childless, 22 (24%) were living in

11 The two items, not included into the agency or communion subscale were “physical healthy and fitness” (original: “Körperlich gesund und fit sein”) and “stability and inner ” (original: “Ruhe und Stabilität”). Method 82 blended families and 31 (34%) in traditional families. The average age of adult child‐ ren was 37 years (M = 36.7, SD = 4.1), and for parents it was years (M = 64.2, SD = 6.9).

Adult children had higher education levels than their parents. For example, about 77 percent had a university entrance or university diploma, compared to 37 percent of parents. Regarding marital status, 63 percent of adult children were married, com‐ pared to 73 percent of parents. In addition, 24 percent of the older generation re‐ ported to be divorced or widowed. Based on the revised version of the symptom checklist (SCL‐90‐R)12, the average number of health symptoms indicated overall good physical wellbeing of parents (M = 2.17, SD = .97; Hessel, Schumacher, Geyer, &

Brähler, 2001).

The 92 dyads represented only a selection of each the adult child and parent samples. Inclusion into the dyadic sample depended upon two conditions: (a) the adult child had to include the parent into the ego‐centered social network; and (b) the parent had to participate in the study. Additionally, only same sex dyads were inte‐ grated. For that reason, sub‐samples were compared to other participants of their generation who were not selected to be in the sub‐sample. This is to account for poss‐ ible selection effects with regard to demographical characteristics. Comparisons of the sub‐sample adult children (n = 92) and the remaining participants of the family study (n = 250) found very few differences, and no variations occurred between par‐ ents of the selected dyads (n = 92) and the remaining parents who participated in the mailing study (n = 107). Two tables containing characteristics in more detail can be found in the Appendix A.

Upon closer examination of differences between the adult child sub‐sample and the non‐selected group, no differences were found on current family situation, in terms of religious affiliation, or characteristics of the partnership. Yet, members of the

12 Consisting of 12 different symptoms, such as ʺheadacheʺ or ʺback painʺ, this scale evaluated indi‐ vidual health problems for the last seven days. Intensity of symptoms was rated on a 7‐point scale ranging from 1 (not at all) to 7 (very much). Internal consistency of the scale was sufficient ( = .80).

Higher scores indicate higher frequency of symptoms (i.e., worse health). Method 83 sub‐sample showed to be about 1.5 years younger, t (340) = 2.198, p < .05, d = .30, have higher education levels, highest school qualification: X2 (3) = 9.801, p < .05; education in years: t (340) = ‐2.730, p < .01, d = .33, and report higher incomes, t (321) = ‐2.578, p <

.05, d = .31. In addition, fewer participants in the non‐selected group reported high family constancy, 51%, X2 (1) = 7,140, p < .01. That is, more adult children with an in‐ stable family of origin either did not name their parent(s) in their networks, or their parents did not participate in the study. This selection must be regarded in the data interpretation process.

5.2.3.2 Intergenerational Relations across Time: The Follow‐up Sample (B2)

The target sample of the family study was assessed a second time one year after the initial interview. Of the 175 individuals that participated a second time, 144 (56% fe‐ male) named their parents as social network partners at both investigations. They constitute the current sample.

At the first assessment selected adult children were about 37 years of age (M =

37.2, SD = 4.5), somewhat highly educated (M = 16.3, SD = 3.6 years), and not reli‐ gious (71%). The average partnership duration was 10.8 years (SD = 5.2) and 65 per‐ cent of participants were married. The majority reported to have a stable family of origin (68% high family constancy). Participants that were not included into this sub‐ sample did not differ from other persons at the first assessment regarding demo‐ graphic characteristics such as sex, age, religion, or marital and parental status, and reported general perceptions of closeness, reciprocity or conflict. However, those in‐ cluding their parents at both assessments were higher educated, t (340) = ‐2.97, p <

.01, d = .33, were less likely to belong to a blended and more likely to live in a tradi‐ tional family setting, χ2 (3) = 12.71, p < .01, and they showed to be less likely to have a family of origin with a structure of high constancy, χ2 (1) = 15.74, p < .001. Also, there were no differences with respect to global self‐esteem. Method 84

5.2.4 Measures on Intergenerational Relationships

Dyadic analyses will contrast and combine measures of the adult children, i.e. partic‐ ipants of the main study, and their parents. Investigations on relationship develop‐ ment across time will include identical items of the main and follow‐up investiga‐ tion. Since measures of the target sample were described in detail in section 5.2.2 and those used with their parents and in the follow‐up study were very similar to the first investigation, this part was shortened at several points. In general, both additional studies included selected demographic characteristics, individual and relationship features of selected social interaction partners (old parents: six relationships; follow up: five relationships), as well as several outcome and background variables. One difference to the main investigation was that the method of data assessment, since mailing questionnaires were utilized. Analyses of the sub‐samples will only be based on a selection of variables that are presented in the following paragraphs.

Selected Relationship Partners of Old Parents. The assessment of old parents con‐ sidered only six relationships: current partner, adult child (who took part in the main study), corresponding child in law, grandchild, one long‐time friend and one (for‐ mer) colleague. Relationship partners were presented each after another and partici‐ pants had to give basic demographical information. General data was similar to so‐ cial network aspects of adult children: name, sex, age, duration of knowing each oth‐ er, geographical distance, contact frequency and conflict frequency. On average, par‐ ents named 4.7 relationship partners (SD = 1.1).

Child‐Parent Relationship across Time. The follow‐up study on main targets in‐ cluded five selected relationships: mother, father, partner, colleague, and friend. The used sub‐sample consists of 144 adult children that evaluated the relationship to 236 parents at both investigations. Of old parents 136 (58%) were mothers. The average age of parents was about 64 years (M = 63.5, SD = 7.1).

Relationship Quality. The assessment of relationship quality was always iden‐ tical to the main study (see 5.2.2.2). However, handling of the items differed slightly.

With regard to perceived closeness, the two items were linear transformed to range Method 85 between 1 (very close) and 0 (very distant), with numbers in between representing feelings in between. Analogous to the main procedure, at both instances the items were then summed to represent a single index of closeness.

Table 5.11 Descriptive statistics, Internal Consistencies, Inter‐Correlations, and Dyadic Similar‐ ity in Key Variables

Descriptive Statistics Inter‐Correlations ICC n k M SD Min Max  CL RE Dyadic Sample Closeness 182 2 0.74 0.21 0 1.00 .69 .42 Reciprocity 182 2 0.75 0.22 0 1.00 .65 .39 .25

Self‐esteem 184 6 0.77 0.17 .08 1.00 .80 .10 .00 .17 Follow‐up Sample T1 / Closeness 236 2 0.68 0.22 0 1.00 .72 Reciprocity 236 2 0.69 0.25 0 1.00 .75 .26 /

Self‐esteem 236 6 0.76 0.17 .25 1.00 .81 .09 ‐.03 / T2 / Closeness 236 2 0.48 0.16 .03 .99 .76 Reciprocity 236 2 0.70 0.24 .17 1.00 .82 .27 /

Self‐esteem 236 6 0.75 0.16 .17 1.00 .81 .05 .10 /

Note. n = number of individuals, k = number of items, M = mean, SD = standard deviation, Min = min‐ imum, Max = maximum,  = Cronbach’s alpha, ICC = Intraclass correlation as indicator of dyadic simi‐ larity between adults and old parents in 92 dyads (Gonzalez & Griffin, 2002; Snijders & Bosker, 1999). Bold intercorrelations were significant at p < .05. Closeness, reciprocity and conflict are specific ratings of the relationship to the dyadic partner in con‐ trast self‐esteem is a global measure (Marsh & O’Neil, 1984). All variables have been transformed to a range 0 to 1: Closeness 0 ‘not close at all’ – 1 ‘very close’, Reciprocity (4‐point scale) 0 ‘not reciprocal’ – 1 ‘reciprocal’, Conflict (5‐point scale) 0 ‘none/rarely’ – 1 ‘very often’, Self‐esteem (7‐point scale) 0 ‘very low’ – 1 ‘very high’.

Since analysis with two reciprocity dummies produced no additional insight, a more simply handling of reciprocity observations was utilized. Analyses of inter‐ generational relationships will not differ whether the participant or the other person reported more investment in the relationship, but instead both bipolar reciprocity items were recoded to a 4‐point unipolar scale with higher numbers indicating higher levels of reciprocity in ties. Subsequently, linear transformations were con‐ Method 86 ducted resulting in items with a range from 1 (reciprocal) to 0 (non‐reciprocal). As with closeness, the two reciprocity items were combined to a single index of reciproc‐ ity. Table 5.11 summarizes important descriptive statistics, internal consistency, and inter‐correlations on the two indexes and both outcome variables across the sub‐ samples.

Outcome Variable. Analysis of intergenerational relations considers one out‐ come: global self‐esteem as a general indicator of self‐evaluation. The assessment of old parents and at the follow‐up study included the global Self‐Esteem measure of

Marsh and O’Neil (1984), which was already described in section 5.2.2.3. Again, in‐ ternal consistency of the scale was sufficient.

Dyadic Similarity. Because one part of data examination will handle the dyadic relationship as main level of analysis, clustering of individuals within a dyad has to be considered in an intraclass correlation (ICC) to indicate dyadic similarity or re‐ semblance between relationship partners (Gonzalez & Griffin, 2002; Snijders & Bosk‐ er, 1999; cf. also section 5.3.1). ICC computations for each variable were based on multilevel unconditional means models, conducted in Mplus, and are shown in the last row of Table 5.11. Similarity between the two generations was found to be rela‐ tively high at perception of closeness. In contrast, perception of reciprocity and self‐ esteem was only reasonably similar between generations. Levels of resemblance were similar to those reported for couple relations (cf. Neyer & Voigt, 2004), with generally higher degrees of similarity in perceived relationship quality than in personality cha‐ racteristics. Table 5.11 gives an overview on all considered variables with descriptive statistics, information on internal consistencies, intercorrelations, and dyadic similar‐ ity in measures for the entire sample of adult children and their parents.

5.3 Procedures and Methods of Statistical Analyses

Prior sections introduced the two major samples and corresponding measures. Over‐ all, the assessment of social relationships was the focus of both studies. Next, differ‐ Method 87 ent levels of data, strategies and statistical packages are introduced that will guide upcoming data analyses.

5.3.1 Data Structures

With regard to relationship research, different levels of data assessment and evalua‐ tion can be followed: (a) individual level, one person’s perception of his/her relation‐ ship partner (e.g., feelings, behaviors, etc.) or the quantitative structure of the entire network (e.g., size or composition), (b) dyad as level, mutual perceptions of both ac‐ tors to find out about the potential interdependency in the tie, and (c) hierarchical level, consideration of the whole of an individual’s social network members. The fol‐ lowing parts will describe these levels of data analyses in more detail.

Individual Data Structure. Specifically for descriptive analyses regarding specif‐ ic relationship partners, the individual’s perspective will be taken. Furthermore, as‐ pects of social network structure, such as size or number of friends and family, are conducted at this level of analysis. Also, in both upcoming structures, dyadic and hierarchical, the individual level is pivotal.

Dyadic Data Structure. Because relationship partners affect each others though‐ ts, behaviors, and feelings, information from each individual about the relationship are expected to be co‐existent. That is, measurements are nonindependent and ap‐ plied statistical methods have to take this into account by treating the dyad as unit of analysis instead of the individual. The actor‐partner interdependence model (APIM;

Cook & Kenny, 2005; Cook & Snyder, 2005; Kenny, Kashy, & Cook, 2006) is one pos‐ sibility to consider individuals as being nested within dyads and in that to estimate both, individual and dyadic factors at once. Central to the APIM analysis are the two components actor and partner effect (cf.,Figure 4). Generally defined, actor effects measure the effect of a person’s predictor on their own outcome (e.g., child on child).

Partner effects, however, estimate the effect of a person’s predictor on her/his part‐ ner’s outcome (e.g., child on parent). In that, it measures a form of interdependence between dyadic partners (Cook & Kenny, 2005; Kenny et al., 2006). There are two Method 88 important correlation coefficients in this analysis. First, it is the correlation between predictors to control for actor and partner effects. The second important correlation coefficient is between the two residual variances of the outcomes. This estimate indi‐ cates the correlation between unexplained variance components after removing the covariance by partner effects. It could possibly be a sign of the existence of other dyadic effects such as a family‐level factor.

Perceived Actor‐Effect Perceived UC Reciprocity Child Conflict Child Partner‐Effect

Partner‐Effect Perceived Perceived UP Reciprocity Parent Actor‐Effect Conflict Parent

Figure 4. The Actor‐Partner Interdependence Model (APIM, Based on Cook & Kenny, 2005, p. 102). Child = Person A; Parent = Person B; UC = Residual (Un‐ explained) Proportion of Child; UP = Residual (Unexplained) Proportion of Parent

Dyadic estimations will be conducted with Mplus version 5.0 (Muthen & Mu‐ then, 1998‐2006) using a structural equation modeling (SEM) approach. This ap‐ proach allows the direct translation of the model shown in Figure 4 into one model that is estimated and tested (Cook & Kenny, 2005; Kenny & Cook, 1999; Kenny et al.,

2006). There is an equation for each actor of the dyad. Residual variances are esti‐ mated for each equation and are allowed to correlate. In addition, independent va‐ riables are allowed to correlate. Another advantage of SEM is the ability to directly compare parameter estimates. That is, the actor effect of child and parent could be tested and statistically evaluated whether they are equal or not and in that indicating who is more influential in the relationship. Differences are found out by comparing the chi‐square goodness‐of‐fit value for the unrestricted model and the model, where parameters were restricted or forced to be equal. If fit indices of the two models dif‐ fer substantially, forcing parameters to equal worsened the model fit and thus, pa‐ rameters have been shown to differ significantly. Method 89

In addition, the Mplus output includes the intraclass correlation coefficient

(ICC) for both dependent variables. In dyadic analysis the ICC is a measure of simi‐ larity between relationship partners (Kenny, Kashy, & Cook, 2006). Similar to Pear‐ son’s correlation, intraclass correlations vary, between ‐1 (characteristics of dyadic partners are opposite to each other) to 0 (characteristics of dyadic partners are dissi‐ milar) to +1 (characteristics of dyadic partners are similar or alike). In that, this indi‐ cator must be distinguished from the ICC indicator in multi‐level modeling, which is described in the following section on hierarchical data structure.

Hierarchical Data Structure. Ego‐centered social networks represent social ties to several individuals (alters) at the relationship level (or level 1) that are nested within one ego at the individual level (or level 2). This relationship structure enables the study of phenomenon at different levels, first, on the relationship level (e.g., strength of a tie, effects of contact frequency, interplay among several relationships), second, on the individual level (e.g., gender of ego, personality, etc.), and third, inte‐ ractions between the two levels. However, this data structure puts a particular chal‐ lenge on statistical methods used: the interdependence between observations is vi‐ olated. Former investigations either neglected these dependencies by using ordinary least square (OLS) strategies, or they frequently used aggregation (i.e. average or sum of a feature across all relations of one ego) and disaggregation (i.e. separate ana‐ lyses of each relation) to handle the hierarchical structure. But in all three proceed‐ ings researchers have to handle potential model misspecification or loss of informa‐ tion (Van Duijin, Van Busschbach, & Snijders, 1999).

Within the last decades the development of new statistical strategies and pro‐ cedures opened up the possibility to handle these complex data structures. Multile‐ vel analysis or hierarchical linear modeling does account for the nested composition and variability of hierarchical clustering structures (Snijders & Bosker, 1999). In the case of ego‐centered networks, it allows random residuals at both levels, between relationships (level‐one) and between individuals (level‐two). The multilevel model will be introduced briefly, however general and more detailed textbooks on multile‐ Method 90 vel modeling are by Raudenbush and Bryk (2002) or Snijders and Bosker (1999). In addition, a number of empirical articles using multilevel modelling with regard to ego‐centered social networks give an example of statistical possibilities and advan‐ tages (e.g., Duijin et al., 1999; Fingerman et al., 2004).

The concept of multilevel modeling may be visualized like nested regression equations, in which estimates of the first level become outcomes at level‐two. Statis‐ tical packages, such as Mplus or HLM, estimate only one equation including all le‐ vels of analyses. In the following, a multi‐level random‐coefficient model with two levels is demonstrated, analyzing ego‐centered networks that include i = 1,…, nj rela‐ tionships (or, level‐one units) nested with j = 1,…, J individuals (or, level‐two units).

The within, or level‐one model, is formulated for one outcome Yij (e.g., perception of closeness, CL) with one explanatory variable Xij (e.g., degree of genetic relatedness,

GR). The following equation describes the level‐one model:

Yij =0j + 1j Xij + rij CLij =0j + 1j (GR)ij + rij,

where 0j and 1j are random level‐one coefficients, referred to as level‐one random intercept and random slope, respectively. The term rij is the level‐one error term with a residual variance of 2. At level‐two, level‐one parameters vary across individuals:

0j = γ00 + u0j 1j = γ10 + u1j,

where γ00 and γ10 are level‐two coefficients and called fixed effects. γ00 is the average of the dependent variable across all level‐two units (e.g., mean perceived closeness to relationship partners with zero genetic overlap13 across the whole sample). γ10 is the average slope between the dependent and the explanatory variable (e.g., variation of perceived closeness across different degrees of genetic relatedness), and should be

13 The predictor is in its natural metric (not centered), because zero is a meaningful and frequently represented value. Therefore, the intercept refers to the category of non‐kin relationships. Additional centering definitions are grand‐mean, group‐mean or theoretically chosen locations (cf., Raudenbush

& Bryk, 2002). Method 91

interpreted as unstandardized regression coefficient. Both, u0j and u1j represent level‐ two random error terms with variances of 02 and 12, respectively. This model may be expanded by the inclusion of additional relationship predictors or individual level variables. In ego‐centered social networks, level‐one models estimate intraindividual variations and level‐two models interindividual variations.

A first step of multilevel modeling is the estimation of an empty model (also called intercept‐only or totally unconditional model, cf., Nezlek, Schröder, & Schütz,

2006; Tabachnick & Fidell, 2007). Residual variances of the estimated model are used to calculate the intraclass (also “intra‐respondent” Duijin et al., 1999, p. 194) correla‐ tion (ICC). In multi‐level analysis, the ICC is an indicator of the proportion on how much variation in the outcome variable is present in the relationship level (level‐one) and in the individual level (level‐two). It is an indicator of whether multilevel model‐ ing is required or not. The ICC I can be defined as

2 2 2 I =  /  +  , where 2 is the between‐individual (level‐two) variance and 2 is the within‐ individual (level‐one) variance (Raudenbush & Bryk, 2002; Snijders & Bosker, 1999;

Tabachnick & Fidell, 2007). In multilevel models, the ICC can be zero or positive, with higher values implying higher interdependence between errors, which means that the clustering matters and has to be considered in data analysis. A higher ICC increases the power of analysis (Kreft & DeLeeuw, 1998), especially with regard to level‐two and cross‐level effects.

In addition, variances of the null (or previous) model can be related to the cur‐ rently fitted model to calculate an index of the proportion of variance explained through including additional predictors. This auxiliary statistic may be used at level‐ one and level‐two (cf., Raudenbush & Bryk, 2002, p.74, 79). Comparison of two esti‐ mated multi‐level models, e.g. null model (M0) with a more restricted model (M1), is conducted by using a scaled difference chi‐square statistic (Satorra & Bentler, 2001).

This analysis tests for differences in the random part of the multi‐level models. Out‐ put information of Mplus, i.e. the loglikelihood values (L0 and L1, respectively), Method 92

number of parameters (p0 and p1, respectively), as well as scaling correction factors

(c0 and c1, respectively), is needed for the computation of the chi‐square difference test (TRd) by the following formula (c.f., Muthen & Muthen, 2008, Mplus homepage)

TRd = ‐2*(L0 ‐ L1)/cd.

This comparison is used to support the model selection process. However, model building processes are guided by theoretical considerations and insights.

5.3.2 Statistical Software Packages

Two different statistical packages were used to handle the given data structures. At first, data preparation and variable transformation were conducted with the Statistic‐ al Package for Social Sciences (SPSS), version 16.0. In addition, SPSS has been used to picture demographic structures and compare group mean differences using t‐Tests as well as one‐way ANOVAs, depending on the number of groups considered. At some points in the dissertation, correlation or regression analyses are used to further en‐ lighten the relationship between one dependent variable and one vs. a number of independent variable(s).

With regard to multilevel and dyadic data, Mplus version 5.0 has been used to perform model estimations (Muthen & Muthen, 1998‐2006). Therefore, SPSS data sheets are written out as ʹ.datʹ file which Mplus is able to read directly. The program is utilized to conduct structural equation modeling as well as multi‐level analyses.

Hence, hierarchical as well as dyadic data analyses have been conducted with this software package. Results 93

6 RESULTS

The results part focuses on three major aspects: First, it takes a critical look on one global and two domain‐specific indicators of self‐esteem. It explores shapes of deve‐ lopmental trajectories across age groups, the differential interplay, and variations in context sensitivity. Second, it examines differences in social network and relationship regulation characteristics among age groups and between childless individuals in contrast to traditional and blended family parents. Third, it observes the functionali‐ ty of relational patterns, concentrating on network characteristics but also exploring specific relationship dyads. In general, results of the online study and the family study are intertwined and discussed by topic.

6.1 Self‐Esteem across Life Span and Family Situations

The first section deals with individual measures of self‐esteem. It thereby examines a global and two domain‐specific indicators, performance and social self‐esteem. Both the online study and the family study investigated into the assessment of self‐esteem, regarded as a construct of self‐evaluation and an internal monitor of relational value, at different levels and domains. The empirical studies aimed at two issues: (a) ex‐ amination of age differences in mean‐level and correlative relationships between global versus domain‐specific measures of self‐esteem (online study), and (b) explo‐ ration of contextual effects with respect to former and present family structures on different self‐esteem measures (family study). Based on the research questions and hypotheses, this first part of the result section will be parted in two sections, ap‐ proaching age characteristics first and contextual effects second.

6.1.1 Online Study: Effects of Age on Self‐Esteem

Across the entire sample the majority of persons reported relatively high values of global self‐esteem (M = .71, SD = .22). This also showed to be true with regard to per‐ Results 94 formance state self‐esteem (M = .75, SD = .18). Results indicated a high confidence into individual abilities and competences. In contrast, the mean‐level of social state self‐esteem (M = .56, SD = .27) was lower and suggested that on average people were fairly sensitive and concerned about their public impressions (cf., Table 6.1).

Several explorative control analyses of mean‐level differences with regard to selected demographic features were conducted across the entire sample (cf., Table

6.1). Results of a number of t‐tests stated no substantial effects in global and domain‐ specific self‐esteem measures with respect to (a) gender, global self‐esteem: t (550) =

.76, n.s., d = .09; performance state self‐esteem: t (551) = 1.25, n.s., d = .17; social state self‐esteem: t (551) = 1.72, n.s., d = .15, (b) partnership status, global self‐esteem: t

(550) = .33, n.s., d = .05; performance state self‐esteem: t (551) = .23, n.s., d = .05; social state self‐esteem: t (551) = ‐.64, n.s., d = .04, or (c) parental status, global self‐esteem: t

(550) = ‐1.06, n.s., d = .09; performance state self‐esteem: t (551) = ‐.44, n.s., d = .05; so‐ cial state self‐esteem: t (551) = ‐1.28, n.s., d = .11.

6.1.1.1 Age‐Associated Patterns of Self‐Esteem Measures

One key advantage of this sample is the already mentioned age variation across the entire adulthood life. To examine age differences in global and domain‐specific self‐ esteem measures, a number of univariate analyses of variance were conducted. The grouping variable differentiates between five age clusters. The distinguished clusters represent theoretically important age groups (Arnett, 2000; Levinson, et al., 1978, in

Lemme, 2006), beginning with early adulthood (ages 17‐29, n = 154), then young adulthood (30‐44, n = 97), middle adulthood (45‐59, n = 123), late adulthood (60‐74, n

= 150), up to a group of old age participants (75‐86, n = 33). This reduction was cho‐ sen for reasons of parsimony, specifically since conducted analyses with eight age groups based on decades (cf., Roberts & Delvecchio, 2000) led to very similar results.

Table 6.1 shows means, standard deviations, and effect sizes of all three self‐esteem Results 95 measures by age group. Figure 5 indicates mean level differences in shapes of age‐ related trajectories for global, performance, and social self‐esteem.14

Table 6.1 Means, Standard Deviations, and Effect Sizes (ES) of Global Trait and Domain‐ Specific State Self‐Esteem Measures by Gender, Partnership Status, Parental Status, and across Age Groups

Global Trait Performance State Social State

Self‐Esteem Self‐Esteem Self‐Esteem M SD ES M SD ES M SD ES

Whole sample 0.71 0.22 0.75 0.18 0.56 0.27 Gender Male 0.72 0.21 d = .09 0.77 0.18 d = .17 0.59 0.27 d = .15 Female 0.70 0.22 0.74 0.18 0.55 0.27 Partnership status Single 0.71 0.21 d = .05 0.75 0.18 d = .05 0.55 0.27 d = .04 Partnership 0.70 0.23 0.74 0.19 0.56 0.27 Parental status Childless 0.70 0.23 d = .09 0.74 0.19 d = .05 0.55 0.27 d = .11 Parent 0.72 0.21 0.75 0.18 0.58 0.26 Age groups 17‐29 (n = 148) 0.65a 0.24 η2 = .07 0.69a 0.18 η2 = .07 0.42a 0.26 η2 = .18 30‐44 (n = 95) 0.64a 0.25 0.70a 0.21 0.45a 0.27 45‐59 (n = 123) 0.71ab 0.21 0.78b 0.18 0.63b 0.24 60‐74 (n =149) 0.79b 0.17 0.80b 0.16 0.67b 0.23 75‐86 (n = 33) 0.76ab 0.18 0.75ab 0.17 0.68b 0.23

Note. Bold faced ES indicate substantial group differences, with p < .05. Means in given column with different subscribes differ significantly from other means according to Scheffe post hoc tests, with p < .05. Results on global self‐esteem implied substantial variations across age groups,

F (4, 543) = 10.34, p < .001, η2 = .07. Global self esteem was at its lowest in early and young adulthood. Those two age groups did not differ from each other (d = .04).

Mean levels were higher in the middle adulthood, indicated by a medium effect size in comparison to young adulthood (d = .30). In the following age group, self‐esteem again showed higher average scores. Individuals in late adulthood reported substan‐ tially higher global self‐esteem compared with early (d = .67), young (d = .70), and middle adulthood (d = .42). In the oldest age group between 75 and 86, again lower

14 When interpreting findings it is important to keep in mind that results are based on cross‐sectional data only. To illustrate the found trajectories, developmental descriptions such as “increase” or “de‐ crease” are only used to describe the found pathway across the eight age groups and not to indicate actual change within one individual. Results 96 mean levels were reported. Yet, mean levels indicate that only a minor difference in global self‐evaluation occurs between old age and late adulthood (d = .17).

0,9

0,8

0,7 Level 0,6 Mean 0,5

Global Trait Self‐esteem 0,4 Performance State Self‐esteem Social State Self‐esteem 0,3 Early Adulthood Young Adulthood Middle Adulthood Late Adulthood Old Age (17‐29) (30‐44) (45‐59) (60‐74) (75‐86)

Age Groups

Figure 5. Trajectories of Cross‐Sectional Data on Global Trait Self‐ Esteem as well as Performance and Social State Self‐Esteem across Seven Age Groups

The additionally assessed domain‐specific sub‐scales of state self‐esteem were utilized to expand the comprehension of the self‐esteem construct and its age‐related trajectory. Concerning both included domains, substantial variations were found across the composed age groups, performance: F (4, 543) = 10.30, p < .001, η2 = .07; so‐ cial: F (7, 543) = 28.99 p < .001, η2 = .18. Patterns of age‐associated patterns differed between performance and social state self‐esteem. Self‐evaluations regarding per‐ formance were very similar to the pattern of the global trait measure. Starting out at a slightly higher average than the global measure in early adulthood, mean‐levels of performance state self‐esteem remained firm throughout young adulthood (d = .05).

In contrast, middle adulthood was characterized by substantially higher mean scores of performance self‐esteem compared with early (d = .5) and young adults (d = .41).

Highest mean levels were reported in late adulthood. But they were only slightly above scores in (d = .12). Similar to the global measure, lower values oc‐ Results 97 curred in old age. Yet the difference was only medium compared to late adulthood (d

= .30).

Results of social state self‐esteem illustrated a different pattern across the em‐ pirically observed age groups. For one thing, mean levels in early adults’ social self‐ esteem happened to be the lowest, compared to performance and global self‐esteem measures. Actually, they showed to be well below the midpoint of the scale. Results indicated a high sensitivity to social contexts, feedback and appearances in early adulthood. For another thing, differences between age groups showed to be stronger in social self‐evaluations compared to the two other measures. More specifically, so‐ cial self‐esteem was only slightly higher in young adulthood (d = .11), but differed substantially with respect to levels of middle adulthood. Between age 45 and 59 indi‐ viduals reported significantly higher social self‐esteem scores than in early (d = .84) and young adulthood (d = .70). In late adulthood, ones more, higher means were re‐ ported (d = .17). In old age, social self‐esteem remained at the same level (d = .04). In this age group mean levels drew comparatively close to the performance and global self‐esteem scores. Thus, social self‐esteem obtained a fairly high level, specifically compared to the starting point in early adulthood.

In summary, results revealed that in early adulthood social state self‐esteem appeared to be well below levels of global and performance self‐esteem. Despite gradually higher mean scores in all three indicators, social self‐esteem showed strongest mean level differences. Thereby it drew closer to the global and perform‐ ance indicator throughout middle and late adulthood. Lower mean levels in old age appeared only in global and performance self‐esteem. Social self‐esteem scores re‐ mained at the same level in the two oldest age groups. Results opened up questions on possible changing relationships among the self‐esteem measures and potential age differential predictors of the diverse indicators. Results 98

6.1.1.2 Interplay of Self‐Esteem Measures

Bivariate correlations indicated that all three self‐esteem measures are substantially related to each other, global and performance self‐esteem: r = .67, p < .001; global and social self‐esteem: r = .58, p < .001; performance and social self‐esteem: r = .51, p < .001.

More detailed correlations supported the anticipated age specific variation of the as‐ sociation (Table 6.2).

Table 6.2 Age Specific Intercorrelations between Global Trait and Domain‐Specific State Self‐ Esteem Measures

Entire Sample Age Groups Early Young Middle Late Old Age Adulthood Adulthood Adulthood Adulthood (n=557) (n=148) (n=95) (n=123) (n=149) (n=33)

Global and Performance .67 .72 .64 .65 .60 .69 Self‐Esteem Global and Social .58 .51abc .60ab .66a .50bc .22c Self‐Esteem Performance and Social .51 .34a .54ab .57b .43ab .42ab Self‐Esteem

Note. Correlations in given row with different subscribes differ significantly from other correlations with p < .05 (DeCoster & Leistco, 2005). Substantial correlations are bold (p < .05).

The relationship between global and performance self‐esteem showed to be highly substantial and robust among all age groups. The high correlation implied that people’s global self‐evaluation was strongly connected to perceptions of capabil‐ ity and mastery. In contrast, the correlation between global and social self‐esteem varied considerably throughout the life span. Comparison of correlations accen‐ tuated an exceptionally high association in middle age, which became gradually lower and even insignificant in old age. That is, global self‐esteem mirrored funda‐ mental aspects of people’s social self‐evaluation only in their young and middle adulthood. From then on the association became smaller. Global and social self‐ esteem seemed to reflect different aspects of self‐evaluation in old age.

The interconnectedness of performance and social self‐esteem was different once more. Even that the overall tendency resembled the pattern of the correlation between global and social self‐esteem, distinctions showed to a lower extent. Only in Results 99 early adulthood associations between the two state measures were significantly low‐ er compared to middle adulthood. The then following smaller associations in late adulthood and old age were not as low as the one in early adulthood. Nevertheless, results indicated fewer overlap between the two domain‐specific measures compared to global and performance self‐esteem.

6.1.1.3 Possible Reasons and Functions of Developmental Trajectories

Differences in age specific variations suggested that global and domain‐specific self‐ esteem measures might serve diverse functions across time and life circumstances. A series of sequential regression analyses addressed the variation in developmental trajectories of self‐esteem measures across adulthood. Investigations were conducted to determine if demographic variables and indicators of positive functioning im‐ prove prediction of the diverse self‐esteem measures. Demographic characteristics were specified in the methodological section (cf., 5.1.1). In general, mean‐levels of life satisfaction and subjective health illustrated a content and fit sample (cf., 5.1.2.3).

A first sequence conducted control analyses with selected demographic va‐ riables. Regression analyses included interaction effects between age and gender, school qualification, partnership status, and parental status (all variables dummy coded). Predicting global self‐esteem, none of the main or interaction effects pro‐ duced an adjusted R2 greater than 1 percent, and none of them was statistically sub‐ stantial. The same pattern of results was true with regard to social state self‐esteem, and all but one demographic variable indicated no predictive value concerning per‐ formance self‐esteem. Pertaining to the latter, the only significant interaction effect occurred, namely between age and partnership status. Despite a relatively small amount of explained variance in the final step (adjusted R2 = .06), the interaction ef‐ fect improved the model substantially, FINC (1, 545) = 8.922, p < .01. Results proposed that older spouses reported higher levels of performance self‐esteem than older sin‐ gles (β = .17, p < .01). Possible three‐, four‐, and five‐way interactions (e.g., Age x Results 100

Gender x University degree) were tested as well. However, none of the terms with age improved the model substantially.

The second set of regression analyses addressed effects of life satisfaction and subjective health as a function of age on the three measures of self‐esteem. All possi‐ ble two‐ and three‐way interactions of the well‐being variables with age were tested.

Presented models contain only those interaction effects with age that showed to be substantial. Including gender in any of the regression analyses did not alter results.

Therefore, only the simple models are presented. Results are displayed in Table 6.3.

Predicting global self esteem, R showed to be different at the end of each step.

In the final step of adding the interaction effect between age and life satisfaction R2 improved substantially, FINC (1, 541) = 10.49, p < .001. Results indicated that about 58 percent of variability in global self esteem was predicted by age of participant, life satisfaction, subjective health, and the interaction effect between age and life satisfac‐ tion (adjusted R2 = .58). The pattern of results suggests that about 51 percent of varia‐ bility in global self‐esteem is explained by life satisfaction. Age contributed addition‐ al 5 percent; subjective health and the interaction term of age and life‐satisfaction each added 1 percent to the model.

1

0,9 Esteem ‐ 0,8 Self Low Age High Age 0,7 Global

of

0,6 Mean

0,5 Low Life Satisfaction High Life Satisfaction

Figure 6. Interaction Plotted From Life Satisfaction and Age Regressed on Global Self‐Esteem (Regarding Both Predictors ʹHighʹ and ʹLowʹ Is Based on +/‐ 1 SD)

Global self‐esteem showed to be higher in older and more satisfied people, as well as those that reported a better subjective health status. With regard to the inter‐ Results 101 pretation of the negative interaction effect, results were plotted according to Aiken and West (1991). Figure 6 indicates that age did not matter with regard to high life satisfaction. However, younger people who were low in life satisfaction also reported substantially lower levels of global self‐esteem compared to older people with low life satisfaction. This result suggested that the interconnectedness between people’s general evaluation of their life and of their self was lower in later adulthood com‐ pared to younger cohorts.

Table 6.3 Regression of Age and Two Indicators of Well‐Being on Trait and State Self‐Esteem Measures

Global Self‐Esteem Predictors B SE β t sr2 (incr.)

Age .001 .00 .09 3.23 .05 Life Satisfaction .73 .03 .68 21.94 .51 Subjective Health .14 .04 .09 ‐3.11 .01 Age x Life Satisfaction ‐.01 .00 ‐.09 ‐3.24 .01 R = .76; R2 = .58; adjusted R2 = .58 F (4, 541) = 185.26

State Self‐Esteem Performance B SE β t sr2 (incr.) Age .001 .00 .11 3.05 .05 Life Satisfaction .46 .03 .52 13.65 .29 Subjective Health .10 .04 .09 2.43 .01 R = .59; R2 = .35; adjusted R2 = .34 F (3, 543) = 96.01

State Self‐Esteem Social B SE β t sr2 (incr.) Age .005 .00 .33 9.08 .16 Life Satisfaction .44 .05 .33 8.48 .14 Subjective Health .22 .06 .13 3.45 .02 R = .56; R2 = .31; adjusted R2 = .31 F (3,543) = 82.99

Note. Both interaction effects have been substantial above and beyond all possible two‐ and three‐way interactions. There were no gender effects, thus simple models are presented.

What about the prediction of the state measures of self‐esteem? With regard to the performance subscale, results pointed out that all predictors explained substan‐ tial portions in the dependent variable. However, there were no significant interac‐ tion effects between age and one of the well‐being measures. The final model without any substantial interaction effects explained 34 percent of the variability in perfor‐ Results 102 mance state self‐esteem (adjusted R2 = .34). The examination of predictors showed the diversity in the amount of explained variability between them. The performance self‐ esteem was again largely predicted by the life satisfaction measure (29%). However, this was almost only half of the variability that this predictor explained in the global trait measure. All other measures were very similar in their characteristics. Age add‐ ed 5 percent, and subjective health predicted an additional 1 percent. Considering the pattern of coefficients, again older individuals, people that evaluated their lives more positively, and the ones that judged themselves to be in a good health state, reported higher levels of performance self‐esteem.

The last dependent variable was the social state self‐esteem. Similar to the above described results, all independent variables explained a substantial amount of variability. Yet again, no interaction effects with regard to age were substantial. The final model explained about 31 percent of the variability of social self‐esteem (ad‐ justed R2 = .31). Highest amounts of variability in social self‐esteem were explained by age (16%). The older people became, the less self‐conscious and social insecure they described themselves. Life satisfaction added another 14 percent, and subjective health predicted about 2 percent. Being older, more satisfied with their life, and in good shape, people reported higher levels of social self‐esteem.

Summing up results on self‐esteem measures across the life span, results indi‐ cated two important characteristics: First, global trait and domain‐specific state measures of self‐esteem all varied across the lifespan, indicating differences between mean level trajectories of people’s global, performance and social self‐evaluation.

Particularly in old age results supported discrepancies between global and social self‐esteem. Second, the variability in all three measures was explained by indicators of positive functioning. But patterns of results fluctuated, suggesting diversity in functions of specific aspects of people’s self‐evaluation. Results support hypothesis 1 on age effects. Results 103

6.1.2 Family Study: Effects of Family Situations on Self‐Esteem

The first part of this section represents a number of control analyses. Some features of family contexts such as motivational structures were expected to serve as differen‐ tial variables between family types. Analyses are utilized to validate the sampling strategy of the research project. This section is followed by results on effects of family situation testing hypothesis 2.

Characteristics of motivated and involuntary childless couples were compared to mothers and fathers. Traditional parents and parents of blended families were se‐ lected because represent family context that are characterized by more (traditional family) or less (blended family) stable and traditional structures. This age homoge‐ neous but structural diverse sample was particularly useful to test contextual effects on self‐evaluation. One advantage of the family study is that always both partners participated. Even though no explicit gender hypothesis is formulated, gender is treated as control variable. Substantial effects will be reported and discussed.

Validation Check 1: Differential Characteristics in Motivational Characteristics. Atti‐ tudes about children and their (anticipated) effects on people’s lives was an indicator expected to differentiate among motivated and involuntarily childless individuals.

Univariate analyses of variance showed that motivated childless individuals substan‐ tially differed from all three family situations regarding anticipated consequences of children on emotional stability, F (3, 338) = 108.292, p < .001, η2 = .49, and the amount of personal constraints, F (3, 338) = 13.114, p < .001, η2 = .10.

Concerning emotional stability, the average mean of blended family and tradi‐ tional parents was nearly four times higher than the one of motivated childless people. Involuntarily childless judged the achievement of emotional stability through having children to be significantly less important than those who are parents but substantially more important than motivated childless individuals. Traditional and blended family parents did not differ in this motivational characteristic. Anticipation of personal constraints was highest in the motivated childless group and substantial‐ ly lower in the remaining family types. No differences were detected among the Results 104 group of involuntary childless individuals, blended families and traditional parents.

Results regarding social acceptance indicated a trend, but no significant difference between family types, F (3, 338) = 2.636, p = .05, η2 = .02. Subsequent post‐hoc tests failed to produce considerable differences, although findings showed the expected pattern. The direction of results was supported by subsequently calculated t‐test be‐ tween the two types of childlessness, t (150) = ‐2.608, p < .05, d = .41. Involuntarily childless judged the perception of social acceptance of being parents as more impor‐ tant compared to motivated childless individuals. No differences occurred in percep‐ tions of extent of support, F (3, 338) = 1.815, n.s.

Validation Check 2: Differential Characteristics in Life Goals. In addition to motiva‐ tions towards children differences in pursued life goals were expected. A multiva‐ riate analysis of variance demonstrated substantial variations between family types for both life goal characteristics ʹagency orientedʹ and ʹcommunion orientedʹ. Com‐ pared to the group of motivated childless, traditional families, blended families and involuntary childless couples had higher levels of communion orientation in the pur‐ suit of life goals, F (3, 338) = 29.924, p < .001, η2 = .21. Motivated childless individuals were less oriented toward life goals that focus on family‐life and stability. In contrast, in agency oriented life goals traditional parents differed substantially from motivated childless individuals, F (3, 338) = 4.706, p < .01, η2 = .04. Involuntarily childless and blended families indicated no difference to any of the two other family types.

In sum, control analyses validated the sample structure. Motivated childless individuals were found to perceive parenthood as a major constrain. Parenthood was not associated with positive effects, such as social acceptance or emotional maturity.

They were also less communion, but more agency‐oriented in their life aspirations.

Differences were not as strong between involuntary childlessness and parenthood.

Instead, findings suggested that involuntary childless individuals were somehow tangling between desired and current status. Their expectations about parenthood were associated with more emotional stability, social acceptance, and to a lesser ex‐ tent, life constraints, thus being different from motivated childless, but similar to Results 105 parents. In addition, they showed to be communion oriented, but did not differ from motivated childless individuals with regard to agency‐orientation.

Effects of Family Situation on Global Self‐Esteem. Descriptive results of the family study indicated a fairly high global self‐esteem (M = .76, SD = .16). No effects oc‐ curred with respect to gender, t (340) = .36, n.s., d = .08, marital status, t (340) = ‐.17, n.s., d = 0, or parental status of participants, t (340) = ‐1.56, n.s., d = .18. Further con‐ siderations of the constancy within the family of origin illustrated again that there were no substantial differences in global self‐evaluation, t (340) = ‐.56, n.s., d = .06.

This was also true for the differentiated present family situations, F (3, 338) = 1.56, n.s., η2 = .01. Motivated or involuntary childless individuals and blended family or traditional parents did not diverge from each other with respect to their general self‐ esteem. In addition, no interaction effects occurred between variables.

Table 6.4 Means, Standard Deviations, and Effect Sizes (ES) of Global Trait and Domain‐ Specific State Self‐esteem Measures by Gender, Partnership Status, Parental Status, Family Constancy and Family Situations in the Follow‐Up Sample (N = 175)

Global Trait Performance State Social State

Self‐Esteem Self‐Esteem Self‐Esteem M SD ES M SD ES M SD ES

Whole sample 0.75 0.17 0.80 0.14 0.62 0.23 Gender Male 0.75 0.17 d = .06 0.80 0.14 d = .07 0.60 0.21 d = .18 Female 0.76 0.17 0.81 0.14 0.64 0.24 Partnership status Steady partnership 0.74 0.17 d = .12 0.78 0.15 d = .21 0.63 0.23 d = .09 Marriage 0.76 0.17 0.81 0.14 0.61 0.22 Parental status Childless 0.73 0.17 d = .24 0.78 0.15 d = .21 0.58 0.23 d = .31 Parent 0.77 0.17 0.81 0.14 0.65 0.22 Family Type Motivated Childless 0.73 0.15 η2 = .02 0.79 0.13 η2 = .02 0.65ab 0.25 η2 = .07 Involuntary Childless 0.72 0.18 0.78 0.16 0.52a 0.26 Blended Family 0.77 0.19 0.82 0.14 0.65ab 0.27 Traditional Family 0.77 0.16 0.81 0.13 0.66b 0.28 Family constancy High 0.75 0.18 d = .00 0.80 0.15 d = .00 0.59 0.22 d = .30 Low 0.75 0.16 0.80 0.12 0.67 0.30 Note. Bold faced ES indicate substantial group differences, with p < .05. Means in given column with different subscribes differ significantly from other means according to Scheffe post hoc tests, with p < .05. Results 106

Domain‐specific Evaluations. In contrast to the initial assessment, the follow‐up sample (1b, p. 71) of the family study also included domain‐specific indicators of self‐ esteem. Therefore, this smaller sample (N = 175) will be used to analyze contextual effects in social and performance self‐esteem. To complete the picture, global self‐ esteem was simultaneously analyzed as well. Descriptive results are shown in Table

6.4. Similar to previous findings, no gender or partnership status effects occurred in respect of all three indicators. As hypothesized, closer considerations of social state self‐esteem showed a number of substantial context effects. At first, findings on the parental status emphasized that parents have a higher mean level in their social self‐ esteem compared to childless people, t (168) = ‐1.99, p < .05, d = .31. To disentangle the effect of childlessness a univariate analysis of variance with family type was con‐ ducted. Results specified that only the involuntary childless but not the motivated childless individuals showed a lower social self‐esteem, i.e. a higher social sensitivity for relational feedback (Table 6.4). In addition to the effect of the current family situa‐ tion, the social self‐esteem indicator detected a substantial difference between par‐ ticipants from high versus low family constancy backgrounds, t (169) = ‐2.44, p < .02, d = .30.

Figure 7. Differences in Social Self‐Esteem by Family Constancy and Gender

Low family constancy was associated with higher social self‐esteem, i.e. less concern for the social impression on others. Additional control analyses revealed a Results 107 substantial interaction effect between family constancy and gender. Only women with a family background characterized by low stability showed substantially higher levels of social self esteem compared to all remaining groups, gender: F (1, 166) =

4.94, p < .05, η2 = .03; family constancy: F (1, 166) = 4.73, p < .05, η2 = .03; gender x fam‐ ily constancy: F (1, 166) = 13.16, p < .001, η2 = .07 (cf., Figure 7). This gender effect was not anticipated. It will be regarded in more detail in the discussion part.

6.1.3 Summary: Self‐Esteem across Life Span and Family Situations

Results of the online study support differences in age‐related patterns of self‐esteem measures. The differentiation of global and social indicators shows in cross‐sectional pathways across the life span as well as in correlative and regression analyses, par‐ ticularly with respect to late adulthood and old age. In contrast, global and perform‐ ance self‐esteem are very similar. Hypothesis 1 is supported.

Analyses of the family study indicate a higher sensitivity of the social self‐ esteem indicator for contextual effects. In contrast to global and performance meas‐ ures, only social self‐esteem detected differences between family situations: Involun‐ tary childless individuals show lower and women with a low family constancy back‐ ground illustrate higher social self‐esteem. Results confirm hypothesis 2.

6.2 Social Networks and Relationship Regulation across Life Span and Family

Situations

The second results section focuses on personal networks and relationship regulation processes. The result presentation starts out with the online study. Analyses focus on age‐related patterns of personal networks and their structural aspects, such as size and composition. This examination is followed by multi‐level analyses that deal with the quality and regulative aspects of relationships in different age groups. In a second step, the family study is addressed. Again, the investigation focuses on struc‐ tural and qualitative characteristics, observing contextual effects of family situation. Results 108

6.2.1 Online Study: Effects of Age on Social Networks and Relationship Regulation

This section explores age‐specific patterns of ego‐centered social networks and rela‐ tionship regulation processes. Statistical analyses differentiate between five age groups: early adulthood, young adulthood, middle adulthood, late adulthood, and old age.

6.2.1.1 Structure of Ego‐Centered Social Networks across Age Groups

Ego‐centered social networks contained about 11 to 12 relationship partners (M =

11.5, SD = 7.8). The cross‐sectional data did not illustrate an age‐related variation of network size. Although the number of relationship partners was lower in old age, the univariate analysis of variance indicated no significant differences, F (47, 552) = 1.08, n.s., η2 = .008 (Table 6.5).

More detailed analyses pointed out an age group specific structure or compo‐ sition in ego‐centered social networks. Conducting a sequence of univariate analyses of variance, results indicated differences in characteristics with regard to kin (par‐ ents: η2 = .51; children: η2 = .08; close kin: η2 = .03; distant kin: η2 = .03), elective kin

(partner: η2 = .07; in‐law/step‐family: η2 = .04), and non‐kin relationships (colleagues:

η2 = .04; acquaintances: η2 = .05). Table 6.5 shows group means and standard devia‐ tions as well as F ratios for the total network and the specific relationship types. Only two kinds of social ties remained constant in number throughout: siblings (η2 = .01) and old friends (η2 = .02). Both types may be considered to play a significant role in people’s journey through adult life.

Ego‐centered networks in early adulthood were more likely to include parents and romantic partners. In contrast, they were less likely to include in‐law or step‐ family members. Also the subsumed type of acquaintances appeared less often in this group’s social networks. In young adulthood, results indicated a high proportion of participants to name their parents, their romantic partner, and about two col‐ leagues to be part of the network. Yet, the likelihood that personal networks contain a child, or biological, and elective kin was very low. In fact, the group’s affiliations Results 109 seemed to be focused on their core family of origin and their occupational connec‐ tions, but not on building their own family.

Table 6.5 Descriptive Statistics of Ego‐Centered Social Networks by Age Groups

Age groups Early Young Middle Late Old Age

Adulthood Adulthood Adulthood Adulthood

(age 17‐29) (age 30‐44) (age 45‐59) (age 60‐74) (age 75‐86)

M (SD) M (SD) M (SD) M (SD) M (SD) η2

Network Size 11.89 (7.86) 11.02 (7.09) 11.49 (7.68) 11.91 (8.39) 9.09 (6.60) .01 r = .5 Parents M 1.53a (0.67) 1.31a (0.74) 0.67b (0.72) 0.09c (0.28) 0.00c (0.00) .51 Siblings M 0.91 (0.84) 0.94 (0.90) 0.90 (1.12) 0.69 (1.02) 0.67 (1.08) .01 Children M 0.10ab (0.41) 0.02a (0.20) 0.07ab (0.34) 0.29b (0.76) 0.73c (1.40) .08 r = .25 Close Kin M 1.30ab (1.82) 1.10a (1.46) 1.54ab (1.59) 1.91b (1.79) 1.64ab (2.09) .03 r ≤ .125 Distant Kin M 0.42ab (1.15) 0.15a (0.49) 0.27ab (0.63) 0.57b (1.12) 0.52ab (1.06) .03 r = .0 Partner M 0.57a (0.56) 0.64a (0.56) 0.46ab (0.55) 0.32bc (0.47) 0.12c (0.33) .07 In‐Laws/Step‐F M 0.87a (1.35) 0.96a (1.45) 1.36ab (2.09) 1.79b (2.46) 1.52ab (1.91) .04 Colleagues M 1.45ab (2.15) 1.77a (2.17) 1.21ab (1.67) 1.00b (1.38) 0.39b (0.83) .04 Acquaintances M 1.75a (2.60) 1.32a (1.60) 2.46ab (3.49) 3.07b (3.63) 1.73ab (2.18) .05 Old Friends M 3.00 (3.44) 2.80 (3.01) 2.55 (2.69) 2.18 (2.85) 1.79 (2.36) .02

Note. Bold faced η2 produced substantial F ratios, with p < .05. Means in given row with different sub‐ scribes differ significantly from other means according to Scheffe post‐hoc tests with p < .05.

Increasing age reduced the likelihood of parents in social networks. Further‐ more, slight changes became noticeable regarding the mean‐number of children and kin as relationship partners, which increased, and simultaneously, the probability to name a partner, or colleagues decreased. Indeed, social networks of middle adult‐ hood appeared to somehow represent a pattern, not being all that different from the younger or the older age groups.

In late adulthood, changes in network composition were particularly evident in numbers of kin and elective kin. Number of included children, close, and distant kin, but also in‐law and step‐family members showed to be higher compared to younger age groups. Furthermore, likelihood to list colleagues was lower and to name acquaintances was higher. Networks seemed to become more characterized by family relationships in late adulthood. In old age, social networks in general became smaller, yet, not differing substantially from the other groups. Change in size was Results 110 specifically due to a continued lower mean level of parents, spouses, and colleagues.

At the same time, the number of children included into social networks was smaller again. Importantly, the amount of named close and distant kin as well as non‐genetic family members remained at a comparatively high level. Thus, results indicated the importance of people’s biological and non‐biological family in late life.

In summary, results on personal networks and their composition indicated a relatively stable number of affiliations that accompany people throughout their adult life. However, a substantial amount of fluctuation appeared in types of included so‐ cial relationships. Indeed, age group specific networks seemed to create a picture of current and arising developmental tasks, goals and demands.

6.2.1.2 Relationship Regulation Characteristics in Ego‐Centered Social Networks across

Age Groups

Based on the framework of lifelong relationship regulation (cf., 3.2), descriptors of relationship quality are understood as regulative mechanisms. These mechanisms are expected to vary intra‐relationally, intra‐individually and inter‐individually. The current cross‐sectional data did not provide information on intra‐relational change.

However, the ego‐centered social network structure has the potential to focus on in‐ ter‐individual differences in intra‐individual variations of closeness and reciprocity perceptions. A basic relationship measurement model is introduced, including rela‐ tionship specific perceptions of emotional closeness, reciprocity, and degree of ge‐ netic relatedness, and then age group specific effects of relationship regulation are modelled. Previously, descriptive statistics of perceived closeness and reciprocity are presented.

Emotional Closeness and Perceptions of Reciprocity. An analysis of variance was conducted on perceived closeness in personal networks. Independent variables were Results 111 relationship type (nine different relationships15) and age group (early, young, middle, late adulthood, and old age). Perceptions of emotional closeness showed substantial variation across relationship types, F (8, 259916) = 39.26, p < .001, η2 = .11, and age groups, F (4, 2599) = 15.96, p < .001, η2 = .02. Additionally, the interaction between the two independent variables showed a significant effect, F (31, 2599) = 3.57, p < .001, η2

= .04 (complete descriptive statistics in Appendix B, Table B.1).

Two major patterns were found. At first, emotional closeness differentiated across relationship types, mainly subdividing the entire group of affiliations into three parts: (1) romantic partners as the closest relationship, (2) core family members, close kin and old friends as a second still very close group, and (3) all remaining elec‐ tive and non‐kin ties as the least close relationships. Second, results outlined that in late adulthood and old age named relationships were on average described as closer than in younger age groups. Results remained stable even after controlling for effects of gender and parental status.

Considerations of reciprocity dealt with the two distinct variables of perceived over‐benefit and perceived under‐benefit in social relationships. With regard to both variables, analyses examined effects of relationship type (9) and age group (5), as well as the interaction term between these two variables. Perceptions of under‐ benefit were substantially affected by the evaluated relationship type, F (8, 2599) =

4.01, p < .001, η2 = .01, age group, F (4, 2599) = 13.98, p < .001, η2 = .02, and their interac‐ tion, F (31, 2599) = 2.18, p < .001, η2 = .03. Results on over‐benefit were smaller, but still significant, relationship type: F (8, 2599) = 2.83, p < .01, η2 = .01; age group: F (4, 2599)

15 The relationship type ʹchildʹ was excluded from the following analyses, since the number of named children in personal networks was very low. In that, variations across only nine groups were exam‐ ined: parents, siblings, close kin, distant kin, partner, in‐law‐/step family, colleagues, acquaintances, and old friends.

16 Different number of degrees of freedom is due to the aggregation of the network data to avoid an artificial inflation of effects. Results 112

= 6.56, p < .001, η2 = .01; relationship type × age group: F (31, 2599) = 1.80, p < .01, η2 =

.02 (complete descriptive statistics in Appendix B, Table B.1).

Observed patterns showed to be particularly interesting with regard to age differential variations of perceived reciprocity. The general consideration of mean level variation illustrated that in middle adulthood, late adulthood, and even in old age people reported more under‐beneficial ties, where targets invest more than the other relationship partner. This pattern did not occur in romantic partnerships and to close kin. These two relationship types illustrated higher scores of under‐benefit throughout adulthood but lower levels in old age. Furthermore, the three oldest age groups showed the least over‐beneficial relationships. That is, the older people are the fewer social ties are regarded as beneficial. This was particularly true in relation‐ ships to parents, distant kin, siblings, and in‐law and step family members. Relation‐ ships to close kin were regarded as beneficial in younger and older age groups, but this was not true with respect to middle adulthood.

In summary, examinations of relationship qualities across personal networks suggest that there was a substantial variation in terms of both perceptions of close‐ ness and of reciprocity. This variation occurred regarding relationship types and age groups, suggesting that relationship regulation characteristics were actively adjusted to current personal and environmental situations. The upcoming section addresses relationship perceptions as intra‐individual regulatory mechanism in social net‐ works.

Relationship Measurement Model. The following model captures the perception of closeness as a function of the perception of reciprocity and the degree of genetic relatedness to each relationship partner. With respect to reciprocity, the two aspects of over‐ and under‐benefit are included as single dummy coded variables. Further‐ more, the interaction term of genetic relatedness and over‐benefit and under‐benefit, respectively, were included in the last step of building up the model. Since for all predictor variables zero is a meaningful category, they were entered uncentered into the multilevel equation (cf., Nezlek, Schröder, & Schütze, 2006). Results 113

Stepwise inclusion of predictor variables allowed keeping track of distinct model improvements. At first, genetic relatedness was included leading to an im‐ proved model and explaining 7% of the variance proportion compared to the uncon‐ ditional model, TRd = 225.00, p < .001. Adding the two dummy coded variables as indicators for over‐ and under‐benefit, again, the model enhanced substantially, TRd

= 281.01, p < .001, with 12% additional variance explained. The last step included the two interaction effects between the genetic relatedness indicator and the two reci‐ procity measures. These two interaction effects did not improve the model substan‐ tially, TRd = 17.57, n.s., but explained 2% additional variance. For theoretical reasons, the two indicators will be kept in the model to explore possible age group efects. Al‐ together, predictors explained about 21% of perceived closeness.

Perceptions of emotional closeness across social network partners were sub‐ stantially influenced by the degree people were genetically related with each other and by perceptions of reciprocity. First, there was a strong positive association be‐ tween closeness and genetic relatedness (γ10 = .75, p < .001). Second, irrespective of the direction of imbalance, closeness and perceived imbalance in non‐kin were nega‐ tively correlated (under‐benefit: γ20 = ‐.11, p < .05; over‐benefit: γ30 = ‐.10, p < .05).

Third, a negative effect showed with respect to the interaction term between percep‐ tions of under‐benefit and genetic relatedness (γ40 = ‐.29, p < .05). In contrast, the inte‐ raction effect of under‐benefit and degree of genetic relatedness was not significant

(γ50 = ‐.19, n.s.). Thus, the online sample shows a strong kinship orientation. Percep‐ tions of imbalance are associated with lower emotional closeness. This was true with respect to kin and non‐kin.

Effects of Age. To examine age group specific relationship regulation parame‐ ters, the measurement model was conducted with five alternating dummy coded va‐ riables. In each case, one age group served as reference category (i.e., this dummy variable was not included). This course of action produced age specific estimates (cf.,

Table 6.6) and tested for substantial differences between groups. Age group variable improved the model substantially, TRd = 86.01, p < .001. Results support the assump‐ Results 114 tion that not only relationship perceptions but also regulative strategies vary across the life span.

Findings with respect to the intercept indicated that in old age perceptions of closeness to reciprocal non‐kin relationship partners was substantially higher (γ00 =

.56, p < .05) compared to any other group. In contrast, perceptions of imbalance in non‐kin ties, irrespective of feelings of over‐ or under‐benefit, had relatively little effects on emotional closeness across age groups. Only in late adulthood, people re‐ ported lower amounts of closeness in under‐beneficial relationships (γ20 = ‐.17, p <

.05) and by that differed substantially from adults of middle adulthood (γ20 = .03, n.s.). The latter reported substantially less closeness in over‐beneficial non‐kin ties

(γ30 = ‐.19).

Table 6.6 Parameter Estimates (PE) of Relationship Measurement Two‐Level Regression Model on Perceived Closeness to Social Network Partners in Five Different Age Groups

Early Young Middle Late Old

Adulthood Adulthood Adulthood Adulthood Age PE SE PE SE PE SE PE SE PE SE

Fixed Effects Intercept 0.09a 0.05 ‐0.00a 0.08 ‐0.02a 0.07 0.05a 0.05 0.56b 0.11 Slopes γ10 Degree of genetic relatedness 0.46a 0.19 0.68ac 0.26 0.66ac 0.25 1.38b 0.20 1.32bc 0.33 γ20 Perceived under‐benefit ‐0.06ab 0.06 ‐0.14ab 0.10 0.03b 0.08 ‐0.17a 0.06 ‐0.33ab 0.17 γ30 Perceived over‐benefit ‐0.11 0.08 ‐0.02 0.10 ‐0.19 0.09 ‐0.04 0.07 ‐0.37 0.23 γ40 Under‐benefit x genetic relatedness ‐0.54 0.27 ‐0.41 0.36 ‐0.25 0.31 ‐0.30 0.28 ‐0.33 0.44 γ50 Over‐benefit x genetic relatedness 0.09a 0.27 ‐0.77b 0.32 ‐0.23ab 0.41 0.60a 0.43 ‐0.24ab 0.91

Random Effects 02 Intercept 0.25 0.02 12 Degree of genetic relatedness 1.03 0.31 22 Perceived under‐benefit 0.22 0.05 32 Perceived over‐benefit 0.20 0.05 42 Under‐benefit x genetic relatedness 1.47 0.63 52 Over‐benefit x genetic relatedness 0.24 0.66 2 0.50 0.02

Model fit LL ‐7653.85 CF 1.07 FP 52 TRd1 86.01 p < .001 Note. Substantial parameters are in bold face. Parameter estimates with different subscribes differ sub‐ stantially at p < .05. 1Comparisions were made to the final measurement model.

Results 115

With respect to kin differentiation, results on all five family stages supported the positive association between emotional closeness and genetic relatedness. Across adulthood, people illustrated a substantial kinship preference. Findings in late adult‐ hood (γ10 = 1.38, p < .05) and old age (γ10 = 1.32, p < .05) indicated an even stronger kinship orientation compared to the younger cohorts. The interaction term between under‐benefit and genetic relatedness was only substantially negative related to closeness perceptions in early adulthood (γ40 = ‐.54, p < .05). However, no significant difference between age groups occurred. With respect to the interaction between over‐benefit and genetic relatedness, young adults reported substantially less close‐ ness (γ50 = ‐.77) than early (γ50 = .09) and late adults (γ50 = .60) did.

Comparison of relationship regulation mechanisms suggests that across age groups people show differential patterns of regulative strategies. Specifically, in late adulthood and old age people feel closer to their kin and indicate higher kin prefe‐ rence, thus distinguishing kin and non‐kin substantially more than younger cohorts.

With respect to reciprocity, findings did not produce a clear pattern but accentuate the general negative effect of imbalance on closeness perceptions in non‐kin as well as in kin relationships.

6.2.2 Family Study: Effects of Family Situation on Social Relationships

Upcoming paragraphs will discuss social networks and relationship regulation processes in the sample of the family study. First, analyses focus on ego‐centered networks and their structural aspects. This examination is followed by descriptive and multi‐level analyses that deal with the quality of relationships. Focus of analyses is the effects of family situation, thus the comparison between motivated and invo‐ luntary childlessness, and parents of blended or traditional families.

6.2.2.1 Structure of Ego‐Centered Social Networks across Family Situations

Composition of Ego‐Centered Social Networks. It was hypothesized that motivated and involuntary childless individuals differ from individuals who are parents in compo‐ Results 116 sitional features of their ego‐centered social networks. A univariate analysis of vari‐ ance on network size suggested substantial differences in number of named relation‐ ship partners, F (3, 338) = 6.28, p < .001, η2 = .05 (see Table 6.7). Family types with chil‐ dren, both blended and traditional, reported equal numbers of network members (M

= 14.82, SD = 8.47; M = 14.82, SD = 8.77, respectively). They both differed substantially from motivated childless individuals, who reported least numbers of social affiliates

(M = 10.23, SD = 6.33), but not from involuntary childless people (M = 12.94, SD =

8.23). This pattern was somehow surprising, because previous research associated involuntary childlessness with alienation and decreasing numbers of network mem‐ bers (Houghton & Houghton, 1987, Sabatelli et al., 1988). On the contrary, motivated childless individuals have not been expected to have smaller networks. These differ‐ ences suggested the need for more detailed analyses with regard to network compo‐ sitions and characteristics.

Thus, variations of quantity among eleven relationship categories were ex‐ amined. Again, a univariate analysis of variance was conducted for differing num‐ bers of generated affiliates in each relationship category among family types. Results are displayed in Table 6.7. Numbers of generated kin and elective kin in particular showed substantial variation.

Concentrating on genetic kin, motivated childless individuals least often inte‐ grated their own parents as part of their social networks (M = 1.12, SD = .84), differ‐ ing from the group of involuntary childless individuals (M = 1.46, SD = .74; F (3, 338)

= 2.60, p = .05, η2 = .02). Numbers of listed parents in social networks of blended and traditional family individuals did not differ substantial from any childless individu‐ als. With regard to the number of named children, parents and childless individuals differed clearly, F (3, 338) = 91.86, p < .001, η2 = .45. Lifelong childlessness of both partners was a sample condition for motivated and involuntary childless couples.

Traditional and blended family parents very much resembled each other in this con‐ text (M = 1.59, SD = 1.10 and M = 1.56, SD = 1.24, respectively). There was also a high similarity regarding number of named close kin such as grand parents or nephews. Results 117

However, findings on distant kin showed substantial variation, F (3, 338) = 2.79, p <

.05, η2 = .02. Even though post‐hoc tests failed to point out substantial differences be‐ tween family types, separately conducted t‐tests supported the descriptive pattern.

Motivated childless participants named least distant kin in their social networks (M =

.23, SD = .62), compared to all three remaining family groups, involuntary childless: t

(150) = ‐2.598, p < .05, d = .43; blended family: t (174) = ‐2.816, p < .01, d = .42; tradition‐ al family: t (176) = ‐2.735, p < .01, d = .40.

Table 6.7 Differences between Familial Situations in Average Number of Overall Size of Social Networks as well as in Specific Relationship Types

Motivated Involuntary Blended Traditional

Childless Childless Family Family (n = 82) (n = 70) (n = 94) (n = 96) M (SD) M (SD) M (SD) M (SD) η2

Network Size 10.23a (6.33) 12.94ab (8.23) 14.82b (8.47) 14.82b (8.77) .05 r = .5 Parents 1.12a (0.84) 1.46b (0.74) 1.27ab (0.75) 1.32ab (0.69) .02 Siblings 0.74 (0.86) 0.86 (0.87) 1.09 (0.95) 0.97 (0.90) .02 Children 0.0a (0.00) 0.0a (0.00) 1.56b (1.24) 1.59b (1.10) .45 r = .25 Close kin 0.78 (0.94) 1.17 (1.69) 1.18 (1.44) 1.16 (1.53) 01 r = .125 Distant kin 0.23 (0.62) 0.60 (1.04) 0.61 (1.11) 0.61 (1.20) .02 r = 0 Partner 0.78 (0.42) 0.73 (0.54) 0.82 (0.41) 0.77 (0.42) .01 Step‐kin 0.11a (0.39) 0.10a (0.46) 0.51b (0.80) 0.07a (0.26) .12 In‐laws 1.00a (1.84) 2.07b (2.81) 1.91ab (2.15) 2.16b (2.58) .04 Colleagues 1.57 (2.05) 2.04 (2.20) 1.56 (2.00) 1.49 (1.95) .01 Acquaintances 1.66 (1.83) 1.93 (2.47) 2.11 (2.96) 2.70 (3.09) .02 Old friends 2.23 (3.43) 1.99 (2.37) 2.20 (2.32) 1.98 (2.48) .00

Note. Bold faced η2 produced substantial F ratios, with p < .05. Means in given row with different sub‐ scribes differ significantly from other means according to Scheffe post hoc tests with p < .05.

With regard to elective kin, about 75 to 80 percent of participants named their romantic partner in their social network. A finding that was highly similar among family types, F (3, 338) = .566, n.s, as was the number of old friends, F (3, 338) = .219, n.s. Compositional differences were found with regard to number of in‐law relation‐ ships, F (3, 338) = 4.26, p < .01, η2 = .04, and of step‐family members, F (3, 338) = 14.6, p

< .001, η2 = .11. Concerning the latter group, expectedly, blended family members most frequently included step‐kin in their networks (M = .51, SD = .80). All other re‐ Results 118 maining family contexts generated only very few step‐relationship partners (moti‐ vated childless: M =.11, SD = .39; involuntary childless: M = .10, SD = .46; traditional families: M = .07, SD = .26). This difference may be attributed to family structures themselves. This is a different matter in the context of included in‐law relationships.

Even that motivated childless and blended parents have been reported to be married less frequently, the duration of partnership did not differ at least compared to invo‐ luntary childless couples. Nevertheless, ego‐centered networks of motivated child‐ less individuals contained on average only one in‐law relation (M = 1.00, SD = 1.84).

Those of involuntary childless and traditional parents enclosed about two on average

(involuntary childless: M = 2.07, SD = 2.81; traditional family: M = 2.16, SD = 2.58).

Blended family parents did not differ substantially from any group (M = 1.91, SD =

2.15). No differences were found in the average number of non‐kin relationship part‐ ners, colleagues: F (3, 338) = 1.159, n.s.; acquaintances: F (3, 338) = 2.431, n.s. There were no gender or interaction effects between gender and family type.

Structure of Relationships. A respective interest concerned the specific features of relationship partners such as their marital and parental status. Furthermore, gend‐ er constellations, geographical proximity, and frequency of contact were included into analyses, since these are also important structural variables (Hollstein, 2001;

Neyer & Lang, 2003). Analyses are conducted with cross‐tabs and univariate analyses of variance. With regard to the first method, overall difference tests are based on Chi‐

Square statistics and between family types on adjusted standardized residuals.

Results suggest heterogeneous patterns with regard to gender, marital and pa‐ rental status (Table 6.8). Motivated childless individuals name more male and less female relationship partners, and they have less relationship partners that are mar‐ ried or have kids themselves. This is particularly evident in contrast to traditional parents, where the patterns were found to be vice versa throughout. Involuntary childless individuals and blended family parents were somewhere in‐between those two groups. Only with regard to the number of married relationship partners, par‐ ents in blended families were more similar to motivated childless individuals. Sur‐ Results 119 prisingly, involuntary childless individuals were not less likely to have parents in their social networks. Results on additional characteristics, such as contact frequency, produced no clear patterns and effect sizes were very small (cf., Appendix C, Table

C.1).

Table 6.8 Differences between Family Types in Characteristics of Social Network Partners

Motivated Involuntary Blended Traditional

Childless Childless Family Family

N % N % N % N % Χ2 (3)

Gender Male 413(+) 49 410 45 653 47 608(‐) 43 10.30, p<.05 Female 425(‐) 51 496 55 730 53 811(+) 57 Marital Status1 Not married 474(+) 57 413 46 685(+) 51 570(‐) 40 61.10, p<.001 Married 365(‐) 43 493 54 699(‐) 49 849(+) 60 Parental Status2 Not a Parent 407(+) 49 341 38 478 35 420(‐) 30 84.50, p<.001 Parent 432(‐) 51 565 62 906 65 1000(+) 70

Note. (‐/+) Indicator that the value contributes significantly (positive or negative) to the overall chi‐ square value with (absolute value of) adjusted standardized residuals being greater than 2.00. 1Effects remained robust if controlled for participant’s marital status. 2Effects remained robust if controlled for participant’s number of biological children.

Three instances seem particularly important when summarizing findings on structural characteristics: First, motivated childless individuals reported substantially smaller networks. In particular, they named less kin and elective kin, but did not ac‐ cumulate more non‐kin relationships. Second, motivated childless individuals are less likely to have females, married individuals and parents as relationship partners.

And third, involuntary childless individuals give the impression to tangle between the characteristics of motivated childless individuals and parents, frequently not be‐ ing different from the one or the other side. Thus the following concern is if, in addi‐ tion to structural variations, there are also differences between family types with re‐ gard to relationship qualities and regulation strategies. Results 120

6.2.2.2 Relationship Regulation Characteristics in Ego‐Centered Social Networks across

Family Situations

Similar to the analysis conducted with the online study, emotional closeness, percep‐ tions of over‐ and under‐benefit, and the degree of genetic relatedness to each rela‐ tionship partner are integrated into a relationship measurement model. After explor‐ ing general patterns, family specific analyses are conducted. Apriori, descriptive analyses are reported to illustrate relationship and family specific perceptions of closeness and reciprocity.

Emotional Closeness and Perceptions of Reciprocity. Perceptions of emotional closeness varied substantially across eleven selected relationship types, F (10, 2084) =

124.82, p < .001, η2 = .3717. The reported overall pattern was similar to the online study.

All family situations pointed out that they maintained closest relationships to roman‐ tic partners. Parents reported in addition the child‐parent tie as exceptionally close.

Relationships to selected kin and elective kin, i.e. parents, siblings, and step‐kin, as well as old friends, were also described to be closer than the overall mean. Close and distant kin are another subset, followed by the relationships to in‐laws as well as the two remaining non‐kin groups of colleagues and acquaintances as least close ties.

An analysis of variance was conducted on perceived closeness with relation‐ ship type (eleven different relationships) and family type (motivated childless, invo‐ luntary childless, blended family, and traditional family) as independent variables.

Family contexts demonstrated only a few substantial differences in perceptions of closeness. Findings pointed out differences in perceived closeness regarding siblings,

F (3, 222) = 319, p < .05, η2 = .04, close kin, F (3, 178) = 3.32, p < .05, η2 = .05, and step‐ kin, F (3, 46) = 10.53, p < .001, η2 = .41. According to Scheffé post‐hoc tests, traditional family members (M = ‐.17, SD = .84) described relationships to siblings significantly less close compared to involuntary childless individuals (M = .28, SD = .58). The re‐

17 This and upcoming analyses of variance on perceived closeness and reciprocity have been con‐ ducted with all named relationship partners aggregated across eleven relationship dimensions, to avoid artificial inflation of effects. Results 121 maining familial contexts did not differ from both groups. With reference to close kin relations, Scheffe post‐hoc tests failed to show substantial effects between family groups. However, mean patterns demonstrated motivated childless individuals to feel closest (M = .02, SD = .88), in particular compared to traditional parents (M = ‐.42,

SD = .75). Considering step‐relations blended family (M = .45, SD = .72) and moti‐ vated childless individuals (M = .39, SD = .52) reported higher amounts of perceived closeness in contrast to traditional parents (M = ‐1.10, SD = .54) according to Scheffe post‐hoc tests. This last finding may be interpretable only to a limited extent, because numbers of generated step‐kin were very low except for blended family members.

Findings on perceived reciprocity picture the diversity between relationship specific observations of balance vs. under‐benefit vs. over‐benefit. Descriptive statis‐ tics (cf., Table 5.9) already indicated that the majority of relationships are reciprocal.

Otherwise, people perceive more relationships as costly then beneficial. The con‐ ducted analyses of variance on under‐benefit pointed out a substantial main effect of relationship type, F (10, 2053) = 15.02, p < .001, η2 = .07, but not on family situation, F

(3, 2053) = 1.39, n.s., η2 = .00, and no substantial interaction effect occured, F (28, 2053)

= 1.38, n.s., η2 = .02. Results for over‐benefit showed to be very similar in pattern. On‐ ly a substantial main effect occurred regarding relationship type, relationship type: F

(10, 2053) = 10.89, p < .001, η2 = .05; family situation: F (3, 2053) = 1.40, n.s., η2 = .00; re‐ lationship type × family situation: F (28, 2053) = 1.01, n.s., η2 = .01. Effects were stable if controlled for gender of participant.

Perceptions of under‐benefit were commonly occurring in kin and elective kin ties. Relationships to children were reported as costliest, followed by step‐kin, sibl‐ ings, in‐laws, and close kin. Romantic partner relationships were named least often.

Perceived over‐benefit was most likely in relationship to parents. Furthermore, rela‐ tionships to elective kin and non‐kin were described as beneficial as well. The rela‐ tionship to the spouse was the only type of affiliation that was more often described as being over‐beneficial. Results 122

Outlining results of perceived closeness and reciprocity, findings demonstrate that both qualities differ among relationship dimensions. However, family situations showed only few differences in relationship perceptions across types of affiliation.

This was particularly true pertaining to reciprocity perceptions.

Relationship Measurement Model. The following model integrates perceptions of emotional closeness as a function of the degree of genetic relatedness and percep‐ tions of over‐ and under‐benefit across social network partners. Again, predictor va‐ riables were entered uncentered (cf., Nezlek, Schröder, & Schütze, 2006). By includ‐ ing all level‐one key variables stepwise, a conditional multilevel model was built up predicting closeness perceptions across ego‐centered social networks. As first predic‐ tor the degree of genetic relatedness accounted for an additional variance proportion of 10%, TRd = 271.80, p < .001, compared to the unconditional closeness model. The two dummy coded reciprocity variables, indicating perceptions of over‐ and under‐ benefit, explained additional 11%, TRd = 284.36, p < .001, and the remaining interac‐ tion effects lead to further explanations of 4% and improve the model substantially,

TRd = 57.18, p < .001. The final model accounts for 23 percent of variability in per‐ ceived closeness.

Fixed effects indicate that the perception of closeness among social network partners was substantially associated with the degree of genetic relatedness and the perceived degree of reciprocity. Genetic relatedness was associated with higher amounts of closeness in relationships (γ10 = .85, p < .05). With regard to relationships with non‐kin, perceptions of under‐ but not of over‐benefit were connected to lower amounts of perceived closeness (γ20 = ‐.34, p < .05). Furthermore, substantial parame‐ ter estimates for both interaction terms illustrate: (a) perceptions of under‐benefit and genetic relatedness are positively associated with closeness (γ40 = .48, p < .05), and (b) perceptions of over‐benefit and genetic relatedness are negatively linked to perceived closeness (γ50 = ‐.37, p < .05).

To elucidate these effects, Figure 8 shows variations in closeness perception for social relationship partners with different degrees of genetic relatedness as a function Results 123 of perceived reciprocity, under‐ and over benefit. Overall, trajectories of the three slopes confirmed the positive association between perceived closeness and genetic relatedness as previously reported by Neyer and Lang (2003). People felt closer to biological kin than to non‐kin. Hence, they showed clear patterns of kinship orienta‐ tion. The monitoring of reciprocity across different social network partners appeared to affect this relationship substantially. For non‐kin social interactions (i.e., degree of genetic relatedness is ʹ0ʹ), high‐benefit relationships did not differ from reciprocal relationships in closeness. Yet, high‐costs in non‐kin relations seemed to decrease closeness perception. However, this cost‐benefit effect changes with regard to kin relationships. Trajectories imply that despite of high‐costs individuals felt close to kin. Yet, with regard to high‐benefits the effect of kinship orientation was less strong.

The slope on over‐beneficial relationships showed the least steep trend across bio‐ logical network partners.

0

Figure 8. Mean Perceived Emotional Closeness as a Function of Genetic Relatedness and Perceived Under‐ and Over‐Benefit in Social Networks

Effects of Family Situation. To estimate specific coefficients for each of the famil‐ ial contexts, alternating dummy coded variables for three of the four family types were included as level‐2 predictors into the relationship measurement model. This procedure allows for the fourth family type to serve as contrast group. Thus, the analysis provides family specific estimates and substantial differences between them.

Investigations illustrated considerable variations in relationship regulation patterns among family situations (Table 6.9). At first, the general found effect of kinship orien‐ Results 124

tation was substantial in all but one of the family contexts (Involuntary childless: γ10

= 0.85, p < .01; blended family: γ10 = 0.86; p < .001, traditional family: γ10 = 1.19, p <

.001). Only motivated childless individuals did not show a substantial association between genetic relatedness and perceptions of closeness (γ10 = 0.21, p > .05). In that, they substantially differed from traditional parents.

Table 6.9 Parameter Estimates (PE) of Relationship Measurement Two‐Level Regression Model on Perceived Closeness to Social Network Partners in Four Different Familial Situa‐ tions

Motivated Involuntary Blended Traditional

Childless Childless Family Family PE SE PE SE PE SE PE SE

Fixed Effects Intercept ‐0.09 0.07 ‐0.09 0.07 ‐0.03 0.07 ‐0.00 0.07 Slopes γ10 Degree of genetic relatedness ‐0.21a 0.31 ‐0.85ab 0.25 ‐0.86b+ 0.18 ‐1.19b 0.22 γ20 Perceived under‐benefit ‐0.37a 0.09 ‐0.40a 0.10 ‐0.12b 0.08 ‐0.51a 0.09 γ30 Perceived over‐benefit ‐0.03 0.11 ‐0.22 0.10 ‐0.03 0.10 ‐0.01 0.12 γ40 Under‐benefit x genetic relatedness ‐0.10 0.45 ‐0.11 0.34 ‐0.38 0.24 ‐0.72 0.30 γ50 Over‐benefit x genetic relatedness ‐0.08a+ 0.41 ‐0.07a+ 0.35 ‐0.37ab 0.30 ‐0.84b+ 0.32

Random Effects 02 Intercept ‐0.25 0.03 12 Degree of genetic relatedness ‐1.11 0.32 22 Perceived under‐benefit ‐0.20 0.05 32 Perceived over‐benefit ‐0.35 0.09 42 Under‐benefit x genetic relatedness ‐1.75 0.58 52 Over‐benefit x genetic relatedness ‐0.70 0.07 2 ‐0.53 0.02

Model fit LL ‐5500.33 CF ‐1.07 FP ‐46 TRd1 ‐58.32 p < .001 Note. Substantial parameters are in bold face. Parameter estimates with different subscribes differ sub‐ stantially at p < .05.

Second, costly relationships were examined. All but one family category showed a negative association with closeness perceptions regarding non‐kin rela‐ tionship partners (motivated childless: γ20 = ‐0.37, p < .001; involuntary childless: γ20 =

‐0.40, p < .001; traditional family: γ20 = ‐0.51, p < .001). Only blended family members illustrated no substantial main effect (γ20 = ‐0.12, n.s.). The interaction term between Results 125 under‐benefit and genetic relatedness proved to be substantial solitary in the context of traditional parents (γ40 = 0.72, p < .05). They reported higher perception of close‐ ness to biological kin despite perceptions of under‐benefit. Perceptions of higher costs led to the disappearance of kinship preference in involuntary childless (γ40 =

0.11, n.s.) and blended families (γ40 = 0.38, n.s.). Thus only traditional parents showed the effect of the overall model that despite imbalance kin relationships are closer than non kin ties.

The third reciprocity perspective captured beneficial relationships. They in‐ clude all ties where participants reported to receive more than they give. In this con‐ text, involuntary childless individuals felt less close to non‐kin (γ30 = ‐0.22, p < .05).

This was not true for any of the remaining family situations (motivated childless: γ30

= ‐0.03, n.s.; blended family: γ30 = ‐0.03, n.s.; traditional family: γ30 = ‐0.01, n.s.). Only traditional parents showed a substantial interaction effect of genetic relatedness and over‐benefit, reporting less closeness in high‐benefit kin ties (γ50 = ‐0.84, p < .05). This effect was not found in any of the other family categories (motivated childless: γ50 =

0.08, n.s.; involuntary childless: γ50 = 0.35, n.s.; blended family: γ50 = ‐0.37, n.s.). Find‐ ings suggest that perceptions of over‐benefit did either not affect or were negatively associated with emotional closeness in kin and non‐kin ties.

Figure 9. Mean Perceived Emotional Closeness To Non‐Kin (left) and Kin (right) as a Function of Perceived Under‐ and Over‐Benefit in Ego‐Centered Social Networks among Four Family Situations

Figure 9 shows family specific parameter estimates displayed as bars. Two fea‐ tures appear to be particularly important. For one thing, motivated childless indi‐ Results 126 viduals showed lowest amounts of regulatory differentiation in closeness percep‐ tions regarding genetic relatedness and felt reciprocity across ego‐centered social network partners. For another thing, ones more involuntary childless individuals tangle between the group of motivated childless and families. They illustrate a clear pattern of kinship orientation that is similar to traditional and blended family par‐ ents. Otherwise they demonstrate high sensitivity for imbalance in non‐kin ties and are comparatively similar to motivated childless individuals with regard to effects of imbalance in kin relationships.

Summing up, findings on relationship regulation characteristics pointed out substantial differences between familial contexts in terms of closeness regulation. In the majority of family situations genetic relatedness was significantly associated with perceptions of closeness. Only motivated childless individuals showed no substantial kinship orientation. The consideration of perceived under‐benefit and over‐benefit in relationships added substantial information, but produced no clear pattern. Percep‐ tions of non‐reciprocity differentially affected emotional closeness to non‐kin and kin ties.

6.2.3 Summary: Social Networks and Relationship Regulation across Life Span and

Family Situations

Findings on social network and relationship regulation show that structure and composition vary with respect to age and family situations. Results on age groups suggest that the importance of kin and elective kin increases in older age. Further‐ more, substantial variations were pictured across the life span with respect to rela‐ tionship regulation mechanisms. This is specifically true in general closeness percep‐ tions of non‐kin and patterns of kinship preference. Late adulthood and old age are characterized by higher values in both characteristics. Thus, findings support hy‐ pothesis 3.

Summarizing findings of the family study, hypothesis 4 is only partly sup‐ ported. As expected, motivated childlessness is associated with less (elective) kinship Results 127 presence within social networks. In contrast, involuntary childless individuals do not differ from traditional parents. However, it proves false, that differential kinship presence would be compensated by non‐kin relationships. In this context no varia‐ tion occurs between family types. Relationship regulative patterns illustrate differen‐ tial characteristics among family situations. Motivated childless individuals show no substantial kinship preference. They do not describe higher amounts of closeness to kin compared to non kin relationship partners. Furthermore, perceptions of imbal‐ ance seem to be differentially related to closeness perceptions. Involuntarily childless individuals are the only family type to report negative effects in emotional closeness to non‐kin regardless of direction of non‐reciprocity. Against expectation, they do not show substantial effects regarding imbalance and genetic relatedness. Substantial kinship orientation even in under‐beneficial ties is shown only by traditional parents and marginal effects by blended family parents.

6.3 Functionality of Relationships – Relational and Individual Self‐Evaluation

The third part of the result section focuses on the interplay between self‐evaluative outcomes and perceptions of closeness and reciprocity. At first, investigations ex‐ amine two relationship specific indicators, communal and mutual esteem, in the con‐ text of ego‐centered social networks. Again, data of the online study provides the possibility of age group specific analyses. Subsequently, two sub‐samples of the family study are explored in terms of specific dyadic relationship regulative patterns and their effect on evaluative outcomes. In this context, global and domain‐specific self‐esteem measures were selected as individual level characteristics. Analyses focus on the adult child‐parent dyad and the relationship between spouses.

6.3.1 Online Study: Relational Self‐Esteem in Social Networks

Communal and mutual‐esteem are the two indicators of relational self‐esteem. De‐ scriptive statistics of both measures are explored to illustrate variations within and Results 128 across social networks. Subsequent analyses focus on the association between rela‐ tional self‐esteem and relationship characteristics. Again, analyses explore variations across different relationship types and age groups.

6.3.1.1 Differential Esteem: Communal‐ and Mutual‐Esteem across Social Relationships

Communal‐Esteem. Each personal relationship was judged with regard to the per‐ ceived relevance of the attributed esteem. Thus, the measure taped on the relation‐ ship‐specific importance of each social network partner. It was expected that this re‐ lational value differs between relationships. Furthermore, relationship‐specific evaluations should vary with respect to contextual characteristics. To indicate this variability, a univariate analysis of variance with relationship type (9) and age group

(5) was conducted. The chosen procedure was equal to the analyses regarding per‐ ceived closeness and reciprocity. A relationship specific aggregated version of the network data was used and the few named children were excluded. Results re‐ mained stable even after controlling for effects of gender and parental status. There‐ fore the simple model is presented.

Results pointed out substantial main effects for relationship type, F (8, 2456) =

14.89, p < .001, η2 = .05, and age group, F (4, 2456) = 3.59, p < .01, η2 = .01. Furthermore, the interaction term produced a substantial effect, F (31, 2456) = 2.28, p < .001, η2 = .03

(cf. Appendix B, Table B.2). Post‐hoc tests on relationship type indicated that partici‐ pants described the perception of communal‐esteem to be most important regarding to the partner. Relationships to parents, close kin, old friends and siblings were also regarded as particularly relevant. Less importance was expressed with respect to felt appreciation of colleagues, in‐law and step‐family members, as well as acquaintances and distant kin. The main effect of age group produced no substantial post‐hoc ef‐ fects, but mean‐levels indicated a general higher relevance of communal‐esteem across the adult life. Results on the interaction effect of communal‐esteem illustrated a diverse pattern. Significant variation showed in parental relationships, affiliations to close kin and in‐law or step‐family members. Young and middle aged adults de‐ Results 129 scribed ties to parents as less important compared to evaluations in early adulthood.

Close kin were most important in old age. Concerning elective kin, higher relevance was illustrated across adulthood.

Mutual‐Esteem. The indicator of mutual‐esteem taped on the evaluated reci‐ procity in feelings of esteem or appreciation. Again, participants had to judge this perception with regard to each relationship partner. Following analyses differen‐ tiated between feelings of higher‐esteem and lower‐esteem. In higher‐esteem relation‐ ships the relationship partner is regarded to appreciate the relationship more than oneself. In contrast, in lower‐esteem relationships one regards the relationship with higher‐esteem than the other person. Two separate analyses were conducted at the previously described aggregated relationship level. Again, Results remained stable even after controlling for effects of gender and parental status.

Results on higher‐esteem illustrated a non‐significant main effect for relation‐ ship type, F (8, 2456) = .99, p > .05, η2 = .00, but a substantial main effect for age group,

F (4, 2456) = 4.62, p < .01, η2 = .01. Decreasing mean‐levels throughout adulthood sug‐ gested that older adults perceived less over‐appreciated relationships than younger people did. In addition, the interaction term was found to be substantial, F (31, 2456)

= 1.63, p < .05, η2 = .02. Results pointed out significant variability across age groups in parental and close kin relations, in affiliations with colleagues and, at the verge of significance, in relationships to elective kin (cf., Appendix B, Table B.2). Means of higher‐esteem were found to be lower in all family relationships in older age groups.

In contrast, the pattern with regard to colleagues showed a higher relevance until late adulthood, followed by very low scores in old age.

The examination of the lower‐esteem demonstrated no main effects, relation‐ ship type: F (8, 2456) = 1.29, p > .05, η2 = .00; age group: F (4, 2456) = 2.09, p > .05, η2 =

.00, but a substantial interaction effect, F (31, 2456) = 1.51, p < .05, η2 = .02. Again, pa‐ rental and close kin relationships showed to vary substantially with fairly similar mean‐level trajectories. The older people were, the more likely they seemed to appre‐ ciate relationships to social partners more than vice versa (cf., Appendix B, Table Results 130

B.2). The relationship to siblings indicated a certain amount of variation, as well.

However, the pattern was less clear. Relationships seemed to be less mutual‐ esteemed in middle and late adulthood, but more so in younger and older age.

Analyses on characteristics of relational self‐esteem suggest variations in dif‐ ferent relationship types and across the life span. In particular relationships to family members, biological and elective kin, differ on the amount of relational value. Col‐ leagues show only a small amount of variation. Relationships to friends or acquain‐ tances seem to be fairly stable throughout – each at its own level. Results support the first part of hypothesis 5a.

6.3.1.2 Effect of Relationship Quality: Relationship Regulation Characteristics and Rela‐

tional Self‐Esteem

Relationship characteristics, emotional closeness, perceived reciprocity and genetic relatedness, were expected to be substantial predictors of relational self‐esteem. Up‐ coming analyses distinguished the three major relationship groups defined by the theoretical framework of relationship regulation: biological kin, elective kin, and non‐kin. Multilevel random coefficient models were estimated for each of the type in single data sets, containing the specific relationship type only. Predictor variables included closeness perceptions, under‐ and over‐benefit, plus the two way interac‐ tion between closeness × under‐benefit and the one of closeness × over‐benefit. All three facets of relational esteem were estimated in single models.

Prediction of Communal‐Esteem. Table 6.10 shows findings with respect to all three relationship systems on communal‐esteem. Results on biological kin suggested that closeness perceptions are highly associated to the evaluation of relevance (γ10 =

2.26, p < .001). The closer people felt to the respective kin, the more they emphasized the esteem of this person. This was true even with respect to close but unbalanced family relationships (Closeness × under‐benefit: γ40 = 0.67, p < .01, closeness × over‐ benefit: γ50= 0.64, p < .01). In less close kin relationships feelings of imbalance were linked to lower communal‐esteem, irrespective of perceptions of higher or lower Results 131

costs (Under‐benefit: γ20 = ‐0.56, p < .001, over‐benefit: γ30 = ‐0.46, p < .01).The three predictors plus the two interactions improved the model substantially.

Table 6.10 Parameter Estimates (PE) of Multilevel Model on Perceived Communal‐Esteem to Social Network Partners in Three Relationship Systems

Perceived Communal‐Esteem to Biological Kin Elective Kin Non‐Kin (n = 1850) (n = 926) (n = 3077) PE SE PE SE PE SE Fixed Effects Intercept1 0.25 0.05 0.13 0.06 ‐0.06 0.04 Slope γ10 Perceived Closeness 2.26 0.18 2.37 0.25 2.59 0.12 γ20 Perceived under‐benefit ‐0.56 0.14 ‐0.63 0.17 ‐0.15 0.10 γ30 Perceived over‐benefit ‐0.46 0.17 ‐0.60 0.24 ‐0.24 0.09 γ40 Closeness × under‐benefit 0.67 0.20 1.10 0.24 0.24 0.17 γ50 Closeness × over‐benefit 0.64 0.23 1.23 0.32 0.46 0.15

Model Fit Loglikelyhood ‐1884.12 ‐973.31 ‐2797.97 Scaling correction Factor 1.33 1.35 1.55 Free Parameters 27 21 27 TRd 1002.43 562.67 1048.90 p < .001 < .001 < .001 Add. explained Level‐1 Variance 75% 77% 60% Note. Displayed are unstandardized PE of MLM. Bold PE are substantial at p < .05. All level‐1 predic‐ tors that include ʹ0ʹ as a meaningful value are included uncentered, i.e., only perceived closeness is group‐mean centered. 1Presensted are z‐values with respect to communal‐esteem.

Results on elective kin were quite similar to patterns in biological kin. Yet two aspects showed to be different. At first, elective kin were generally described to carry less relevance (γ00 = 0.13, p < .05). Second, the positive effect of the interaction be‐ tween closeness and imbalance was even stronger. Unstandardized effects were about twice the size compared to those in biological kin (Closeness × under‐benefit:

γ40 = 1.10, p < .001, closeness × over‐benefit: γ50 = 1.23, p < .001). These results on non‐ kin family members suggest that regardless of non reciprocity high closeness is asso‐ ciated to high relevance of this person for one’s self‐evaluation. Again, predictors yielded to explain a substantial part of the communal‐esteems variance.

Patterns in non‐kin yielded similarities and differences compared to the two previous groups. The intercept indicated less relevance of these relationships for people’s relational self‐esteem (γ00 = ‐0.06, n.s.). However, the positive effect of felt Results 132

closeness showed to be very strong again (γ10 = 2.59, p < .001). Another discrepancy occurred with regard to perceptions of under‐benefit. The feeling of high costs did not have any effect on evaluations of communal‐esteem (γ20 = ‐0.15, n.s.). This was also true regarding the interaction effect with closeness (γ40 = 0.24, n.s.). In contrast, feeling over‐benefited led to less relevance of this relationship for one’s self evalua‐ tion (γ30 = ‐0.24, p < .001). But the interaction effect with closeness was again positive‐ ly related to communal esteem (γ50 = 0.46, p < .05). One possible explanation for these effects could be that in under‐beneficial relationships the relationship partner is per‐ ceived as role model for example with more desired abilities. Therefore, one does look up to this person and perceptions of relevance are important for people’s self‐ evaluation. This is not true with respect to people that one feels superior to already.

Age Differences. To analyze age effects, specific models for each relationship system were run with four alternating dummy coded variables (for each age group minus one reference category). Since the number of elective kin in personal networks was relatively small, models were unable to estimate stable coefficients. Therefore,

Table 6.11 shows findings separately for perceived communal‐esteem as a function of age in kin and in non‐kin relationships, only.

The kin specific model on communal‐esteem improved substantially with the included level‐2 predictors, TRd = 56.16, p < .001. The influence of closeness and reci‐ procity perceptions on the evaluation of communal‐esteem varied substantially among the defined age groups. In early adulthood results indicated only one sub‐ stantial effect, namely with respect to closeness perception (γ10 = 2.53, p < .05). The appreciation of a biological family member was more important the closer early adults felt to the person. In young adulthood this effect of closeness perceptions re‐ mained relatively stable (γ10 = 1.95, p < .05). In addition, this group emphasized to feel a lower need for appreciation in under‐benefited relationships (γ20 = ‐0.53, p < .05).

Middle adulthood was similar to the young adults regarding the positive ef‐ fect of closeness (γ10 = 2.26, p < .05) and the negative effect of high‐cost relationships

(γ20 = ‐0.63, p < .05) on communal‐esteem. A comparable negative effect occurred on Results 133

over‐beneficial relationships (γ30 = ‐0.79, p < .05). Thus, imbalance in either direction was associated to lower communal‐esteem to kin. The interaction between closeness and over‐benefit instead showed to be positive (γ50 = 1.20, p < .05).

Table 6.11 Parameter Estimates (PE) of Multilevel Model on Perceived Communal‐Esteem to Social Network Partners in Different Age groups in Kin and Non‐Kin Relationships

Perceived Communal‐Esteem To Biological Kin in Different Age groups Early Young Middle Late Old

Adulthood Adulthood Adulthood Adulthood Age PE SE PE SE PE SE PE SE PE SE Fixed Effects Intercept1 0.19a 0.07 0.18a 0.11 0.18a 0.10 0.45b 0.09 1.07c 0.15 Slope γ10 Perceived closeness 2.53a 0.21 1.95a 0.32 2.26a 0.39 2.06a 0.32 ‐2.34b 1.45 γ20 Perceived under‐benefit ‐0.37a 0.22 ‐0.53a 0.27 ‐0.63a 0.31 ‐0.91a 0.29 ‐3.17b 0.48 γ30 Perceived over‐benefit ‐0.22a 0.22 ‐0.48a 0.29 ‐0.79a 0.31 ‐0.68a 0.47 ‐3.00b 0.53 γ40 Closeness x under‐benefit 0.41a 0.31 0.55a 0.36 0.78a 0.41 1.01a 0.37 3.48b 0.67 γ50 Closeness x over‐benefit 0.27a 0.32 0.73a 0.40 1.20a 0.43 0.73a 0.60 3.23b 0.65

Model Fit Loglikelyhood ‐1856.31 Scaling correction Factor 1.17 Free Parameters 52 TRd 56.16 p < .001 To Non‐Kin in Different Age groups Early Young Middle Late Old

Adulthood Adulthood Adulthood Adulthood Age PE SE PE SE PE SE PE SE PE SE Fixed Effects Intercept1 0.01 0.07 ‐0.01 0.11 ‐0.10 0.12 ‐0.10 0.11 ‐0.18 0.16 Slope γ10 Perceived closeness 2.72 0.19 2.43 0.47 2.67 0.67 2.32 0.92 2.63 0.77 γ20 Perceived under‐benefit 0.32 1.17 ‐0.11 1.58 ‐0.19 0.43 ‐0.63 0.71 ‐0.57 0.93 γ30 Perceived over‐benefit ‐0.03 1.57 ‐0.38 1.13 ‐0.35 0.26 ‐0.32 0.79 ‐0.49 0.33 γ40 Closeness x under‐benefit ‐0.46a 1.91 0.12ab 3.04 0.30ab 0.80 0.94ab 1.26 1.16b 1.96 γ50 Closeness x over‐benefit 0.05 3.25 0.59 2.07 0.69 0.92 0.67 2.28 1.23 0.84

Model Fit Loglikelyhood ‐2780.24 Scaling correction Factor 3.03 Free Parameters 52 TRd 85.56 p < .001 Note. Displayed are unstandardized PE of MLM. Bold PE are substantial at p < .05. All indicators in italic font are level‐2 predictors. All level‐1 predictors that include ʹ0ʹ as a meaningful value are in‐ cluded uncentered, i.e., only perceived closeness is group‐mean centered. 1 Communal‐esteem was z‐ standardized prior to analyzes.

Participants of the late adulthood group emphasized once more the specific role of close kin relationships regarding the perception of appreciation and esteem Results 134

(γ10 = 2.06, p < .05). Communal‐esteem was negatively associated with high cost rela‐ tions (γ20 = ‐0.91, p < .05) but positively with the interaction of closeness (γ40 = 1.01, p <

.05). The feeling of over‐benefit was unrelated to perceptions of communal‐esteem. In old age results became somehow more distinctive. The intercept indicates higher communal‐esteem in general. Negative effects of unbalanced relationships were par‐ ticularly strong (under‐benefit: γ20 = ‐3.17, p < .05, over‐benefit: γ30 = ‐3.00, p < .05).

This association turned around into a highly positive relationship in unbalanced close kin ties (closeness x under‐benefit: γ40 = 3.48, p < .05, closeness x over‐benefit: γ50

= 3.23, p < .05). Thus, results indicate that the esteem of either under‐ or over‐ beneficial close family members was perceived to be important again.

The non‐kin model substantially improved by including the age group dum‐ my variables, TRd = 85.56, p < .001 (lower part of Table 6.11). However, differential effects between groups were scarce. Most notably, the positive effect of closeness was enhanced across age groups. This finding suggested the stable importance of close non‐kin relationships for people’s self‐evaluation. The perception of communal‐ esteem was important throughout adult life. Characteristics of reciprocity did not have any substantial effects on the perception of the communal‐esteem in these rela‐ tionships. Nevertheless, one pattern differed among the groups. The interaction ef‐ fect between closeness and under‐benefit was negative associated to communal es‐ teem in early adulthood (γ40 = ‐0.46, n.s.) and by that differed substantially from the positive effect in old age (γ40 = 1.16, n.s.). Across the adult life the relevance of the perceived esteem of close, high‐cost ties with neighbors or colleagues increased. A similar effect surfaced with respect to over‐beneficial, close non‐kin relationships but failed to reach significance.

Prediction of Mutual‐Esteem. Logistic multilevel regressions had to be con‐ ducted, since indicators of higher‐ and lower‐esteem were categorical outcomes.

Models explored the likelihood of perceived higher or lower‐esteem (vs. balance) to the respective relationship partner as a function of felt closeness, under‐benefit, over‐ benefit, and the two corresponding interaction terms. Dichotomous outcomes were Results 135 coded 0 (ʹmutual‐esteemʹ) and 1 (ʹhigher‐esteemʹ or ʹlower‐esteem, respectively). Hie‐ rarchical generalized linear models (HGLM) were chosen as appropriate modeling framework (Raudenbush & Bryk, 2002). First analyses indicated that random slopes models were too complex to be estimated within the three relationship systems. To keep all variables, i.e. information, in the equation, random intercept models were conducted, only. Results are presented in Table 6.12.

Table 6.12 Parameter Estimates (PE) and Odds Ratios (OR) of HGLM on Perceived Mutual‐ esteem to Social Network Partners in Three Relationship Systems

Perceived Mutual‐Esteem to Biological Kin Elective Kin Non‐Kin (n = 1850) (n = 926) (n = 3077) PE OR PE OR PE OR Mutual vs. Higher‐Esteem γ10 Perceived Closeness ‐1.15 0.32 ‐0.52 0.60 ‐1.59 0.20 γ20 Perceived under‐benefit 1.69 5.42 1.18 3.27 0.57 1.77 γ30 Perceived over‐benefit 2.85 17.22 2.42 11.21 1.68 5.39 γ40 Closeness x under‐benefit ‐2.04 0.13 ‐1.53 0.22 ‐0.57 0.56 γ50 Closeness x over‐benefit ‐1.68 0.19 ‐1.30 0.27 ‐0.32 0.73

Mutual vs. Lower‐Esteem γ10 Perceived Closeness ‐1.82 0.16 ‐0.67 0.51 ‐1.37 0.25 γ20 Perceived under‐benefit 2.32 10.22 2.61 13.55 1.98 7.22 γ30 Perceived over‐benefit 0.34 1.41 0.48 1.62 0.10 1.10 γ40 Closeness x under‐benefit ‐1.05 0.35 ‐1.69 0.19 ‐0.43 0.65 γ50 Closeness x over‐benefit ‐0.48 0.62 ‐0.58 0.56 ‐0.22 0.81

Note. Positive substantial coefficients and odds ratios greater than 1 indicate that an increase on the variable increased the probability of the relationship to be imbalanced in esteem relative to perceived mutuality. Negative substantial coefficients and odds ratios smaller than 1 indicate that an increase on this variable decreased that probability. Bold PE are substantial at p < .05. All level‐1 predictors that include ʹ0ʹ as a meaningful value are included uncentered, i.e., only perceived closeness is group‐mean centered.

The first comparisons contrasted mutuality in esteem and higher‐esteem, thus relationships where people describe to receive more appreciation than they return.

Inclusion of the five predictors improved the model substantially with respect to all three relationship systems, kin: TRd = 88.77, p < .001; elective kin: TRd = 29.41, p <

.001; non‐kin: TRd = 91.03, p < .001. In kin relationships closeness per se was not asso‐ ciated to perceptions of higher‐esteem. The probability to perceive higher‐esteem increased significantly in under‐ (γ20 = 1.69, p < .05) and over‐benefited relationships

(γ30 = 2.85, p < .05). Likelihood decreased in close, imbalanced ties (closeness × under‐ Results 136

benefit: γ40 = ‐2.04, p < .05, closeness × over‐benefit: γ50 = ‐1.68, p < .05). In elective kin, the probability to belong to the higher‐esteem group only increased if people felt over‐benefited (γ30 = 2.42, p < .05). Thus, in this context they perceive higher apprecia‐ tion from the other person than they return. Amounts of closeness or higher costs did not affect the evaluation in elective kin ties. A similar pattern of results was found for non‐kin relationships. Over‐beneficial relationships were associated with higher‐ esteem (γ30 = 1.68, p < .05). In addition, particularly close non‐kin ties were less likely to be perceived as a higher‐esteem relationship (γ10 = ‐1.59, p < .05). In non‐kin rela‐ tionships, higher‐esteem was associated with low closeness and perceptions of over‐ benefit.

The second dummy coded variable taped on the differentiation between mu‐ tuality and lower‐esteem, thus social ties where perceptions of appreciation of indi‐ viduals exceed perceived reward of relationship partners. In all three relationship systems models improved to a substantial extent, kin: TRd = 148.80, p < .001; elective kin: TRd = 61.49, p < .001; non‐kin: TRd = 183.47, p < .001. Overall, less substantial ef‐ fects occurred. Still result patterns differed among relationship systems, confirming the differential effect of relational self‐evaluation. In kin relationships perceptions of lower‐esteem were less likely in close ties (γ10 = ‐1.82, p < .05), but perceptions of un‐ der‐benefit increased the probability (γ20 = 2.32, p < .05). Over‐benefit and interaction terms produced no substantial effects. In elective kin ties perceptions of lower‐esteem were particularly related to under‐beneficial relationships (γ20 = 2.61, p < .05).

Though, the corresponding interaction term indicated that close, under‐beneficial ties were less likely to be perceived as lower‐esteem (γ40 = ‐1.69, p < .05). Costly elective kin relationships increased the likelihood of lower‐esteem, but simultaneous percep‐ tions of closeness decreased the probability. Results in non‐kin ties showed that close relations were less (γ10 = ‐1.37, p < .05) and under‐beneficial ties were more likely (γ20

= 1.98, p < .05) to be rated as lower‐esteem relationships. Results 137

Age differential analyses were not conducted, because models were estimated with fixed slopes only. Moreover, age group specific estimates would not extend the knowledge above and beyond the already conducted analyses of variance.

In sum, emotional closeness and perceived reciprocity demonstrated a strong predictive value on relational self‐esteem indicators. Findings on communal‐esteem suggested differential patterns of both relationship qualities across the distinguished relationship contexts. Overall, perceived communal‐esteem is strongly related to emotional closeness. Meaningful relationship partner affect self‐evaluation. Unexpec‐ tedly, reciprocity perceptions showed a substantial (negative) association with rela‐ tionship‐specific self‐evaluations in biological and elective kin. In non‐kin, only over‐ benefit produced a substantial effect. A number of age group specific variations be‐ came obvious. Cross‐level effects suggested that the association between communal‐ esteem and perceptions of reciprocity in kin increased in older age groups. Less effect occurred in elective and non‐kin ties. Indicators of mutual‐esteem support the result that perceptions of reciprocity affect self‐evaluation characteristics in kin ties, indicat‐ ing higher likelihood for mutual appreciation in reciprocal ties. Effects were not as strong for elective and non‐kin. In contrast, perceptions of closeness decreased the probability to belong to one of the imbalanced‐esteem groups. This effect was only found in non‐kin relationships and for lower‐esteem in kin ties. Results confirm hy‐ pothesis 5a.

6.3.1.3 Level of Data: Relational and Individual Self‐Esteem Indicators

Relational self‐esteem is a construct that addresses self‐evaluation within specific relational contexts, thus at the relationship level. In contrast, global and domain‐ specific self‐esteem assess tendencies of self‐evaluation at the individual level. The following analyses considered these different levels of data assessment.

Interplay with Communal‐Esteem. Relationship specific models of communal‐ esteem were expanded with all three individual self‐esteem scales. Thus, models tested possible effects of individual on relationship specific self‐evaluation tenden‐ Results 138 cies. All three variables were included as level‐2 indicators to affect intercept and random slope variables. Stepwise deletion of non‐significant estimates led to the final model, presented in Table 6.13. In any cases, main level‐2 effects on the intercept are controlled for even if they were not substantial.

Findings on biological kin showed that none of the individual characteristics were associated with the evaluation of relevance in general or produced cross level effects. Regarding biological kin, perceptions of communal‐esteem were independent of people’s trait or state self‐esteem level. They did not explain substantial variability in the dependent variable of communal‐esteem.

Table 6.13 Parameter Estimates (PE) of Multilevel Model on Perceived Communal‐Esteem to Social Network Partners in Three Relationship Systems with Cross Level Effects with Trait and State Self‐Esteem, Life Satisfaction and Subjective Health

Perceived Communal‐Esteem to Kin Elective Kin Non‐Kin (n = 1850) (n = 918) (n = 3067) PE SE PE SE PE SE Fixed Effects Intercept1 0.25 0.05 0.14 0.07 ‐0.05 0.04 Slope γ10 Perceived closeness 2.26 0.18 2.34 0.26 2.59 0.12 γ20 Perceived under‐benefit ‐0.56 0.14 ‐0.67 0.17 ‐0.18 0.10 γ30 Perceived over‐benefit ‐0.46 0.17 ‐0.57 0.33 ‐0.24 0.09 γ40 Closeness x under‐benefit 0.67 0.20 1.12 0.23 0.27 0.17 γ50 Closeness x over‐benefit 0.64 0.23 1.14 0.41 0.44 0.15 Level‐2 Main‐/Cross‐Level Effects γ01 Communal‐esteem × Performance self‐esteem 0.24 0.18 γ01 Communal‐esteem × Social self‐esteem 0.47 0.20 ‐0.26 0.14 γ11 Closeness × Performance self‐esteem 1.25 0.49 γ21 Under‐benefit × Social self‐esteem ‐0.60 0.21 ‐1.04 0.31 γ41 Closeness x under‐benefit x Social self‐esteem 1.54 0.51

Model Fit Loglikelihood ‐1884.12 ‐960.58 ‐2771.41 Scaling correction Factor 1.33 1.50 1.49 Free Parameters 27 24 33 TrD / 9.41 44.26 p < .05 < .001 Note. Displayed are unstandardized PE of MLM. Bold PE are substantial at p < .05. All level‐1 predic‐ tors that include ʹ0ʹ as a meaningful value are included uncentered, i.e., only perceived closeness is group‐mean centered. 1Presented are z‐values with respect to communal‐esteem.

In contrast, the relevance of elective kin varied substantially by individual characteristics. Two effects occurred and improved the previous model substantially, Results 139

TRd = 9.41, p < .05. Social state self‐esteem showed a main (γ01 = 0.47, p < .05) and an interaction effect with perceived under‐benefit (γ21 = ‐0.60, p < .05). Higher social con‐ fidence is associated with higher perceptions of communal‐esteem in elective kin and with a stronger negative effect of under‐benefit. Thus, results suggest that higher so‐ cial self‐esteem increases social sensitivity to under‐benefit in elective kin.

The relevance of perceiving esteem in non‐kin relationships showed no main but three cross‐level effects with closeness, under‐benefit, and the interaction effect of these two variables. Performance self‐esteem increased the effect of closeness on communal‐esteem (γ11 = 1.25, p < .05). People with high confidence in their abilities and achievements illustrated an enhanced effect of closeness on self‐evaluation in non‐kin. As in elective kin relations, social self‐esteem showed a negative effect on under‐benefit (γ21 = ‐1.06, p < .05). Thus, people with high social self‐esteem regarded the esteem of under‐beneficial ties with less relevance, i.e. they were more sensitive to reciprocity violations. In addition, social self‐esteem indicated a positive cross‐ level effect with the interaction between closeness and under‐benefit. The originally insignificant effect became substantial if persons had more social self‐assurance (γ41 =

1.54, p < .05). Higher social self‐esteem was associated to higher perceptions of com‐ munal‐esteem in under‐beneficial and close non‐kin relationships. Thus, high percep‐ tions of closeness seemed to reverse the previously described negative cross‐level effect between social self‐esteem and under‐benefit.

Findings illustrate that performance and particularly social self‐esteem were related to effects of closeness, under‐benefit, and their interaction on communal‐ esteem. Yet again, differential effects were found for relationship systems. Percep‐ tions of communal‐esteem to kin were unaffected by individual trait measures.

Interplay with Mutual‐Esteem. Since models on mutual‐esteem were estimated with fixed slope only, analyses focused on intercept effects of individual measures.

As in the previous section, all level‐2 variables were included into the model at once and non significant effects were deleted to receive a parsimonious final model. Results 140

Effects of level‐2 characteristics on kin relationships were not existent in per‐ ceptions of higher‐esteem, but occurred in lower‐esteem evaluations. People with a higher global self‐esteem were generally less likely to judge kin relationships as low‐ er‐esteem ties (γ01 = ‐1.03, p < .05). The additional predictor improved the overall quality of the model, TRd = 10.31, p < .01. In relationships to non‐biological family members, no substantial influence of level‐2 indicators showed in lower‐ and higher esteem. With regard to the last relationship system, results demonstrated a strong similarity to patterns in kin ties. Again, the global self‐esteem measure influenced the likelihood to rate a relationship as lower‐esteem. In fact, a higher global self‐esteem decreased the probability to rate non‐kin social partners as lower‐esteem relation

(γ01= ‐2.42, p < .001). The model fit was significantly enhanced through this additional predictor, TRd = 41.86, p < .001.

Summarizing findings on communal and mutual‐esteem, three aspects seem most important: First, communal‐esteem accomplished to differentiate between se‐ lected types of relationship partners and showed variations by age groups. Effects on mutual esteem were less distinctive. Patterns of mutuality in relational esteem showed a substantial variation across the life span particularly in relationships to kin and elective kin.

Second, multilevel regression analyses supported a differential pattern of closeness and reciprocity in the prediction of perceived esteem. A strong positive effect of closeness was confirmed among the differentiated relationship systems: the closer the relationship, regardless of kinship status, the higher communal‐esteem.

With respect to mutual esteem, closeness decreased the likelihood of lower‐ and higher esteem only in non‐kin. Perceptions of over‐ and under‐benefit varied in an unexpected way. Results suggest that there is a higher sensitivity for reciprocity threats in kin and elective kin relationships, i.e. communal esteem decreased in un‐ der‐ and over‐benefit ties. Simultaneously, perceptions of high closeness and high imbalance were related to higher communal esteem. These effects were low or insig‐ nificant in non‐kin ties. Patterns in lower‐ and higher esteem mainly support these Results 141 findings. Likelihood of mutuality in esteem was lower in imbalanced kin and, to lesser extends, in elective kin relationships. Age differential effects were only con‐ ducted for communal‐esteem. They supported a throughout strong effect of close‐ ness. In the context of kin relations this effect disappeared but only in old age. In con‐ trast, effects of reciprocity increased in older age groups. These two aspects support hypothesis 5a.

Third, interplay with individual self‐esteem indicators suggests that relational self‐esteem shows generally few associations with individual self‐esteem, confirming hypothesis 5b. Most effects on communal‐esteem are found for social self‐esteem.

High social self‐esteem enhanced some of the found patterns. Importantly, individu‐ al characteristics only affected communal‐esteem to elective and non‐kin relation‐ ships. Individual characteristics did not have any effect on higher‐esteem, but a few on lower‐esteem. Higher global self‐esteem decreased the likelihood of lower‐esteem perceptions in kin and non‐kin relationships.

6.3.2 Family Study: Self‐Evaluation in Adult Child‐Parent Ties

What are the dynamics and interdependencies of self‐evaluative outcomes and closeness and reciprocity in specific relationship contexts such as intergenerational relations? Does family constancy affect this interplay? Before answering these dyadic research questions, descriptive analyses illustrate individual relationship perceptions of adult child and old parent, respectively. Examinations include comparisons to ag‐ gregated social network characteristics and between the two generations. Considera‐ tions of dyadic interdependencies are supplemented by cross‐lagged analyses across a 1‐year interval. Group comparisons focus on relationship regulation patterns in the light of high versus low family constancy. Gender is integrated in a number of analy‐ ses, particularly the first part, but does not hold a key interest in the argumentation. Results 142

6.3.2.1 Effects of Family of Origin on Individual Relationship Perceptions of Adult Children

and Old Parents

Perspective of Adult Child. Adult children described relationships to their old parents to be close (M = .72, SD = .22) and fairly reciprocal (M = .70, SD = .24). Descriptions of emotional closeness and perceived reciprocity to parents did not differ between sons and daughters, closeness: t (89) = .21, n.s., d = .05; reciprocity: t (89) = ‐.66, n.s., d = .13

(c.f., Appendix C, Table C.2). In contrast, perceptions of relationship quality are af‐ fected by family constancy. Adult children with instable structures in the family of origin reported less closeness to their parents, when compared to children of stable family structures, t (89) = 2.98, p < .01, d = .58. There were no differences with regard to reciprocity perceptions. No interaction effects occurred between gender and fam‐ ily constancy.

The particular strength of the current study is the social network approach, which possesses the ability to contrast specific relationship perceptions with overall characteristics of the ego‐centered network. One goal is to test whether relationship characteristics are either (to a certain extent) undifferentiated, trait‐like descriptions or specific, targeted perceptions. In order to accomplish this comparison, within‐ person aggregations were computed of emotional closeness and perceived reciproci‐ ty across all named social network partners of the adult child18. These variables served as comparison level in the following analyses. In general, the adult child’s relationship perception to the parent differed considerably from overall means in both relationship qualities emotional closeness and perceived reciprocity (cf. Table

6.14). Children described the relationship to their parents as closer, t (90) = 4.38, p <

.001, d = .45, but less reciprocal, t (90) = ‐2.67, p < .01, d = .25, compared to the network ratings. Effect sizes suggest a stronger effect of closeness with regard to differences between the averaged relationship perception and the relationship to the parent.

18 This sub‐sample of 92 adult children reported an average network size of M = 15.2, SD = 7.7. Results 143

With regard to family constancy, differential effects were only stable in high but not in low family constancy (cf., lower part of Table 6.14). Adult children from stable family contexts reported a closer relationship to their parent compared to the other relationships of the network, t (59) = 5.75, p < .001, d = .70. At the same time, relationships were less reciprocal in relation to aggregated reciprocity perceptions, t

(59) = ‐2.62, p < .05, d = .36. Characteristics of adult children with instable family struc‐ tures appeared to be different. They did not differentiate the relationship to their parents from the average perception of their social affiliations, neither in terms of closeness, t (30) = .63, n.s., d = .13, nor in terms of reciprocity, t (30) = ‐1.02, n.s., d = .19.

Table 6.14 Descriptive Statistics of Relationship Specific Qualities to Parent and Aggregated Grand Means of Closeness and Reciprocity for Adult Child

Closeness Reciprocity M SD t‐ratio M SD t‐ratio

Whole Sample (n=92) Parent 0.72 0.22 4.38, p < .001 0.70 0.24 ‐2.67, p < .01 Network 0.63 0.12 (.43) 0.76 0.10 (.20)

Family Constancy High Parent 0.76 0.18 5.75, p < .001 0.69 0.23 ‐2.62, p < .05 (n=60) Network 0.64 0.12 (.41) 0.77 0.09 (.26)

Low Parent 0.63 0.26 0.63 0.70 0.25 ‐1.02 (n=32) Network 0.60 0.10 (.43) 0.75 0.13 (.13)

Note. M = Mean, SD = Standard deviation. t‐ratio is based on paired t‐tests. Parent = Indicator for per‐ ceived relationship to old parent, Network = Grand mean aggregates of relationship perceptions across the whole network of adult child. Numbers in parenthesis placed below t‐ratio indicate paired correlations, with bold faced indicators being significant at p < .05. *** p < .001, ** p < .01, * p <. 05

In sum results support the assumption of a stable effect of family constancy on the child‐parent relationship perception of adult children. High family constancy is associated with more closeness, compared to perceptions in low family constancy, and in contrast to the within‐person aggregates of closeness perceptions across the social network. Low family constancy is linked to less intimate child‐parent relation‐ ships that do not differ from overall aggregates of network closeness. Effects on reci‐ procity demonstrate less balance in child‐parent relations compared to network cha‐ racteristics. This difference in reports was only found in high family constancy. Results 144

However, the violation of reciprocity (less balance) in high family constancy does not go along with a lowered closeness.

Perspective of Old Parent. The parental description of the relationship to their adult children pictured a social tie that was perceived close (M = .76, SD = .19) and mutual (M = .80, SD = .20, cf. Appendix C, Table C.3). No differences were found in relationship perceptions in reference to family constancy. There were also no gender variations in reported amounts of closeness and reciprocity. No interaction effects between gender and family constancy appeared.

Table 6.15 Descriptive Statistics of Relationship Specific Qualities to Adult Child and Aggre‐ gated Grand Means of Closeness and Reciprocity for Old Parent

Closeness Reciprocity M SD t‐ratio M SD t‐ratio

Whole Sample (n=92) Child 0.76 0.19 7.44, p < .001 0.80 0.20 ‐0.23 Network 0.66 0.14 (.78) 0.81 0.14 (.73)

Family Constancy High Child 0.77 0.19 5.76, p < .001 0.80 0.21 ‐0.10 (n=60) Network 0.68 0.15 (.77) 0.80 0.14 (.68)

Low Child 0.74 0.19 4.69, p < .001 0.82 0.18 ‐0.31 (n=32) Network 0.64 0.13 (.80) 0.82 0.14 (.82)

Note. M = Mean, SD = Standard deviation. t‐ratio is based on paired t‐tests. Child = Indicator for per‐ ceived relationship to adult child, Network = Grand mean aggregates of relationship perceptions across the reported six selected relationship partner of old parent with average size of 4.7 (SD = 1.1, Min = 1, Max = 6) network partners. Numbers in parenthesis placed below t‐ratio indicate paired cor‐ relations that are all substantial with p < .001.

In a second step, the within‐person aggregates are used in comparative analy‐ sis. Paper‐pencil questionnaires of parents did not contain the assessment of the en‐ tire ego‐centered network, but only six selected relationship partners19. For the up‐ coming analysis it will be important to bear in mind that aggregates were based on no more than an average of 4.7 (SD = 1.1) relationship partners. Nevertheless, within‐

19 Selected relationship partners were current partner, if existent, child, that took part in the main study, corresponding child‐in‐law and grand child, long term friend, (former) colleague. Full account of assessment can be found in section 5.2.4. Results 145 person means are expected to contain an adequate variation of kin and non‐kin rela‐ tions. Thus, they serve as valid comparison variable and contrast reported relation‐ ship qualities to adult children. First analyses included the whole sample of parents.

Paired t‐tests revealed that parents differentiated between their adult children and other relationship partners (cf. Table 6.15), describing the relationship as substantial‐ ly closer, t (90) = 7.44, p < .001, d = .84. Though, they did not report any discrepancy in terms of reciprocity, t (90) = ‐.23, n.s., d = .07.

Differential analyses revealed that parents described the relationship to their adult child as more intimate compared to the within‐person aggregates regardless of low or high family constancy, high family constancy: t (59) = 5.76, p < .001, d = .74; low family constancy: t (30) = 4.69, p < .001, d = .86 (cf. Table 6.15). Analyses did not show any differences in the perception of reciprocity. Old parents regarded their re‐ lationships to be fairly balanced in terms of give and take. Perceptions did not vary among relationship types or by contextual differences.

In sum, results indicate parental relationship perceptions to be, above all, cha‐ racterized by a high amount of closeness. This quantity clearly differentiates the pa‐ rental relation from other network partners. Reciprocity does not serve as differential characteristic. Most importantly, perceptions of parents are not influenced by family constancy. Relationship descriptions illustrate a high amount of closeness and overall balance regardless of stable or instable patterns in the family of origin.

Comparison of Perceptions of Adult Child and Old Parent. The upcoming question was whether family constancy yielded stronger effects on the younger generation or else, did parents perceive relationships to their children as more positive in general?

Thus, direct comparisons between the two generations were conducted.

Results are shown in Figure 10. No differences existed in perceptions of close‐ ness, t (180) = ‐1.33, n.s., d = .20. The implied effect was not substantial but was in the expected direction. Parents tended to consider the relationship to be closer than their adult children. With respect to reciprocity, significant variations occurred, t (180) = ‐

3.33, p < .01, d = .45. Old parents described the relationship to be more reciprocal as Results 146 compared to adult children. Across the entire sample, results indicated an effect of positivity in reciprocity for the older generation (cf., Carstensen & Mikels, 2005).

Figure 10. Group Differences in Relationship Percep‐ tions by Generation of Adult Children and Old Par‐ ents

To examine the stability of effects, the next step of analysis included family constancy, gender and within‐person aggregates (as general tendency) as additional explanation variables. Two sequential regression models were conducted with per‐ ceived closeness and reciprocity each as dependent variable. Predictors were in‐ cluded stepwise. In addition, a dummy coded variable was included to represent the generational affiliation. Table 6.16 displays standardized regression coefficients for the two dependent variables at each step of analysis.

For perceived closeness, the first step confirmed the non‐significant effect of generation affiliation ( = .10, n.s.). The variable did not contribute significantly to the overall model, F (1, 180) = 1.76, n.s. The next step added family constancy and gend‐ er. Variables improved the model substantially, F (2, 178) = 4.43, p < .05, adjusted R2 =

4%. The improvement was specifically due to the effect of family constancy. As last predictor, the within‐person mean of closeness was included. The variable served the function of a control variable. It may rule out that effects are only due to general ten‐ dencies of relationship perceptions. The adjusted R2 of model three indicated that Results 147 altogether, 22 percent of variability in the perception of closeness were explained by the four independent variables, F (1, 177) = 42.09, p < .001.

Table 6.16 Standardized Regression Coefficients Indicating the Effect of Generation, Family Constancy, Grand‐Mean Variable of Respective Outcome, and Gender on Perceived Closeness and Reciprocity in Intergenerational Relations

Closeness Reciprocity Model  p  p

1 Generation (Parent=1) .10 n.s. .24 < .01

2 Generation (Parent=1) .10 n.s. .24 < .01 Family Constancy (High = 1) .22 < .001 ‐.03 n.s. Gender (Female = 1) .10 n.s. ‐.01 n.s.

3 Generation (Parent=1) .04 n.s. .17 < .05 Family Constancy (High = 1) .19 < .01 ‐.02 n.s. Gender (Female = 1) .02 n.s. .04 n.s. Grand Mean of respective DV .44 < .001 .41 < .001

Adjusted R2 F Change Adjusted R2 F Change 1 .004 1.76 .05 11.09, p < .01

2 .04 4.43, p < .01 .04 0.07

3 .22 42.09, p < .001 .20 36.14, p < .001

Note. Substantial parameters are in bold face. Grand mean of respective DV = Grand mean aggregates of relationship perceptions across the reported social network of the child and the six selected rela‐ tionship partner of old parent for the respectively dependent variable of the regression analysis.

The amount of perceived closeness in intergenerational relations was substan‐ tially predicted by two variables: first, family constancy yielded a substantial effect.

Relationship partners of the low family constancy group reported less closeness as compared to the high constancy group ( = .19, p < .01). Second, perceived closeness in intergenerational relations was associated with the amount of closeness people reported across the entire network ( = .44, p < .001). The closer one felt to all other relationship partners, the closer the relationship within the child‐parent relation was described. There were no effects of generation affiliation or gender as well as no inte‐ raction effect between the predictors. Results suggest that the perception of closeness in intergenerational relations did not vary by age, as would be assumed by positivity effect. What is important about these findings, however, is that stability of structures Results 148 in the family of origin showed to be a significant predictor of perceptions. This effect was stable above and beyond overall network characteristics.

With regard to reciprocity, generation affiliation, led to the expected substan‐ tial contributions of variability, F (1, 180) = 11.09, p < .01. Parents perceived relation‐ ships more balanced than their adult children ( = .24, p < .01). The inclusion of fami‐ ly constancy and gender in the second step did not improve the model, F (2, 178) =

.067, n.s. The final model, with all four predictors included, explained about 20 per‐ cent of variability in perceived reciprocity, F (1, 177) = 36.14, p < .001. Standardized coefficients of model three indicated that the observation of mutuality in intergenera‐ tional relations depended on generation affiliation ( = .17, p < .05) and the general tendency of reciprocity perceptions across the network ( = .41, p < .001). No effects of gender or family constancy appeared and no significant interaction effects showed.

Results supposed a positivity effect in reciprocity perceptions of intergenerational relations that remained substantial above and beyond the overall tendency of reci‐ procity perceptions across the network.

In sum, results indicate the existence of differences between generations with regard to perceptions of reciprocity but not closeness. The inclusion of additional control variables demonstrated that effects of generation affiliation remained stable, indicating a specific relationship effect. Closeness perceptions, in contrast, were asso‐ ciated with family constancy.

6.3.2.2 Effects of Family of Origin on Dyadic Self‐ and Other‐Related Closeness and Recip‐

rocity Perceptions

Prior analyses focused on one or the other side of the relationship, or compared per‐ ceptions of dyadic partners. However, relationship research emphasizes that rela‐ tionship partners affect each others thoughts, feelings and behaviors (Cook & Kenny,

2006). This interdependency is regarded in actor‐partner interdependence models Results 149

(APIM)20 that take the dyad as level of analysis. Following models analyze whether one’s own perceptions of relationship quality to the respectively other relationship partner are somehow interacting with one’s own global self‐esteem (i.e. actor effects).

In addition, the models enable the interpretation of whether the relationship percep‐ tion of the dyadic partner is connected with one’s own self‐evaluation (i.e. partner effects). Family constancy is expected to affect the structure of interdependence be‐ tween child and parent.

Global Self‐Esteem as Regulatory Outcome. Descriptive statistics of global self‐ esteem measures point out that both generations indicate positive self‐evaluations

(adult child: M = .75, SD = .18; old parent: M = .78, SD = .17). Characteristics do not differ considerably by generational affiliation, t (182) = ‐1.18, d = .17, or family con‐ stancy, t (182) = .663, d = .12.

APIM’s included all main and all possible interaction effects (closeness with reciprocity for child and parent each, closeness of child with closeness of parent, re‐ ciprocity of child with reciprocity of parent). Only significant terms were kept and will be reported, subsequently. The second row of Table 6.17, referred to as ʹModel 1ʹ, illustrates the final APIM with standardized parameter estimates conducted for the entire sample of 92 dyads. The table presents estimates for all estimated paths: four actor effects, two partner effects, and the significant interaction term between emo‐ tional closeness of adult child and old parent. R2 captures the explained variability in self‐esteem of both dyadic partners. The lower section of the table illustrates the model fit of the final APIM. This section also includes the drawn comparisons with more restricted versions21 of the same model. In all tested instances the decline in

20 For all analyses below, it is important to bear in mind that measures were assessed at one point in time. The implication of causality is not suitable in this context. Instead, paths have to be addressed in a manner of correlative associations between predictors and outcomes.

21 Selected paths of the final model were fixed to be equal to test significant differences of those esti‐ mates. Declined model fit in more restricted models emphasized substantial variation between paths, and thus, facilitated interpretation of findings. Results 150 model fit indicated that estimates differed substantially (cf. Table 6.17 bottom sec‐ tion).

Table 6.17 Standardized Parameter Estimates of Actor‐Partner Interdependence Models of Closeness and Reciprocity Regressed on Global Self‐Esteem in Adult Child ‐ Parent Dyads

Model 1 Model 2 Whole Sample Family Constancy High Low N = 92 N = 60 N = 32 PE PE PE d2

Actor‐Effects Closeness Child  Self‐Esteem Child 0.09 0.14 ‐0.03 .18 Reciprocity Child  Self‐Esteem Child ‐0.03 ‐0.03 0.00 .05 Closeness Parent  Self‐Esteem Parent 0.29 0.33 0.31 .01 Reciprocity Parent  Self‐Esteem Parent ‐0.18 ‐0.20 ‐0.10 .07 Partner Effects Closeness Child  Self‐Esteem Parent 0.01 ‐0.15 0.17 .28 Closeness Parent  Self‐Esteem Child ‐0.07 ‐0.24+ 0.57 .64 Interaction terms Closeness Child*Closeness Parent  Self‐Esteem Parent 0.28 0.17 0.55 .00 Closeness Child*Closeness Parent  Self‐Esteem Child ‐0.15 0.04 0.05 .04 R2 SE child 0.03 0.05 0.27 SE Parent 0.10 0.11 0.13 Model Fit: (df) = Chi2 Final Model (2) = 0.55, n.s. (4) = 2.06, n.s. Fixed Actor Effects Closeness (4) = 107.59, p < .001 (8) = 105.19, p < .001 Fixed Actor Effects Reciprocity (4) = 176.19, p < .001 (8) = 182.04, p < .001 Fixed Partner Effects Closeness (4) = 147.11, p < .001 (8) = 138.54, p < .001 Fixed Interaction Effects (4) = 19.22, p < .01 (8) = 9.39, n.s.1

Note. Substantial parameters are in bold face, with p < .05 (+ p < .01). Four interaction effects (CL/RE for child and parent each, CL between child and parent, RE between child and parent) have been in‐ cluded in model 1 but only the reported effect showed significant results. For the significant interac‐ tion effects main effects are controlled for. 1 Even that the Chi‐Square test is not significant, a decline in model fit became obvious by checking additional fit indices (nested model: CFI = .87, RMSEA = .06). 2Effect sizes are based on unstandardized parameters.

Results of model one suggest two aspects: At first, relationship qualities are stronger connected to the self‐esteem of parents. This association is demonstrated by the significant actor effect between perceived closeness and self‐esteem of the parent

(PE = .29, p < .05). Fixing both actor effects of closeness to be equal declined the model fit. Thus, the association with parental self‐esteem was substantially stronger than it was in adult children. Second, the significant interaction term points out the assumed interdependence between the two generations (PE = .28, p < .05). It captures the inter‐ Results 151 play of children’s and parent’s closeness perception and is significantly associated with parental self‐evaluation. Parental self‐esteem was higher if both relationship actors perceive the intergenerational tie as particularly close.

Since the measure of self‐esteem in this study was only a global estimator of self‐evaluation, the explained amount of variance was unexypected, particularly in parents. Perceptions of closeness and reciprocity explained only 3 percent in the children’s but 10 percent in the parents’ self‐esteem. This result may serve as indica‐ tor of the relative importance of child‐parent relations in self‐evaluations of older parents.

Model two reflects variations between high and low family constancy, includ‐ ing this characteristic as grouping variable (cf., ʹModel 2ʹ in Table 6.17). Results pointed out discrepancies in patterns of relationship interdependence between the two types of family constancy. In dyads with high family constancy the significant actor effect in parents reappeared. Higher amounts of perceived closeness were asso‐ ciated with positive self‐evaluations (PE = .33, p < .05). This effect was not found for adult children (PE = .14, n.s., model fit of restricted model: χ2 (8) = 105.188, p < .001).

In addition, an only marginal significant partner effect appeared between closeness perception of parent and self‐esteem of adult child (PE = ‐.24, p < .1). Amount of closeness, described by the parent, seemed to be marginally connected to the child’s self‐evaluation. Unexpectedly, results supposed only a minor degree of interdepend‐ ence between the two relationship partners.

Dyads of low family constancy showed a different pattern. Unanticipated, there were no significant actor effects in this sub group. Patterns of self‐reliance were not supported. In fact, results indicated a significant partner effect between perceived closeness of parent and self‐esteem of child (PE = .57, p < .05). Self‐evaluation of the child was substantially associated with parental perceptions of closeness. Further‐ more, the interaction term between perceived closeness in parents and in adult child‐ ren was associated with self‐evaluations of parents (PE = .55, p < .05). This result sug‐ gested that in low family constancy parental self evaluations became more positive Results 152 when both parent and child perceived the intergenerational tie as close. These two paths represented partner effects, indicating interdependence between adult children and old parents. This pattern was contrary to the hypothesized self‐reliance in insta‐ ble family surroundings.

Considering the explained variances in the two types of family constancy, the perception of closeness and reciprocity were better predictors in dyads of low family constancy. Explained variances were rather similar in parents of the two types, with

11 percent in high and 13 percent in low family constancy. Yet, adult children dif‐ fered highly. Explained variance in high family constancy was only 5 percent for the adult child, instead closeness and reciprocity perceptions explained 27 percent for the children of low family constancy. Results were contrary to hypothesis 6b. Adult children in low constancy families particularly were expected to be independent of the child–parent relationship characteristics in their self‐evaluations.

6.3.2.3 Effects of Family of Origin on Relationship Regulation Characteristics across Time

Relationships are expected to have stable as well as malleable aspects. Variations eventually become obvious across time. Therefore, the last hypothesis on intergen‐ erational relationships addressed the stability and interplay between global self‐ esteem and relationship characteristics of the adult child across a 1‐year period.

Cross‐lagged analyses were conducted for adult children of low and high family constancy. Indicators of the first assessment predicted relationship specific percep‐ tions of closeness, reciprocity and self‐esteem at the follow‐up study. Stability of all three measures was considerably high. Nevertheless, change occurred. The predic‐ tion of change differed substantially between the two groups, leading to the separa‐ tion of analyses. It has to be noted that models of both family contexts do not show a perfect model fit. This may be due to the relatively small sample sizes. But results pointed out an interesting pattern which may not be unreported. At any rate, find‐ ings need to be replicated with a bigger sample. Results 153

High Family Constancy. Results indicated high stability of the relationship cha‐ racteristics (closeness: PE = .59, p < .001; reciprocity: PE = .48, p < .001) and of the global self esteem indicator (PE = .67, p < .001) across a one year time period. Pertain‐ ing to the change in emotional closeness and perceptions of reciprocity to the parent, self‐esteem did not have any predictive value in high family constancy settings. But relationship quality affected the adult child’s global self‐evaluation at time 2. Percep‐ tions of reciprocity in the parent‐child relationship at time 1 led to higher self‐esteem one year later (PE = .13, p < .05). In contrast, closeness did not have an effect on self‐ esteem at the follow‐up assessment (cf., Figure 11). The model showed a sufficient model fit, χ2 (4) = 6.31, p > .05, CFI = .99, RMSEA = 0.058 with 90% CI .00 ‐ .14.

High Family Constancy Low Family Constancy

Closeness .59 Closeness e Closeness .55 Closeness e .13 ‐.07 ‐.02 .02 ‐.00 .23

.08 Self‐Esteem .67 Self‐Esteem e .12 .60 Self‐Esteem .72 Self‐Esteem e .08 .20+ ‐.01 .13 .02 ‐.09 .05

Reciprocity .48 Reciprocity e Reciprocity .49 Reciprocity e

Χ2 (4) = 6.31, p = .18, CFI = .99, RMSEA = .058 (90% CI .00 ‐ .14) Χ2 (4) = 4.81, p = .31, CFI = .99, RMSEA = .056 (90% CI .00 ‐ .20)

Figure 11. Standardized Parameter Estimates of Path Analysis Predicting Closeness, Reciprocity, and Self‐esteem at Follow‐up in High and Low Family Constancy

Low Family Constancy. Similar to findings in high family constancy, both rela‐ tionship qualities (closeness PE = .55, p < .001; reciprocity PE = .49, p < .001) and glob‐ al self‐esteem (PE = .72, p < .001) showed to be highly stable across time. However, findings illustrated differences in the prediction of change. In adults with an instable family background, global self‐esteem substantially affected relationship perceptions at the follow‐up assessment. The more positive the self‐evaluation at time 1 the closer the adult child described its relationships to the parent one year later (PE = .23, p <

.05). Results on reciprocity perceptions suggested a similar effect, but failed to be substantial (PE = .20, p < .10). In contrast, self‐esteem of adult children at time 2 was unaffected by perceptions of emotional closeness and reciprocity at the first investi‐ Results 154 gation. Again, the model demonstrated a sufficient model fit, χ2 (4) = 4.81, p > .05, CFI

= .99, RMSEA = 0.056 with 90% CI .00 ‐ .20. Findings for both types of family constan‐ cy are presented in Figure 11.

In sum, quality in intergenerational relationships across time is affected by re‐ lationship specific as well as individual characteristics. Yet, patterns of findings differ as a function of stability in the family of origin. In high family constancy, relationship specific characteristics are associated with self‐evaluation at time two. In contrast, global self‐esteem is related to closeness and reciprocity in low family constancy.

Thus, individual characteristics affect relational change but not vice versa.

6.3.3 Family Study: Self‐Evaluation in Romantic Partnerships

Looking at heterosexual couple relationships, questions that occur are: What are in‐ terdependencies within dyads of different family structure? Do perceptions of close‐ ness and reciprocity affect the self‐esteem of individuals and their partners? To an‐ swer these questions, a dyadic sub‐sample is utilized for two reasons: (1) to explore mean level differences in relationship qualities between genders and across time, and

(2) to examine dyadic dependencies between spouses. Finally, spousal relationships are examined across time to address stability and change in characteristics of roman‐ tic relationship perceptions and self‐evaluation.

6.3.3.1 Effects of Situation on Relationship Perceptions in Romantic Relationships

Throughout the analyses of the social network, the relationship to the partner showed to be special ‐ particularly close and reciprocal. This previous observation is supported by results of this sub‐sample once more. Women and men alike describe the relationship to their spouse as very close and reciprocal. As shown in Table 6.18, at both assessment points men and women did not differ pertaining to degrees of emotional closeness and perceived reciprocity, closeness T1: t(51) = 1.56, n.s.; T2: t(51)

= .155, n.s.; reciprocity T1: t(51) = .162, n.s.; T2: t(51) = ‐.319, n.s. Results 155

Table 6.18 Descriptive Statistics of Relationship Perceptions in the Couple Sub‐Sample at T1 and T2 (N = 52 couples)

T1 T2 M SD M SD d

Closeness Men .95 .08 .90 .12 .49 Women .92 .14 .89 .14 .21 d .26 .07

Reciprocity Men .87 .20 .89 .14 .11 Women .87 .21 .90 .16 .16 d 0 .07 Note. Bold indicators are substantial at p < .01.

Mean‐level comparisons across time indicated one substantial difference with respect to closeness perception, t(51) = 3.40, p < .01, d = .49. Although intimacy was still at a high level, men reported less closeness to their spouse in the follow‐up. No substantial differences were found regarding closeness in women, t(51) = 1.67, n.s., or regarding reciprocity, men: t(51) = ‐.64, n.s.; women: t(51) = 1.56, n.s. There were no substantial effects of family type or parental status.

6.3.3.2 Effects of Family Situation on Dyadic Interdependencies in Romantic Partnerships

Similar to analyses in intergenerational settings, interdependencies within stable partnerships are expected. In the following dyadic analyses, emotional closeness and perceptions of reciprocity at the first assessment are regressed on the three self‐ esteem indicators at the follow‐up investigation: global self‐esteem, performance self‐ esteem, and social self‐esteem. Within all models, predictors were allowed to be cor‐ related to control for the intercorrelation of qualities at t1.

Self‐Esteem Indicators. Before conducting dyadic analyses, a brief descriptive overview on the self‐esteem indicators of the sub‐sample is given. Throughout, men and women report high levels of performance and global self‐esteem alike. No dif‐ ferences between genders appeared in global self‐esteem, T1: t(51) = .74, n.s.; T2: t(50)

= .85, n.s., and performance self‐esteem, T2: t(50) = .76, n.s. With respect to social self‐ esteem findings at the follow‐up investigation suggest a moderate level. Again, no difference occurred between males and females, t(50) = ‐.36, n.s. (cf., Table 6.19). Con‐ Results 156 siderations of global self‐esteem across time support high stability of the individual trait measure, males: t(51) = 1.60, n.s., females: : t(50) = .98, n.s.

Table 6.19 Descriptive Statistics of Self‐Esteem Measures in the Couple Sub‐Sample at T1 and T2 (N = 52 couples)

T1 T2 M SD M SD d

Global Self‐Esteem Men .78 .14 .76 .16 .19 Women .77 .16 .75 .19 .16 d .05 .05

Performance Self‐ Men .80 .15 Esteem1 Women .81 .14 d .05

Social Self‐ Esteem1 Men .61 .20 Women .63 .25 d .03 Note. Bold indicators are substantial at p < .01. 1Performance and social self‐esteem were not assessed at T1.

Dyadic Analyses. Starting out with an APIM on global self‐esteem, results indi‐ cated only small effects between relationship qualities and the global regulative out‐ come one year later. Within the full blown model22 only one substantial effect oc‐ curred. A further simplification of the model did not improve the model. Results suggest that in women only higher perceptions of reciprocity at time 1 led to higher self‐esteem at the follow‐up assessment (PE = .31, p < .05). In contrast, men indicated a non substantial but by tendency negative association between reciprocity and self‐ esteem (PE = ‐.18, n.s.). No interdependency occurred between partners (i.e., no sub‐ stantial partner effects). The differentiation by parental status did not lead to addi‐ tional insights. None of the paths were substantial.

With respect to the family situation, traditional and non‐traditional contexts were distinguished. The final model produced a good model fit, χ2 (8) = 4.75, p = .78;

CFI = 1.00, RMSEA = .00. Again, the actor effect between reciprocity and self‐esteem in women was substantial, but only for non‐traditional women (PE = .42, p < .05 vs.

22 Full blown models have no degrees of freedom. Thus no model fit can be reported. Results 157 traditional women: PE = .17, n.s.). Perceptions of reciprocity increased global self‐ esteem of women in non‐traditional settings, such as motivated childlessness and blended families. In both family situations men supported findings of the general model (non‐traditional: PE = ‐.04, n.s. vs. traditional: PE = ‐.16, n.s.). With respect to performance self‐esteem, none of the paths were substantial. Thus, no further analys‐ es occurred.

Finally, APIMs with social self‐esteem as outcome were conducted. The full‐ blown model showed a different pattern compared to global self‐esteem. Reciprocity perceptions of men affected their own social self‐esteem (PE = ‐.32, p < .05) and the social self‐esteem of their spouses (PE = .28, p < .05). Surprisingly, higher reciprocity perception in men at time one decreased their own social self‐esteem at the follow‐up but increased the social self‐esteem in their spouse. The substantial partner effect emphasizes substantial interdependency between partners pertaining to the regula‐ tion of social self‐evaluation. Women did not show any of these patterns (Reciprocity actor effect: PE = .03, n.s.; partner effect: PE = ‐.02, n.s.).

The consideration of parental status in the full‐blown model indicated that on‐ ly childless males have a positive effect on their women’s social self‐evaluation (PE =

.56, p < .05). In parents, this partner effect was not substantial (PE = ‐.28, n.s.). The partner effect in childless couples accentuates a higher interdependence compared to parents. No additional effects occurred. Further considerations of family situation did not yield substantial results.

In sum, APIM analyses generally showed only very few substantial effects be‐ tween relationship perceptions in romantic partnerships and self‐esteem indicators.

Occurring effects suggested that perceptions of reciprocity in spouses were more im‐ portant for self‐evaluations throughout. However, reciprocity showed to be particu‐ larly important for women. In global self‐esteem reciprocity perceptions of non‐ traditional women affect their own self‐evaluation. In contrast, childless men’s reci‐ procity perceptions were associated with their spouse’s social self‐esteem. Findings are only based on a small dyadic sample and call for further replication. Results 158

6.3.3.3 Effects of Family Situation on Relationship Regulation Characteristics in Romantic

Partnerships across Time

The last part of this result section examines the stability and change in relationship perceptions and global self‐esteem in steady romantic relationships. In cross‐lagged models, emotional closeness, perceived reciprocity and global self‐esteem at time one were regressed on indicators at the follow‐up assessment. Predictors at the first as‐ sessment were allowed to be correlated. Domain‐specific indicators were not in‐ cluded since they were not part of the first assessment in the family study.

Starting out with a full blown model, stepwise deletion of non‐substantial ef‐ fects led to a final model with good model fit, χ2 (3) = 2.20, p = .53, CFI = 1.00, RMSEA

= .00. All three indicators showed to be relatively stable across the one year period

(Closeness: PE = .49, p < .05; reciprocity: PE = .47, p < .05), with highest constancy in the trait measure of global self esteem (PE = .73, p < .05). But there was also consider‐ able change. Global self‐esteem at time one increased perceptions of reciprocity at the follow‐up assessment (PE = .18, p < .05). Change in self esteem was only marginally predicted by higher reciprocity perceptions (PE = .11, p < .10) and not by emotional closeness (PE = .10, n.s). None of the other predictors yielded substantial effects on closeness perceptions. However, residuals of closeness and self‐esteem showed to be significantly associated at the follow‐up measure (cf. Figure 12). Including context variables, such as family situation or gender, decreased the model fit substantially.

R2 = .24

Closeness .49 Closeness e .15 ‐.10 .15 R2 = .53

.10 Self‐Esteem .73 Self‐Esteem e .09 .18 ‐.00 .05 .11+ R2 = .25

Reciprocity .47 Reciprocity e

Χ2 (3) = 2.20, p = .53, CFI = 1.00, RMSEA = .00 (90% CI .00 ‐ .15)

Figure 12. Standardized Parameter Estimates of Path Analysis Predicting Closeness, Reciprocity, and Global Self‐Esteem at Follow‐up in Roman‐ tic Partnerships Results 159

In sum, the trait measure of global self‐esteem and the two relationship indica‐ tors showed to be relatively stable in romantic partnerships across a one year period.

Prediction of change occurred only between perceived reciprocity and global self‐ esteem. High self‐esteem increased perceptions of reciprocity substantially. Further‐ more, perceptions of reciprocity also appeared to raise the global self‐esteem across time. No contextual effects were found, thus hypothesis 7b is only partially sup‐ ported.

6.3.4 Summary: Functionality of Relationships – Relational and Individual Self‐

Evaluation

A differential relevance in communal‐esteem is found among relationship partners and in specific age groups. Indicators of mutual‐esteem illustrate fewer effects. Emo‐ tional closeness and perception of reciprocity demonstrate strong predictive value with respect to relational self‐esteem indicators. Closeness is positively related to re‐ lational esteem and enhances the likelihood of mutuality in esteem perceptions. Im‐ balance in social relationships seems to affect communal and mutual esteem to a higher extend in kin and elective kin relations. There are substantial variations by age groups. Thus, results on relational self‐esteem support hypothesis 5a. Considera‐ tion of individual trait measures indicates only few effects with communal‐esteem.

As expected, social self‐esteem shows most associations. However, esteem in kin‐ relationships is unaffected by trait and state self‐esteem. With respect to mutual‐ esteem only general effects on the intercept are tested. The indicator shows few sub‐ stantial results. Hypothesis 5b is supported.

Dyadic analyses on intergenerational relationships between adult children and parents in the family study partially support hypotheses 6a. Results show that with respect to relationship perceptions effects of family constancy are more distinctive in the younger generation. Parents perceive the relationship as generally more reciproc‐ al, but not as closer. Hypothesis 6b is not supported. Dyadic interdependence shows an unanticipated pattern between adult children and old parents. Low family con‐ Results 160 stancy dyads are not more self‐reliant compared to those with high family constancy.

In general, interdependency between children and parents is found to be relatively low. The adult child’s relationship perception across time supports hypothesis 6c.

Closeness, reciprocity and self‐esteem indicate the expected characteristic of stability and change across the one year period. Variations occur in the hypothesized differen‐ tial pattern between family types. In high family constancy, relationship perceptions are associated with self‐esteem. Thus, relationship perceptions affect the adult child’s self‐evaluation. In low family constancy, the reverse pattern shows. Self‐esteem is related to relationship perceptions one year later. Results imply a tendency of self‐ reliance in relationship regulation strategies.

As hypothesized, romantic partners are highly important throughout. Find‐ ings on dyadic interdependency suggest the importance of reciprocity perceptions in spousal relationships. This is particularly true for non‐traditional family situations with respect to global self‐esteem. Unexpectedly, traditional family situations are not associated with a differential pattern that implies the significance of closeness per‐ ceptions. In social self‐evaluations, childless partnerships in general emphasized the importance of reciprocity perceptions. Thus, hypothesis 7a is only partially sup‐ ported. This is also true with respect to hypothesis 7b. Although all indicators show the expected pattern of stability and change; and reciprocity perceptions prove to be important for self‐esteem across time, results do not confirm any contextual effects.

Motivated and involuntarily childless couples as well as both parental family situa‐ tions do not illustrate any differences. Thus, reciprocity seems to have a differential effect in the dyadic relationship regulation among diverse family situations. Howev‐ er, relationship change is not affected by contextual differences. Results 161

6.4 Summary of Findings in Relation to Hypotheses

Table 6.20 summarizes all results of the present study. It relates them back to the ini‐ tial hypotheses and indicates whether hypotheses were confirmed or not. If hypothe‐ ses are not or only partially supported by empirical results, a note briefly summa‐ rizes in what way the results differ from predictions.

Table 6.20 Summary of Results

Hypothesis Data supported hypothesis?

Self‐Esteem across Life Span and Family Situations (1) Effects of Age. Age‐associated patterns of global and domain‐specific Yes self‐esteem measures vary across the life span. The differentiation between global and social indicators increases in late adulthood and old age. Indicators of positive functioning, such as life satisfac‐ tion and subjective health, differentially predict global and domain‐ specific self‐esteem indicators. (2) Effects of Family Situation. Domain‐specific self‐esteem, particularly Partially: the indicator of social self‐esteem, show higher sensitivity to differ‐ Performance self‐esteem did not ential family characteristics than global self‐esteem. show differential family effects.

Social Networks and Relationship Regulation across Life Span and Family Situations (3) Effects of Age. Personal network structures and relationship regula‐ Yes tion characteristics differ across age groups, showing higher pro‐ portion of kin network partners and kinship preference in older age. (4) Effects of Family Situation. Social networks of motivated childless Partially: individuals are less oriented toward kin relationships and more Involuntary childless individuals focused on cooperative relationships. In contrast, involuntarily do not differentiate under‐ and childless individuals will show structures similar to parents, being over‐benefit among genetic kin, oriented more toward kin relationships and less toward non‐kin but only in non‐kin. Blended fami‐ relationships. Motivated childless individuals will demonstrate lies show more similarity in recip‐ lower levels of kinship preference, compared to the involuntarily rocity regulation to traditional childless and parents. The differentiation in reciprocity perception families. between kin and non‐kin is highest in involuntary childlessness and traditional families and lowest in motivated childlessness and blended families.

Functionality of Relationships – Relational and Individual Self‐Evaluation (5 a) Effect of Relationship Quality. Self‐evaluations of perceived relational Yes esteem in relationships are relationship and age specific. Mechan‐ isms of closeness and reciprocity perceptions in social networks differentially predict the relevance of relationship partners. (5 b) Effect of Level of Data. Relational self‐esteem as a relationship specific Yes indicator of self‐evaluation differs from individual self‐esteem measures. Only small interdependencies occur between the two levels. Strongest effects are expected with social self‐esteem. (Table continues) Results 162

Hypothesis Data supported hypothesis? (6 a) Effect of Family of Origin. Relationship quality for adult children Partially: with low consistency in the family of origin is characterized by less Reciprocity perceptions do not closeness and more imbalance than that for adult children with differ between adult children of high consistency. This difference will be less noticeable in the older high and low family constancy. generation. Older parents perceive intergenerational relations as Positivity does not exist in close‐ more positive than adult children, regardless of family constancy. ness perceptions. (6 b) High constancy in the family of origin is associated with stronger No: relational interdependence. In contrast, low family constancy in‐ There is more dyadic inter‐ creases self‐regulative strategies. dependence in child‐parent ties of low family constancy. (6 c) Relationship perceptions of adult children are characterized by Yes stability and change across a one year period. In high family con‐ stancy change is associated with relationship perceptions. In low family constancy, adult children rely on self‐regulative strategies. (7 a) Effect of Family Type. In general, the romantic partner is highly im‐ Partially: portant across time and regardless of family type or parental status. No dyadic effects show with re‐ The degree of dyadic interdependence between relationship quali‐ spect to emotional closeness in ties and self‐esteem will depend more on reciprocity perceptions involuntary childlessness and for motivated childless and blended family couples, but more on traditional families. Few family emotional closeness in spousal ties of involuntarily childlessness context effects occur. and traditional families. Effects differ with respect to global and domain‐specific self‐esteem. Most effects occur regarding social self‐esteem. (7 b) Relationship perceptions and self‐esteem of romantic partners are Partially: characterized by stability and change across time. In motivated Self‐evaluation in spousal relation‐ childless individuals and blended families, change is related to ships is generally more related to perceptions of reciprocity, and in involuntary childless individuals perceived reciprocity and not and traditional parents it varies as a function of emotional close‐ emotional closeness ness. Discussion 163

7 DISCUSSION

This study provides new insight on two issues: first, diversity as well as age‐ and context‐related patterns of self‐esteem, and second, the connectedness of self‐ evaluation and social relationship regulation. The new knowledge on the above is‐ sues is then used to explore personal networks and social relationship characteristics across age groups and in the specific family context of motivated and involuntary childlessness. The following discussion addresses three main aspects: (a) self‐esteem across life span and family situations, (b) social networks and relationship regulation across life span and family situations, and (c) the functionality of relationships re‐ garding individual and relational self‐evaluation. It is organized as follows: First, the main findings of the empirical studies are summarized and each is followed by an integrative discussion in the context of the existing empirical and theoretical litera‐ ture. Next, the discussion focuses on some final remarks about the relational aspect of self‐evaluation and the specific phenomenon of childlessness. The study concludes with a discussion of limitations and on possibilities for future research projects.

7.1 Different Facets of Self‐Evaluation across Life Span and Family Situations

Findings on self‐esteem suggest that there are differences between global and social self‐esteem in younger and older age groups. Age‐associated patterns of social self‐ esteem differ from global and performance indicators. Intercorrelations and regres‐ sion analyses suggest a differential meaning of social and global self‐esteem across the life span. As hypothesized, only social self‐esteem differs because of family situa‐ tions. Mean‐level distinctions occurred in the structure of the family of origin be‐ tween high and low constancy as well as between involuntary childlessness and oth‐ er family types.

The present study supports patterns of global self‐esteem with respect to an age‐diverse German online sample. The overall results were similar to the results of Discussion 164

Robins and colleagues (2002), yet there was one exception. The decline in the oldest group of the Robins et al. sample (age 80‐90) was much steeper than in the current study. Indeed, this difference could be due to age differences between samples. Old age in the current study ranged from age 75 to 86 and from 80 to 90 in Robins’s sam‐ ple. Other possible explanations could be the high variability of self‐esteem in old age (Trzesniewski et al., 2003) or the overall positive state of health of the current sample until old age.

Global self‐esteem is a widely used and discussed measure in personality and social psychology. Research studies regularly concentrate on self‐esteem as a trait like surface personality construct. Only within the last two decades, personality re‐ search increasingly addressed stability and change in core personality constructs

(e.g., Allemand, Zimprich, & Martin, 2008; Roberts, Robins, Trzesniewski, & Caspi,

2003; Roberts, Walton, & Viechtbauer, 2006; Robins, Fraley, Roberts, & Trzesniewski,

2001). In this context, Trzesniewski and colleagues (Robins et al., 2002; Trzesniewski et al., 2003) conducted one meta‐analytical and multi cross‐sectional investigations to provide evidence of stability and change in global self‐esteem across the life span.

Similar to core personality traits, mean‐levels of self‐esteem increase and stabilize throughout adulthood. Yet, findings showed a robust curvilinear trend with a de‐ cline in old age.

Another important finding refers to results that extend and advance our knowledge of domain‐specific patterns of self‐esteem. Performance and social self‐ esteem illustrated a differential outline of self‐esteem across the life span. This is par‐ ticularly true if comparing global and social self‐esteem. Age‐related patterns of the global measure showed high similarity to the performance indicator. Social self‐ esteem clearly differed, mostly in the youngest and oldest age groups. The outline across five age groups indicated that social self‐esteem is very low in early and young adulthood. Mean levels were higher for middle and late adulthood.

Early adulthood is characterized by vast changes. Individuals are confronted with a new autonomy. They leave their parental homes, go to college or get their first Discussion 165 job. They meet numerous new people and have to cope with unfamiliar situations and challenges. Thus it seems adaptive to be open and sensitive to social feedback. In young and middle adulthood, core personality traits and global self‐esteem show increasing stability (Roberts & DelVecchio, 2000; Trzesniewski et al., 2003). This may contribute to higher social self‐confidence in middle adulthood. This contribution of global self‐esteem was supported by age‐specific intercorrelations between global and social self‐esteem. Particularly in young and middle adulthood, this correlation suggests a stronger association between global and social as well as performance and social self‐esteem. Thus mastery and positive evaluation of life, self, and performance outcomes seem to play a pivotal role in young and middle adulthood. The high pre‐ dictive value of general life satisfaction in global and performance self‐esteem sup‐ port this point. Moreover, the interaction effect between age and life satisfaction indi‐ cated that, particularly in young adulthood, low life‐satisfaction is associated with low levels of global self‐esteem. Results for young and middle adulthood suggest not only a continuous convergence of mean levels, but also a stronger overlap of social and global self‐esteem in these age groups.

A key finding in this context was the pattern of social self‐esteem in old age. It was hypothesized that social and global self‐esteem increasingly diverge in older age. Two results supported this assumption. First, in contrast to global and perfor‐ mance mean‐levels, social self‐esteem scores did not differ between late adulthood and old age. Both age groups reported the same levels of social confidence. In con‐ trast, global and performance self‐esteem mean levels were lower in old age. Second, low and even insignificant intercorrelations occurred between social and global self‐ esteem in the last two age groups. Patterns suggest that global and social self‐esteem tap into different aspects of the self‐evaluation mechanism. In contrast, the high asso‐ ciation between global and performance measure across the life span indicate that global self‐esteem is more likely to assess performance specific aspects. A possible explanation behind this differentiation could be that in old age the decline in mental and physical health may contribute to perceptions of decline in such skills as intellec‐ Discussion 166 tual capacity. It seems adaptive when people differentiate between performance and social self‐esteem in later adulthood and prevent negative self‐evaluations with re‐ spect to abilities and contexts. Results are consistent with findings on well‐being

(e.g., Diener, Lucas, & Scollon, 2006; Gerstorf, Ram, Estabrook, Schupp, Wagner, &

Lindenberger, 2008). Despite numerous losses and gradual deterioration, there is no negative age‐difference of social self‐esteem, indicating a human capability to adapt to changing life conditions.

The increasing differentiation may also be due to changed compositions of so‐ cial networks and characteristics of interpersonal relationships in old age. Variations are expected to be based on age‐related differences in information processing and shifts in future time perspective (cf., Antonucci & Akiyama, 1995; Carstensen; 1992;

Fingerman et al., 2004; Lang & Carstensen, 1994, 2002). Yet the general ability of in‐ dividuals to function within social contexts remains stable and important. Thus the differentiation of social self‐esteem from global and performance self‐evaluation may be considered as an adaptive process. It represents one possibility to maintain a posi‐ tive evaluation of the self, which has been shown to be a self‐relevant goal through‐ out the life span (Tesser, 2003). Further studies are needed to replicate these findings on self‐esteem differentiation.

To date, the majority of psychological studies assess only global self‐esteem using the ten‐item Rosenberg scale (Gray‐Little, Williams, & Hancock, 1997). Studies continue to measure self‐esteem in this manner, despite the suggestion in several lines of research that self‐esteem is a multi‐dimensional construct (i.e., Crocker &

Luhtanen, 2003; Crocker & Park, 2004; Fleming & Courtney, 1984; Heatherton & Po‐ livy, 1991; Schütz, 2000a). For example, Crocker and colleagues (Crocker & Luhtanen,

2003; Crocker & Park, 2004) suggest that beside characteristics (high or low), consid‐ erations have to focus on contingencies of self‐esteem. Contingencies or domains are seen as possibilities to prove a specific worth, but they also incorporate an increased likelihood of , particularly if important domains are not accomplished (Crocker

& Luhtanen, 2003). Findings of this study point out a need for domain‐specific re‐ Discussion 167 search. It could improve the understanding of stability and change in self‐evaluation, particularly in old age.

Finally, a third finding verifies social self‐esteem as more sensitive indicator of self‐evaluation. This indicator reflects contextual differences. Involuntary childless individuals show lower social self‐esteem compared to all other family types. Thus, they are more concerned about the impression they make in social contexts and show higher levels of self‐consciousness. This result aligns with previous findings of social insecurity and shyness in the context of involuntary childlessness (e.g., Houghton &

Houghton, 1987; Sabatelli et al., 1988). Having a strong, yet unaccomplished desire for children indicates that people would love to be parents but have not achieved that status. Perceptions of social exclusion (from the parental group) may be directly associated with lower self‐esteem (Leary, 2003a; Baumeister & Leary, 1995; Leary et al., 1998). Results at several levels suggest that involuntarily childless individuals live in an ʺin‐betweenʺ status and therefore lower levels of social self‐esteem are a rea‐ sonable result.

There are possible alternative explanations that should be ruled out. First, lower self‐esteem could be due to higher stressors with respect to social interactions or emotion regulation in involuntarily childless people (e.g., Exley & Letherby, 2001;

Greil et al., 1988). However, results show no differences in global or performance self‐esteem. Also, family types do not differ with respect to general life‐satisfaction.

Second, involuntarily childless individuals indicated that social acceptance is one im‐ portant aspect of parenthood. They seem to put emphasis on a domain of self‐ evaluation that they can not sufficiently accomplish (Crocker & Luhtanen, 2003;

Crocker & Park, 2004). Thus, social self‐evaluation becomes less positive and less confident. In contrast, motivated childless individuals report no such desire. Instead they emphasize that children decrease personal freedom and are not an essential as‐ pect of identity. Although some empirical investigations find tendencies of segrega‐ tion and stigmatization (Bernardi, 2003; Park, 2002), motivated childless couples in‐ tentionally choose this life style (Carl, 2002). Social context will adapt to this decision. Discussion 168

Thus, general tendencies of self‐evaluation are positive and not distinguishable from parents in this sample.

An unexpected finding in this context occurred with respect to the family of origin. Despite a main effect of family constancy, the interaction effect indicates that gender is a defining factor. That is, only women with a background of low family constancy have higher social self‐esteem. This was true compared to all reference groups. The theoretical argumentation suggested that low family constancy is gener‐ ally associated with less social interdependence and higher self‐reliance. Yet, results confirmed this relationship only for women. What would be a possible explanation for this result? Higher social self‐esteem is understood as higher self‐confidence and less concern about social impressions on others. The fact that only women show this tendency may be associated with evolutionary argumentations. The maturation and development of daughters has been shown to be particularly effected by family in‐ stability (Ellis, 2004; Ellis & Garber, 2000; Neberich et al., in press). Experiences dur‐ ing childhood and adolescence seem to shape relationship and self‐perceptions until adulthood specifically of female offspring. This may explain the singular effect for women.

Summary. Results for the first set of hypotheses do the following (a) confirm differences in age‐related patterns between selected self‐esteem measures, (b) suggest that social self‐esteem taps more into differential aspects of self‐evaluation than glob‐ al and performance self‐esteem, and (c) indicate a high sensitivity of social self‐ esteem to reflect contextual differences. Feedback of social interaction partners ap‐ pears to have a fundamental effect on the continuous interplay between self‐ evaluative and self‐regulative processes. A shift of self‐defining contingencies is im‐ portant to maintain a positive self across age groups and family situations. Discussion 169

7.2 Social Networks and Relationship Regulation Characteristics: Differential

Patterns Pertaining to Age Group and Family Situation

The second complex of results illustrates similarities and differences between social network structures and relationship regulation characteristics. As expected the num‐ ber of kin and elective kin varied substantially across life span and family contexts.

In addition, they show differential characteristics of kinship preference. Only a few effects occur in the regulation of over‐ and under‐benefit. Findings support and ex‐ tend prior research on the topic.

Age‐related Effects. Unexpectedly, there are no substantial differences between age groups in overall network size. Although a certain drop is noticeable in older age, variation within the group is too large to clearly distinguish it from the remain‐ ing age groups. Network composition, however, varies substantially. In early and middle aged adults, networks are characterized by more diversity: kin, in‐law family, colleagues, and friends. In late adulthood and old age, kin and elective kin become principal network partners. Such age related changes in personal network structures have been described before in German (Neyer & Lang, 2003), French (Bidart & Lave‐ nu, 2005), American (Fingerman et al., 2004), Japanese (Antonucci et al., 2004), and

Chinese samples (Fung, Stoeber, Yeung, & Lang, 2008). Findings of the current study also showed that two specific relationships remained important and stable in their presence across the entire adult life: ties to siblings and to long‐term friends. This may be due to the age‐similarity with these relationship partners (long shared life‐ time), to the specific role of siblings in support structures (Campbell, Connidis, &

Davies, 1999), or to the previously described relationship quality with siblings and long‐term friends (e.g., full‐blown attachment figure in Doherty & Feeney, 2004). So‐ cial network structures demonstrate that people remained socially embedded until late in life. Only the nature of their network partners changed with age. Differential patterns show the expected patterns.

In addition, general structures of relationship regulation mechanisms show age related differences. Kinship preference, as one indicator of closeness regulation, Discussion 170 is generally prevailing but increases over the life span. Mean‐level differences among the five age groups show substantial discrepancies until late adulthood and stabiliza‐ tion in old age. In late adulthood and old age individuals describe relationships to biological kin as closer, with an even more pronounced distinction from non‐kin.

Thus older people not only have more relationships with kin, but they also describe those relationships as more significant. Genetic relatedness is a substantial predictor and cursor of social relationship regulation. Findings replicate earlier results by Ney‐ er and Lang (2003).

Effects of perceived over‐ and under‐benefit are less clear in their descriptive patterns and their association with emotional closeness. Thus the discussion of the results starts out with the consideration of descriptive patterns. Descriptive findings suggest that the majority of social interactions are balanced, followed by costly and then beneficial relations. Patterns in older age groups were particularly surprising.

These participants report a higher number of relationships with under‐beneficial ex‐ change, i.e. where they give more than they get, compared to younger age groups. In contrast, fewer relationships were described as over‐beneficial. At least two possible explanations could account for this unexpected tendency. First, the good health sta‐ tus of this study’s older participants suggests that they are still independent and not reliant on support by others. Second, support exchanges have a general tendency to benefit the younger generation more than the older generation (Buss, 2004).

Fewer substantial effects occur with respect to the monitoring of reciprocity in social networks. Imbalance is generally associated with less closeness. Younger age groups show more sensitivity to non‐reciprocity in ties to family and older age groups in relations to non‐kin. Different explanations are possible. First, people strive for balanced relationships in general. They purposefully try to avoid both providing and receiving over‐benefits (Uehara, 1995).

Second, age effects for younger age groups may be due to network structures.

Perceptions of reciprocity are more likely within (family) generations than across them (Ross et al., 2005). Younger people not only have a diverse social network, but Discussion 171 they may also be particularly involved with same‐age non‐kin partners. Thus they perceive high levels of reciprocity in general. Their relationships to kin may be more often across generations, making perceptions of imbalance more likely. In addition, help from family‐members could be perceived as threatening in specific situations and age groups (Ackerman & Kenrick, 2008). Early and young adults tend to estab‐ lish an autonomous life style. They may want to start their own family, and be inde‐ pendent of parental support.

Third, older individuals are likely to adapt to changing personal and physical conditions. For example, decline in cognitive functioning may decreases brain capaci‐ ty for reciprocity monitoring. Thus social ties to close relationship partners are pre‐ ferred since exchange patterns differ ‐ or are even dissolved ‐ in this context (Clark et al., 1994). This study shows a lower presence of non‐kin relations and a focus on

(elective) kin ties in social networks of older people. At the same time, negative cor‐ relations with closeness suggest that sensitivity to imbalance remains high for non‐ family members. Older people tend to end unbalanced relationships with non‐kin, retaining only established and lasting social relationships. Yet, they remain in non‐ reciprocal kin ties (Ikkink & van Tilburg, 1998, 1999). Thus they represent the result of the adaptation and attainment process which is driven by the norm of reciprocity

(Perugini, Gallucci, Presaghi, & Ercolani, 2003). Recent findings additionally imply that it is not only the general action of reciprocation that matters. The meaning of social exchange has to be considered to fully understand the dynamics of perceived reciprocity in social interaction (Keysar, Converse, Wang, & Epley, 2008).

Contextual Effects. The family situation has substantial effects on network size and structure as well as relationship regulation. Motivated childless individuals have significantly smaller networks, compared to all other family types. They list fewer kin and elective kin as social network partners. They also do not show the expected increase in non‐kin relationships to colleagues or acquaintances. In short, results present a decrease of kin contact with no compensating increase of non‐kin ties. Ob‐ served differences are not only due to the absence of children, because involuntarily Discussion 172 childless individuals do not differ substantially from parents of traditional or blended families. Also the availability of in‐laws does not account for the difference, because all participants live in stable romantic partnerships, which, except for tradi‐ tional parents, do not differ in duration. Prior studies on older childless adults report similar findings (Dykstra, 2006): childlessness did not increase sociability but simply indicated limited family contact.

The social relationships of involuntarily childless individuals are similar to those of parents and of motivated childless alike. With respect to size they do not differ from any family type. In composition they are more likely than motivated childless individuals to name kin relationships such as parents. In this context, they are similar to traditional families. Yet, the absence of their own children makes them obviously ʺdifferentʺ from this group. Results contradict the few previous studies that associated infertility with social distinction (Bernardi, 2003). Involuntarily child‐ less couples demonstrate social embeddedness. Nevertheless, in comparison to older childless individuals (Dykstra, 2006), where motivational structures are often un‐ known or uncertain, there is a noticeable similarity with network structures of young and middle‐aged motivated childless individuals. Thus involuntarily childless people that remain childless may experience an adaptation process. In old age their social networks are expected to become comparable to those in this study’s moti‐ vated childless individuals.

Only very few differences among family types occur in perceptions of close‐ ness and reciprocity among social relationships. First, involuntarily childless indi‐ viduals describe enhanced emotional closeness to their siblings. This is particularly true compared to traditional families. Findings may be an effect of indirect reproduc‐ tion, where a sibling represents the closest biological ancestor. Second, an unex‐ pected result occurs regarding close family members. Motivated childless people de‐ scribe relationships to close kin as more significant than do all reference groups. It suggests that despite fewer kin relations in their social networks this group may Discussion 173 maintains only particularly important and meaningful ties. This finding supports the active part of relationship regulation.

Results support hypothesized differences in kinship preference but are less indicative in effects of reciprocity monitoring. Kinship preference is a distinctive re‐ gulative characteristic of involuntary childlessness, traditional and blended family parents. They feel substantially closer to kin than to non‐kin network partners. In contrast, motivated childless individuals do not show a substantial association be‐ tween emotional closeness and degree of genetic relatedness. They do not make dis‐ tinctions in closeness perceptions as a function of genetic similarity. Results support previous findings of lower kinship orientation (Neyer & Lang, 2003) but only in the context of motivated and not involuntary childlessness. Thus they confirm the diffe‐ rentiation between the two types of childlessness.

Patterns are less distinctive regarding sensitivity of imbalance in kin and non‐ kin. Motivated childless individuals are once more unaffected by genetic (dis‐) simi‐ larity. There is no substantial association between perceived closeness and perceived over‐ or under‐benefit in kin ties. Possible reasons are an open question: Have non‐ reciprocal kin ties simply been ended? Do they put a lot of effort into the achieve‐ ment and perception of reciprocity? Is genetic relatedness simply not a meaningful predictor in this context? Missing the actual need to form a family seems to affect regulative mechanisms. As indicated by results of the validation check, behaviors are generally more driven by agentic life goals, such as self‐fulfillment and indepen‐ dence, than by communal life goals for family and children (cf., also Carl et al., 2000).

In this line, kinship effects on social perceptions and behaviors seem to decrease.

Contrary to expectations, involuntarily childless individuals demonstrate high similarity to the motivated group in this context. Perceptions of under‐ and over‐ benefit with kin do not affect emotional closeness. This is not true for traditional or blended family parents. Possible explanations for this pattern may be found in evolu‐ tionary theory. The biological family constitutes an important institution for involun‐ tarily childless individuals. Beyond seeking meaningful and intimate relationships, Discussion 174 involuntarily childless couples depend on their families to carry on the family’s ge‐ netic heritage. Thus they may be more likely to accept non‐reciprocal relationships in both directions: either being the one to over‐benefit and generally sustain the family, or being grateful for the support the family gives in the demanding situation of infer‐ tility (Sabatelli et al., 1988; Voß et al., 1994).

In non‐kin ties perceptions of under‐benefit are related to less emotional closeness. This effect occurred in all but blended families. Relationships to step‐ family members, particularly step‐children, will likely alter reciprocity monitoring in this family type. In addition, involuntarily childless individuals generally tend to feel less close to imbalanced non‐kin ties. Negative effects occurred regardless of percep‐ tions of over‐ or under‐benefit, the opposite of the situation with kin ties. This high sensitivity to imbalance could be due to restricted resources that are available besides the engagement into their primary goal “family”.

Summary. Bringing together findings on different age groups and the family study highlights differential characteristics in closeness regulation. Kinship orienta‐ tion seems to strengthen throughout adulthood, varying among age groups. This pattern of variety depends not only on age‐related changes, but also on contextual and motivational aspects. This is supported by findings on motivated childlessness.

Results are further illustrated by social network compositions. Middle aged and el‐ derly as well as parents and involuntarily childless individuals specify more kin and elective kin as relationship partners. In contrast, reciprocity monitoring does not produce a clear pattern. Like age specific findings, family characteristics show either consistency or only minor differences in reciprocity monitoring. Evidence of few ef‐ fects in network composition confirms this pattern. Whether results mirror the exist‐ ing characteristic or are based on methodological limitations is open to question. This discussion is carried on in section 7.6. Discussion 175

7.3 Social Relationships Differentially Affect Self‐Evaluation across the Life

Span and in Specific Relationship Dyads

The third set of findings indicates that social relationship partners are a source of perceived relational self‐esteem, but vary substantially with respect to perceived amounts of esteem. Relational value of interaction partners is substantially affected by relationship qualities of emotional closeness and perceptions of reciprocity. Yet specific ties are differentially affected by relationship qualities. Closeness perceptions are more relevant for self‐evaluation in child‐parent ties. Reciprocity is a better pre‐ dictor of self‐esteem in romantic partnerships. Contextual characteristics of family situations do not establish the expected strong differential pattern.

7.3.1 Calibration of Relational Self‐Esteem

Network partners differ in their significance for self‐evaluation. Meaning varies by relationship type. Discrepancies are particularly strong in communal‐esteem and less strong in mutual‐esteem. Romantic partners, if existent, are most important. Close kin, siblings, and long‐term friends are described as meaningful as well. Least value is given to acquaintances. Results indicate the differential importance of feedback on self‐evaluation.

Leary and colleagues (Leary et al., 1998) show that the self‐esteem sensor is less sensitive to extremely negative or positive feedback, compared to moderate val‐ ues. ʺParticipants seemed to be no more bothered by extremely negative reactions

[…], and […], exceedingly accepting responses from others evoked no greater posi‐ tive responses than moderate acceptanceʺ (p. 1298). Authors suggest that variations in the middle tiers of acceptance and rejection are more important in everyday life.

The credibility of feedback may also be higher, particularly regarding unknown in‐ teraction partners in experimental designs. Explorations of relational self‐esteem need to account for possible relationship‐specific baselines or a differential sensibility to certain relationship partners. In this context, upcoming research should ask such Discussion 176 questions as: Do meaningful relationship partners start out at a different level per se, or are people more sensitive to extreme feedback in closer, steady relationships?

7.3.2 Relationship Quality and Relational Self‐Esteem

Results on biological, elective, and non‐kin, confirmed that the closer the relationship the more important it is for the individual’s self‐evaluation. Unexpected are findings regarding the predictive value of perceived under‐ and over‐benefit. The sensitivity to and negative effect of non‐reciprocity is particularly strong in kin and elective kin relations. The strong negative link occurred in both indicators of relational self‐ esteem: unbalanced relationships to kin are (a) less relevant for self‐evaluations and

(b) less likely to be perceived as mutual in esteem. However the perception of close‐ ness interferes with this link. It gives a different direction to self‐evaluative patterns.

Close biological and elective kin ties are regarded as important for personal esteem regardless of non‐reciprocity. In addition, they are more likely to be perceived as mu‐ tually relevant for both relationship partners. Results support and extend empirical knowledge on sociometer theory (Denissen et al., 2008) suggesting that not only per‐ ceptions of closeness but reciprocity as well affect self‐evaluation.

Meaningful others positively affect relational self‐evaluation. But reciprocity intertwines with this pattern, supporting theoretical assumptions of relational self‐ esteem. Individually, non‐reciprocity has a strong negative effect but seems to be

“buffered” by emotional closeness. Surprising is the particular occurrence of substan‐ tial effects in the situation of elective kin. Not only are effects of non‐reciprocity more enhanced, the “buffering‐effect” of closeness is significantly higher compared to kin ties. This supports assumptions of the theoretical framework of lifelong relationship regulation (Lang & Neyer, 2003), where elective and biological kin are distinguished particularly through differential influences of perceived reciprocity. In elective kin, results confirm a higher sensitivity for (im‐)balance.

In non‐kin relations only over‐beneficial ties appear to be less relevant. Invest‐ ing more than one gets was associated with less relational esteem. In unbalanced re‐ Discussion 177 lationships with colleagues or neighbors that are described as particularly meaning‐ ful, closeness seems to serve a buffering function again. Singular feelings of ʺsupe‐ riorityʺ decrease the relevance for relational self‐evaluation – with the exception of close non‐kin relationships. These relationships may represent a context of ʺbeing neededʺ, thus negative exchange ratios are accepted to boost self‐evaluation. The in‐ crease of relational self‐esteem is likely contingent on feelings of inclusion and posi‐ tive relational value (Leary et al., 1995) combined with positive overall evaluations of the relationship itself.

Inclusion of age groups supports the suggestion that age matters to the relev‐ ance of family relationships. Age‐related patterns show higher relevance of biological kin in late adulthood and old age. Thus findings support socio‐emotional selectivity theory (Carstensen, 1992) at three levels: older individuals (a) include more kin in their networks, (b) describe kin as particularly meaningful, (c) assign higher relev‐ ance to kin regarding relational self‐evaluation. Age‐differential effects of closeness and reciprocity were less distinctive. The increase of relevance is enhanced by close‐ ness perceptions except for old age. Imbalance reduces the relevance of kin relations for self‐evaluation, but perceptions of closeness evidently serve as “buffer” in this context. Regardless of high levels of support that people in old age give and receive, the relevance of relationship partners for self‐evaluation seems to be predominantly stabilized by high emotional closeness. This is particularly true for older age groups, where closeness regulation seems to serve as important characteristic to adapt to shifting exchange patterns (cf., also Ikkink & van Tilburg, 1998). In this context it has to be acknowledged that age specific analyses show a substantial increase of variabil‐ ity. Specifically, older age groups are displaying high variability in coefficients, which suggests that effects hold true for some people more than others.

7.3.3 Effects of Level of Data: Relational and Individual Self‐Esteem.

Relational self‐esteem shows a number of substantial associations with domain‐ specific state and global trait self‐esteem. Patterns indicate a beneficial effect of gen‐ Discussion 178 eral positive self‐evaluation. Higher social, performance, and global self‐esteem en‐ hance perceived communal esteem and reduce likelihood to perceive relationships as one‐sided. Thus high self‐esteem may enhance people’s ability to use social relation‐ ship partners to boost their relational value (Murray et al., 2001). In addition, people with high social self‐esteem show an increased sensitivity to reciprocity violations in relationships with elective and non‐kin. High social self‐esteem seems to promote social self‐confidence which in turn enhances striving for reciprocal interactions. In contrast, people with low trait self‐esteem emphasize acceptance by interaction part‐ ners (Anthony, Wood, & Holmes, 2007) and may even allow non‐reciprocal social interactions.

As expected, the level of social self‐esteem clarifies as most influential domain.

Overlap seems higher between social and relational self‐esteem compared to global as well as performance indicators, thus effects are not due to differential sensitivity of state and trait measures. Rather characteristics of indicators drive the interplay be‐ tween measures of relational and individual level. Results suggest that relational self‐ esteem is partly rooted in more general, individual characteristics. Similar to the so‐ ciometer, relational self‐esteem, defined as dynamic and flexible process, has a stable, trait‐like baseline (Leary & Baumeister, 2000). This gives a certain amount of stability but allows for change as well. Even with respect to such core personality characteris‐ tics as the Big Five, change is expected. For example, Sturaro, Denissen, van Aken, &

Asendorpf (2008) show effects of relationship conflict on personality change in early adulthood. This emphasizes the potential flexibility of personality characteristics.

The baseline of (relational) self‐esteem is particularly formed and affected by pre‐ vious experiences in social interactions, for example attachment patterns (Srivastava

& Beer, 2005).

Additional personal characteristics of the individual and the relationship af‐ fect self‐evaluative processes and thus should be considered in future investigations.

Even people who deny the influence of social approval and disapproval on self‐ esteem demonstrate reactivity to interpersonal evaluation (Leary et al., 2003). Reac‐ Discussion 179 tivity to social cues is additionally associated with general levels of neuroticism (De‐ nissen & Penke, 2008). Highly neurotic people are more sensitive to social exclusion and social stressors. Further considerations of core personality traits, motives, etc. may lead to a broader understanding of associations among relational self‐esteem and individual characteristics or motivational processes.

7.3.4 Self‐Evaluation in Specific Dyadic Relationship Contexts

Adult Child‐Older Parent Relationships. Intergenerational dyads confirm generational differences in relationship perceptions as well as contextual effects of family con‐ stancy. First, older parents tend to describe the parent‐child tie as closer and more reciprocal. Thus a more positive evaluation is noticeable in the older generation (cf.,

Mather & Carstensen, 2005). Second, family constancy in particular influences rela‐ tionship perceptions of adult children. It supports the long‐term effect of low family constancy or family incompleteness on the child’s relationship perceptions (cf., Bel‐ sky et al., 1991a; 1991b; Ellis, 2004).

Dyadic interdependence between relationship perceptions and self‐esteem are unexpectedly low. Only closeness serves as a valid predictor for global self‐esteem and its important role is supported in adult child‐old parent dyads, but relational interdependence is contrary to expectations. Mutual dependency is higher in low family constancy, compared to high family constancy where no partner effects occur.

Findings with the same sub‐sample that concentrated on relationship‐specific conflict perceptions show the expected dyadic interdependence in high family constancy and patterns of self‐reliance in low family constancy (Wagner, Wrzus, Neyer, & Lang,

2008). What could be possible explanations of the differential patterns? First, it is demonstrated that global self‐esteem is a partial indicator of performance self‐ evaluation. Motivated childless individuals, who are more likely to have a low fam‐ ily constancy background in the current study, are also more oriented toward agentic life goals. Thus occupational success and general performance levels could be more important in gaining acceptance (and closeness) in parent‐child relations of low fam‐ Discussion 180 ily constancy. Second, self‐esteem is an individual characteristic and therefore con‐ trary to the relationship‐specific conflict frequency. Methodological problems may account for a certain proportion of the pattern as well.

The additional consideration of cross‐lagged effects is able to shed some light on causal order of change in variables while controlling for stability. Effects in both directions are possible: quality on self‐esteem and vice versa (Leary et al., 1995;

Murray et al., 2000; Denissen et al., 2008). Yet previous studies on parent‐child rela‐ tionships support the first direction (quality on self‐esteem; Asendorpf & van Aken,

2003; Roberts & Bengtson, 1993, 1996). Asendorpf and van Aken (2003) even explic‐ itly reject a self‐esteem effect on the relationship quality. Only core trait personality characteristics supported this causality in samples of adolescents and young adults.

The specific prediction in this intergenerational context hypothesized differences in patterns between types of family constancy. Consistent with expectations, relation‐ ship quality affects self‐esteem in children with high family constancy backgrounds.

The reverse effect is found in low family constancy. The adult child’s self‐esteem in‐ creases perceptions of closeness and tends to raise perceptions of reciprocity. This differential pattern supports the assumption of self‐reliance in low family constancy.

Perceptions of low stability reduce the relationship effect on the adult child’s self‐ evaluation across time.

At least four reasons exist for differential patterns of results within dyads vs. over time. First, social relationships always rely on three terms: the actor, the partner and the relationship. All three terms may change across time, but current cross‐ lagged effects consider only the actor’s perspective. Second, results were not con‐ trolled for possible confounding variables (e.g., life events) between the first and sec‐ ond investigation that may have changed self‐evaluative and/or relationship struc‐ tures. Third, the time between investigations may be too short or too long to capture a valid picture of change in self‐evaluation. Finally, global self‐esteem may not be sensitive enough to reflect the expected relationship specific evaluation process. Discussion 181

Romantic Partnership. Result patterns indicate that perceptions of reciprocity are the strongest predictive relationship quality between spouses. They seem to be particularly important for the woman’s self‐evaluation. Consistent with expectations, perceptions of reciprocity increase global self‐esteem in women of non‐traditional family backgrounds. Thus analyses support previous findings on motivated childless women that attach particular importance to a constant dyadic exchange with their spouse (Houseknecht, 1979; Carl et al., 2000).

With respect to social self‐esteem, reciprocity perceptions in childless men in‐ creased their spouse’s self‐consciousness (i.e. women become more self‐confident).

Childless women or partnerships have been shown to approve equality in their rela‐ tionships (Lamprecht, Wagner, & Lang, 2008). Thus reciprocity may stabilize their social standing as independent persons (Carl, 2002). The substantial partner effect in general suggests higher interdependency in childless couples, which may compen‐ sate for weaker social support by other social affiliations (Ishii‐Kuntz & Seccombe,

1989). However small sample sizes do not allow distinct dyadic analyses of moti‐ vated and involuntary childlessness. Subsequent research should further explore this issue.

Cross‐lagged effects in partnerships demonstrate stability and change alike.

Stability coefficients indicate that not only self‐esteem, but also relationship percep‐ tions, have a trait‐like component. Data supports the exceptional role of reciprocity in spousal ties. Only perceptions of reciprocity show an association with global self‐ esteem across time. Hypothesized effects of relationship quality on self‐esteem reached only marginal significance, but are substantial in the reverse direction.

Global self‐esteem indicates positive effects on perceived reciprocity. Results are not consistent with expectations of sociometer theory that focuses on effects from quality on self‐esteem, but similar patterns have already been reported in the context of ro‐ mantic partners (Murray et al., 2000).

Contrary to expectations, no contextual differences emerged regarding either family of origin or current family situations. Patterns appear to be fairly stable across Discussion 182 all family conditions. Regardless of the perspective, e.g. evolutionary, attachment, the romantic partner is highly relevant. Diary data has demonstrated the exceptional role of romantic partners for state self‐esteem across days (Denissen, et al., 2008). Es‐ tablished spousal relationships of the current study may illustrate relatively univer‐ sal patterns of heterosexual partnerships. Marital relationships might constitute a specific dyadic pattern, such that overall similarities remain more striking than dif‐ ferences (Abbey, Andrews, & Halman, 1995). Thus contextual patterns could be less important for the internal functioning of stable romantic partners. On the other hand, findings may be based on statistical conditions: (a) results are based on only a small sub‐sample of couples and no differential analyses were conducted among all four family situations, (b) measured time‐lag (1 year in this study) could be too long or too short to assess reliable patterns, (c) global self‐esteem may be not sensitive enough to tap into relationship specific evaluation patterns.

Summary. Besides the relationship type, emotional closeness and perceived re‐ ciprocity demonstrate a strong effect on the self‐evaluative process. Particularly emo‐ tional closeness serves an important role in conditions of perceived imbalance to maintain positive self‐evaluations in social ties. Age‐related patterns indicate higher self‐evaluative relevance of kin and elective in old age. This fact supports the special role of family ties in late adulthood. Dyadic and cross‐lagged analyses illustrate rela‐ tionship specific patterns of self‐evaluation. Distinct relationship qualities character‐ ize self‐evaluation in parental ties and partnerships. Yet, contextual effects of family of origin and current situation appear to be less distinguishing.

7.4 There Is a Relational Nature to Self‐Evaluation – But That Is Not the Whole

Story

Perceptions of being liked and accepted by others lead to higher self‐esteem (Bau‐ meister & Leary, 1995; Leary et al., 1995; Srivastava & Beer, 2005). In daily function‐ ing this is particularly true for interactions with friends and romantic partners Discussion 183

(Denissen et al., 2008). In the reverse direction, high self‐esteem increases relation‐ ship well‐being with romantic partners (Murray et al., 2000) and perceptions of social support in general (Gracia & Herrero, 2004). Thus empirical results have already broadened assumptions of sociometer theory (Baumeister & Leary, 1995). Relational self‐esteem is proposed as a relationship‐specific monitor that captures the relevance of each social interaction partner. Empirical results support hypotheses on differen‐ tial characteristics of communal and mutual‐esteem across relationship types and age groups. Relationship specific self‐evaluation processes are essentially guided by per‐ ceptions of emotional closeness and reciprocity. Captured in the mechanism of rela‐ tional self‐esteem, the monitoring and processing of self‐relevant interpersonal cues enables people to function in complex social conditions. At the same time, relational self‐esteem has a trait‐like baseline. People’s general tendency to assess themselves in social contexts affects relationship specific evaluations in particular. The theoretical outline accentuates the regulative nature and developmental potential of self‐esteem in connection with age, former experiences, current contextual conditions, and pre‐ vailing demands. Thus findings indicate that self‐evaluation is certainly partially so‐ cial in nature.

This is not, however, a totally new idea. The ʺself is relationalʺ (Andersen &

Chen, 2002, p. 619) is the central argument in the theoretical outline of the relational self. Andersen and Chen (2002) suggest that individuals have a number of represen‐ tations of the self in relation to significant others. Existing representations, available in memory and chronically accessible, will trigger earlier interpersonal experiences that help to manage interactions with new relationship partners. Yet, relational self‐ esteem focuses additionally on the differential relevance of and perceived quality with relationship partners and the monitoring process that arises between them. This enduring regulative process enables the individual to evaluate, adapt, and/or stabi‐ lize certain characteristics within one relationship, across different social network partners, within changing contextual conditions, and amid demanding life chal‐ lenges. Discussion 184

However, the initial question asked: how relational is self‐evaluation? Is there more to it than its social nature? Or does the entire process of self‐evaluation and self‐regulation rely on social cues? Strong evidence exists for a vastly positive impact of social interactions on people’s well‐being and personal functioning (Denissen et al., 2008; Fiori, Antonucci, & Cortina, 2006; Fingerman, Miller, & Charles, 2008).

However, besides social relationships and their effect on self‐evaluation, additional, important facets of self‐perception and self‐evaluation must be considered. Being socially embedded and well‐regarded is a fundamental need and human motivation, but it may not be regarded as the only feature.

7.5 Motivated and Involuntary Childlessness – Differentiation versus Conver‐

gence in Reproduction and Post‐Reproduction Times

The separate consideration of motivated and involuntary childless couples proves that differences exist in social network structures and relationship regulation charac‐ teristics between the two types. Furthermore, explorations of relationship qualities and dyadic patterns establish the impact of family constancy and current family status. At the same time, results provide evidence of a considerable number of simi‐ larities as well, which opens the question of a process of convergence across adult‐ hood.

Motivated Childlessness. Beyond the (few) demographic and motivational dif‐ ferences, the current project demonstrates that motivated childless individuals are generally less family oriented than all reference groups. Three findings show this: (a) motivated childless designate fewer biological and elective kin members as social network partners, (b) they show no substantial kinship preference, and (c) they are less oriented toward communal goals and illustrate no positive motivation towards having children. Social networks of motivated childless individuals do not include more non‐kin such as colleagues or neighbors. One implication is that they have smaller networks. Thus the consideration of ego‐centered network data supports Discussion 185 prior findings that suggest no compensational social activity in (motivated) chil‐ dlessness.

Relationship regulation characteristics are generally less distinctive. There is no substantial interplay among emotional closeness, genetic similarity and perceived reciprocity. The motivated childless lack kinship preference and show no differential effects of reciprocity perceptions. Based on knowledge of enhanced strivings for in‐ dependence and career orientation (Callan, 1987; Carl, 2002; Carl et al., 2000; Kan,

2007), motivated childlessness is hypothesized to be associated with high monitoring of recourses and a strong motivation for equilibrium in social contexts. This effect occurs only in dyadic analyses. Self‐esteem of motivated childless women increases in the light of their own and their spouse’s perceptions of reciprocity. The withdraw‐ al from family ties is generally not compensated for by non‐family relations. Rather, motivated childless couples emphasize the specific nature of their romantic partner‐ ship (Burman & de Anda, 1986; Gillespie, 2003; Lawrence, Rothman, Cobb, Rothman,

& Bradbury, 2008). Longitudinal results in child‐parent dyads additionally support this characteristic of self‐reliance.

Involuntary Childlessness. The most notable aspect of involuntary childless in‐ dividuals is their stable ʺin‐betweenʺ position. They belong neither to the group of parents nor to the motivated childless. This position is mirrored by characteristics of the network size and composition, and by relationship regulation strategies. In con‐ trast to motivated childlessness, involuntary childless individuals name a number of kin similar to that of parents, describe relationships with their siblings as particularly close, and show substantial kinship orientation. With respect to most other structural features and regarding most regulatory mechanisms they are very similar to moti‐ vated childless individuals.

Results suggest that there are differences and similarities between involuntary and motivated childless alike. Although involuntary childlessness is expected to be a livelong issue (Letherby, 2002a, 2002b; Kotter‐Grün, 2008), a certain alignment be‐ tween the two types of childlessness has to be expected. To remain capable and satis‐ Discussion 186 fied, people generally strive for primary control (Heckhausen, 1997; Heckhausen &

Schulz, 1995). As described in the life span theory of control (Heckhausen & Schulz,

1995), selecting and actively working towards a goal even if the help of others is ne‐ cessary, illustrates an active, regulative behavior. However, when passing a deadline such as the biological capability to bear children, goal disengagement and self‐ protective strategies must be applied to stabilize well‐being in childless women

(Heckhausen et al., 2001). Thus involuntarily childless people, particularly post‐ deadline women but also men in stable partnerships with infertile women, must cen‐ ter on their child‐free life style and engage in new, different life goals. Thus moti‐ vated and involuntary childless individuals will become more similar in late adult‐ hood. This convergence may also be mirrored by the often neglected differentiation between the two types in older samples (e.g., Koropeckyj‐Cox & Call, 2007; Wenger,

Dykstra, Melkas, & Knipscheer, 2007).

Results describe the disparity between two superficially similar life situations.

Motivated and involuntary childless individuals not only differ with respect to their parental desires, but social networks and relationship regulation strategies also show dissimilar patterns. Furthermore, dyadic analyses assume that social relationships serve different functions in the process of self‐evaluation. At the same time, chil‐ dlessness has to be considered as transitional status in young and middle adulthood.

Infertility is still associated with the and striving for parenthood. In motivated childlessness contextual or individual changes such as a new partner or a drastic life event may shift people’s motivations and initiate different behaviors. If staying childfree, an adaptation to the unwanted life style constitutes a necessary process to maintain psychological functioning. Longitudinal analyses of childless individuals from reproductive to post‐reproductive ages could enhance knowledge of time frames and patterns of assimilation, but also of other possible differences between the two types. In addition, single childless men and women should be included.

Lacking the stable and meaningful relationship to a spouse, they may show different social and behavioral characteristics. Discussion 187

7.6 Limitations of the Study

This section focuses on limitations and weaknesses of the present study. They are important to consider in generalizing results. Furthermore, they have important im‐ plications for future research investigations. In the following section, emphasis is placed on seven caveats: (a) representativity of sample structures, (b) gender, (c) per‐ ception vs. regulation of reciprocity, (d) elaboration of relational self‐esteem, (e) reli‐ ance on self‐report measures, (f) statistical analyses, and (g) need for longitudinal research.

Representativity of Sample Structures. In both main studies a certain degree of selectivity within the samples must be assumed. With respect to the online study, selectivity is very likely, particularly in older age groups, although an increasing number of German citizens uses the internet and the sample shows a diversity of educational levels and family situations. Initial analyses comparing the online with an old age community sample showed satisfactory congruence. Further comparisons are needed.

For the family study, recruiting motivated and involuntarily childless couples proved to be very difficult. Thus the replication of findings is important. This is also true because fertility attitudes can change across time (Mitchell & Gray, 2007). Fur‐ thermore, single childless men and women have to be included in empirical research projects to broaden the understanding of the phenomenon. Another issue is the size of the sample. Specifically the sub‐samples of the family study are rather small. With‐ in both dyadic analyses the differentiation of all four family types was not possible because of the insufficient number of dyads within each group. Although general groups do provide several important findings, results need to be replicated, specifi‐ cally with bigger samples.

Gender. Gender of participants was not a major subject of the current project.

Yet, the theme calls out for some final notes. To begin with, it is important to keep in mind that all analyses of results for network characteristics and relationship percep‐ tions were controlled for gender of participant and gender of relationship partner. Discussion 188

The consideration did not alter the pattern of findings. However, two unexpected findings occurred with respect to social self‐esteem. First, women of low family con‐ stancy were less self‐conscious than all other reference groups. Second, in dyadic analyses on partnerships, reciprocity indicated a positive effect on women’s self‐ evaluation. Both patterns rely on the smaller sub‐sample of the follow up study.

Thus, it was not possible to conduct more elaborative analyses. Further investigation will amplify an understanding of these patterns.

Perception vs. Regulation of Reciprocity. The relationship specific evaluation of perceived reciprocity produced an item which tilted in the direction of reciprocal re‐ lationships. In general, people strive for balanced social relationships. To attain reci‐ procity however, people have to engage in a monitoring process. It is a cognitive ef‐ fort to remember, to observe, and to manage patterns of previous and future ex‐ change situations across a number of social relationship partners. The graphical items used were not able to assess this observational process. They only captured the anticipated final result of the monitoring mechanism. Future investigations could assess reciprocity perceptions across several days, e.g. diary study, to tap into daily regulatory processes across a certain number of relationship partners, or they could provoke a threat to reciprocity to assess possible global or relationship specific regu‐ latory strategies or behaviors.

Elaboration of Relational Self‐Esteem. Three issues have to be considered with re‐ spect to this newly introduced construct: First, similar to the reciprocity indicators, mutual self‐esteem lacked a certain amount of variation and only random‐intercept models with fixed slopes could be examined. Follow‐up studies should vary me‐ thods of assessment. Findings need to be replicated for different samples and condi‐ tions. Second, no information on causality can be drawn from these analyses: Does closeness lead to relevance or vice versa? Which relationship developments or changes occur across time and contexts? Dyadic analyses found only a limited inter‐ dependence between specific relationship perceptions and individual self‐esteem.

Further elaborations are needed to understand the associations and causal directio‐ Discussion 189 nality. Third, a distinction from other known relationship‐specific measures is neces‐ sary to further connect and frame the construct within social relationship research.

Relational self‐esteem extends and advances our knowledge on self‐evaluation com‐ pared to global and domain‐specific measures. Thus, it may open up a way to broa‐ den theoretical and empirical investigation and to focus on regulative, not only de‐ scriptive, structures of or behind self‐esteem.

Reliance on Self‐Report Measures. Presented findings are all based on question‐ naire data. They offer introspective self‐reports from participants. Although this me‐ thod is a primary tool of psychological research to learn about individual feelings, behaviors and thoughts, it pays “remarkably little attention to the important things that people do” (Baumeister, Vohs, & Funder, 2007, p. 396). Baumeister and col‐ leagues suggest that both methods – subjective self‐report and behavioral observa‐ tions – seem to be important in understanding and answering research questions.

To reduce the weaknesses of self‐reporting the current study used several me‐ thods: First, a pilot study was conducted to ensure reliability and validity of meas‐ ures. Second, besides simple verbal items, relationship perceptions were assessed using pictorial items. Third, results were replicated within different samples and age groups. Fourth, dyadic analyses considered both sides of a relationship and captured the interdependent dynamic of social relationships. Fifth, intra‐individual network analyses were based on multi‐level models to reduce artificial correlations and to take data structures into account. Future research might apply experimental designs or behavioral observations such as differential consequences of making and receiving compliments within different social relationships.

Statistical Analysis. Given the different data structures, a variation of analysis strategies had to be utilized. When the sample size was insufficient to conduct family specific analyses, more general differentiations were made, e.g. childless vs. parents, to indicate contextual effects on a broader level. Bigger samples are needed to con‐ firm the patterns noted. Furthermore, sometimes the number of variables considered, and thus the number of estimated effects, seemed to destabilize the models. One in‐ Discussion 190 dicator is the high standard error that sometimes appears, particularly regarding in‐ teraction effects.

Need for Longitudinal Research. Apart from the small follow‐up sample in the family study, which indicates a fair amount of change across a 1‐year period in rela‐ tionship perceptions, all results are based on cross‐sectional data only. This limitation is particularly significant with respect to the different age‐associated patterns drawn from the online sample. Trzesniewski and colleagues (Trzesniewski et al., 2003) indi‐ cated the lack of (longitudinal) studies on self‐esteem in old age. Empirical research has to replicate the higher variability in global self‐esteem and explore possible pat‐ terns in domain‐specific self‐esteem measures. Moreover, there is a high probability of cohort effects, since the construct was differentially important across the last dec‐ ades (Swann, Chang‐Schneider, & Larsen McClarty, 2007). Topics to be raised may include: What provokes the strong difference in mean levels of social self‐esteem be‐ tween young and middle adulthood? What possible contextual conditions or mod‐ erating variables define change in social self‐evaluation?

7.7 Outlook

The preceding section made several suggestions for the continuation and improve‐ ment of future empirical investigations. This last part concentrates on three content‐ related subjects that point to potential directions of research: (a) cognitive processes of/behind relational self‐esteem, (b) cross‐cultural comparison of contributions of closeness regulation and reciprocity monitoring to relational self‐esteem, and (c) transition and adaptation of relational self‐esteem in motivated and involuntary childlessness across time.

Cognitive Processes. What cognitive processes lie behind social self‐evaluation?

Srivasta and Beer (2005) found that conscious and unconscious aspects were inde‐ pendent predictors of self‐evaluation processes. People do not know what others really think of or about them. Thus self‐evaluation regularly builds on cues or stim‐ Discussion 191 uli. Possible research questions in this context are: Is self‐evaluation more strongly mediated by conscious or unconscious cues? Is the mediation process dependent on the relationship partner? To explore potential patterns, experimental studies could vary the type (e.g., verbal, facial expressions), content (e.g., self‐relevant, irrelevant), and connotation (e.g., positive, neutral, ambivalent) of feedback cues. At the same time it would be possible to control (or contrast) the type (e.g., partner, foreigner, friend) and context (e.g., leisure activity, competitive test) of social interactions.

Cross‐cultural Comparison. Considerations of the self in different cultures showed substantial evidence of striking differences in nature and construction. These differences likely have implications for cognition, emotion, and motivation (Markus

& Kitayama, 1991; 2003). In this context, two suggestions are possible for relational self‐esteem: first, a discrepancy in processes of relational self‐evaluation is likely among cultures, i.e., different cues may be perceived and processed. Second, rele‐ vance of relationships and levels of influence may vary among cultures, e.g., how much are individuals affected by the feedback from their mothers versus their friends. Prior cross‐cultural research on social network characteristics illustrates simi‐ larities and differences between German and Hong Kong Chinese samples (Fung et al., 2008). Furthermore, recent Hong Kong Chinese network data suggested that moderate levels of negative exchange are positively associated with emotional close‐ ness, particularly in kin relationships (Fung, Yeung, Li, & Lang, in subm.). Findings on relational self‐esteem supported the positive relational valence of close imbal‐ anced kin relationships. Thus accepting moderate negative exchange may be func‐ tional in the light of positive relational feedback that people perceive in a specific social relationship. Further empirical projects could address the topic and function of relational self‐esteem in different cultural contexts.

Transitions and Adaptation. Humans have a huge variety of social relationships, which change over time depending on contextual conditions (Lang & Heckhausen,

2006). They may support or constrict individual development. According to the life span theory of control (Heckhausen & Schulz, 1995), individuals try to change their Discussion 192 environment to meet their needs. They also attempt to adapt personal goals, motiva‐ tions, and emotions according to contextual requirements. Thus, changed prospects, such as involuntary childlessness, may initiate discussion on future goals or alterna‐ tives – particularly for childless couples. At least three possible courses remain open:

(1) adoption, (2) finding a new, fertile partner, or (3) adapting to a fulfilling childfree life style. All three decisions hold different challenges. All three have different impli‐ cations for social networks and relationship regulation strategies. Longitudinal re‐ search could address questions on: Who chooses which life course? How do social networks transform? What changes are found in patterns of relational self‐ evaluation?

References 193

8 REFERENCES

Aarssen, L. W., & Tzipporah Altman, S. (2006). Explaining below‐replacement fertili‐

ty and increasing childlessness in wealthy countries: Legacy drive and the

ʹtransmission competitionʹ hypothesis. Evolutionary Psychology, 4, 290‐302.

Abbey, A., Andrews, F. M., & Halman, L. J. (1995). Provision and Receipt of Social

Support and Disregard: What Is Their Impact on the Marital Life Quality of

Infertile and Fertile Couples? Journal of Personality and Social Psychology, 68(3),

455‐469.

Abele, A. E., & Wojciszke, B. (2007). Agency and Communion From the Perspective

of Self Versus Others. Journal of Personality and Social Psychology, 93(5), 751‐763.

Ackerman, J. M., & Kenrick, D. T. (2008). The costs of benefits: Help‐refusals high‐

light key trade‐offs of social life. Personality and Social Psychology Review, 12(2),

118‐140.

Ackerman, J. M., Kenrick, D. T., & Schaller, M. (2007). Is friendship akin to kinship?

Evolution and Human Behavior, 28(5), 365‐374.

Ainsworth, M. S. (1989). Attachments beyond infancy. American Psychologist, 44(4),

709‐716.

Allemand, M., Zimprich, D., & Martin, M. (2008). Long‐term correlated change in

personality traits in old age. Psychology and Aging, 23(3), 545‐557.

Andersen, S. M., & Chen, S. (2002). The Relational Self: An Interpersonal Social ‐

Cognitive Theory. Psychological Review, 109(4), 619‐645.

Anderson, D. Y., & Hayes, C. L. (1996). Gender, identity, and self esteem: A new look at

adult development. New York: Springer. References 194

Anthony, D. B., Wood, J. V., & Holmes, J. G. (2007). Testing sociometer theory: Self‐

esteem and the importance of acceptance for social decision‐making. Journal of

Experimental Social Psychology, 43(3), 425‐432.

Antonucci, T. C., & Akiyama, H. (1995). Convoys of Social Relations: Family and

Friendships within a Life Span Context. In R. Blieszner & V. H. Bedford (Eds.),

Handbook of Aging and the Family (pp. 355‐371). Westport, Conneticut: Green‐

wood Press.

Antonucci, T. C., Akiyama, H., & Takahashi, K. (2004). Attachment and close rela‐

tionships across the life span. Attachment & Human Development, 6(4), 353‐370.

Antonucci, T. C., Lansford, J. E., & Akiyama, H. (2001). Impact of positive and nega‐

tive aspects of marital relationships and friendships on well‐being of older

adults. Applied Developmental Science, 5(2), 68‐75.

Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of Other in the Self Scale and

the structure of interpersonal closeness. Journal of Personality and Social Psycho‐

logy, 63(4), 596‐612.

Asendorpf, J. B. (1999). Psychologie der Persönlichkeit. 2. überarbeitete und aktualisierte

Auflage. Berlin: Springer.

Asendorpf, J. B. (2004). Psychologie der Persönlichkeit. 3. überarbeitete und aktualisierte

Auflage. Berlin: Springer.

Asendorpf, J. B., & van Aken, M. A. G. (2003). Personality‐Relationship Transaction

in Adolescence: Core Versus Surface Personality Characteristics. Journal of Per‐

sonality, 71(4), 629‐666.

Asendorpf, J. B, & Wilpers. (1998). Personality effects on social relationships. Journal

of Personality and Social Psychology, 74(6), 1531‐1544.

Auferkorte‐Michaelis, N., Metz‐Göckel, S., Wergern, J., & Klein, A. (2006). Junge El‐

ternschaft und Wissenschaftskarriere. Wie kinderfreundlich sind Wissenschaft und

Universitäten? .Unpublished manuscript, Dortmund. References 195

Bagwell, C. L., Bender, S. E., Andreassi, C. L., Kinoshita, T. L., Montarello, S. A., &

Muller, J. G. (2005). Friendship quality and perceived relationship changes

predict psychosocial adjustment in early adulthood. Journal of Social and Per‐

sonal Relationships, 22(2), 235‐254.

Bakan, D. (1966). The Duality of Human Existence: Isolation and Communion in Western

Man. Boston: Beacon Press.

Bartholomew, K., & Horowitz, L. M. (1991). Attachment Styles Among Young

Adults: A Test of a Four‐Category Model. Journal of Personality and Social Psy‐

chology, 61(2), 226‐244.

Batson, C. D., & Shaw, L. L. (1991). Evidence for altruism: Toward a pluralism of pro‐

social motives. Psychological Inquiry, 2(2), 107‐122.

Baumeister, R. F., & Vohs, K. D. (2003). Self‐Regulation and the Executive Function of

the Self. In M. R. Leary & J. P. Tangney (Eds.), Handbook of Self and Identity (pp.

197‐217). New York, NY, US: Guilford Press.

Baumeister, R. F., Campbell, J. D., Krueger, J. I., & Vohs, K. D. (2003). Does high self‐

esteem cause better performance, interpersonal success, , or health‐

ier lifestyles? Psychological Science in the Public Interest, 4(1), 1‐44.

Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal

attachments as a fundamental human motivation. Psychological Bulletin, 117(3),

497‐529.

Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatened egotism to

violence and aggression: The dark side of high self‐esteem. Psychological Re‐

view, 103(1), 5‐33.

Baumeister, R. F., Vohs, K. D., & Funder, D. C. (2007). Psychology as the science of

self‐reports and finger movements: Whatever happened to actual behavior?

Perspectives on Psychological Science, 2(4), 396‐403. References 196

Befu, H. (1980). Structural and Motivational Approaches to Social Exchange. In K. J.

Gergen, M. S. Greenberg & R. H. Willis (Eds.), Social exchange. Advances in The‐

ory and Research (pp. 197‐214). New York: Plenum Press

Belsky, J. (1997). Theory testing, effect‐size evaluation, and differential susceptibility

to rearing influence: The case of mothering and attachment. Child Development,

68(4), 598‐600.

Belsky, J., Jaffee, S. R., Caspi, A., Moffitt, T., & Silva, P. A. (2003). Intergenerational

Relationships in Young Adulthood and Their Life Course, Mental Health, and

Personality Correlates. Journal of Family Psychology, 17(4), 460‐471.

Belsky, J., Steinberg, L., & Draper, P. (1991a). Childhood experience, interpersonal

development, and reproductive strategy: An evolutionary theory of socializa‐

tion. Child Development, 62(4), 647‐670.

Belsky, J., Steinberg, L., & Draper, P. (1991b). Further reflections on an evolutionary

theory of socialization. Child Development, 62(4), 682‐685.

Bengtson, V. L. (2001). Beyond the nuclear family: The increasing importance of mul‐

tigenerational bonds (The Burgess Award Lecture). Journal of Marriage & the

Family, 63(1), 1‐16

Bernardi, L. (2003). Channels of on reproduction. Unpublished manu‐

script, Rostock.

Bianchi, S. M. (2006). Mothers and Daughters ʹDo,ʹ Fathers ʹDonʹt Doʹ Family: Gender

and Generational Bonds. Journal of Marriage and Family, 68(4), 812‐816.

Bidart, C., & Lavenu, D. (2005). Evolutions of personal networks and life events. So‐

cial Networks, 27(4), 359‐376.

Birditt, K. S., Fingerman, K. L., & Almeida, D. M. (2005). Age Differences in Exposure

and Reactions to Interpersonal Tensions: A Daily Diary Study. Psychology and

Aging, 20(2), 330‐340. References 197

Bishop, J. A., & Inderbitzen, H. M. (1995). Peer acceptance and friendship: An inves‐

tigation of their relation to self‐esteem. The Journal of Early Adolescence, 15(4),

476‐489.

Björklund, D. F., & Pellegrini, A. D. (2002). The Origins of Human Nature: Evolutionary

Developmental Psychology. Washington DC: American Psychological Associa‐

tion.

Blieszner, R., & Adams, R. G. (1992). Adult friendship. Newbury Park, CA: Sage.

Borchardt, A., & Stöbel‐Richter, Y. (2004). Die Genese des Kinderwunsches bei Paaren ‐

eine qualitative Studie. Materialen zur Bevölkerungswissenschaft. Wiesbaden: Bun‐

desinstitut für Bevölkerungsforschung.

Borgenhoff Mulder, M. (1998). The demographic transition: are we any closer to an

evolutionary explanation? Trends in Ecology & Evolution, 13(7), 266‐270.

Bowlby, J. (1969). Attachment. New York: Basic Books.

Bowlby, J. (1983). Verlust: Trauer und Depression. Frankfurt am Main: Fischer.

Brähler, E., Stöbel‐Richter, Y., & Schumacher, J. (2001). Für und Wider eines eigenen

Kindes: Der Leipziger Fragebogen zu Kinderwunschmotiven (LKM). Diagnos‐

tica, 47, 96‐106.

Brandtstädter, J. (1999). The Self in Action and Development: Cultural, Biosocial, and

Ontogenetic Bases of Intentional Self‐Development. In J. Brandtstädter & R. M.

Lerner (Eds.), Action and Self‐Development. Theory and Research Through the Life

Span (pp. 37‐65). Thousand Oaks: Sage Publications Inc.

Brandtstädter, J. (2001). Entwicklung, Intentionalität, Handeln. Stuttgart: Kohlhammer.

Brennan, K. A., & Bosson, J. K. (1998). Attachment‐style differences in attitudes to‐

ward and reactions to feedback from romantic partners: An exploration of the

relational bases of self‐esteem. Personality and Social Psychology Bulletin, 24(7),

699‐714. References 198

Bucx, F., van Wel, F., Knijn, T., & Hagendoorn, L. (2008). Intergenerational contact

and the life course status of young adult children. Journal of Marriage and Fam‐

ily, 70(1), 144‐156.

Burman, B., & de Anda, D. (1985). Parenthood or nonparenthood: A comparison of

intentional families. Lifestyles, 8(2), 69‐84.

Buss, D. M. (2004). Evolutionäre Psychologie. München: Pearson Studium.

Callan, V. J. (1982). How Do Australians Value Children? A Review and Research

Update Using the Perceptions of Parents and Voluntarily Childless Adults

Journal of Sociology, 18(3), 384‐398.

Callan, V. J. (1983). Childlessness and partner selection. Journal of Marriage & the Fam‐

ily, 45(1), 181‐186.

Callan, V. J. (1987). The Personal and Marital Adjustment of Mothers and of Volun‐

tary and Involuntary Childless Wives. Journal of Marriage & the Family, 49(4),

847‐856.

Campbell, L. D., Connidis, I. A., & Davies, L. (1999). Sibling ties in later life: A social

network analysis. Journal of Family Issues, 20(1), 114‐148.

Carl, C. (2002). Gewollt Kinderlose Frauen und Männer. Psychologische Einflußfaktoren

und Verlaufstypologien des generativen Verhaltens. Frankfurt: Verlag für akade‐

mische Schriften.

Carl, C., Bengel, J., & Strauß, B. (2000). Gewollte Kinderlosigkeit aus psychologischer

Perspektive. Reproduktionsmedizin, 2000(16), 28‐36.

Carstensen, L. L. (1992). Social and emotional patterns in adulthood: Support for so‐

cioemotional selectivity theory. Psychology and Aging, 7(3), 331‐338.

Carstensen, L. L., Isaacowitz, D. M., & Charles, S. T. (1999). Taking time seriously: A

theory of socioemotional selectivity. American Psychologist, 54(3), 165‐181. References 199

Carstensen, L. L., & Löckenhoff, C. E. (2003). Aging, Emotion, and Evolution: The

Bigger Picture. In P. Ekman, J. J. Campos, R. J. Davidson & F. B. M. de Waal

(Eds.), Emotions inside out: 130 years after Darwinʹs: The expression of the emotions

in man and animals. (pp. 152‐179). New York, NY, US: New York University

Press.

Carstensen, L. L., & Mikels, J. A. (2005). At the Intersection of Emotion and Cogni‐

tion: Aging and the Positivity Effect. Current Directions in Psychological Science,

14(3), 117‐121.

Carver, C. S., & Scheier, M. F. (1998). On the Self‐Regulation of Behavior. New York,

NY, US: Cambridge University Press.

Chasiotis, A. (1999). Kindheit und Lebenslauf. Untersuchungen zur evolutionären Psycho‐

logie der Lebensspanne. Bern: Hans Huber.

Cialdini, R. B., Brown, S. L., Lewis, B. P., Luce, C., & Neuberg, S. L. (1997). Reinter‐

preting the ‐altruism relationship: When one into one equals oneness.

Journal of Personality and Social Psychology, 73(3), 481‐494.

Clark, M. S., & Grote, N. K. (1998). Why arenʹt indices of relationship costs always

negatively related to indices of relationship quality? Personality and Social Psy‐

chology Review, 2(1), 2‐17.

Clark, M. S., Helgeson, V. S., Mickelson, K., & Pataki, S. P. (1994). Some cognitive

structures and processes relevant to relationship functioning. In R. S. Wyer, Jr.

& T. K. Srull (Eds.), Handbook of social cognition, Vol. 1: Basic processes; Vol. 2:

Applications (2nd ed., pp. 189‐238). Hillsdale, NJ England: Lawrence Erlbaum

Associates, Inc.

Clark, M. S., & Mills, J. (1979). Interpersonal attraction in exchange and communal

relationships. Journal of Personality and Social Psychology, 37(1), 12‐24. References 200

Clark, M. S., & Mills, J. (1993). The Difference Between Communal and Exchange Re‐

lationships: What It Is and Is Not. Personality and Social Psychology Bulletin, 19,

684‐691.

Clark, M. S., Mills, J., Reis, H. T., & Rusbult, C. E. (2004). Interpersonal Attraction in

Exchange and Communal Relationships Close relationships: Key readings. (pp.

245‐256). Philadelphia, PA, US: Taylor & Francis.

Clark, M. S., Oullette, R., Powell, M. C., & Milberg, S. (1987). Recipientʹs mood, rela‐

tionship type, and helping. Journal of Personality and Social Psychology, 53(1), 94‐

103.

Clark, M. S., & Taraban, C. (1991). Reactions to and Willingness to Express Emotions

in Communal and Exchange Relationships. Journal of Experimental Social Psy‐

chology, 27, 324‐336.

Cook, W. L., & Kenny, D. A. (2005). The Actor‐Partner Interdependence Model: A

model of bidirectional effects in developmental studies. International Journal of

Behavioral Development, 29(2), 101‐109.

Cook, W. L., & Snyder, D. K. (2005). Analyzing Nonindependent Outcomes in Cou‐

ple Therapy Using the Actor‐Partner Interdependence Model. Journal of Family

Psychology, 19(1), 133‐141.

Coromina, L. s., & Coenders, G. (2006). Reliability and validity of egocentered net‐

work data collected via web: A meta‐analysis of multilevel multitrait mul‐

timethod studies. Social Networks, 28(3), 209‐231.

Crocker, J., & Park, L. E. (2004). Reaping the Benefits of Pursuing Self‐Esteem With‐

out the Costs? Reply to DuBois and Flay (2004), Sheldon (2004), and Pyszczyn‐

ski and Cox (2004). Psychological Bulletin, 130(3), 430‐434.

Crocker, J., & Luhtanen, R. K. (2003). Level of self‐esteem and contingencies of self‐

worth: Unique effects on academic, social and financial problems in college

students. Personality and Social Psychology Bulletin, 29(6), 701‐712. References 201

Cross, S. E., Gore, J. S., & Morris, M. L. (2003). The Relational‐Interdependent Self‐

Construal, Self‐Concept Consistency, and Well‐Being. Journal of Personality and

Social Psychology, 85(5), 933‐944.

Daniluk, J. C., & Tench, E. (2007). Long‐Term Adjustment of Infertile Couples Fol‐

lowing Unsuccessful Medical Intervention. Journal of Counseling & Develop‐

ment, 85(1), 89‐100.

Datta, P., Marcoen, A., & Poortinga, Y. H. (2005). Recalled early maternal bonding

and mother‐ and self‐related attitudes in young adult daughters: A cross‐

cultural study in India and Belgium. International Journal of Psychology, 40(5),

324‐338.

Deci, E. L., La Guardia, J. G., Moller, A. C., Scheiner, M. J., & Ryan, R. M. (2006). On

the Benefits of Giving as Well as Receiving Autonomy Support: Mutuality in

Close Friendships. Personality and Social Psychology Bulletin, 32(3), 313‐327.

Denissen, J. J. A., & Penke, L. (2008). Neuroticism predicts reactions to cues of social

inclusion. European Journal of Personality(22), 497‐517.

Denissen, J. J. A., Penke, L., Schmitt, D. P., & van Aken, M. A. G. (2008). Self‐Esteem

Reactions to Social Interactions: Evidence for Sociometer Mechanisms Across

Days, People, and Nations. Journal of Personality and Social Psychology, 95(1),

181‐196.

Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction With

Life Scale. Journal of Personality Assessment, 49, 71‐75.

Diener, E., Lucas, R. E., & Scollon, C. N. (2006). Beyond the hedonic treadmill: Revis‐

ing the adaptation theory of well‐being. American Psychologist, 61(4), 305‐314.

Doherty, N. A., & Feeney, J. A. (2004). The composition of attachment networks

throughout the adult years. Personal Relationships, 11(4), 469‐488. References 202

DʹOnofrio, B. M., Turkheimer, E., Emery, R. E., Harden, K. P., Slutske, W. S., Heath,

A. C., Madden, P. A. F., & Martin, N. G. (2007). A genetically informed study

of the intergenerational transmission of marital instability. Journal of Marriage

and Family, 69(3), 793‐809.

Dunn, J. (2002). The adjustment of children in stepfamilies: Lessons from community

studies. Child and Adolescent Mental Health, 7(4), 154‐161.

Duschek, K., & Wirth, H. (2005). Kinderlosigkeit von Frauen im Spiegel des Mikrozensus.

Dyer, D. (2000). Interpersonal Relationships. London: Routledge.

Dyer, S. J. (2007). The value of children in African countries‐‐Insights from studies on

infertility. Journal of Psychosomatic Obstetrics & Gynecology, 28(2), 69‐77.

Dykstra, P. A. (2006). Off the Beaten Track: Childlessness and Social Integration in

Late Life. Research on Aging, 28(6), 749‐767.

Ellis, B. J. (2004). Timing of Pubertal Maturation in Girls: An Integrated Life History

Approach. Psychological Bulletin, 130(6), 920‐958.

Ellis, B. J., & Garber, J. (2000). Psychosocial antecedents of variation in girlsʹ pubertal

timing: Maternal depression, stepfather presence, and marital and family

stress. Child Development, 71(2), 485‐501.

Ellis, B. J., McFadyen‐Ketchum, S., Dodge, K. A., Pettit, G. S., & Bates, J. E. (1999).

Quality of early family relationships and individual differences in the timing

of pubertal maturation in girls: A longitudinal test of an evolutionary model.

Journal of Personality and Social Psychology, 77(2), 387‐401.

Exley, C., & Letherby, G. (2001). Managing a disrupted lifecourse: Issues of identity

and emotion work. Health: An Interdisciplinary Journal for the Social Study of

Health, Illness and Medicine, 5(1), 112‐132.

Feeney, B. C. (2007). The dependency paradox in close relationships: Accepting de‐

pendence promotes independence. Journal of Personality and Social Psychology,

92(2), 268‐285. References 203

Fehr, E., & Gächter, S. (2002). Altruistic punishment in humans. Nature, 415(6868),

137‐140.

Ferring, D., Filipp, S.H., & Schmidt, K. (1996). Die ”Skala zur Lebensbewertung”:

Empirische Skalenkonstruktion und erste Befunde zu Reliabilität, Stabilität

und Validität. Zeitschrift für Differentielle und Diagnostische Psychologie, 17, 141‐

153.

Fetchenhauer, D., & Bierhoff, H.‐W. (2004). Altruismus aus evolutionstheoretischer

Perspektive. Zeitschrift für Sozialpsychologie, 35(3), 131‐141.

Fingerman, K. L. (1998). Tight lips?: Aging mothers’ and adult daughters’ responses

to interpersonal tensions in their relatiosnships. Personal Relationships, 5, 121‐

138.

Fingerman, K. L. (2003). Mothers and Their Adult Daughters: Mixed Emotions, Enduring

Bonds. New York: Springer.

Fingerman, K. L., & Hay, E. L. (2002). Searching under the streetlight?: Age biases in

the personal and family relationships literature. Personal Relationships, 9(4),

415‐433.

Fingerman, K. L., Hay, E. L., & Birditt, K. S. (2004). The best of ties, the worst of ties:

Close, problematic, and ambivalent social relationships. Journal of Marriage and

Family, 66(3), 792‐808.

Fingerman, K. L., Miller, L., & Charles, S. (2008). Saving the best for last: How adults

treat social partners of different ages. Psychology and Aging, 23(2), 399‐409.

Fiori, K. L., Antonucci, T. C., & Cortina, K. S. (2006). Social Network Typologies and

Mental Health Among Older Adults. Journals of Gerontology: Series B: Psycho‐

logical Sciences and Social Sciences(1), P25‐p32.

Fiske, A. P. (1992). The four elementary forms of sociality: framework for a unified

theory of social relations. Psychological Review, 99(4), 689‐723. References 204

Fiske, A. P., & Haslam, N. (1996). Social cognition is thinking about relationships.

Current Directions in Psychological Science, 5(5), 137‐142.

Fiske, A. P., Haslam, N., & Fiske, S. T. (1991). Confusing one person with another:

What errors reveal about the elementary forms of social relations. Journal of

Personality and Social Psychology, 60(5), 656‐674.

Fleming, J. S., & Courtney, B. E. (1984). The dimensionality of self‐esteem: II. Hierar‐

chical facet model for revised measurement scales. Journal of Personality and

Social Psychology, 46(2), 404‐421.

Foa, E. B., & Foa, U. G. (1976). Resource theory of social exchange. In j. S. Thibaut, J.

Spence, & R. Carson (Eds.), Contemporary topics in social psychology. Morris‐

town, NJ: General Learning Press.

Fooken, I. (1997). Intimität auf Abstand. Familienbeziehungen und soziale Netzwer‐

ke. In D. I. f. Fernstudienforschung (Ed.), Funkkolleg Altern (Vol. 14, pp. 1‐42).

Tübingen.

Freund, A. M. (1995). Die Selbstdefinition alter Menschen. Inhalt, Struktur und Funktion.

Studien und Berichte 61. Berlin Max‐Planck‐Institut für Bildungsforschung.

Fung, H. H., Stoeber, F. S., Yeung, D. Y.‐l., & Lang, F. R. (2008). Cultural Specificity of

Socioemotional Selectivity: Age Differences in Social Network Composition

Among Germans and Hong Kong Chinese. Journal of Gerontology, 63B(3), 156‐

164.

Fung, H. H., Yeung, D. Y., Li, B., & Lang, F. R. (in subm.). The benefits of negative

social exchanges for emotional closeness.

Gable, S. L., & Reis, H. (1999). Now and then, them and us, this and that: Studying

relationships across time, partner, context, and person. Personal Relationships,

6, 415‐432.

Gallucci, M., & Perugini, M. (2000). An experimental test of a game‐theoretical model

of reciprocity. Journal of Behavioral Decision Making, 13(4), 367‐389. References 205

Gerstorf, D., Ram, N., Estabrook, R., Schupp, J. r., Wagner, G. G., & Lindenberger, U.

(2008). Life satisfaction shows terminal decline in old age: Longitudinal evi‐

dence from the German Socio‐Economic Panel Study (SOEP). Developmental

Psychology, 44(4), 1148‐1159.

Gillespie, R. (2001). Contextualizing voluntary childlessness within a postmodern

model of reproduction: implications for health and social needs. SAGE Publica‐

tions, 21(2), 139.

Gillespie, R. (2003). Childfree and feminine: Understanding the gender identity of

voluntarily childless women. Gender & Society, 17(1), 122‐136.

Gloger‐Tippelt, G., Grimmig, R., & Gomille, B. (1993). Der Kinderwunsch aus psycholo‐

gischer Sicht. Opladen: Leske und Budrich.

Gnoth, C. (2004). Zur Definition und Prävalenz von Subfertilität und Infertilität. Jour‐

nal für Reproduktionsmedizin und Endokrinologie, 1(4), 272‐278.

Gnoth, C., Godehardt, D., Godehardt, E., Frank‐Herrmann, P., & Freundl, G. (2003).

Time to pregnancy: results of the German prospective study and impact on

the management of infertility. Fertility and Sterility, 80(Supplement 3), 25‐25.

Gonzalez, R., & Griffin, D. (2002). Modeling the personality of dyads and groups.

Journal of Personality, 70(6), 901‐924

Gosling, S. D., Vazire, S., Srivastava, S., & John, O. P. (2004). Should We Trust Web‐

Based Studies? A Comparative Analysis of Six Preconceptions About Internet

Questionnaires. American Psychologist, 59(2), 93‐104.

Gouldner, A. W. (1960). The Norm of Reciprocity: A Preliminary Statement. American

Sociological Review, 25(2), 161‐178.

Gracia, E., & Herrero, J. (2004). Personal and situational determinants of relationship‐

specific perceptions of social support. Social Behavior and Personality, 32(5), 459‐

476. References 206

Gray‐Little, B., Williams, V. S. L., & Hancock, T. D. (1997). An Item Response Theory

Analysis of the Rosenberg Self‐Esteem Scale. Personality and Social Psychology

Bulletin, 23(5), 443‐451.

Greve, W. (2000). Die Psychologie des Selbst ‐Konturen eines Forschungsthemas. In

W. Greve (Ed.), Psychologie des Selbst. Weinheim: Beltz.

Greil, A. L., Leitko, T. A., & Porter, K. L. (1988). Infertility: His and Hers. Gender &

Society, 2(2), 172‐199.

Griffin, D. W., & Bartholomew, K. (1994). Models of the self and other: Fundamental

dimensions underlying measures of adult attachment. Journal of Personality and

Social Psychology, 67(3), 430‐445.

Grundy, E. (2005). Reciprocity in relationships: Socio‐economic and health influences

on intergenerational exchanges between Third Age parents and their adult

children in Great Britain. British Journal of Sociology, 56(2), 233‐255.

Grundmann, M. (1996). Historical context of father absence: Some consequences for

the family formation of German men. International Journal of Behavioral Devel‐

opment, 19(2), 415‐431.

Hakim, C. (2003). A New Approach to Explaining Fertility Patterns: Preference The‐

ory. Population and Development Review, 29(3), 349‐374.

Hamilton, W. D. (1964). The genetical evolution of social behaviour. II. Journal of

Theoretical Biology, 7(1), 17‐52.

Hannover, B. (2000). Das kontextabhängige Selbst oder warum sich unser Selbst mit

dem sozialen Kontext verändert. In W. Greve (Ed.), Psychologie des Selbst (pp.

15‐36). Weinheim: Beltz.

Hartup, W. W. (1998). The company they keep: Friendships and their developmental

significance. In M. E. Hertzig & E. A. Farber (Eds.), Annual progress in child psy‐

chiatry and child development: 1997 (pp. 63‐77). Philadelphia, PA, US: Brun‐

ner/Mazel. References 207

Hartup, W. W., & Stevens, N. (1997). Friendships and Adaptation in the Life Course.

Psychological Bulletin, 121(3), 355‐370.

Heatherton, T. F., & Polivy, J. (1991). Development and Validation of a Scale for

Measuring State Self‐Esteem. Journal of Psychology and Social Psychology, 69,

895‐910.

Heckhausen, J. (1997). Developmental Regulation Across Adulthood: Primary and

Secondary Control of Age‐Related Challenges. Developmental Psychology, 33(1),

176‐187.

Heckhausen, J., & Schulz, R. (1995). A life‐span theory of control. Psychological Re‐

view, 102(2), 284‐304.

Heckhausen, J., Wrosch, C., & Fleeson, W. (2001). Developmental regulation before

and after a developmental deadline: The sample case of ʹbiological clockʹ for

childbearing. Psychology and Aging, 16(3), 400‐413.

Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.

Helfferich, C., & Küppers‐Chinnow, M. (1996). Verbeitung von und Umgang mit

Fruchtbarkeitsstörungen. In B. f. g. Aufklärung (Ed.), Kontrazeption, Konzepti‐

on, Kinder oder keine. Dokumentation einer Expertentagung (pp. 113‐136). Köln:

BZgA.

Hendrick, S. S., Hendrick, C., & Adler, N. L. (1988). Romantic relationships: Love,

satisfaction, and staying together. Journal of Personality and Social Psychology,

54(6), 980‐988.

Henning, K., & Strauß, B. (2002). Psychologie und psychosomatische Aspekte der

ungewollten Kinderlosigkeit. Zum Stand der Forschung. In B. Strauß (Ed.),

Ungewollte Kinderlosigkeit. Psychologische Diagnostik, Beratung und Therapie (pp.

15‐33). Göttingen: Hogrefe. References 208

Hessel, A., Schumacher, J., Geyer, M. & Brähler, E. (2001). Symptom‐Checkliste SCL‐

90‐R: Testtheoretische Überprüfung und Normierung an einer bevölkerungs‐

repräsentativen Stichprobe. Diagnostica, 47 (1), 27‐39.

Hill, S. E., & Buss, D. M. (2006). The Evolution of Self‐Esteem. In M. H. Kernis (Ed.),

Self‐esteem issues and answers: A sourcebook of current perspectives (pp. 328‐333).

New York, NY, US: Psychology Press.

Hinde, R. A. (1993). Auf einem Weg zu einer Wissenschaft zwischenmenschlicher

Beziehungen. In A. E. Auhagen & M. von Salisch (Eds.), Zwischenmenschliche

Beziehungen (pp. 7‐12). Göttingen: Hogrefe.

Hird, M. J., & Abshoff, K. (2000). Women without children: A contradiction in terms?

Journal of Comparative Family Studies, 31(3), 347‐366.

Hoffman, L. W., & Hoffman, M. L. (1973). The Value of Children to Parents. In J. T.

Fawcett (Ed.), Psychological Perspectives on Population (pp. 19‐76). New York:

Basic Books.

Hollstein, B. (2001). Grenzen sozialer Integration. Zur Konzeptioin informeller Beziehungen

und Netzwerke. Opladen: Leske & Budrich.

Hollstein, B. (2003). Netzwerkveränderungen verstehen. Zur Integration von struk‐

tur‐ und akteurstheoretischen Perspektiven. Berliner Journal für Soziologie, 13,

153‐174.

Holmes, J. G. (2002). Interpersonal expectations as the building blocks of social cogni‐

tion: An interdependence theory perspective. Personal Relationships, 9(1), 1‐26.

Homans, G. C. (1961). Social behavior. Its elementary forms. New York: Harecourt, Brace

& World.

Houghten, D., & Houghten, P. (1987). Coping with Childlessness. London: Unwin Pa‐

perbacks.

Houseknecht, S. K. (1979). Childlessness and marital adjustment. Journal of Marriage

& the Family, 41(2), 259‐265. References 209

Ikkink, K. K., & van Tilburg, T. (1998). Do older adults’ network members continue

to provide instrumental support in unbalanced relationships? Journal of Social

and Personal Relationships, 15, 59‐75.

Klein Ikkink, K., & van Tilburg, T. (1999). Broken ties: reciprocity and other factors

affecting the termination of older adultsʹ relationships. Social Networks, 21(2),

131‐146.

Ikkink, K. K., van Tilburg, T., & Knipscheer, K. C. P. M. (1999). Perceived instrumen‐

tal support exchanges in relationships between elderly parents and their adult

children: Normative and structural explanations. Journal of Marriage & the Fam‐

ily, 61(4), 831‐844.

Ishii‐Kuntz, M., & Seccombe, K. (1989). The impact of children upon social support

networks throughout the life course. Journal of Marriage & the Family, 51(3),

777‐790.

Jordan, C., & Revenson, T. A. (1999). Gender differences in coping with infertility: A

meta‐analysis. Journal of Behavioral Medicine, 22(4), 341‐358.

Jung, J. (1990). The role of reciprocity in social support. Basic and Applied Social Psy‐

chology, 11(3), 243‐253.

Kan, M. Y. (2007). Work Orientation and Wivesʹ Employment Careers: An Evaluation

of Hakimʹs Preference Theory. Work and Occupations, 34, 430‐462.

Kenny, D. A., & Cook, W. (1999). Partner effects in relationship research: Conceptual

issues, analytic difficulties, and illustrations. Personal Relationships, 6(4), 433‐

448.

Kenny, D. A., Kashy, D. A., & Cook, W. (2006). Analyzing Mixed Independent Vari‐

ables: The Actor‐Partner Interdependence Model. In D. A. Kenny, D. A. Kashy

& W. Cook (Eds.), Dyadic data analysis. New York: Guilford Press.

Kernis, M. H. (2003). Toward a Conceptualization of Optimal Self‐Esteem. Psychologi‐

cal Inquiry, 14(1), 1‐26. References 210

Keysar, B., Converse, B. A., Wang, J., & Epley, N. (2008). Reciprocity is not give and

take. Psychological Science, 19(12), 1280‐1286.

Kirkpatrick, L. A., & Ellis, B. J. (2001). An evolutionary‐psychological approach to

self‐esteem: Multiple domains and multiple functions. In G. Fletcher & M.

Clark (Eds.), The Blackwell handbook of social psychology: Vol. 2: Interpersonal proc‐

esses (pp. 411‐436). Oxford, England: Blackwell..

Kirkpatrick, L. A., & Ellis, B. J. (2006). The Adaptive Functions of Self‐Evaluative

Psychological Mechanisms. In M. H. Kernis (Ed.), Self‐esteem issues and answers:

A sourcebook of current perspectives (pp. 334‐339). New York, NY, US: Psychol‐

ogy Press.

Kirkpatrick, L. A., Waugh, C. E., Valencia, A., & Webster, G. D. (2002). The functional

domain specificity of self‐esteem and the differential prediction of aggression.

Journal of Personality and Social Psychology, 82(5), 756‐767.

Kogovsek, T., & Ferligoj, A. k. (2005). Effects on reliability and validity of egocen‐

tered network measurements. Social Networks, 27(3), 205‐229.

Koropeckyj‐Cox, T., & Call, V. R. A. (2007). Characteristics of older childless persons

and parents: Cross‐national comparisons. Journal of Family Issues, 28(10), 1362‐

1414.

Kotter‐Grün, D. (2008). Development and Functionality of Life Longings: The Case of In‐

voluntary Childlessness. Dissertation Thesis, FU Berlin

Kowalcek, I. (1998). Reproduktionsmedizin und Psychosomatik: Gegensatz, Wider‐

spruch oder Annäherung? Reproduktionsmedizin, 14, 275‐281.

Krackhardt, D., & Kilduff, M. (1999). Whether close or far: Social distance effects on

perceived balance in friendship networks. Journal of Personality and Social Psy‐

chology, 76(5), 770‐782.

Krebs, D. L. (1991). Altruism and : A false dichotomy? Psychological Inquiry,

2(2), 137‐139. References 211

Krebs, D. L., & Hesteren, F. V. (1994). The Development of Altruism: Toward an In‐

tegrative Model. Developmental Review, 14(2), 103‐158.

La Guardia, J. G., & Ryff, C. (2003). Self‐Esteem Challenges: Comment. Psychological

Inquiry, 14(1), 48‐51.

Lamprecht, J., Wagner, J., & Lang, F. R. (2008). Kinder? Küche? Karriere? – Dring‐

lichkeit des Kinderwunsches, Geschlechtsrollenorientierung und Aufgaben‐

verteilung in kinderlosen Partnerschaften. Zeitschrift für Entwicklungspsycholo‐

gie und Pädagogische Psychologie. 40, 112‐123.

Lang, F. R. (2001). Regulation of Social Relationships in Later Adulthood. Journal of

Gerontology, 56B(6), 321‐326.

Lang, F. R. (2003). Gestaltung und Regulation sozialer Beziehungen im Lebenslauf:

Eine entwicklungspsychologische Perspektive. Berliner Journal für Soziologie,

2/03, 175‐195.

Lang, F. R. (2005). Die Gestaltung sozialer Netzwerke im Lebenslauf. In U. Otto & P.

Bauer (Eds.), Mit Netzwerken professionell zusammenarbeiten. Band 1: Soziale

Netzwerke in Lebenslauf‐ und Lebenslagenperspektiven (pp. 41‐63). Tübingen:

DGVT Verlag.

Lang, F. R. (2007a). Motivation, Selbstverantwortung und Beziehungsregulation im

mittleren und höheren Erwachsenenalter. In H. W. Wahl & H. Mollenkopf

(Eds.), Alternsforschung am Beginn des 21. Jahrhunderts. Alterns‐ und Lebenslauf‐

konzeptionen im deutschsprachigen Raum. Heidelberg: Akademische Verlagsans‐

talt.

Lang, F. R. (2007b). Regulating Interpersonal Contexts Across the Lifespan: Psychological

mechanisms of linked lives? Paper presented at the Linked lives and self‐

regulation. Lifespan ‐ Life Course: Is it really the same?. References 212

Lang, F. R., & Carstensen, L. L. (1994). Close emotional relationships in late life: Fur‐

ther support for proactive aging in the social domain. Psychology and Aging,

9(2), 315‐324.

Lang, F. R., & Carstensen, L. L. (2002). Time counts: Future time perspective, goals,

and social relationships. Psychology and Aging, 17(1), 125‐139.

Lang, F. R., Diewald, M., & Wend, V. (2004). Skala zu Lebenszielen. IDUN‐Projekt.

Lang, F. R., & Heckhausen, J. (2006). Developmental Changes of Motivation and In‐

terpersonal Capacities Across Adulthood: Managing the Challenges and Con‐

straints of Social Contexts. In C. Hoare (Ed.), The Oxford Handbook of Adult De‐

velopment and Learning (pp. 149‐166). Oxford: Oxford University Press.

Lang, F. R., & Neyer, F. J. (2005). Soziale Beziehungen als Anlage und Umwelt: Ein

evolutionspsychologisches Rahmenmodell der Beziehungsregulation. Zeit‐

schrift für Soziologie der Erziehung und Sozialisation, 25(2), 162‐177.

Lang, F. R., & Neyer, F. J. (2008). Soziale Beziehungen als Anlage und Umwelt. In F. J.

Neyer & F. M. Spinath (Eds.), Anlage und Umwelt. Neue Perspektiven der Verhal‐

tensgenetik und Evolutionspsychologie (pp. 173‐192). Stuttgart: Lucius & Lucius.

Lang, F. R., Neyer, F. J., & Asendorpf, J. B. (2005). Entwicklung und Gestaltung sozia‐

ler Beziehungen. In S.‐H. Filipp & U. M. Staudinger (Eds.), Enzyklopädie der

Psychologie. Band C/V/6. Entwicklungspsychologie des mittleren und höheren Er‐

wachsenenalters. Göttingen: Hogrefe.

Lang, F.R., Neyer, F. J., Wrzus, C., & Wagner (2005). RELATE: Study Manual. Unpub‐

lished Manuscript.

Lang, F. R., Wagner, J., & Neyer, F. J. (in subm.). Interpersonal contexts across the

lifespan: A theory of relationship regulation. To appear in: Advances in Life

Course Research. References 213

Lawrence, E., Rothman, A. D., Cobb, R. J., Rothman, M. T., & Bradbury, T. N. (2008).

Marital Satisfaction Across the Transition to Parenthood. Journal of Family Psy‐

chology, 22(1), 41‐50.

Leary, M. R. (1999). Making sense of self‐esteem. Current Directions in Psychological

Science, 8(1), 32‐35.

Leary, M. R. (2003a). Interpersonal Aspects of Optimal Self‐Esteem and the Authentic

Self: Comment. Psychological Inquiry, 14(1), 52‐54.

Leary, M. R. (2003b). Commentary on Self‐Esteem as an Interpersonal Monitor: The

Sociometer Hypothesis. Psychological Inquiry, 14, 270‐274.

Leary, M. R. (2004a). The Sociometer, Self‐Esteem, and the Regulation of Interper‐

sonal Behavior. In R. F. Baumeister & K. D. Vohs (Eds.), Handbook of Self‐

Regulation. Researach, Theory, and Applications (pp. 373‐391). New York: The

Guilford Press.

Leary, M. R. (2004b). The Function of Self‐Esteem in Terror Management Theory and

Sociometer Theory: Comment on Pyszczynski et al. (2004). Psychological Bulle‐

tin, 130(3), 478‐482.

Leary, M. R. (2007). Motivational and Emotional Aspects of the Self. Annual Review of

Psychology, 58, 317‐344.

Leary, M. R., & Baumeister, R. F. (2000). The nature and function of self‐esteem: Soci‐

ometer theory. In M. P. Zanna (Ed.), Advances in experimental social psychology

(Vol. 32, pp. 1‐62). San Diego: Academic Press.

Leary, M. R., Cottrell, C. A., & Phillips, M. (2001). Deconfounding the effects of

dominance and social acceptance on self‐esteem. Journal of Personality and So‐

cial Psychology, 81(5), 898‐909.

Leary, M. R., Gallagher, B., Fors, E., Buttermore, N., Baldwin, E., Kennedy, K., Mills,

A. (2003). The invalidity of disclaimers about the effects of social feedback on

self‐esteem. Personality and Social Psychology Bulletin, 29(5), 623‐636. References 214

Leary, M. R., Haupt, A. L., Strausser, K. S., & Chokel, J. T. (1998). Calibrating the so‐

ciometer: The relationship between interpersonal appraisals and the state self‐

esteem. Journal of Personality and Social Psychology, 74(5), 1290‐1299.

Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995). Self‐Esteem as an

Interpersonal Monitor: The Sociometer Hypothesis. Journal of Personality and

Social Psychology, 68(3), 518‐530.

Lehnart, J., & Neyer, F. J. (2006). Should I Stay or Should I Go? Attachment and Per‐

sonality in Stable and Instable Romantic Relationships. European Journal of Per‐

sonality, 20(6), 475‐495.

Lemme, B. H. (2006). Development in Adulthood. Boston: Pearson.

Letherby, G. (2002a). Challenging dominant discourses: Identity and change and the

experience of ʹinfertilityʹ and ʹinvoluntary childlessness.ʹ. Journal of Gender

Studies, 11(3), 277‐288.

Letherby, G. (2002b). Childless and bereft?: Stereotypes and realities in relation to

ʹvoluntaryʹ and ʹinvoluntaryʹ childlessness and womanhood. Sociological In‐

quiry, 72(1), 7‐20.

Lewinter, M. (2003). Reciprocities in caregiving relationships in Danish elder care.

Journal of Aging Studies, 17(3), 357‐377.

Lin, L.‐W. (2004). Intergenerational Interdependence: Mid‐Life Couplesʹ Help Ex‐

change in a Three‐Generational Model. Family & Consumer Sciences Research

Journal, 32(3), 275‐290.

Lin, I. F. (2008). Consequences of parental divorce for adult childrenʹs support of

their frail parents. Journal of Marriage and Family, 70(1), 113‐128.

Lippe, H. v. d. (2002). Vater werden, Vater sein, Vater bleiben. Dokumentation einer

Fachtagung der Heinrich‐Böll‐Stiftung und des Forum Männer in Theorie und

Praxis der Geschlechterverhältnisse. Berlin. References 215

Lippe, H. v. d., & Bernardi, L. (2006). Zwei deutsche Ansichten über Kinder und Kar‐

riere. Demographische Forschung. Aus Erster Hand, 3, 1‐2.

Lorber, J., & Bandlamudi, L. (1993). The dynamics of marital bargaining in male in‐

fertility. Gender & Society, 7(1), 32‐49.

Love, K. M., & Murdock, T. B. (2004). Attachment to Parents and Psychological Well‐

Being: An Examination of Young Adult College Students in Intact Families

and Stepfamilies. Journal of Family Psychology, 18(4), 600‐608.

Lowenstein, A., Katz, R., & Gur‐Yaish, N. (2007). Reciprocity in Parent‐Child Ex‐

change and Life Satisfaction among the Elderly: A Cross‐National Perspective.

Journal of Social Issues, 865‐883.

Lüscher, K. (2005). Looking at ambivalences: The contribution of a “Newold” View of

intergenerational relations to the study of the life course. Towards an Interdisci‐

plinary Perspective on the Life Course Advances in Life Course Research, 10, 95‐131.

Lüscher, K., & Pillemer, K. (1998). Intergenerational AmbivalenceA New Approach

to the Study of Parent‐Child Relations in Later Life. Working Paper.

Luke, M. A., Maio, G. R., & Carnelley, K. B. (2004). Attachment models of the self and

others: Relations with self‐esteem, humanity‐esteem, and parental treatment.

Personal Relationships, 11(3), 281‐303.

Luttbeg, B., Borgerhoff Mulder, M., & Mangel, M. (2000). To Marry Again or Not ‐ A

Dynamic Model for Demographic Transition. In L. Cronk, N. A. Chagnon &

W. Irons (Eds.), Adaption and Human Behavior: an Anthropological Perspective

(pp. 345‐368). Hawthorne, New York: Aldine de Gruyter.

Lutz, W. (2006). Fertility rates and future population trends: will Europe’s birth rate

recover or continue to decline? International Journal of Andrology, 29, 25‐33.

Mandemakers, J. J., & Dykstra, P. A. (2008). Discrepancies in parentʹs and adult

childʹs reports of support and contact. Journal of Marriage and Family, 70(2),

495‐506. References 216

Marcoen, A. (2005). Parent Care: the core component of intergenerational relation‐

ships in middle and late adulthood. European Journal of Ageing, 2, 208‐212.

Markus, H., & Wurf, E. (1987). The dynamic self‐concept: A social psychological per‐

spective. Annual Review of Psychology, 38, 299‐337.

Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition,

emotion, and motivation. Psychological Review, 98(2), 224‐253.

Markus, H. R., & Kitayama, S. (2003). Culture, Self, and the Reality of the Social. Psy‐

chological Inquiry, 14(3), 277‐283.

Marsh, H. W., & OʹNeill, R. (1984). Self Description Questionnaire III: The construct

validity of multidimensional self‐concept ratings by late adolescents. Journal of

Educational Measurement, 21(2), 153‐174.

Marsh, H. W., Trautwein, U., Lüdtke, O., Köller, O., & Baumert, J. (2006). Integration

of Multidimensional Self‐Concept and Core Personality Constructs: Construct

Validation and Relations to Well‐Being and Achievement. Journal of Personal‐

ity, 74(2), 403‐456.

Mather, M., & Carstensen, L. L. (2005). Aging and motivated cognition: The positivity

effect in attention and memory. Trends in Cognitive Sciences, 9(10), 496‐502.

Matthews, R., & Matthews, A. M. (1986). Infertility and Involuntary Childlessness.

Journal of Marriage & the Family, 48(3), 641‐649.

McMullin, J. A., & Cairney, J. (2004). Self‐esteem and the intersection of age, class,

and gender. Journal of Aging Studies, 18(1), 75‐90.

Meulemann, H. (2006). Religiösität: Immer noch die Persistenz eines Sonderfalls. Aus

Politik und Zeitgeschichte, 15‐22, 30‐31.

Mikulincer, M., & Shaver, P. R. (2001). Attachment theory and intergroup : Evi‐

dence that priming the secure base schema attenuates negative reactions to

out‐groups. Journal of Personality and Social Psychology, 81(1), 97‐115. References 217

Mikulincer, M., & Shaver, P. R. (2005). Attachment Security, Compassion, and Altru‐

ism. Current Directions in Psychological Science, 14(1), 34‐38.

Mikulincer, M., Shaver, P. R., Gillath, O., & Nitzberg, R. A. (2005). Attachment, Care‐

giving, and Altruism: Boosting Attachment Security Increases Compassion

and Helping. Journal of Personality and Social Psychology, 89(5), 817‐839.

Miller, W. B. (1992). Personality Traits and Developmental Experiences as Antece‐

dents of Childbearing Motivation. Demography, 29(2), 265‐285.

Mitchell, D., & Gray, E. (2007). Declining fertility: Intentions, attitudes and aspira‐

tions. Journal of Sociology, 43(1), 23‐44.

Murray, S. L. (2005). Regulating the Risks of Closeness: A Relationship‐Specific Sense

of Felt Security. Current Directions in Psychological Science, 14(2), 74‐78.

Murray, S. L., Holmes, J. G., & Griffin, D. W. (2000). Self‐Esteem and the Quest for

Felt Security: How Perceived Regard Regulates Attachment Processes. Journal

of Personality and Social Psychology, 78(3), 478‐498.

Murray, S. L., Holmes, J. G., MacDonald, G., & Ellsworth, P. C. (1998). Through the

looking glass darkly? When self‐doubts turn into relationship insecurities.

Journal of Personality and Social Psychology, 75(6), 1459‐1480.

Murray, S. L., Bellavia, G., Feeney, B., Holmes, J. G., & Rose, P. (2001). The contingen‐

cies of interpersonal acceptance: When romantic relationships function as a

self‐affirmational resource. Motivation and Emotion, 25(2), 163‐189.

Murray, S. L., Rose, P., Bellavia, G. M., Holmes, J. G., & Garrett Kusche, A. (2002).

When Rejection Stings: How Self‐Esteem Constrains Relationship‐

Enhancement Processes. Journal of Personality and Social Psychology, 83(3), 556‐

573. References 218

Murray, S. L., Rose, P., Holmes, J. G., Derrick, J., Podchaski, E. J., Bellavia, G., et al.

(2005). Putting the Partner Within Reach: A Dyadic Perspective on Felt Secu‐

rity in Close Relationships. Journal of Personality and Social Psychology, 88(2),

327‐347.

Muthen & Muthen (1998‐2006). Mplus. Software.

Nash, A. (1988). Onogeny, Ohylogeny, and Relationshipy. In S. W. Duck (Ed.), Hand‐

book of Personal Relationships (pp. 121‐141). Chichester: John Wiley & Sons.

Nauck, B. (1989). Intergenerational relationships in families from Turkey and Ger‐

many. European Sociological Review, 5(3), 251‐274.

Nauck, B. (2001). Der Wert von Kindern für ihre Eltern. Kölner Zeitschrift für Soziologie

und Sozialpsychologie, 53(3), 407‐436.

Nave‐Herz, R. (1988). Kinderlose Ehen. Eine empirische Studie über die Lebenssituation

kinderloser Ehepaare und die Gründe für ihre Kinderlosigkeit. Weinheim: Juventa.

Nave‐Herz, R. (2002). Family Change and Intergenerational Relationships in Ger‐

many. In R. Nave‐Herz (Ed.), Family change and Intergenerational Relations in

Different Cultures. Band 9: Familie und Gesellschaft (pp. 215‐248). Würzburg: Er‐

gon.

Nave‐Herz, R. (2004). Ehe‐ und Familiensozilogie. Eine Einführung in die Geschichte, theo‐

retische Ansätze und emprische Befunde. Weinheim: Juventa.

Neberich, W., Lehnart, J., Penke, L., & Asendorpf, J. B. (in press). Family of Origin, Age

at Menarche, and Reproductive Strategies: A Test of Four Evolutionary‐

Developmental Models.

Neiss, M. B., Sedikides, C., & Stevenson, J. (2002). Self‐esteem: A behavioural genetic

perspective. European Journal of Personality, 16(5), 351‐368.

Neyer, F. J. (1997). Free recall or recognition in collecting egocentered networks: The

role of survey techniques. Journal of Social and Personal Relationships, 14(3), 305‐

316. References 219

Neyer, F. J., & Asendorpf, J. B. (2001). Personality‐relationship transaction in young

adulthood. Journal of Personality and Social Psychology, 81(6), 1190‐1204.

Neyer, F. J., & Lehnart, J. (2005). Relationships Matter In Personality Development:

Evidence From An 8‐Year Longitudinal Study Across Young Adulthood. Jour‐

nal of Personality, 75 (3), 535‐568.

Neyer, F. J., & Lang, F. R. (2003). Blood is thicker than water: Kinship orientation

across adulthood. Journal of Personality and Social Psychology, 84(2), 310‐321.

Neyer, F. J., & Voigt, D. (2004). Personality and Social Network Effects on Romantic

Relationships: A Dyadic Approach. European Journal of Personality, 18(4), 279‐

299.

Neyer, F. J., Wrzus, C., Wagner, J., & Lang, F. R. (in subm.). Principles of Relationship

Differentiation.

Nezlek, J. B., Schröder, M., & Schütz, A. (2006). Mehrebenenanalysen in der psycho‐

logischen Forschung: Vorteile und Möglichkeiten der Mehrebenenmodellie‐

rung mit Zufallskoeffizienten. Psychologische Rundschau, 57(4), 213‐223.

Park, K. (2002). Stigma management among the voluntarily childless. Sociological Per‐

spectives, 45(1), 21‐45.

Park, K. (2005). Choosing Childlessness: Weberʹs Typology of Action and Motives of

the Voluntarily Childless. Sociological Inquiry, 75(3), 372‐402.

Park, L. E., Crocker, J., & Mickelson, K. D. (2004). Attachment Styles and Contingen‐

cies of Self‐Worth. Personality and Social Psychology Bulletin, 30(10), 1243‐1254.

Parrott, T. M., & Bengtson, V. L. (1999). The effects of earlier intergenerational affec‐

tion, normative expectations and family conflict on contemporary exchanges

of help and support. Research on Aging, 21(1), 73‐105.

Perrig‐Chiello, A., & Höpflinger, F. (2005). Aging parents and their middle‐aged

children: demographic and psychological challenges. European Journal of A‐

geing, 2, 183‐191. References 220

Perugini, M., Gallucci, M., Presaghi, F., & Ercolani, A. P. (2003). The Personal Norm

of Reciprocity. European Journal of Personality, 17(4), 251‐283.

Posada, G. n., Jacobs, A., Carbonell, O. A., Alzate, G., Bustamante, M. R., & Arenas,

A. (1999). Maternal care and attachment security in ordinary and emergency

contexts. Developmental Psychology, 35(6), 1379‐1388.

Pyszczynski, T., Greenberg, J., Solomon, S., Arndt, J., & Schimel, J. (2004). Why Do

People Need Self‐Esteem? A Theoretical and Empirical Review. Psychological

Bulletin, 130(3), 435‐468.

Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical Linear Models. Applications and

Data Analysis Methods. Thousand Oaks: Sage Publications.

Rawn, C. D., Vohs, K. D., & Finkel, E. J. (2006). The Importance of Self‐Regulation for

Interpersonal Functioning Self and relationships: Connecting intrapersonal and in‐

terpersonal processes (pp. 15‐31). New York, NY US: Guilford Press.

Roberts, R. E. L., & Bengtson, V. L. (1993). Relationships with parents, self‐esteem,

and psychological well‐being in young adulthood. Social Psychology Quarterly,

56(4), 263‐277.

Roberts, R. E. L., & Bengtson, V. L. (1996). Affective ties to parents in early adulthood

and self‐esteem across 20 years. Social Psychology Quarterly, 59(1), 96‐106.

Roberts, B. W., & DelVecchio, W. F. (2000). The Rank‐Order Consistency of Personali‐

ty Traits From Childhood to Old Age: A Quantitative Review of Longitudinal

Studies. Psychological Bulletin, 126(1), 3‐25.

Roberts, B. W., Robins, R. W., Trzesniewski, K. H., & Caspi, A. (2003). Personality

Trait Development in Adulthood. In J. T. Mortimer & M. J. Shanahan (Eds.),

Handbook of the Life Course (pp. 579‐598). New York: Kluwer Academic.

Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of Mean‐Level

Change in Personality Traits Across the Life Course: A Meta‐Analysis of Lon‐

gitudinal Studies. Psychological Bulletin, 132(1), 1‐25. References 221

Robins, R. W., Fraley, R. C., Roberts, B. W., & Trzesniewski, K. H. (2001). A longitu‐

dinal study of personality change in young adulthood. Journal of Personality,

69(4), 617‐640.

Robins, R. W., Hendin, H. M., & Trzesniewski, K. H. (2001). Measuring global self‐

esteem: Construct validation of a single‐item measure and the Rosenberg Self‐

Esteem Scale. Personality and Social Psychology Bulletin, 27(2), 151‐161.

Robins, R. W., & Trzesniewski, K. H. (2005). Self‐Esteem Development Across the

Lifespan. Current Directions in Psychological Science, 14(3), 158‐162.

Robins, R. W., Trzesniewski, K. H., Tracy, J. L., Gosling, S. D., & Potter, J. (2002).

Global self‐esteem across the life span. Psychology and Aging, 17(3), 423‐434.

Rosenberg, M., Schooler, C., Schoenbach, C., & Rosenberg, F. (1995). Global self‐

esteem and specific self‐esteem: Different concepts, different outcomes. Ameri‐

can Sociological Review, 60(1), 141‐156.

Rosenstiel, L. v., Spiess, E., Stengel, M., & Nerdinger, F. W. (1984). auf Kinder?

Höchstens 1. Psychologie heute, 4, 20‐31.

Ross, H., Stein, N., Trabasso, T., Woody, E., & Ross, M. (2005). The quality of family

relationships within and across generations: A social relations analysis. Inter‐

national Journal of Behavioral Development, 29(2), 110‐119.

Rossi, A. S., & Rossi, P. H. (1990). Of human bonding. Parent‐child relations across the life

course. New York: Aldine de Gruyter.

Rudolph, A., Schröder‐Abé, M., Schütz, A., Gregg, A. P., & Sedikides, C. (2008).

Through a glass, less darkly? Reassessing convergent and discriminant valid‐

ity in measures of implicit self‐esteem. European Journal of Psychological Assess‐

ment, 24(4), 273‐281.

Rudolph, A., Schütz, A., & Schröder‐Abé, M. (2008). Development and Validation of a

German‐Language Version of the State Self‐Esteem Scale. Paper presented at the

14th European Conference on Personality. References 222

Rudolph, A., Schütz, A., & Schröder‐Abé, M. (2008). Validation of a German Version of

the State Self‐Esteem Scale. Paper presented at the 29th International Congress

of Psychology.

Rusbult, C. E. (1983). A longitudinal test of the investment model: The development

(and deterioration) of satisfaction and commitment in heterosexual involve‐

ments. Journal of Personality and Social Psychology, 45(1), 101‐117.

Rusbult, C. E., Verette, J., Whitney, G. A., Slovik, L. F., & Lipkus, I. (1991). Accom‐

modation processes in close relationships: Theory and preliminary empirical

evidence. Journal of Personality and Social Psychology, 60(1), 53‐78.

Sabatelli, R. M., Meth, R. L., & Gavazzi, S. M. (1988). Factors Mediating the Adjust‐

ment to Involuntary Childlessness. Family Relations, 37(3), 338‐343.

Sahlins, M. (1972). Stone Age Economics. Chicago: Aldine Atherton.

Sam, D. L., Peltzer, K., & Mayer, B. (2005). The Changing Values of Children and

Preferences Regarding Family Size in South Africa. Applied Psychology: An In‐

ternational Review, 54(3), 355‐377.

Santos, J. D., & Levitt, M. J. (2007). Intergenerational relations with in‐laws in the con‐

text of the social convoy: Theoretical and practical implications. Journal of So‐

cial Issues, 63(4), 827‐843.

Satorra, A., & Bentler, P. (2001). A scaled difference chi‐square test statistic for mo‐

ment structure analysis. Psychometrika, 66(4), 507‐514.

Schmeeckle, M., Giarrusso, R., Feng, D., & Bengtson, V. L. (2006). What Makes Some‐

one Family? Adult Childrenʹs Perceptions of Current and Former Stepparents.

Journal of Marriage and Family, 68(3), 595‐610.

Schmitt, C., & Winkelmann, U. (2005). Wer bleibt kinderlos? Sozialstrukturelle Daten zur

Kinderlosigkeit von Frauen und Männern. Berlin: German Institut for Economic

Research. References 223

Schneewind, K. A., Vaskovics, L. A., Backmund, V., Gotzler, P., Rost, H., Salih, A., et

al. (1994). Optionen der Lebensgestaltung junger Ehen und Kinderwunsch. Zweiter

Projektbericht. Schriftenreihe des Bundesministeriums für Familie und Senioren.

Stuttgart: Kohlhammer.

Schneewind, K. A., Vaskovics, L. A., Gotzler, P., Hofmann, B., Rost, H., Schlehlein, B.,

et al. (1996). Optionen der Lebensgestaltung junger Ehen und Kinderwunsch. Ver‐

bundstudie‐Endbericht. Schriftenreihe des Bundesministeriums für Familie, Senioren,

Frauen und Jugend. Stuttgart: Kohlhammer.

Schneider, B. H., Atkinson, L., & Tardif, C. (2001). Child‐parent attachment and chil‐

drenʹs peer relations: A quantitative review. Developmental Psychology, 37(1),

86‐100.

Schröder‐Abé, M., Rudolph, A., & Schütz, A. (2007). High implicit self‐esteem is not

necessarily advantagous: Discrepancies between explicit and implicit self‐

esteem and their relationship with expression and psychological health.

European Journal of Personality, 21(3), 319‐339.

Schütz, A. (1996). Fragebogen zum Zustands‐Selbstwertgefühl. Eine deutsche Version der

State‐Self Esteem Scale von Heatherton und Polivy, 1991.Unpublished manuscript,

Bamberg.

Schütz, A. (2000a). Das Selbstwertgefühl als soziales Konstrukt: Befunde und Wege

der Erfassung. In W. Greve (Ed.), Psychologie des Selbst (pp. 189‐207). Wein‐

heim: Beltz.

Schütz, A. (2000b). Psychologie des Selbstwertgefühls: Von Selbstakzeptanz bis Arroganz.

Stuttgart: Kohlhammer.

Schulz, R. (1996). Die Reziprozität als konstitutives Netzwerkmerkmal. Zeitschrift für

Bevölkerungswissenschhaften, 21, 263‐280. References 224

Schwarz, B., & Trommsdorff, G. (2005a). Intergenerationaler Austausch von Unters‐

tützung und Reziprozität im Kulturvergleich. In A. Steinbach (Ed.), Generati‐

ves Verhalten und Generationenbeziehungen (pp. 199‐212). Wiesbaden: Verlag für

Sozialwissenschaften.

Schwarz, B., & Trommsdorff, G. (2005b). The relation between attachment and inter‐

generational support. European Journal of Ageing, 2(3), 192‐199.

Schwarz, B., Trommsdorff, G., Albert, I., & Mayer, B. (2005). Adult Parent‐Child Rela‐

tionships: Relationship Quality, Support, and Reciprocity. Applied Psychology:

An International Review, 54(3), 396‐417.

Shamana, F. N., & DeCherny, A. H. (1995). Infertility: A Historical Perspective. In W.

R. Keye, R. J. Chang, R. W. Rebar & M. R. Soules (Eds.), Infertility: Evaluation

and Treatment (pp. 3‐7). Philadelphia: W.B: Saunders Company.

Shmotkin, D. (1999). Affective Bonds of Adult Children with Living Versus Deceased

Parents. Psychology and Aging, 14, 473‐482.

Shulman, S., Cohen, O., Feldman, B., & Mahler, A. (2006). Emerging adult men and

their mothers in divorced families: A typology of relationship patterns. Journal

of Social and Personal Relationships, 23(3), 465‐481.

Simpson, J. A., Collins, W. A., Tran, S., & Haydon, K. C. (2007). Attachment and the

experience and expression of emotions in romantic relationships: A develop‐

mental perspective. Journal of Personality and Social Psychology, 92(2), 355‐367.

Slade, A. (1987). Quality of attachment and early symbolic play. Developmental Psy‐

chology, 23(1), 78‐85.

Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis. An introduction to basic and

advanced multilevel modeling. London: Sage Publications.

Somers, M. D. (1993). A comparison of voluntarily childfree adults and parents. Jour‐

nal of Marriage & the Family, 55(3), 643‐650. References 225

Srivastava, S., & Beer, J. S. (2005). How Self‐Evaluations Relate to Being Liked by

Others: Integrating Sociometer and Attachment Perspectives. Journal of Perso‐

nality and Social Psychology, 89(6), 966‐977.

Statistisches Bundesamt (2006). Kinderlosigkeit von Akademikerinnen im Spiegel des Mik‐

rozensus. Wiesbaden.

Statistisches Bundesamt (2007a). Bevölkerung. Statistisches Jahrbuch.

Statistisches Bundesamt (2007b). Geburten und Kinderlosigkeit in Deutschland. Ergebnis‐

se der Sondererhebung 2006. Wiesbaden.

Stöbel‐Richter, Y., & Brähler, E. (2006). Ausgewählte Fakten zum politischen Lamen‐

to über Deutschlands sinkende Kinderzahlen. Reproduktionsmedizin und Endok‐

rinologie, 3(5), 307‐314.

Sturaro, C., Denissen, J. J. A., van Aken, M. A. G., & Asendorpf, J. B. (2008). Person‐

environment transactions during emerging adulthood: The interplay between

personality characteristics and social relationships. European Psychologist,

13(1), 1‐11.

Swann, W. B., Jr., Chang‐Schneider, C., & Larsen McClarty, K. (2007). Do peopleʹs

self‐views matter? Self‐concept and self‐esteem in everyday life. American Psy‐

chologist, 62(2), 84‐94.

Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston,

MA: Allyn & Bacon/Pearson Education.

Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High Self‐Control Predicts

Good Adjustment, Less Pathology, Better Grades, and Interpersonal Success.

Journal of Personality, 72(2), 271‐324.

Tesser, A. (2003). Self‐evaluation. In M. R. Leary & J. P. Tangney (Eds.), Handbook of

self and identity (pp. 275‐290). New York, NY US: Guilford Press.

Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York: John

Wiley and Sons. References 226

Thibaut, J. W., & Riecken, H. W. (1955). Some determinants and consequences of the

perception of social causality. Journal of Personality, 24, 113‐133.

Tölke, A., & Diewald, M. (2003). Insecurities in employment and occupational careers

and their impact on the transition to fatherhood in western Germany. Demo‐

graphic Research, 9(3), 41‐67.

Townsend, J. W. (2003). Reproductive behavior in the context of global population.

American Psychologist, 58(3), 197‐204.

Trivers, R. L. (1971). The evolution of reciprocal altruism. The Quarterly Review of Biol‐

ogy, 46, 35‐57.

Trzesniewski, K. H., Donnellan, M. B., & Robins, R. W. (2003). Stability of Self‐Esteem

Across the Life Span. Journal of Personality and Social Psychology, 84(1), 205‐220.

Uehara, E. S. (1995). Reciprocity reconsidered: Gouldnerʹs ʹmoral norm of reciprocityʹ

and social support. Journal of Social and Personal Relationships, 12(4), 483‐502.

Ulrich, M., & Weatherall, A. (2000). Motherhood and infertility: Viewing motherhood

through the lens of infertility. Feminism & Psychology, 10(3), 323‐336.

Umberson, D. (1992). Relationships between adult children and their parents: Psy‐

chological consequences for both generations. Journal of Marriage & the Family,

54, 664‐674. van Duijn, M. A. J., van Busschbach, J. T., & Snijders, T. A. B. (1999). Multilevel

analysis of personal networks as dependent variables. Social Networks, 21(2),

187‐209. van Duijn, M. A. J., & Vermunt, J. K. (2006). What is special about social network

analysis? Methodology: European Journal of Research Methods for the Behavioral

and Social Sciences, 2(1), 2‐6.

Van Lange, P. A. M., De Bruin, E. M. N., Otten, W., & Joireman, J. A. (1997). Devel‐

opment of prosocial, individualistic, and competitive orientations: Theory and

preliminary evidence. Journal of Personality and Social Psychology, 73(4), 733‐746. References 227

Vermunt, J. K., & Kalmijn, M. (2006). Random effects models for personal networks.

Methodology: European Journal of Research Methods for the Behavioral and Social

Sciences, 2(1), 34‐41.

Vigil, J. M., & Geary, D. C. (2006). and Community Background and Varia‐

tion in Womenʹs Life‐History Development. Journal of Family Psychology, 20(4),

597‐604.

Vohs, K. D., & Ciarocco, N. J. (2004). Interpersonal functioning requires self‐

regulation. In R. F. Baumeister & K. D. Vohs (Eds.), Handbook of self‐regulation:

Research, Theory and Applications (pp. 392‐407). New York: Guilford Press.

Vohs, K. D., & Finkel, E. J. (2006). Self and Relationships. Connecting Intrapersonal and

Interpersonal Processes. New York: Guilford Press.

Voland, E., & Beise, J. (2005). „The husband’s mother is the devil in the house” Data

on the impact of the mother‐in‐law on stillbirth mortality in historical

Krummhörn (1750‐1874) and some thoughts on the evolution of postgenera‐

tive female life. In E. Voland, A. Chasiotis & W. Schiefenhövel (Eds.), Grand‐

motherhood: The evolutionary significance of the second half of female life (pp. 239‐

255). New Brunswick: Rutgers University Press.

Voss, K., Markiewicz, D., & Doyle, A. B. (1999). Friendship, marriage and self‐esteem.

Journal of Social and Personal Relationships, 16(1), 103‐122.

Voß, A., Soeffner, H. G., Krämer, U., & Weber, W. (1994). Ungewollte Kinderlosigkeit als

Krankheit. Die sozialen Funktionen und die sozialen Folgen reproduktionsmedizini‐

scher Behandlungen. Forschungsberichte des Landes Nordrhein‐Westfalen, Nr. 3248.

Opladen.

Wagner, J., Wrzus, C., Neyer, F. J., & Lang, F. R. (2008). Regulatory Processes in int‐

ergenerational relationships: Adult children and their parents. Talk given at

the 20th meeting of the International Society for the Study of Behavioural De‐

velopment, Wuerzburg, FRG, July 2008. References 228

Welsh, W. M., & Stewart, A. J. (1995). Relationships between woman and their par‐

ents: Implications for midlife well‐being. Psychology and Aging, 10, 181‐190.

Wenger, G. C., Dykstra, P. A., Melkas, T., & Knipscheer, K. C. P. M. (2007). Social

embeddedness and late‐life parenthood: Community activity, close ties, and

support networks. Journal of Family Issues, 28(11), 1419‐1456.

Wieselquist, J., Rusbult, C. E., Foster, C. A., & Agnew, C. R. (1999). Commitment,

pro‐relationship behavior, and trust in close relationships. Journal of Personality

and Social Psychology, 77(5), 942‐966.

Wilson, E. K., & Koo, H. P. (2006). The Relationship Context: Its Effects on Low‐

Income Womenʹs Desire for a Baby. Journal of Marriage and Family, 68(5), 1326‐

1340.

Wischmann, T. (2006). Unerfüllter Kinderwunsch – Stereotype und Fakten Reproduk‐

tionsmedizin und Endokrinologie, 3(4), 220‐225.

Zeifman, D., & Hazan, C. (1997). Attachment: The bond of pair‐bonds. In J. A. Simp‐

son & D. T. Kenrick (Eds.), Evolutionary social psychology (pp. 237‐263). Mah‐

wah, New Jersey: Lawrence Erlbaum Associates.

Zeigler‐Hill, V. (2006). Discrepancies Between Implicit and Explicit Self‐Esteem: Im‐

plications for and Self‐Esteem Instability. Journal of Personality,

74(1), 119‐143.

Zeigler‐Hill, V., & Showers, C. J. (2007). Self‐Structure and Self‐Esteem Stability: The

Hidden Vulnerability of Compartmentalization. Personality and Social Psycho‐

logy Bulletin, 33(2), 143‐159.

Appendix: Table of Contents

9 APPENDIX

Appendix A Family Study: Additional Sample Descriptions ...... 9‐2

Appendix B Online Study: Additional Result Tables ...... 9‐4

Appendix C Family Study: Additional Result Tables ...... 9‐6

Appendix D Online Study: Methods ...... 9‐7

Appendix E Family Study: Study‐Specific Methods ...... 9‐8

Appendix 9-2

Appendix A Family Study: Additional Sample Descriptions

Table A.1. Sample Comparison between selected sub‐sample (N = 92) and remain‐ ing participants (N = 250) in target sample of family study

Target Sample of Family Study Differences Member of Child‐Parent Sub‐Sample Yes No (n = 92) (n = 250) M (SD) M (SD) Age 36.7 (4.1) 38.1 (5.3) t (340) = 2.198*

Income (weighted) 1432 (729) 1218 (640) t (321) = ‐2.578* Education (in years) 16.5 (3.8) 15.2 (4.0) t (340) = ‐2.730**

Partnership duration (in years) 10.7 (5.5) 10.4 (5.5) n.s. N % N % Religion evangelic 22 24 50 20 n.s. catholic 7 8 22 9 other 4 4 13 5 none 59 64 165 66 Highest School Qualification1 Hauptschule 3(‐) 3 27(‐) 11 X2 (3) = 9.801* Realschule 18 20 68 27 Abitur 23 34 63 25 University 48(+) 52 92(‐) 37 Employment full‐time 53 58 155 62 n. s. part‐time 19 21 45 18 none 20 22 50 20 Current Family Status partnership 34 37 89 36 n.s. marriage 58 63 161 64 Family Constancy High 62(+) 67 128(‐) 51 X2 (1) = 7.140** Low 30(‐) 33 122(+) 49 Note. * p < .05, ** p < .01 Appendix 9-3

Table A.2. Sample Comparison between selected sub‐sample of old parents (N = 92) and remaining participants (N = 107) of the family’s study old parents

Old Parents Sample of Family Study Differences Member of Child‐Parent Sub‐Sample Yes No (n = 92) (n = 107) M (SD) M (SD) Age (SD) 64.2 (6.9) 64.6 (7.4) n.s.

Partnership duration (SD) 32.3 (16.1) 35.7 (13.4) n.s. N % N % Religion evangelic 32 35% 36 34% n.s. catholic 10 11% 9 9% other 1 1% 3 3% none 48 53% 58 55% Highest School Qualification1 None 2 2% 1 1% n.s. Hauptschule 22 25% 39 38% Realschule 31 35% 29 28% Abitur 11 12% 2 7% University 23 26% 27 26% Employment full‐time 16 18% 14 14% n. s. part‐time 5 6% 8 8% none 70 77% 82 79% Current Family Status partnership/new marriage 14 15% 21 20% n.s. old marriage 60 65% 72 67% divorced 10 11% 2 2% widowed 8 8% 12 11%

Appendix 9-4

Appendix B Online Study: Additional Result Tables

Table B.1 Means, Standard Deviations and Partial Eta Squares in Perceived Closeness, Under‐Benefit and Over‐Benefit by Relationship Type and by Age groups

Relationship Type Effects In‐law/ Ac‐ Age group Close Distant step Col‐ quain‐ Old Effects Parent Sibling kin kin Partner family league tance friend η2 Closeness Early M .67 .64 .52 .47 .86 .47 .50 .49 .66 .25 Adulthood (SD) (.22) (.21) (.23) (.24) (.18) (.25) (.20) (.20) (.16)

Young M .53 .59 .66 .46 .81 .39 .44 .43 .62 .24 Adulthood (SD) (.27) (.24) (.24) (.28) (.19) (.21) (.20) (.22) (.16)

Middle M .60 .56 .70 .47 .75 .50 .49 .47 .62 .15 Adulthood (SD) (.25) (.24) (.21) (.23) (.25) (.23) (.21) (.24) (.19)

Late M .68 .63 .73 .50 .85 .57 .55 .46 .59 .25 Adulthood (SD) (.27) (.22) (.19) (.18) (.20) (.19) (.20) (.16) (.17) M / .79 .78 .67 1.00 .66 .67 .54 .69 .24 Old Age (SD) (.19) (.18) (.22) (.00) (.21) (.18) (.19) (.15)

η2 .06 .05 .14 .06 .06 .11 .05 .02 .03

Under‐Benefit Early M .23 .37 .26 .40 .27 .26 .28 .32 .36 .02 Adulthood (SD) (.36) (.46) (.36) (.45) (.45) (.38) (.38) (.38) (.36)

Young M .40 .45 .48 .08 .32 .46 .35 .34 .32 .03 Adulthood (SD) (.46) (.47) (.47) (.17) (.47) (.44) (.40) (.39) (.37)

Middle M .51 .48 .61 .28 .44 .51 .29 .35 .30 .07 Adulthood (SD) (.49) (.47) (.42) (.44) (.49) (.43) (.39) (.39) (.36)

Late M .62 .56 .55 .41 .44 .49 .47 .41 .39 .03 Adulthood (SD) (.51) (.45) (.42) (.47) (.50) (.43) (.44) (.36) (.39) M / .56 .47 .66 .00 .64 .67 .49 .40 .10 Old Age (SD) (.48) (.46) (.42) (.00) (.41) (.37) (.38) (.44)

η2 .08 .03 .09 .08 .03 .06 .05 .02 .01

Over‐Benefit Early M .43 .15 .29 .08 .23 .21 .24 .25 .18 .07 Adulthood (SD) (.44) (.33) (.39) (.26) (.42) (.35) (.37) (.35) (.29)

Young M .33 .20 .24 .36 .28 .26 .16 .19 .22 .02 Adulthood (SD) (.42) (.36) (.36) (.45) (.45) (.37) (.30) (.30) (.34)

Middle M .26 .17 .07 .17 .16 .10 .14 .21 .16 .03 Adulthood (SD) (.43) (.35) (.20) (.35) (.36) (.26) (.30) (.32) (.29)

Late Adult‐ M .08 .13 .10 .17 .25 .17 .13 .27 .21 .03 hood (SD) (.28) (.32) (.26) (.35) (.44) (.33) (.29) (.32) (.34) M / .00 .26 .05 .25 .07 .00 .25 .12 .11 Old Age (SD) (.00) (.41) (.16) (.50) (.23) (.00) (.37) (.31)

η2 .04 .02 .08 .06 .01 .03 .03 .01 .01

Note. Bold faced η2 produced substantial F ratios, with p < .05. Appendix 9-5

Table B.2 Means, Standard Deviations and Partial Eta Squares in Perceived Communal‐esteem, Higher‐esteem and Lower‐esteem by Relationship Type and by Age groups

Relationship Type Effects In‐law/ Ac‐ Age group Close Distant step Col‐ quain‐ Old Effects Parent Sibling kin kin Partner family league tance friend η2 Communal‐esteem Early M .44 .19 ‐.18 ‐.46 .85 ‐.37 ‐.27 ‐.19 .29 .18 Adulthood (SD) (.88) (.87) (.99) (.88) (.75) (1.10) (.77) (.76) (.68)

Young M .04 .04 .23 ‐.78 .76 ‐.67 ‐.08 ‐.45 .16 .17 Adulthood (SD) (1.05) (.97) (.99) (.94) (.98) (.85) (.66) (.83) (.68)

Middle M .19 .02 .27 ‐.62 .64 ‐.31 ‐.15 ‐.35 .14 .10 Adulthood (SD) (1.19) (.96) (1.05) (.75) (.95) (.97) (.94) (.88) (.81)

Late M .00 .13 .49 ‐.42 .97 ‐.02 ‐.06 ‐.34 ‐.01 .15 Adulthood (SD) (1.19) (.99) (.88) (.81) (.77) (.92) (.91) (.76) (.87) M / .53 .74 .01 1.33 .58 .14 ‐.33 .56 .22 Old Age (SD) (.95) (.65) (1.11) (0) (.91) (.48) (.80) (.76)

η2 .03 .01 .08 .04 .02 .09 .01 .01 .02

Higher‐esteem Early M .25 .17 .24 .22 .15 .21 .13 .17 .16 .02 Adulthood (SD) (.39) (.36) (.37) (.38) .36) (.36) (.27) (.32) (.29)

Young M .28 .22 .18 .32 .22 .27 .14 .18 .14 .02 Adulthood (SD) (.42) (.39) (.33) (.46) .41) (.42) (.26) (.33) (.28)

Middle M .11 .10 .09 .24 .16 .09 .17 .14 .12 .02 Adulthood (SD) (.29) (.29) (.25) (.41) (.36) (.22) (.34) (.26) (.24)

Late M .09 .09 .09 .06 .23 .15 .31 .15 .13 .05 Adulthood (SD) (.30) (.27) (.27) (.24) .43) (.31) (.40) (.27) (.26) M / .00 .07 .29 .00 .08 .13 .11 .11 .07 Old Age (SD) (.00) (.25) (.49) / (.20) (.25) (.29) (.22)

η2 .03 .03 .05 .06 .01 .04 .05 .00 .00

Lower‐esteem Early M .14 .26 .13 .22 .25 .19 .22 .21 .20 .02 Adulthood (SD) (.31) (.43) (.27) (.38) (.42) (.36) (.33) (.33) (.30)

Young M .23 .14 .16 .05 .19 .31 .15 .18 .26 .03 Adulthood (SD) (.40) (.33) (.30) (.15) (.40) (.42) (.27) (.32) (.37)

Middle M .30 .34 .28 .10 .32 .31 .17 .21 .15 .04 Adulthood (SD) (.44) (.46) (.40) (.29) (.47) (.41) (.35) (.35) (.31)

Late Adult‐ M .36 .30 .26 .21 .18 .26 .18 .25 .19 .01 hood (SD) (.50) (.42) (.38) (.40) (.39) (.35) (.34) (.38) (.33) M / .13 .29 .21 .00 .30 .13 .37 .26 .04 Old Age (SD) (.35) (.44) (.39) / (.40) (.25) (.44) (.44)

η2 .04 .03 .03 .03 .02 .02 .01 .01 .02

Note. Bold faced η2 produced substantial F ratios, with p < .05. Appendix 9-6

Appendix C Family Study: Additional Result Tables

Table C.1 Differences between family types in social relationship characteristics

Motivated Involuntary Blended Traditional

Childless Childless Family Family F Ratio M SE M SE M SE M SE (3, 4542)

Geographical 2 3.88ab 1.46 3.95a 1.43 3.52b 1.60 3.68a 1.65 5.12; η = .003 Proximity

Frequency of 2 2.37ac 1.38 2.25ab 1.36 2.68c 1.52 2.46b 1.57 8.69; η = .006 Contact

Relationship 2 4.82ab 1.60 4.96a 1.52 4.73b 1.53 4.95a 1.44 6.84; η = .004 Duration

Note. Means in given row with different notations differ significantly from other means according to Scheffe post hoc tests with p < .05. All effects remained robust if controlled for participant’s number of biological children.

Table C.2 Relationship Perception of Adult Child to Old Parent

Closeness Reciprocity M SD t‐ratio M SD t‐ratio Adult Children (n=92) .72 .22 / .70 .24 /

Gender S‐F (n=36) .72 .15 .21 .68 .22 ‐.66 D‐M (n=56) .71 .25 .71 .24

Family Constancy High (n=60) .76 .18 2.98** .69 .23 ‐.09 Low (n=32) .63 .26 .70 .25

Note. S‐F = Son‐Father dyads, D‐M = Daughter‐Mother dyads. There are no significant interaction ef‐ fects between family constancy and gender composition. ** p < .01.

Table C.3 Relationship Perception of Old Parent to Adult Child

Closeness Reciprocity M SD t‐ratio M SD t‐ratio Old Parent (n=92) .76 .19 / .80 .20 /

Gender S‐F (n=36) .73 .19 ‐1.14 .82 .22 .77 D‐M (n=56) .78 .20 .79 .19

Family Constancy High (n=60) .77 .19 .73 .80 .21 ‐.45 Low (n=32) .74 .19 .82 .18

Note. S‐F = Son‐Father dyads, D‐M = Daughter‐Mother dyads. There are no interaction effects between family constancy and gender.

Appendix 9-7

Appendix D Online Study: Methods

Ego‐centered Network Original Translation

Woher kennen sie diese Person? Where do you know this person from? Relationship types Großmutter/Großvater Grandmother/Grandfather Mutter/Vater Mother/Father Tante/Onkel Aunt/Uncle Cousine/Cousin Cousin Schwester/Bruder Sister/Brother Tochter/Sohn Daughter/Son Nichte/Neffe Nice/Nephew Enkelin/Enkel Granddaughter/Grandson Schwägerin/Schwager Sister‐in‐law/Brother‐in‐law Halbschwester/Halbbruder Half‐Sister/Half‐Brother Sonstige Verwandte Other Relatives Andere Familienmitglieder, mit denen Sie nicht Other family members, that you are not genetic verwandt sind related with Partner Partner … von der Arbeit …from work … über eine dritte Person …through another person … aus der Nachbarschaft …from neighbourhood … aus dem Verein (Sport‐, ehrenamtlicher Ver‐ …from institutions (sports club, non‐profit soci‐ ein) ety) … aus dem Dienstleistungssektor (Arzt, Friseur, …from service sector (doctor, hairdresser, insur‐ Versicherungsvertreter) ance agent) … , weil er/ sie mein langjährige/r Freund/in ist …, because he/she is a long‐term friend

Appendix 9-8

Appendix E Family Study: Study‐Specific Methods

Original Translation

Woher kennen sie diese Person? Where do you know this person from? Relationship Types of Großmutter/Großvater Grandmother/Grandfather Ego‐centered Net‐ Mutter/Vater Mother/Father work Assessment Tante/Onkel Aunt/Uncle Cousine/Cousin Cousin Schwester/Bruder Sister/Brother Tochter/Sohn Daughter/Son Nichte/Neffe Nice/Nephew Enkelin/Enkel Granddaughter/Grandson Schwägerin/Schwager Sister‐in‐law/Brother‐in‐law Halbschwester/Halbbruder Half‐Sister/Half‐Brother Sonstige Verwandte Other Relatives Andere Familienmitglieder, mit denen Sie nicht Other family members, that you are not genetic verwandt sind related with Partner Partner … durch die/ von der Arbeit/ aus dem Studium …from work/ from college or university …, weil er/ sie mein langjährige/r Freund/in ist …, because he/she is a long‐term friend … von der Arbeit (Vorgesetzte/r) … from work (superior) … von der Arbeit (Untergebene/r) … from work (subordinate) … über unsere Kinder (andere Eltern, Lehrer, …through our children (other parents, teacher, etc.) etc.) … aus der Nachbarschaft …from neighbourhood … aus dem Verein (Sport‐, politischer, ehrenamt‐ …from institutions (sports club, non‐profit soci‐ licher Verein) ety) … aus dem Dienstleistungssektor (Arzt, Friseur, …from service sector (doctor, hairdresser, insur‐ Versicherungsvertreter) ance agent)

Life Goals Beruflicher Erfolg Professional success (Lang et al., 2004) Liebe & Zuneigung erfahren Experiencing love Körperlich gesund & fit sein Physical health and fitness Spannende & neue Dinge erleben Experiencing new things Unabhängigkeit & Selbstbestimmung Independence and self‐determination Anerkennung & Respekt erfahren Experience of appreciation and respect Meinem Leben einen Sinn geben Seeing a sense in life Anderen vertrauen Trusting people Kinder haben & erziehen Raise and educate children Ruhe & Stabilität Stability and inner peace Harmonisches Familienleben Harmonious family life Selbstverwirklichung Self‐realization Einen großen Freundeskreis pflegen Keeping a large circle of friends

Appendix 9-9

Relationship Index