11.1.4.1 Parent-Driven Effects

Most research on dyadic emotion regulation has focused on the effects of parental emotion regulation on child emotion regulation (i.e. parent-driven effects). The notion that parents can positively or negatively influence their children’s emotional responses and regulation dates back decades (Gottman et al., Reference Gottman, Katz and Hooven1997). As was covered in Part II and Part III of this book, regulating emotions well is an essential faculty of parenting that influences child development through several avenues. The pathways of the effects of family socialization on emotion regulation are outlined in the Tripartite Model (Morris et al., Reference Morris, Silk, Steinberg, Myers and Robinson2007), which details influences that include observation (e.g. modeling), parenting practices (e.g. emotion coaching), and emotional climate of the family (e.g. marital relations). These pathways have been documented empirically since (for a review, see Morris et al., Reference Morris, Criss, Silk and Houltberg2017).

Parent characteristics such as their own attachment styles, levels of stress and social support, and mental health influence familial socialization of emotion regulation. Most relevant to this chapter, parent-driven effects on child emotion regulation could involve the transmission of emotion dysregulation from parent to child (see more on this in Section 11.1.5). Leveraging this directionality, interventions target parenting skills to improve child behavior/emotion regulation (e.g. Rothenberg et al., Reference Rothenberg, Weinstein, Dandes and Jent2019). In sum, models with parent-driven effects view emotion dysregulation as originating from the parent and transmitting to the child through various mechanisms.

11.1.4.2 Child-Driven Effects

Less well characterized is the role that children play in the dyadic nature of emotion regulation (i.e. child-driven effects), though research indicates that children can also modulate the flow of parent emotion and corresponding regulatory strategies. Short term, within a parent–child exchange, a distressed child can evoke an emotionally dysregulated state within the parent. Long term, attributes of the child (e.g. difficult temperament) can negatively affect parenting (Micalizzi et al., Reference Micalizzi, Wang and Saudino2017), perhaps through emotion dysregulation. Patterson’s Coercion Theory (Patterson, Reference Patterson, Dishion, Snyder and Patterson2016) captures early child emotion dysregulation as an evocative factor of negative parent behavior that cascades into a coercive dyadic cycle occurring over many years. Further, Sameroff’s transactional model (Sameroff, Reference Sameroff and Sameroff2009) outlines both bidirectional and recursive effects to relations between caregivers and children that adds to the complex, dynamic multilevel processes involved in emotion regulation development (Olson & Sameroff, Reference Olson and Sameroff2009).

11.1.4.3 Interactive Effects

Since Bell’s (Reference Bell1968) reinterpretation of directions of effects in socialization processes, an accumulating research base indicates that there are mutually interactive dyadic influences observed between parent and child (e.g. Micalizzi et al., Reference Micalizzi, Wang and Saudino2017; Thomas et al., Reference Thomas, Deros, Jain, Jacobs and De Los Reyes2022). Parent–child interactions can evoke intense and complex emotions from both members of the dyad (Hajal et al., Reference Hajal, Teti, Cole and Ram2019) that can result in proximal, reciprocal exchanges of emotions and regulatory strategies as well as future implementation of these strategies. To illustrate, because a child’s emotion regulation abilities result from continuous and reciprocal interactions between the child and their caregiver over time (Sameroff, Reference Sameroff, Evans and Wachs2010), a caregiver’s response to their child’s anger can alter the child’s perception of if/how the expression of anger is acceptable and their subsequent expression of anger. At the same time, the caregiver receives information about if/how they must change their regulatory strategies to influence their child’s regulatory capacity (Chan et al., Reference Chan, Feng, Inboden, Hooper and Gerhardt2022). To this end, increased research attention has been paid to the conceptualization of emotion regulation as a dynamic, dyadic process in recent years (Gates & Liu, Reference Gates and Liu2016; Morris et al., Reference Morris, Cui, Criss, Simmons and Cole2018; Silk, Reference Silk2019; Stone et al., Reference Stone, Mennies, Waller, Ladouceur, Forbes, Ryan, Dahl and Silk2019; Wright & Hopwood, Reference Wright and Hopwood2016).

11.2.1.1 Questionnaires

The most accessible approach to studying interactive dyadic emotion regulation is to administer developmentally appropriate questionnaires to both parents and youth and to employ one of the analytic techniques outlined later in this chapter. If children are too young to self-report, researchers could obtain reports on child emotion regulation from teachers, parents, or researchers depending on the child’s age. Because methodological biases could be introduced if parents report on themselves and their children (Podsakoff et al., Reference Podsakoff, MacKenzie, Lee and Podsakoff2003), different raters for parent and child emotion regulation are preferable.

Measures that assess interpersonal emotion regulation include the Interpersonal Emotion Regulation Questionnaire (Hofmann et al., Reference Hofmann, Carpenter and Curtiss2016), the Interpersonal Regulation Questionnaire (Williams et al., Reference Williams, Morelli, Ong and Zaki2018), and the Emotion Regulation of Others and Self (Niven et al., Reference Niven, Totterdell, Stride and Holman2011). Notably, the psychometric properties of these measures were evaluated among adult samples; administration to children would require further psychometric evaluation. Questionnaires of intrapersonal regulation are more widely implemented and include the Emotion Regulation Questionnaire (Gross & John, Reference Gross and John2003) and the Difficulties in Emotion Regulation Scale (Gratz & Roemer, Reference Gratz and Roemer2004).

Generally, questionnaire methods for emotion regulation are poised to tap trait, rather than state, emotion regulation (Silk, Reference Silk2019). Assessing emotion regulation from parents and children over time would inform stability/change over time and/or how one member’s emotion regulation influences that of the other dyad member over time. Some limitations of self-report measures of emotion regulation include the ability of measures only to provide insight into explicit, conscious processes (i.e. processes for which the individual is aware), reliance on a high degree of the individual’s emotional awareness, retrospective recall, and global, rather than context-specific items. Next, we review other methods researchers use that attempt to assess parent–youth interactions and emotion regulation in vivo.

11.2.1.2 Ecological Momentary Assessment

Ecological momentary assessment (EMA), a form of ambulatory assessment and experience sampling, overcomes some of the limitations of questionnaires. EMA involves the repeated assessment of behaviors and experiences in-the-moment to assess a domain of interest when it occurs in real-time in the natural environment (Bettis et al., Reference Bettis, Burke, Nesi and Liu2022; Shiffman et al., Reference Shiffman, Stone and Hufford2008; Silk, Reference Silk2019). Using EMA, one can assess the fluctuations of a domain over time and across contexts to minimize recall bias and maximize external validity (Shiffman et al., Reference Shiffman, Stone and Hufford2008; Silk, Reference Silk2019). This enables researchers to make inferences into short-term dynamics of the parent–child dyad, yet it is an underused method for assessing parent–adolescent interactions (Keijsers et al., Reference Keijsers, Boele and Bulow2022). With respect to emotion regulation and parenting, EMA can enable parents and/or youth to report on some aspect of their interactions with each other, and/or their emotion regulation throughout the day, with the ability to indicate context (e.g. with peers; Stone et al., Reference Stone, Mennies, Waller, Ladouceur, Forbes, Ryan, Dahl and Silk2019). Several studies using EMA demonstrate the role that parents play in supporting adolescents‘ emotion regulation, based on adolescents’ EMA reports (Silk, Reference Silk2019; Silk et al., Reference Silk, Forbes, Whalen, Jakubcak, Thompson, Ryan, Axelson, Birmaher and Dahl2011; Waller et al., Reference Waller, Silk, Stone and Dahl2014). For example, in adolescents with and without major depressive disorder, the use of EMA to measure daily social interactions after a negative event enabled researchers to determine that adolescents with depression co-ruminate with their parents, which adolescent participants without depression did not do (Waller et al., Reference Waller, Silk, Stone and Dahl2014). Limitations to EMA include potential burden on participants due to the repeated nature of assessments, reliance on self-report, and the potential for missing data (Bettis et al., Reference Bettis, Burke, Nesi and Liu2022). Several recent reviews provide further detail on this method and its relevance to emotion regulation (e.g. Bettis et al., Reference Bettis, Burke, Nesi and Liu2022; Keijsers et al., Reference Keijsers, Boele and Bulow2022; Silk, Reference Silk2019).

11.2.1.3 Interaction Tasks

Some parent–child interactions may occur less frequently, for only a short period of time (Keijsers et al., Reference Keijsers, Boele and Bulow2022), or are difficult to assess with a static survey report, and thus may be better suited to measurement while being induced and observed in a laboratory setting rather than via questionnaires or EMA (e.g. Thomas et al., Reference Thomas, Wilson, Jain, Deros, Um, Hurwitz, Jacobs, Myerberg, Ehrlich, Dunn, Aldao, Stadnik and De Los Reyes2017, Reference Thomas, Jain, Wilson, Deros, Jacobs, Dunn, Aldao, Stadnik and De Los Reyes2019). Furthermore, if the process of how parents and children behave together is the focus, an interaction task is the optimal method to obtain such data. Moreover, behavior and psychophysiology are common facets of emotion regulation (Morris et al., Reference Morris, Cui, Criss, Simmons and Cole2018) and such a task enables measurement of both for dyad members. Accordingly, observational procedures for assessing emotion regulation provide some advantages beyond self-report (Girard & Cohn, Reference Girard and Cohn2016). Methodologically, it is beneficial to have an objective source of data on emotion regulation, separate from survey reports. Survey reports and behavioral observations are not highly correlated, indicating they do not measure the same things (e.g. parent behavior; Hendriks et al., Reference Hendriks, Van der Giessen, Stams and Overbeek2018). Moreover, interaction tasks are often designed to be ecologically valid in order to observe behavior as it may naturally occur between dyads (rather than rely on retrospective recall or insight during survey reports).

Highlighting their importance, even some interventions for childhood behavior problems involve parent–youth interaction tasks. Their inclusion is supported by the role of emotion dysregulation in the context of impairing psychiatric conditions requiring early intervention (Aldao et al., Reference Aldao, Gee, De Los Reyes and Seager2016; Sheppes et al., Reference Sheppes, Suri and Gross2015), and the recognition that parents and youth evoke responses from one another that can become entrenched patterns (Patterson, Reference Patterson, Dishion, Snyder and Patterson2016). These tasks enable clinicians to observe patterns of interaction that may elicit emotional reactions that escalate/exacerbate problem behaviors as part of the assessment process, subsequently informing a treatment plan and skill development. One example is Parent–Child Interaction Therapy (PCIT), a treatment for disruptive child behavior in which clinicians coach parents in vivo how to respond to their children during activities. Child emotion dysregulation has been shown to decrease significantly from pre- to post-treatment (Rothenberg et al., Reference Rothenberg, Weinstein, Dandes and Jent2019).

In laboratory settings, many parent–youth interaction task designs salient to measuring emotion regulation and parenting involve a discussion task focused on problem-solving, planning, or conflict-resolution (Bodner et al., Reference Bodner, Kuppens, Allen, Sheeber and Ceulemans2018; Cui et al., Reference Cui, Morris, Harrist, Larzelere, Criss and Houltberg2015; Donenberg & Weisz, Reference Donenberg and Weisz1997). In the latter situation, topics about which parent–youth dyads have conflict are introduced (e.g. Issues Checklist; Prinz et al., Reference Prinz, Foster, Kent and O’Leary1979), and dyads are asked to come to a resolution within a brief period (e.g. 5–10 minutes). The theoretical underpinnings that inform how researchers observe and code parent and child behavior are diverse (e.g. stress, attachment). One commonality of these phenotypes is that many involve facets of emotion regulation, particularly if the objective of the task is to induce stress or replicate an emotionally intense interaction that is typical for the dyad.

In the context of the interaction task, researchers often measure emotional reaction or intensity, often along a spectrum of some variation of positive affect (warmth/validation) and negative affect (hostility/anger), the psychometric considerations of which are discussed thoroughly elsewhere (Girard & Cohn, Reference Girard and Cohn2016). Finally, interaction task measurement may occur from coding behavior such as body language, content of speech, and facial expressions; psychophysiological responding (e.g. heart rate variability during task; Cui et al., Reference Cui, Morris, Harrist, Larzelere, Criss and Houltberg2015; Thomas et al., Reference Thomas, Jain, Wilson, Deros, Jacobs, Dunn, Aldao, Stadnik and De Los Reyes2019); and/or post-task ratings of emotions experienced during the task (Chaplin et al., Reference Chaplin, Sinha, Simmons, Healy, Mayes, Hommer and Crowley2012; Turpyn et al., Reference Turpyn, Chaplin, Cook and Martelli2015).

In an example study including multiple domains to measure emotion, researchers assessed adolescents‘ psychophysiology (blood pressure, heart rate) during a 10-minute conflict discussion task with parents and asked them to report on their emotions pre- and post-task (Chaplin et al., Reference Chaplin, Sinha, Simmons, Healy, Mayes, Hommer and Crowley2012). Results indicated an inverse relationship between parenting behaviors during the task and adolescents’ emotions, such that lower parental involvement (e.g. offering solutions, setting/explaining rules) and parental support/warmth were associated with greater adolescent anger arousal and greater blood pressure. In another study using the same conflict discussion task, researchers created latent profiles of adolescent emotion regulation during the conflict discussion task, comprised of positive and negative emotional expressions during the conflict task (i.e. coded from observed behavior), self-reports of subjective experiences of anger and anxiety evoked from the task (completed immediately post task), and heart rate reactivity during the task (Turpyn et al., Reference Turpyn, Chaplin, Cook and Martelli2015). Parenting behaviors during the task were measured via an existing coding system by trained staff, with a focus on behaviors that were negative/critical (e.g. mocking, interrupting). Negative/critical parenting was associated with a greater likelihood of adolescents belonging to profiles indicative of less emotion regulation (Turpyn et al., Reference Turpyn, Chaplin, Cook and Martelli2015). It is important to note that although the two aforementioned studies are cross-sectional, the predominant theoretical rationale, which also underpins the analyses and results, reflect parent-driven effects; however, both articles acknowledge that adolescents’ behaviors stemming from emotional reactivity may evoke certain parental responses (i.e. child-driven effects). In sum, parent–youth interaction tasks are a flexible means of assessment that enable derivation of behavior consistent with emotion (dys)regulation, which can then be related to parenting domains and/or youth risk behaviors and psychopathology.

11.2.1.4 Physiological Indices

Although it is beyond the scope of what can be described in detail in this chapter, physiological measurements can be implemented during these interactions. Researchers use physiological measurements, such as of cardiovascular response and brain synchrony/connectivity, to index facets of emotion regulation (Cui et al., Reference Cui, Morris, Harrist, Larzelere, Criss and Houltberg2015; Ratliff et al., Reference Ratliff, Kerr, Misaki, Cosgrove, Moore, DeVille, Silk, Barch, Tapert, Simmons, Bodurka and Morris2021; Reindl et al., Reference Reindl, Gerloff, Scharke and Konrad2018; Turpyn et al., Reference Turpyn, Chaplin, Cook and Martelli2015). These assessments facilitate investigations of processes like synchrony between parents and youth (Abney et al., Reference Abney, daSilva and Bertenthal2021) that are theorized to be a mechanism by which parents help their offspring regulate their emotions (Morris et al., Reference Morris, Cui, Criss, Simmons and Cole2018). As an example, one study found parent–child brain synchrony during a cooperative interaction task that was not present during a competitive interaction task or when completing tasks with a stranger (Reindl et al., Reference Reindl, Gerloff, Scharke and Konrad2018).

11.2.2.1 Study Design

The study research question(s) and how the measures, means, and schedule of administration will fulfill that objective should guide the selection, design, or modification of the measurement instrument and subsequent analysis of data. As a priority, researchers must determine how to define and subsequently assess emotion regulation in the context of parent–youth interactions (discussed next). Then, researchers must decide how to measure interaction, either with a task or statistical inference. In the former case, researchers could implement an existing task (e.g. Chaplin et al., Reference Chaplin, Sinha, Simmons, Healy, Mayes, Hommer and Crowley2012). However, if the research question involves contextual variation of emotion regulation, a sufficient task design is necessary to elicit the behaviors of interest and perhaps include a “control” condition to contrast the conditions under which they expect to observe the target behavior (Thomas et al., Reference Thomas, Jain, Wilson, Deros, Jacobs, Dunn, Aldao, Stadnik and De Los Reyes2019). If the investigation involves ratings of observed behavior, sufficient numbers of research staff will be necessary, as well as a plan to monitor reliability of ratings (for a thorough review on considerations related to observational measurement, see Girard & Cohn, Reference Girard and Cohn2016). Carefully selecting study design can address the long-standing research question as to whether behaviors are stable across situations or depend on situational context (Donenberg & Weisz, Reference Donenberg and Weisz1997; Silk, Reference Silk2019); therefore, depending on the researcher’s interest, study design and task selection should be adjusted accordingly.

11.2.2.2 Defining Emotion Regulation and Its Level of Analysis

Operationally defining “emotion regulation” is crucial. In the case of a task designed to measure emotion regulation, a primary question is how one observes what may be an internal process. Because some adaptive emotion regulation skills are idiosyncratic to the individual and may be unobservable (e.g. deep breathing, reappraisal; Gross & John, Reference Gross and John2003), researchers may instead focus on indicators of lack of emotion regulation or dysregulation. Thus, in many cases of interaction tasks, it has been the manifestation of an emotion that may be excessive (i.e. not adaptive), thereby interfering with the goal of the task (Beauchaine, Reference Beauchaine2015).

A related challenge is deciding what level of analysis emotion regulation will be recorded. Emotion dysregulation may be assessed at the level of behavior (e.g. yelling), or by measurement of physiological functioning, such as cardiovascular indices (e.g. heart rate) as a proxy for emotional reactivity/regulation that can be recorded continuously throughout the interaction and/or to measure synchrony between dyad members (Morris et al., Reference Morris, Cui, Criss, Simmons and Cole2018; Reindl et al., Reference Reindl, Gerloff, Scharke and Konrad2018). Moreover, it is also possible to include a parent–youth interaction task to capture a parenting behavior of interest and associate those behaviors with survey reports of emotion regulation. Finally, an emotion regulation latent variable could be derived from different levels (self-report, behavior, physiology; e.g. Turpyn et al., Reference Turpyn, Chaplin, Cook and Martelli2015).

11.2.2.3 Selecting a Coding Scheme/Emotion Regulation Metric

Once emotion regulation has been defined, researchers must select an appropriate measurement instrument to facilitate inferences about interaction. Examples include survey, EMA, psychophysiology, or coding of observed behavior during an interaction task. Additional nontrivial considerations when coding behavior include having a laboratory audiovisual recording system so behaviors do not have to be coded live, and sufficient storage space and security for these recordings. For psychophysiological indices of emotion regulation, researchers could use wearables like heart rate watches or a BIOPAC system to record sympathetic and parasympathetic nervous system response (Bettis et al., Reference Bettis, Burke, Nesi and Liu2022).

11.2.2.4 Informants

Even when assessing the same construct, parent and youth reports often do not agree, suggesting they may have different perspectives (De Los Reyes et al., Reference De Los Reyes, Thomas, Goodman and Kundey2013, Reference De Los Reyes, Augenstein, Wang, Thomas, Drabick, Burgers and Rabinowitz2015). Although it may complicate the study design, ideally both parents and youth will be included as informants in a study of parent–youth interactive effects of emotion regulation, particularly when investigating questions pertaining to synchrony or coregulation. As discussed previously, shared method variance is of concern if parents report on both themselves and their child. Relatedly, objective measures like observed behavior by trained raters also address informant issues.

In some cases of prior research, adolescents have reported on facets of their own emotion regulation and reported on another domain that allowed inferences into parental role of emotion regulation (Silk, Reference Silk2019; Stone et al., Reference Stone, Mennies, Waller, Ladouceur, Forbes, Ryan, Dahl and Silk2019; Waller et al., Reference Waller, Silk, Stone and Dahl2014). These investigations are informative; however, when possible, including both informants eliminates bias in relying solely on one informant’s perspective. An additional consideration is whether both parent and youth are providing data on the same construct, or whether one dyad member reports on one construct (e.g. parenting), and the other dyad member reports on another construct (e.g. adolescent emotion regulation). To use certain interactive analyses (described in Section 11.3.2.1 Actor–Partner Interdependence Model), data from both members of the dyad on both constructs are required. Relatedly, there may be developmental issues to consider if parents and youth will report on the same emotion regulation domain using different measures. For example, the Difficulties in Emotion Regulation Scale Short Form (Kaufman et al., Reference Kaufman, Xia, Fosco, Yaptangco, Skidmore and Crowell2016) is appropriate for adolescents and adults, but if researchers wanted to measure the same emotion regulation domain in young children and their parents, determining which measures would enable inferences about the same domain across different developmental levels is key to avoiding invalid results due to measurement issues. Finally, studying family emotion regulation and interactive effects (i.e. including all members of a family unit) is exceptionally challenging due to translating the data into a consistent format needed for statistical models, and is therefore typically restricted to dyads (for example, parent and child; for designs integrating mother, father, and child emotion regulation, see, e.g. Kerr et al., Reference Kerr, Rasmussen, Smiley, Buttitta and Borelli2021; Li et al., Reference Li, Li, Wu and Wang2019).

11.2.2.5 Timing

Dyadic emotion regulation transactions occur within interaction as well as across time (Morris et al., Reference Morris, Cui, Criss, Simmons and Cole2018). Within brief interactions, conventional approaches to the study of dyadic emotion regulation aggregate data over time (i.e. take the arithmetic average), which fails to capture the rich and dynamic moment-to-moment fluctuations. To illustrate, by aggregating data, nuance is lost such that one cannot evaluate how parent–child emotion regulation exchanges manifest over time. For example, the effect of a maternal regulation strategy on the child’s subsequent emotional experience or the influence of a child’s emotional experience on parent’s emotion coaching.

Furthermore, depending on the study design, researchers may prefer to sequentially code behavior on a micro level between parents and youth to infer how the sequence reveals the interaction process (e.g. sequential analysis; Bakeman & Gottman, Reference Bakeman and Gottman1997). For research questions involving synchrony or co-regulation for which psychophysiology serves as a proxy of emotion regulation indices, it is important to have accurate time stamps of the start and end of tasks, as well as any notable events to match dyadic processes for analysis. Finally, researchers should consider whether the interaction task will be used once with study participants or readministered at intervals (e.g. yearly). If the latter, it is important to consider youth developmental level and select a task that will enable multiple administrations over the time interval while still appropriate for the youth’s developmental level.

11.2.2.6 Causality

A brief but important note on causality is warranted, given that many theories are premised on the idea of whose emotion (dys)regulation impacts whom. To infer causality, research studies need sufficient temporal ordering, and an ability to measure change, either within-person over time, or before and after an intervention. Furthermore, experimental manipulation via random assignment to conditions helps eliminate alternate explanations (e.g. Thomas et al., Reference Thomas, Jain, Wilson, Deros, Jacobs, Dunn, Aldao, Stadnik and De Los Reyes2019). Although the data may fit a particular analytic method, the use of that method is not what enables researchers to infer causality but rather their research design.

In sum, there are several design, staff, and implementation hurdles to overcome that explain why there are not as many studies on interactive effects of parent–youth emotion regulation as would be useful. Relatedly, it is often difficult to collect large sample sizes when interaction tasks are involved. By reducing complexity without compromising data quality (e.g. automated behavioral coding; Girard & Cohn, Reference Girard and Cohn2016) the likelihood is greater that researchers can scale up data collection and obtain a larger sample size.

11.3.2.1 Actor–Partner Interdependence Model

The application of the Actor–Partner Interdependence Model (APIM; Kenny et al., Reference Kenny, Kashy and Cook2006) permits evaluation of bidirectional effects in the context of interpersonal relationships (Cook & Kenny, Reference Cook and Kenny2005). The APIM tests the effect of one individual’s predictor on their own outcome, as well as on the outcome of their dyadic partner (and vice versa), all within one model (Stas et al., Reference Stas, Kenny, Mayer and Loeys2018). To use the APIM, both dyad members must have data on a predictor and an outcome variable. APIM derives both actor and partner effects, enabling a test of interpersonal and intrapersonal processes. In other words, one can investigate the association between a characteristic (e.g. emotion regulation ability) and an outcome (e.g. depression symptoms) for both dyad members, controlling for the influence of the other. A significant actor effect indicates one’s own characteristics are related to one’s own outcome, whereas a significant partner effect indicates one dyad member’s characteristics are related to another’s outcome. If both partner effects are significant, it suggests bidirectional influence, but one significant partner effect indicates interdependence (Cook & Kenny, Reference Cook and Kenny2005).

APIM models can be implemented using structural equation modeling (SEM) or multilevel modeling (MLM). To assist with analysis and interpretation, several R Shiny apps have been made available by David Kenny and colleagues that allow the investigator to upload a data set and derive output of results, including tables, figures, text summarizing the analysis and results, and code (DyadR)Footnote ¹. Briefly, APIM using SEM is recommended for dyads who are distinguishable, meaning dyad members differ according to an attribute like sex or gender, or family role (e.g. parent, child; Ledermann & Kenny, Reference Ledermann and Kenny2017). Should researchers select this method, there is an R Shiny app (APIM-SEM)Footnote ² that executes the analysis, provides text interpretation of the findings, along with tables, figures, and R code, although it is not necessary that researchers know R programming to use this resource (Stas et al., Reference Stas, Kenny, Mayer and Loeys2018). MLM APIM is recommended when dyad members are not distinguishable (same-sex roommates; Ledermann & Kenny, Reference Ledermann and Kenny2017), which may be less relevant in the study of parent–youth emotion regulation. Best practices for APIM have been reviewed elsewhere (Ledermann & Kenny, Reference Ledermann and Kenny2017; Stas et al., Reference Stas, Kenny, Mayer and Loeys2018). Versions of this model have been adapted for longitudinal data as well (Bolger & Laurenceau, Reference Bolger, Laurenceau, Kenny and Little2013; Savord et al., Reference Savord, McNeish, Iida, Quiroz and Ha2022), one of which includes a Shiny app (L-APIMFootnote ³; Gistelinck & Loeys, Reference Gistelinck and Loeys2019). Longitudinal APIM has recently been applied to parent–child emotion regulation (Boeve et al., Reference Boeve, Beeghly, Stacks, Manning and Thomason2019). This approach stands to enhance our understanding of the unfolding of dyadic parent–child emotion regulation over time.

There are several examples of the APIM implemented to investigate parent–youth emotion regulation. First, given that psychopathology can be a manifestation of persistent emotional dysregulation (Aldao et al., Reference Aldao, Gee, De Los Reyes and Seager2016; Sheppes et al., Reference Sheppes, Suri and Gross2015), in a study of adolescents hospitalized for treatment of acute psychiatric symptoms and their parents, researchers used the APIM to probe the association between self-reports on one’s own difficulty with emotion regulation and depression symptoms (Wolff et al., Reference Wolff, Thomas, Hood, Bettis, Rizzo and Liu2020). There were significant actor effects between difficulty accessing emotion regulation strategies and depression symptoms. Further, there was a significant negative partner effect for parental impulsive emotion regulation and adolescent depressive symptoms, demonstrating interdependence between parents and adolescents. Second, a study investigated the associations between parents’ and adolescents’ reports of sources of parental knowledge with observed parent and adolescent behavior during a conflict discussion task, based on the rationale that how parents and adolescents interact and manage their emotions when resolving conflict (i.e. a goal-directed activity) will be associated with processes of parental monitoring (Thomas et al., Reference Thomas, Deros, Jain, Jacobs and De Los Reyes2022). Behavioral codes came from attachment domains representing behavioral categories that either helps or hurts the dyad’s goal of resolving the conflict topic. Hostile behavior (e.g. mocking) and an attachment domain known as Secure Base Use (adolescent) or Secure Base Provision (parent) (e.g. validation, smiling) were coded by trained staff for each dyad member. Adolescent reports of greater adolescent disclosure about their activities and whereabouts were associated with more secure base behavior exhibited by both adolescents (actor effects) and parents (partner effect), as well as less hostile behavior exhibited by both adolescents (actor effects) and parents (partner effect; Thomas et al., Reference Thomas, Deros, Jain, Jacobs and De Los Reyes2022). These findings support the interdependence of sources of parental knowledge that are related to monitoring, and manifestations of emotion regulation processes during a conflict discussion task (see Figure 11.1, for example APIM figure).

CC BY 4.0 SARAH THOMAS

Figure 11.1 Illustration of cross-sectional Actor–Partner Interdependence Models

Note. (a) Unstandardized parameter estimates of actor and partner effects for adolescent- and parent-reports of Adolescent Disclosure in relation to Secure Base behavior, controlling for the effects of age, sex, and sample source (covariate parameters not included in figure). (b) Unstandardized parameter estimates of actor and partner effects for adolescent- and parent-reports of Adolescent Disclosure in relation to Hostile behavior, controlling for the effects of age, sex, and sample source (covariate parameters not included in the figure). Standard errors are presented in parentheses. * p < 0.05; ** p < 0.01; *** p < 0.001.

11.3.2.2 Other Strategies

The APIM is not appropriate for all dyadic data, such as with only a youth outcome. In those situations, it is important to use an analytic strategy that will account for the nonindependent nature of the data. Options include generalized estimating equations (Hanley et al., Reference Hanley, Negassa, Edwardes and Forrester2003), MLM (Page-Gould, Reference Page-Gould, Cacioppo, Tassinary and Berntson2017), latent growth models for longitudinal data (Muniz-Terrera et al., Reference Muniz-Terrera, Robitaille, Kelly, Johansson, Hofer and Piccinin2017), and SEM. SEM permits evaluation of multivariate causal relations and these methods have significantly enhanced our understanding of the bidirectional nature of parent–child interactive effects (e.g. Micalizzi et al., Reference Micalizzi, Ronald and Saudino2016, Reference Micalizzi, Wang and Saudino2017). A specific model that is widely used is the cross-lagged panel model, a discrete-time structural equation model that can be used to analyze panel data where variables are assessed at least twice over time. The goal of these models is to examine the effect of one variable on the other variable (and vice versa) over time. Furthermore, SEM could be used if a researcher wished to analyze data from two parents and one youth (Kerr et al., Reference Kerr, Rasmussen, Smiley, Buttitta and Borelli2021; Li et al., Reference Li, Li, Wu and Wang2019). Detailed options for analytic strategies have been reviewed elsewhere (e.g. Gates & Liu, Reference Gates and Liu2016; Thorson et al., Reference Thorson, West and Mendes2018).

11.3.2.3 Latent Growth Curve

To measure a behavior over time, latent growth curve models enable assessment of slope and intercept, which can be used to assess the average rate of change in a construct, like emotion regulation. Further, by adding time-varying covariates, one can determine if another variable covaries with the behavior of interest. Consequently, it is also possible to evaluate parent–youth associations related to emotion regulation with growth curve models. To illustrate, researchers used growth curve analyses in a sample of mother–child dyads (half of the mothers were randomized to emotion regulation skills treatment) to investigate how change in maternal emotion regulation over 12 months was related to both the starting point and change in youth emotion regulation (Byrd et al., Reference Byrd, Lee, Frigoletto, Zalewski and Stepp2021). In another example, in the context of an intervention for comorbid adolescent psychiatric disorders and substance use, when assessing past 7-day cannabis use over time during and after an intervention, a time-varying covariate of parental frustration was used to determine how cannabis and parental frustration were related over time (Thomas et al., Reference Thomas, Brick, Micalizzi, Wolff, Frazier, Graves, Esposito-Smytherse and Spirito2020). This analytic technique has important implications for considering how the emotional facets of parent–adolescent relationships can be integral to the fluctuation of target behavior of an intervention over time. Despite the intervention, weekly cannabis use increased across the 1-year follow-up time, and parental ratings of frustration that were higher than their average were associated with greater adolescent cannabis use at baseline, 3-, and 6-month follow-up. This analytic strategy could be used to understand influence over time in parent–youth relationships and emotion regulation.