Unraveling the complexities of parenting adolescents across time(scales) and families Savannah Boele ONE OF A KIND?
ONE OF A KIND? Unraveling the complexities of parenting adolescents across time(scales) and families Savannah Boele
Copyright © 2023 by Savannah Boele All rights reserved. No part of this thesis may be reproduced, stored, or transmitted in any way or by any means without the prior permission of the copyright holder. This research was supported by a grant from the Dutch Council for Scientific Research (NWO-VIDI; 452-17-011) awarded to Loes Keijsers. Printed by: Ridderprint, ridderprint.nl Layout: Penny Dalladol | Ridderprint Cover Design: Joey Roberts | Ridderprint ISBN: 978-94-6483-608-0
ONE OF A KIND? Unraveling the complexities of parenting adolescents across time(scales) and families Echt uniek? De complexiteiten van opvoeden in de adolescentie ontrafelen over verschillende tijdschalen en tussen verschillende families Proefschrift ter verkrijging van de graad van doctor aan de Erasmus Universiteit Rotterdam op gezag van de rector magnificus Prof.dr. A.L. Bredenoord en volgens besluit van het College voor Promoties. De openbare verdediging zal plaatsvinden op donderdag 15 februari 2024 om 13:00 uur door Savannah Boele geboren te Bergeijk.
PROMOTIECOMMISSIE: Promotoren: Prof.dr. L.G.M.T. Keijsers Prof.dr. J.J.A. Denissen Overige leden: Prof.dr. P.W. Jansen Prof.dr. G.J. Overbeek Prof.dr. I. Myin-Germeys Copromotor: Dr. A.D. de Haan
TABLE OF CONTENTS Chapter 1 General introduction 7 Chapter 2 Over-time fluctuations in parenting and adolescent adaptation within families: A systematic review 33 Chapter 3 Testing transactional processes between parental support and adolescent depressive symptoms: From a daily to a biennial timescale 73 Chapter 4 For better, for worse, or both? Testing environmental sensitivity models with parenting at the level of individual families 109 Chapter 5 The direction of effects between parenting and adolescent affective well-being in everyday life is family specific 145 Chapter 6 Like no other? A family-specific network approach to parenting adolescents 169 Chapter 7 Summary and general discussion 197 Appendices References Samenvatting (Summary in Dutch) Curriculum vitae Dankwoord (Acknowledgements) 222 245 251 258
CHAPTER 1 General introduction
Chapter 1 8 Each person is like every other person, each person is like some others, and each person is like no other person. Adapted from Kluckhohn & Murray (1948) Each person is a unique human being, one of a kind, shaped by many idiosyncratic experiences throughout life. One of the primary contexts in which most humans gain lifechanging experiences is the parent-child relationship (Bronfenbrenner, 2005; Sameroff, 2010). While it is universally expected that parents (or other caregivers) have an impact on their children’s everyday functioning and long-term development (Bronfenbrenner, 2005; Granic et al., 2008), and vice versa (Kuczynski & Parkin, 2007), the nature of parentchild influences is assumed to vary across families as a function of the characteristics of the child, the parent, the parent-child relationship, and other contexts (Belsky & Pluess, 2009; Darling & Steinberg, 1993; Soenens et al., 2015). Influences between parents and children may even be idiosyncratic, as every family has its own unique set of characteristics (e.g., the personality of the child and the parents, quality of interparental relationship, culture, religion, etc.) (Bronfenbrenner, 2005) and a unique history of interactions (Lollis & Kuczynski, 1998). Thus, how parents and their children influence each other might be unique to each family, fueling a child’s individual development. Adolescence is a period in which parents are considered to still significantly contribute to their children’s developmental trajectory, for better or worse. Adolescence is the period roughly between 10 and 25 years (Smetana & Rote, 2019; Steinberg, 2014), and is marked by many physiological (e.g., pubertal changes), physical (e.g., growth spurt), neurological (e.g., gradually maturing prefrontal cortex), social (e.g., stronger peer orientation, more independence from parents), and psychological changes and developments (e.g., heightened emotionality, identity formation) (Smetana et al., 2006; Soenens et al., 2019; Steinberg, 2005). Alarmingly, adolescence also seems to be a sensitive period for developing internalizing (e.g., depressive and anxiety symptoms; Hankin et al., 1998; Shorey et al., 2022) and externalizing problems (e.g., substance abuse; (Kessler et al., 2005; Nivard et al., 2017), which can persist into adulthood (Copeland et al., 2009; Kim-Cohen et al., 2003) and have long-lasting effects on other psychosocial outcomes (Clayborne et al., 2019; Copeland et al., 2009). As one of the primary socialization figures, parents play a significant role in navigating their adolescent child through the challenges and opportunities of growing up (Soenens et al., 2019). However, as parenting
1 General introduction 9 adolescents is understood to be a heterogeneous phenomenon (e.g., Bronfenbrenner, 2005; Grusec, 2008), there are likely no universal parenting instructions that will work for every family. Therefore, to help every parent effectively promote the development of their unique adolescent, it is vital to understand how and why parents and adolescents influence each other over time in heterogeneous ways across families. Despite the strong theoretical focus and common knowledge that families function in heterogeneous ways, until recently, empirical studies have mostly been dedicated to identifying general parenting principles or patterns (for a review see Chapter 2; Boele et al., 2020). In doing so, parenting science has established a solid understanding of how families differ in their stable levels of parenting and adolescent functioning. To illustrate, studies have frequently shown that adolescents who experience relatively high levels of parental warmth exhibit fewer internalizing and externalizing problems than adolescents who experience less parental warmth (Pinquart, 2017a, 2017b). However, such grouplevel patterns provide little to no information on how the dynamic processes between parenting and an adolescent’s functioning unfold over time within the same family (Hamaker, 2012; Molenaar, 2004), let alone how such dynamic parenting processes differ from one family to the next. Therefore, parenting science is still in its infancy in understanding the complexities of parenting adolescents: (I) how parents and adolescents influence each other within families, (II) on micro- to macro-timescales, and (III) how and why these influences might vary across families. To address these three pressing knowledge gaps, the overarching goal of this dissertation was to provide insights into the complexities of parenting adolescents across timescales and families. To achieve this goal, this dissertation presents five studies (one systematic review and four empirical studies) that examined how key dimensions of parenting are reciprocally linked to adolescent functioning within families on various timescales and how and why these linkages differ across families. The empirically studied key parenting dimensions are warmth, autonomy support, behavioral control, and psychological control, which will be explained later in this introduction. The empirically studied domains of adolescent functioning include aspects of psychological (i.e., depressive and anxiety symptoms and self-esteem) and affective functioning (i.e., positive and negative affect). In this chapter, I elaborate on the contemporary theoretical understanding of parenting adolescents and how generally applied methods have had a mismatch with these theoretical ideas (Richters, 1997, 2021). I end with the aims, which address the methodological gaps and outline of this dissertation.
Chapter 1 10 1. THEORETICAL UNDERSTANDING OF PARENTING ADOLESCENTS 1.1 Influences between parents and adolescents are proximal processes Parents play a central role in the lives of adolescents. Though children spend progressively less time with their parents during adolescence (R. Larson & Richards, 1991), the parentchild relationship remains one of the primary and unique contexts in which adolescents amass positive and negative experiences (Laursen & Bukowski, 1997). Different theories have described how such experiences with parents contribute to adolescents’ development. That is, bio(psychosocial) ecological models view micro-timescale (e.g., real-time) influences between parents and adolescents as one of the proximal processes (i.e., “enduring forms of interaction in the immediate environment”; Bronfenbrenner, 2005, p. 6) that shape the adolescent’s longer-term development (Bronfenbrenner, 2005; Sameroff, 2010). Importantly, both the objective quality of these proximal influences and the subjective quality perceived by adolescents are viewed as important drivers of their development (Bronfenbrenner, 2005). Similarly, scholars who have adopted a dynamic systems perspective to parenting also propose that reoccurring microtimescale influences shape long-term developmental changes (Granic et al., 2008; Smith & Thelen, 2003). Hence, according to these macro-developmental theories, the answer to understanding how individual adolescents grow and mature lies hidden in the nature of the proximal influences in the parent-adolescent relationship. 1.2 Parenting dimensions that contribute to adolescent functioning Macro-developmental theories thus assume that the quality of influences within the parent-child relationship drives the course of future development (Bronfenbrenner, 2005; Sameroff, 2010; Smith & Thelen, 2003). But what constitutes good-quality parenting in adolescence? Various more general and specific parenting theories have proposed which key parenting dimensions, composed of clusters of practices, are relevant in terms of hindering or supporting adolescent functioning (e.g., Kerr & Stattin, 2000; Rohner et al., 2005; Soenens et al., 2017). Overall, this broad range of work can be synthesized into four key parenting dimensions: warmth, autonomy support, behavioral control, and psychological control (Smetana, 2017; Soenens et al., 2019). These key dimensions have been theoretically and empirically linked to a wide range of important domains of adolescent functioning, such as psychosocial (mal)functioning, including internalizing problems (e.g., depressive and anxiety symptoms and low self-esteem) and externalizing problems (e.g., aggression, delinquency, and substance use). The definitions and
1 General introduction 11 theorized consequences of these four parenting dimensions in terms of adolescents’ functioning and current empirical patterns are described below. Warmth (also known as emotional support) is one of the most frequently studied dimensions of parenting. Parental warmth involves practices such as providing affection, intimacy, security, and responding to children’s emotional needs (Furman & Buhrmester, 1985; Soenens et al., 2017). According to the interpersonal acceptance-rejection theory (IPARTheory; Rohner, 2016) and the self-determination theory (SDT; Ryan & Deci, 2000; Soenens et al., 2019), experiencing parental warmth can satisfy the universal need to feel loved and appreciated by close others (i.e., called “need for relatedness” by the SDT), which is believed to promote psychosocial functioning, such as a better self-esteem. Meta-analytic work has indicated that parental warmth indeed correlates with many domains of adolescent functioning at the group level. Specifically, a large body of work has established that adolescents from families with higher levels of parental warmth show, on average, fewer internalizing (e.g., McLeod, Wood et al., 2007; Pinquart, 2017b; Yap et al., 2014) and externalizing problems (Hoeve et al., 2009; Pinquart, 2017a; Yap et al., 2017) than adolescents from families with lower levels of parental warmth. Whether parental warmth, and all other parenting dimensions described below, exhibits similar associations with adolescent functioning within families is yet to be established (for a systematic review, see Chapter 2; Boele et al., 2020). The second parenting dimension is autonomy support, which includes respecting adolescents’ individuality by recognizing and accepting their perspectives, giving choices, explaining new rules, and encouraging initiative-taking (Soenens et al., 2019). According to the self-determination theory, experiencing autonomy-supportive parenting can satisfy the universal human need for autonomy, which is thought to foster adolescents’ psychosocial functioning (Ryan & Deci, 2000; Soenens et al., 2019). Although parental autonomy support has not gained as much attention as parental warmth, several metaanalyses have similarly indicated that higher levels of autonomy support relate to lower levels of internalizing (McLeod, Weisz, et al., 2007; Pinquart, 2017b) and externalizing problems in adolescence at the group level (Pinquart, 2017a). In other words, adolescents from families who experience more parental autonomy support show, on average, better psychosocial functioning than adolescents whose parents are less autonomy supportive. In addition to parental warmth and autonomy support, providing structure and setting boundaries seems key. This third parenting dimension is behavioral control, involving practices to supervise and regulate adolescents’ behavior, such as rule setting and
Chapter 1 12 actively soliciting information of one’s children’s whereabouts and activities (Barber, 1996; Stattin & Kerr, 2000). At appropriate levels, behavioral control is thought to protect against adolescent malfunctioning, especially by preventing or reducing externalizing problems, such as delinquency (Stattin & Kerr, 2000) and substance use (Koning et al., 2020). However, in more recent work, behavioral control has also been suggested to undermine adolescents’ autonomy and, therefore, hinder adolescents’ psychosocial functioning. That is, when parents exert too much behavioral control or when the domain of control touches on personal domains, such as prohibition of friendships (Kakihara & Tilton-weaver, 2009; Keijsers et al., 2012), parental behavioral control may be hindering rather than helping. Therefore, based on the self-determination theory (Ryan & Deci, 2000), a conceptual distinction has been made between promotive practices that provide structure (e.g., rule setting, monitoring, and feedback) and hindering, harsh practices (e.g., hostility, criticism, and punishment) (Soenens et al., 2019). Meta-analyses that differentiated between “behavioral control” (i.e., rule setting and monitoring) and “harsh control” indeed show that adolescent internalizing and externalizing problems are negatively associated with behavioral control but positively associated with harsh control at the group level (Pinquart, 2017a, 2017b). Thus, adolescents whose parents provide more structure and are less harsh toward them have shown better psychosocial functioning on average than adolescents whose parents provide less structure and are harsher. In addition to behavioral control, a second type of parental control has been identified and labeled as psychological control. The fourth parenting dimension psychological control involves the regulation of adolescents’ thoughts and emotions. This can be done by manipulative parenting practices, including intrusiveness, guilt induction, and love withdrawal (Barber, 1996; Barber et al., 2012). According to the self-determination theory (Ryan & Deci, 2000), psychologically controlling parenting is thought to hinder adolescent psychosocial functioning because it undermines adolescents’ autonomy and competence (Barber, 1996; Soenens et al., 2019). Indeed, meta-analyses indicate that parental psychological control and related practices, such as withdrawal, rejection, and aversiveness, show one of the strongest links with internalizing (McLeod, Weisz, et al., 2007; Pinquart, 2017b; Yap et al., 2014) and externalizing problems at the group level (Hoeve et al., 2009; Pinquart, 2017a). Hence, adolescents who experience relatively high levels of parental psychological control show worse psychosocial functioning on average than adolescents who experience lower levels of parental psychological control. Over the past few decades, theoretical parenting work has attempted to understand which parenting dimensions and how they are linked to the development of adolescent (mal)
1 General introduction 13 functioning. These dynamic processes between parenting and adolescent functioning are characterized by complexity, which requires a further specification of the nature of these dynamic parenting processes. That is, the contemporary understanding of the dynamic processes between parenting and adolescent functioning includes several other metatheoretical elements (i.e., processes unfolding within families, the presence of reciprocity, processes unfolding on various timescales, and heterogeneity of processes across families; see Figure 1) which are elaborated below. Figure 1 Summary of the theoretical understanding of parenting adolescents Note. IPARTheory = Interpersonal acceptance and rejection theory. SDT = self-determination theory. TIES model = temporal interpersonal emotions systems model. 1.3 Within a family: Influences between a parent and their own child To understand how parenting relates to adolescent functioning, the linkages between parenting and adolescent functioning have been theoretically described at two distinct levels: between-family differences and within-family processes (Smetana, 2017). Many studies have been focusing on differences between families. That is, the groundbreaking work of Baumrind (Baumrind, 1971, 1991) distinguished three stable parenting styles that differed between parents and families. These typologies were based on the two
Chapter 1 14 overarching parenting dimensions of responsiveness and demandingness: authoritative (i.e., high responsiveness and demandingness), authoritarian (i.e., low responsiveness, high demandingness), and permissive (i.e., high responsiveness and low demandingness). Later, Maccoby & Martin (1983) added a fourth style, the neglecting style, which describes parents who score low on both responsiveness and demandingness. These parenting styles are differentially linked to adolescent psychosocial functioning; research often demonstrated that adolescents whose parents exert an authoritative parenting style demonstrate better psychosocial functioning than adolescents whose parents exert other parenting styles (Power, 2013). However, this typology of parenting styles at the aggregated level of stable between-family differences has paid little attention to the underlying dynamic processes that explain how parenting contributes to changes in adolescent functioning. To explain why adolescents raised by parents with different parenting styles show different levels of functioning, the integrative parenting model of Darling and Steinberg (1993) distinguished between stable parenting styles and time-variable parenting practices. Parenting style is conceptualized as the stable emotional climate of the parent-child relationship, which can alter the effectiveness of parenting practices. Parenting practices (e.g., responsiveness, rule setting, and intrusiveness) are thought to vary across time and situations and are conceptualized as the direct mechanisms through which parents influence the functioning of their adolescents. In other words, fluctuations in parenting practices over time are thought to explain the changes in adolescent functioning. Hence, this integrative model of parenting made clear (Darling & Steinberg, 1993), although not explicitly described as such, that theorized socialization influences from parenting to adolescent functioning occur within a family, which may differ between families due to variations in parenting styles. Accordingly, different theoretical questions can be formulated regarding the phenomenon of parenting adolescents (see Table 1). One set of questions pertains to stable differences between families, such as differences in parenting styles, and how they relate to interindividual differences in the levels of adolescent functioning. Another set of questions pertains to the motor behind such differences: The dynamic processes by which parenting and adolescent functioning influence one another over time. Although the latter set of questions represents the core theoretical ideas on parenting adolescents, they have rarely been empirically tested because of methodological gaps in parenting science and in developmental psychology more broadly (which will be explained later in this introduction) (Richters, 1997, 2021). To increase the empirical understanding of parenting adolescents,
1 General introduction 15 this dissertation therefore focused on the dynamic processes between (perceived) parenting and adolescent functioning, introducing novel methodological designs that allowed me to do so (e.g., intensive longitudinal methods). 1.4 Reciprocity: Adolescents also influence their parents Theories of human development, such as bio(psychosocial) ecological models (Bronfenbrenner, 2005; Sameroff, 2010) and the dynamic systems perspective (Granic et al., 2008; Smith & Thelen, 2003) emphasize the omnipresence of personenvironment transactions. Person-environment transactions are dynamic and ongoing, with individuals shaped by and shaping their immediate environment. This highlights the importance of considering both the person and their social environment when aiming to understand human development. When applied to parenting adolescents, adolescents might not merely be a product of their parents’ parenting practices, as described in earlier paragraphs (see 1.2 and 1.3), but most likely also exert influences on their parents. Table 1 Matching theoretical questions with methodological design Theoretical questions Concepts Parenting styles Stable levels, trait-like, differences between families Dynamic parenting processes Dynamics, change, over-time fluctuations, time-varying situations Research questions (examples) What characterizes families in which adolescents show relatively worse or better functioning? How do fluctuations in parenting and adolescent functioning predict each other on average within families? And which time-invariant or time-varying characteristics moderate these withinfamily effects? What are the unique dynamics between parenting and adolescent functioning in a specific family? And how do these family-specific parenting processes vary across families? Level of analysis Between-family level (nomothetic) Within-family level Individual-family level (idiographic) Methodological design Design Cross-sectional or longitudinal panel study (years) Longitudinal study (days to years) Intensive longitudinal study (seconds to weeks) Sample size Large Medium to large ≥ 1 (small to large) Time points One or few Few to many Many Analyses (examples) Correlation, regression, CLPM Multi-level regression, RICLPM, DSEM DSEM, GIMME, state space grid Note. (RI-)CLPM = (random intercept) cross-lagged panel model. DSEM = dynamic structural equation modelling. GIMME = group iterative multiple modelling estimation.
Chapter 1 16 Indeed, the concept of reciprocity between parents and adolescents has been integrated into many (but not all) parenting theories, from early (Bell, 1968; Patterson, 1982) to contemporary theories (Granic et al., 2008; Lougheed & Keskin, 2021; Sameroff, 2010). Hence, scholars generally agree that parenting, especially parenting adolescents, can best be operationalized as a two-way street (Kuczynski & Parkin, 2007; Soenens & Vansteenkiste, 2020). Rather than being passive recipients of their parents’ behavior, adolescents exhibit agency and actively choose their responses. For example, adolescents have the agency to decide whether to comply with their parents’ demands (Darling et al., 2007) or to disclose personal information (Smetana, 2008). Additionally, how adolescents function, such as how they behave or feel, is likely to elicit certain parenting behaviors. For instance, externalizing problems may elicit more controlling and rejecting behavior from parents in an attempt to reduce their adolescent’s problem behavior, which may further reinforce the adolescent’s externalizing problems (Patterson & Fisher, 2002). Similarly, adolescents who experience negative emotions or depressive symptoms may elicit a supportive parental reaction or even face parental rejection (Rudolph, 2009). Thus, according to the widely acknowledged theoretical principle of reciprocity, the dynamic processes within families flow not only from parenting to adolescent functioning (i.e., parent-driven influence) but also from adolescent functioning to parenting (i.e., adolescent-driven influence), as depicted in Figure 2 below. 1.5 Timescales: Processes between parenting and adolescent functioning unfold over time at varying timescales As previously described (see 1.3), parenting adolescents is understood as a highly dynamic phenomenon. Parenting practices vary over time and across different situations. For instance, parents need to adjust their behavior to the changing needs of their developing adolescent (Darling & Steinberg, 1993). Additionally, parents’ own psychological functioning (e.g., experience of stress) fluctuates over time, for example because of workrelated experiences, impacting how they behave towards their adolescent (Belsky, 1984). As such, on one day, a parent may behave warmer or more demanding toward their adolescent than on the previous day due to internal or external influences. These over-time fluctuations in parenting practices (or in overarching dimensions) are thought to impact the functioning of adolescents, and vice versa. All earlier described theories on human development and parenting (see 1.1 to 1.4) indeed propose that the dynamic processes between parenting and adolescent functioning unfold over time. For example, psychologically controlling parenting is thought to hamper adolescents’
1 General introduction 17 psychosocial functioning, meaning that a (perceived) increase in parental psychological control precedes a decline in the adolescent’s functioning (Soenens et al., 2019; Soenens & Vansteenkiste, 2010). Moreover, several theories acknowledge that these dynamic parenting processes unfold across various timescales, from micro- to macro-timescales. Macro developmental theories, including bio(psychosocial) ecological models (Bronfenbrenner, 2005; Sameroff, 2010) and the dynamic systems perspective (Granic et al., 2008; Smith & Thelen, 2003), posit that parents and adolescents influence each other on a microtimescale, such as in real-time face-to-face interaction. Micro-timescale influences may instigate longer-term developmental changes in the adolescent (and in the parent) if these micro-timescale influences “occur on a fairly regular basis over extended period of time” (Bronfenbrenner, 2005, p. 6). However, to my knowledge, the length of this “extended period of time” has not been made explicit. Perhaps micro-timescales influences between parents and adolescents need to persist on at least a meso-timescale (e.g., weeks to months) to have an impact on adolescents’ developmental outcomes (e.g., internalizing and externalizing problems). Figure 2 Theoretical model of the dynamic processes between parenting and adolescent functioning within families Parenting Adolescent functioning
Chapter 1 18 Such multiple timescale theoretical frameworks have also been applied to parenting theories. For example, the well-known coercion theory of Patterson (1982) argues that the repetition of hostile parent-child interactions on a micro-timescale impacts the child’s longer-term development of externalizing problem behavior. Furthermore, the recent temporal interpersonal emotion systems (TIES) model (Lougheed & Keskin, 2021) theorizes that parents and adolescents influence each other’s emotional experiences, expressions, and physical arousal at a micro-timescale (e.g., moments or days). At a longer-term (developmental) timescale (e.g., months or years), they influence each other’s psychosocial adjustment, operationalized as stable patterns of emotional or behavioral responses, such as internalizing and externalizing symptoms, or more stable relationship characteristics. Moreover, by means of circular causality, micro-timescale parentadolescent influences may shape parents’ and adolescents’ psychosocial adjustment, and their psychosocial adjustment may also shape the micro-timescale influences. Hence, the contemporary TIES model clearly postulates that parent-adolescent influences unfold on different timescales, depending on the timescale at which the involved constructs fluctuate or develop. 1.6 Heterogeneity: Different families have different process 1.6.1 The quantity of theorized heterogeneity Many theories, including bio(psychosocial) ecological models (Bronfenbrenner, 2005; Sameroff, 2010), the dynamic systems perspective (Granic et al., 2008; Smith & Thelen, 2003), the integrative parenting model (Darling & Steinberg, 1993), and environmental sensitivity models (Belsky & Pluess, 2009), converge on the idea that the nature of the dynamic processes between parenting and adolescent functioning is likely different between families. Nevertheless, theoretical perspectives differ in terms of the quantity of the expected heterogeneity. On the one hand, dynamic parenting processes may vary from subgroup to subgroup due to group-differential characteristics. For instance, due to personality (Belsky & Pluess, 2009), legitimacy beliefs of parental authority (Darling et al., 2007), parenting style (Darling & Steinberg, 1993), or culture (Soenens et al., 2015). Put differently, families who share the same group-differential characteristics (e.g., parenting style) may be influenced in quite similar ways. On the other hand, dynamic parenting processes may be idiosyncratic to each family (Granic et al., 2008; Grusec, 2008). For instance, bio(psychosocial) ecological models suggest that the nature of daily influences between parents and adolescents varies not only due to the characteristics of the developing adolescent but also due to the changing characteristics of the context
1 General introduction 19 (e.g., parents and parent-child relationship) and timing of events (Bronfenbrenner, 2005; Sameroff, 2010). Additionally, there are varying theoretical perspectives on the extent to which the nature of dynamic parenting processes differs across families. On the one hand, scholars may hold a view of “universality without uniformity”. Such a view means that certain parenting practices are expected to be universally beneficial (e.g., warmth and autonomy support) or detrimental (e.g., overcontrol or psychological control) to adolescents’ functioning, although they are not uniform, given that individual differences can exist in the strength of the parenting effect (Soenens et al., 2015). To illustrate this, universality without uniformity is assumed in environmental sensitivity models, as some individuals are expected to be more strongly influenced by the same environment than others (Pluess, 2015). On the other hand, scholars may hold a more relativistic perspective, assuming that the effectiveness of parenting practices may depend on each adolescent due to many moderators (Grusec, 2008). In other words, the same parenting practice might have beneficial effects for one adolescent, but harmful consequences for another adolescent. Relativistic accounts of parenting are evident in bio(psychosocial) ecological models, postulating that children develop and interact with their parents in unique ways because of the complex dynamic interplay between time-varying contexts and the developing individual (Bronfenbrenner, 2005; Sameroff, 2010). Moreover, according to a dynamic systems perspective, each system (here a family) is viewed as having its own unique characteristics and processes, which also change over time (Granic et al., 2008; Smith & Thelen, 2003). Thus, although heterogeneity is widely acknowledged theoretically, whether the nature of the dynamic processes between parents and adolescents is universal (without uniformity) across families or unique to each family has been a topic of debate. 1.6.2 Characteristics which might contribute to heterogeneity In addition to the quantity of the expected heterogeneity, theories also propose different characteristics that might explain why the dynamic processes between parenting and adolescent functioning differ from one family to the next. Here, I distinguish between individual characteristics (i.e., characteristics of the adolescent, parent, and their relationship) and contextual characteristics (i.e., characteristics of the immediate and remote environments). An individual characteristic that has received considerable attention is the adolescent’s personality. Personality may play a role in how adolescents differentially perceive,
Chapter 1 20 interpret, cope with, and respond to parenting practices (Soenens et al., 2015; Soenens & Vansteenkiste, 2020). For instance, according to environmental sensitivity models (Belsky & Pluess, 2009; Pluess, 2015), some individuals are more sensitive to perceive and more responsive to positive and/or negative environmental influences, including parenting practices. This heightened sensitivity and responsivity is believed to be driven by a (partly) innate and stable high trait level of sensitivity to the environment (Greven et al., 2019). The personality trait neuroticism, which is distinct but related to environmental sensitivity (Greven et al., 2019), has also been linked to greater responsivity to the (social) environment, particularly with respect to negative events (Bolger & Schilling, 1991; Suls & Martin, 2005). Because individuals with higher trait levels of neuroticism seem to interpret negative events more negatively and seem less effective in coping with negative events (Suls & Martin, 2005), high-neuroticism adolescents might particularly be more prone to suffer from negative parenting practices. Thus, the personality of the adolescent is understood as a moderating factor in how parenting practices influence an adolescent’s functioning. Although the personality of adolescents has received much attention in the parenting literature, similar arguments also likely apply to the personality of the parent, such that parents with higher levels of environmental sensitivity or neuroticism might be more responsive to the behavior of their adolescent child than parents with lower trait levels. In addition to differences in personality, two other individual characteristics have received considerable attention. The first is adolescents’ legitimacy beliefs of parental authority (Darling et al., 2007; Smetana & Asquith, 1994), which reflect the degree to which adolescents view their parents’ exercise of control over a specific domain (e.g., moral, conventional, and personal) as appropriate for their role as caregivers (Darling et al., 2008). Adolescents’ legitimacy beliefs might be a moderating factor in how controlling practices impact their functioning, such that adolescents with weak legitimacy beliefs might experience more negative reactions to controlling practices (e.g., monitoring) (LaFleur et al., 2016). The second characteristic is the parenting style, which is believed reflect the overall emotional climate within the parent-adolescent relationship. Parenting style is thought to directly alter the effectiveness of specific parenting practices on adolescent functioning, and indirectly by influencing adolescents’ openness to socialization (Darling & Steinberg, 1993). Hence, the stable emotional climate of the parent-adolescent relationship and adolescents’ willingness to accept and obey their parental socialization attempts are key factors that are thought to moderate the effects of parenting practices on adolescent functioning.
1 General introduction 21 Besides individual characteristics, bio(psychosocial) ecological models also place special emphasis on how contextual characteristics might moderate proximal parenting processes (Bronfenbrenner, 2005; Sameroff, 2010). These moderating contextual characteristics can exist at different levels: meso- (e.g., work of parents), exo- (e.g., neighborhood), and macrolevel (e.g., culture) (Bronfenbrenner, 2005). In the parenting literature, it has been suggested that the context might determine the extent to which a parenting practice is perceived to be appropriate or normative. Controlling parental behavior may be appropriate and therefore more effective in reducing adolescent problem behavior in families who live in high-risk environments, such as neighborhoods with high crime rates, whereas the same behavior may be perceived as intrusive by adolescents who live in low-risk environments (Boykin McElhaney & Allen, 2001). Similarly, a parenting practice might be more promotive of adolescent functioning if it is (perceived to be) culturally normative (Lansford, 2022; Soenens et al., 2015). Thus, parent-adolescent influences are also assumed to vary as a function of characteristics outside the parentadolescent relationship, including their immediate and remote environment. 2. THEORY VERSUS METHODS: IDENTIFIED METHODOLOGICAL GAPS Even though established theories, as summarized in Figure 1 and visualized in Figure 2, assume that the dynamic processes between parenting and adolescent functioning occur within a family over time and are reciprocal and heterogeneous across families, hitherto, most of these meta-theoretical elements still lack empirical evidence. Despite the thousands of parenting studies (e.g., Pinquart, 2017a, 2017b), methodological gaps have refrained parenting science from truly examining the theorized dynamic processes between parenting and adolescent functioning. Similar to other fields in psychological science (Moeller, 2022; Molenaar, 2004), there is thus a considerable mismatch between theory and the generally applied methods. What these three methodological gaps are and how this dissertation addresses them is explained below. 2.1 Mind the (first) gap: Between-family methods do not tap into within-family processes The first methodological gap concerns the ecological level at which the theoretical assumptions and inferences from empirical findings are most commonly drawn. Although the theorized dynamic processes between parenting and adolescent functioning are
Chapter 1 22 understood to take place within families (e.g., Darling & Steinberg, 1993), the most common approach to studying these theoretical ideas is with a between-family design (Keijsers, 2016). Between-family approaches often include cross-sectional or longitudinal panel studies, in which one or a few data points per family are collected (see Table 1). Based on such cross-sectional or longitudinal panel data, group-level (between-family level) bivariate associations between parenting and adolescent functioning can be estimated, for example with a correlation or regression coefficient. These between-family estimates indicate relative (rank-order) differences between families (Hamaker, 2012; Keijsers, 2016). For instance, a positive between-family correlation between parental psychological control and adolescent depressive symptoms indicates that adolescents who experienced higher levels of psychological control experienced more depressive symptoms on average than adolescents who experienced lower levels of parental psychological control (see Figure 3; Pinquart, 2017b). Figure 3 A between-family association versus a within-family association Parental psychological control Adolescent depressive symptoms min max min max a a a a a a a a b b b b b b b b b c c c c c c c c Family a Family b Family c Between-family association Within-family associations Note. Within-family correlations of three families based on eight measurements, showing a pattern of “universality without uniformity”. Between-family correlation based on the aggregate of each family. By applying a within-family design, it can be tested how parenting and adolescent functioning are associated at the within-family level. Associations at the within-family level show how over-time fluctuations or changes in parenting and adolescent functioning are related within families. To illustrate, a positive within-family correlation between parental
1 General introduction 23 psychological control and adolescent depressive symptoms indicates that increases in psychological control are related to same-time increases in adolescents’ depressive symptoms within families (see Figure 3). To assess such within-family bivariate associations, (intensive) longitudinal data needs to be analyzed using statistical methods that can disentangle stable between-family variance from over-time within-family variance (see Table 1) (Hamaker et al., 2015; Lucas, 2023). In contrast to between-family analyses that can already be conducted with only one “snapshot” per family, within-family analyses need (few to many) repeated measures per family. Applying a within-family approach thus requires a more intensive and expensive data collection. Moreover, analyzing (intensive) longitudinal data at the within-family level requires advanced statistical techniques, many of which had yet to be developed a decade ago (Asparouhov et al., 2018; Hamaker et al., 2015, 2018), and some still have to be released today (Asparouhov et al., 2017). Because of these practical, financial, and statistical drawbacks, empirical studies applying a within-family design are not yet standard practice in the empirical study of parenting adolescents. Although modern statistical and technological innovations have facilitated the application of within-family designs (Repetti et al., 2015; Van Roekel et al., 2019), within-family studies are still overshadowed by many between-family studies. To illustrate, at the beginning of this PhD project, only 46 studies were available that examined within-family associations between parenting and adolescent functioning (see Chapter 2; Boele et al., 2020). These 46 within-family studies stand in stark contrast to the hundreds of studies included in meta-analyses that assessed between-family correlations (e.g., Pinquart, 2017a, 2017b). Therefore, many meta-theoretical ideas about the dynamic processes between parenting and adolescent functioning (for a summary, see Figure 1) have remained untested at the correct ecological level, the within-family level. This dissertation started to fill this gap by conducting four empirical parenting studies at the within-family level by using six datasets from five different samples (Chapters 3-6). 2.2 The second gap: Dynamics might not generalize across timescales A second gap in the empirical parenting literature concerns the fact that the dynamic processes between parenting and adolescent functioning are often empirically studied on a macro-timescale. That is, despite modern theories of human development and parenting propose that dynamic parenting processes unfold on various timescales (see Figure 2), the limited work of within-family studies has mostly tested how parenting and adolescent
Chapter 1 24 functioning are associated on a (semi-)annual timescale (see Chapter 2: Boele et al., 2020). As both theoretical (e.g., Granic et al., 2008) and methodological work (Voelkle et al., 2018) suggest that associations on one timescale may not generalize to other timescales, whether and how the dynamic processes between parenting and adolescent functioning unfold on different timescales remains still an open question. To avoid a galloping horse fallacy – infer results from one timescale to another (Keijsers & Van Roekel, 2018) –, within-family studies on parenting adolescents should consider various timescales: micro-, meso-, and macro-timescales. Micro-timescales include momentary or daily processes, which can be investigated using Experience Sampling methods or daily diary studies (Myin-Germeys & Kuppens, 2022; Repetti et al., 2015). Meso-timescales can include weekly or monthly processes, and macro timescales can include (semi-)annual processes. To start filling the gap in how the dynamic processes between parenting and adolescent functioning unfold on various timescales, this dissertation examined these dynamics at five time intervals: daily (Chapters 3, 5, & 6), bi-weekly (Chapters 3 & 4), three-monthly, annual, and biennial (Chapter 3). 2.3 The third gap: The dynamics of the average family might not apply to (all) individual families A third gap in the parenting literature is that theories propose heterogeneous dynamic processes between parenting and adolescent functioning, while most applied withinfamily methods that model homogeneous processes. That is, most within-family studies on parenting adolescents, thus far, have estimated the average effects within the sample (see Chapter 2: Boele et al., 2020). For example, studies have estimated average sample effects by applying a random-intercept cross lagged panel model (Hamaker et al., 2015; Keijsers, 2016) or fixed effects in a multilevel regression model (Bai et al., 2017; Coley & Medeiros, 2007). Such average sample effects indicate how fluctuations around the family’s typical level of parenting were on average (concurrently or longitudinally) associated with fluctuations around the family’s typical level of adolescent functioning. The potential of this method is to identify ‘general’ parenting processes or the parenting processes in ‘the average family’. Moreover, a benefit of this approach is that existing data from longitudinal panel studies can be used to reanalyze data at the within-family level (e.g., Keijsers, 2016). However, because many theories agree that the dynamic processes between parenting and adolescent functioning are heterogeneous across families (see Figure 1), heterogeneity should be tested to avoid a “one-size-fits-all” fallacy: interpreting average sample effects as homogeneous processes while they are not (Keijsers & Van Roekel, 2018).
1 General introduction 25 To illuminate potential heterogeneity, the first approach may be to apply a groupdifferential approach. That is, to test whether the dynamic processes between parenting and adolescent functioning differ between subgroups. This can be done by estimating moderator effects or by a multi-group approach in SEM, for instance, which requires a relatively small number of assessments and moderate sample sizes (e.g., t = 30, with N = 300; see Table 1). By applying this approach, one can assess whether the dynamic parenting processes vary from subgroup to subgroup, as several theoretical perspectives have been proposed (see 1.6). For example, in Chapter 2, I assessed whether the average reciprocal time-lagged associations between perceived parental support and adolescent depressive symptoms varied between adolescent boys and girls, or between adolescents with low and high trait levels of neuroticism. Nonetheless, the dynamic processes between parenting and adolescent functioning may still vary among families that belong to the same predefined subgroup (e.g., female vs. male adolescents) (Moeller et al., 2022). Therefore, a more fine-grained approach is warranted to uncover the heterogeneity that potentially hides behind sub-sample averages. Such a fine-grained approach is an idiographic approach. Using an idiographic approach, the dynamic processes between parenting and adolescent functioning are estimated at the individual-family level: for each individual family separately. Although many scholars have stressed that humans function, develop, and interact with their environment in idiosyncratic ways (Bronfenbrenner, 2005; Harris, 2006; Richters, 2021), modeling idiosyncratic dynamic parenting processes in adolescence has rarely been applied in parenting studies. In fact, idiographic studies on parenting adolescents were lacking at the start of this PhD project (see Chapter 2: Boele et al., 2020; but see Molenaar & Campbell, 2009), which might be due to the required intensive longitudinal data and the application of state-of-the-art advanced statistical techniques (see Table 1). To start filling this gap and to examine the dynamic processes between key dimensions of parenting and adolescent functioning at the level of the individual family, this dissertation applied an idiographic (or family-specific) approach in three empirical studies (Chapters 4-6). Specifically, 26 bi-weekly assessments of 256 Dutch families and 100 daily diaries of 159 Dutch families were collected (for an example time series, see Figure 4 below) and analyzed using DSEM (Chapters 4 & 5) and GIMME (Chapter 6). One criticism of the idiographic approach is that it may undermine generalization (Beltz et al., 2016). Because of the detailed family-specific results, it might be challenging to illuminate the extent to which the dynamic parenting processes of individual families are unique or shared by most or some families. However, state-of-the-art methods have
Chapter 1 26 made it possible to search for generalizable principles while using idiographic analysis (Beltz et al., 2016). A so-called bottom-up idionomic approach combines the strengths of two complementary research paradigms: first unit-specific effects are estimated, and then recurring generalizable patterns are identified across units (Moeller et al., 2022; Sanford et al., 2022). Recurrent patterns can be identified by simply describing the variation in unit-specific effects; for instance, by summarizing the proportion of participants for which a particular association was found (e.g., Beyens et al., 2021). Recurrent patterns can also be identified using data-driven procedures, such as with Subgrouping Group Iterative Multiple Model Estimate (S-GIMME; Gates et al., 2017; Lane et al., 2019). Hence, by integrating nomothetic and idiographic approaches, generalizable patterns are empirically established and thus powerfully avoids a one-size-fits-all fallacy. This dissertation took the first steps in detecting such recurring generalizable patterns in family-specific dynamic parenting processes, both using a descriptive (Chapters 4 and 5) and data-driven idionomic approach (Chapter 6). By doing so, this dissertation answered an increasingly loud call for a more differentiated empirical understanding of adolescent development and family functioning (Barbot et al., 2020; Chaku & Beltz, 2022). Figure 4 Timeseries of two adolescents Note. Including daily mean scores for positive affect and perceived parental psychological control across 100 days. Response scale ranged from 0 to 100. See also Chapter 5.
1 General introduction 27 3. AIMS AND OUTLINE OF THIS DISSERTATION To address the three described methodological gaps in the parenting literature (i.e., gaps pertaining to the within-family level, timescales, and heterogeneity across families) and to increase the empirical understanding of the dynamic processes between parenting and adolescent functioning within families, this dissertation had three aims. The first aim was to examine how key dimensions of parenting (i.e., warmth, autonomy support, behavioral control, and ps ychological control) are (on average) reciprocally associated with adolescent functioning at the within-family level. The second aim was to examine within-family associations between parenting and adolescent functioning on various timescales. The third aim was to quantify and explain heterogeneity across families in within-family associations between parenting and adolescent functioning. To achieve these aims, I conducted five studies: one systematic literature review (Chapter 2) and four empirical studies (Chapters 3-6; for a visualization of the empirical studies see Figure 5). These five studies are described below and summarized in Table 2. To obtain an overview of the available empirical insights into the dynamic processes between parenting and adolescent functioning within families, I first conducted a systematic literature review in Chapter 2. Here I summarize the small existing body of studies that examined the within-family linkages between parenting and adolescent functioning. This systematic overview exposed significant gaps in the literature, in terms of the studied dimensions, timescales, and heterogeneity. These gaps were further addressed in four empirical studies. In Chapter 3, a multiple-timescale study is presented. Specifically, I investigated whether the average processes between perceived parental support and adolescent depressive symptoms varied across timescales. I used five datasets with different time intervals, from a daily to a biennial time interval, to estimate cross-lagged effects between parental support and depressive symptoms (see Table 2). Additionally, I examined the potential heterogeneity in these effects by studying average subgroup differences based on adolescent characteristics: differences between girls and boys and between adolescents with low and high levels of trait neuroticism. In Chapter 4, I tested environmental sensitivity models at the individual family level. According to these models, some adolescents are more sensitive to supportive and/or unsupportive parenting than others. I tested whether such different responsivity patterns
www.ridderprint.nlRkJQdWJsaXNoZXIy MTk4NDMw