Interpersonal Emotion Regulation and Depressive Symptoms in Parent-Adolescent Dyads: A Daily-Diary Investigation

Abstract

Deficient parental extrinsic interpersonal emotion regulation (IER, how people regulate others’ emotions) is a known risk factor for adolescent depression. Although IER and depression development are transactional, dyadic processes, previous work has almost exclusively focused on how parental IER is associated with adolescent depression. The association between parental IER and parental depression and the associations between adolescent IER and adolescent and parental depression have received little attention. Moreover, most studies have focused on the regulation of negative but not positive affect. We address these gaps by examining associations between parent and adolescent IER and depressive symptoms using the Actor-Partner Interdependence Model framework. For 28 days, 112 parent-adolescent dyads (12- to 18-year-old adolescents) completed a dyadic daily diary, reporting their own depressive symptoms and IER strategies employed in response to dyad members’ positive and negative affect. Our results, based on 5,442 data points, show that the use of positive- and negative-affect-worsening IER is associated with more depression in the regulator (be it parent or adolescent). Surprisingly, parents’ use of more negative-affect-improving IER was associated with higher levels of their own and adolescents’ depression. Finally, adolescents’ use of positive-affect-improving IER was associated with their own decreased depression. Overall, parents (vs. adolescents) used more negative- and positive-affect-improving extrinsic IER, whereas adolescents used more positive-affect-worsening extrinsic IER. Our results highlight the importance of using dyadic designs in studying depression and IER as well as the need to consider who is regulating, the valence of the affect regulated, and the type of strategy used.

Depression is among the most prevalent and debilitating psychiatric illnesses, with enormous personal, familial, and societal costs (Greenberg et al., 2015). Depressive symptoms dramatically increase during adolescence (e.g., Salk et al., 2016). Alarmingly, the prevalence of depressive symptoms in adolescence has been consistently increasing over the last 20 years (Shorey et al., 2022). A salient risk factor for the development of depression is difficulty in emotion regulation (ER), the capacity to alter how one experiences and expresses emotions to facilitate adaptive functioning (Gross, 1998). Indeed, a vast body of research shows that across the lifespan, depression risk and recurrence is related to more frequent use of affect-worsening ER strategies (e.g., rumination) and less frequent use of affect-improving strategies (e.g., reappraisal; Aldao et al., 2010; Compas et al., 2017).

Theoretical models of ER development posit that emotion (dys)regulation begins in the home via emotion socialization processes, whereby parents teach children to understand, express, and regulate their emotions (Eisenberg et al., 1998; Morris et al., 2007). The emotion socialization coaching process, by which parents directly guide their children’s regulatory efforts, largely overlaps with the construct of interpersonal ER from the adult literature (e.g., Niven et al., 2009; Zaki & Williams, 2013). Interpersonal emotion regulation (IER) is broadly defined as actions performed in a social context to modulate the experience or expression of affect (Niven et al., 2009; Nozaki & Mikolajczak, 2020; Zaki & Williams, 2013). IER can be classified as intrinsic – regulating oneself with another person’s input – or extrinsic – regulating another person (Zaki & Williams, 2013). Extrinsic IER strategies are often categorized according to their typical influence on the other person’s affect as “affect-improving” (also referred to in literature as supportive or adaptive) or affect-worsening (also referred to in literature as unsupportive or maladaptive) strategies (López-Pérez & Pacella, 2021; Niven et al., 2009; Waslin et al., 2023; cf., Cohen & Arbel, 2020). In the past decade, extrinsic IER has been gaining attention as a potential risk factor for depression in adolescents.

Extrinsic IER and Adolescent Depression

Research shows that parents’ use of less affect-improving (e.g., comfort) and more affect-worsening (e.g., punitive) extrinsic IER is related to more depression in children and adolescents (e.g., Breaux et al., 2022; Lougheed et al., 2016; Schwartz et al., 2018). Moreover, parental affect-worsening extrinsic IER longitudinally predicts increased difficulties in regulating sadness in adolescents (e.g., Hale et al., 2023). Intervention studies focusing on improving parental extrinsic IER further support the causal relation between parental IER and child outcomes. For example, Kehoe et al. (2014) showed that following a training intervention that focused on enhancing parents’ awareness, acceptance, and regulation of their child’s emotions, as well as increasing parental empathy towards their child, parents’ IER skills improved. As a result, their adolescent children’s internalizing symptoms decreased.

Most studies on extrinsic IER and depression focused on parental IER and child outcomes. However, IER and depression are transactional and more complex than a unidirectional, parent-directed process (Lougheed et al., 2020; Zaki & Williams, 2013). Indeed, research has increasingly acknowledged that the development of depression is related to both parental and child characteristics. For example, a study following early adolescents (age 11) and their parents over three years showed that adolescents’ depression and anxiety symptoms predicted increases in parent depressive symptoms over time (Johnco et al., 2021). Similarly, a four-year longitudinal study of mother-daughter dyads (15-year-old adolescents) showed that adolescents’ depressive symptoms predict subsequent parental emotion dysregulation (Felton et al., 2021).

Given the evidence that children and adolescents may influence parental depression, examining adolescents’ extrinsic IER of their parents’ affect is warranted. Although no research has directly examined how youths’ extrinsic IER is associated with parental depression, indirect evidence can be derived from studies using behavioral observations of parent-child emotional discussions. Together, these studies demonstrate that across different affective contexts, there is a bi-directional association between parent and child affective responses over time (e.g., Lougheed et al., 2020; Van Bommel et al., 2018). Moreover, it seems that in dyads with younger adolescents (13–14 years old), it is the parent that leads changes in affect, whereas in dyads in which the adolescent is older (17–18 years old), it is the child that leads affective change in the dyad (Main et al., 2016).

The Effect of Extrinsic IER on the Regulator’s Depression

In addition to examining how extrinsic IER is associated with parent and adolescent depressive symptoms, recent advances in research suggest that regulating others’ affect also impacts the regulator’s own affect (for a review, see Cohen & Arbel, 2020). Thus, extrinsic IER may be associated not only with depressive symptoms in the target, but also in the regulator. Similar to findings regarding the effect of IER on the target of regulation, the effect of extrinsic IER on the regulator seem to be partly dependent on the strategy used. Specifically, using affect-improving extrinsic IER is associated with benefits to the regulator such as improved emotional well-being, heightened positive affect, and feelings of intimacy (e.g., Geiger et al., 2023; Horn et al., 2019; Niven et al., 2012; Springstein et al., 2023). Conversely, when making targets feel worse, regulators exhibited prolonged worse emotional well-being (Niven et al., 2012).

Adolescence as a Sensitive Period for the Development of Extrinsic IER and Depression

Most work examining extrinsic IER and depression before adulthood has focused on infancy and early childhood. However, there are several reasons why examining parent-child extrinsic IER and depression during adolescence is crucial. Although children become increasingly independent in their ability to regulate their own affect over time (Gee, 2020), parents still exert critical influence throughout adolescence (Klimes‐Dougan et al., 2007; Laursen & Collins, 2009). Moreover, adolescence is a sensitive window for the development of regulatory skills (Riediger & Klipker, 2014) and a time of increased risk for the onset of depression (Shorey et al., 2022); parents play an important role in both (Barthel et al., 2018; Gee, 2020; Goodman et al., 2020). Adolescence is also a developmental period in which adolescents’ social-cognitive skills become more developed (Weimer et al., 2021), and parent-child relationships become more egalitarian (Smetana et al., 2006). These changes often lead parents to perceive their children more as equals, and parents may be more likely to rely on adolescents for emotional support and regulation than during previous developmental periods. Therefore, examining parent-child IER during adolescence makes it possible to examine how children regulate their parents’ affect.

Although existing literature on IER has focused mostly on adults, accumulating evidence shows that children and adolescents use extrinsic IER in response to others’ affective expressions from a young age. For example, preschool-aged children use physical comfort (e.g., hugs) to improve others’ negative affect (Persson, 2005). During middle childhood and early adolescence, individuals show increased motivation to use extrinsic IER strategies as well as increased use of affective and cognitive (vs. behavioral) strategies (Gummerum & López-Pérez, 2020; López-Pérez et al., 2016; McCoy & Masters, 1985; Rose & Asher, 2004). One study comparing children between 8 and 10 years old showed that older (vs. younger) children use more affect-improving and less affect-worsening extrinsic IER (López-Pérez & Pacella, 2021). Finally, a study examining adolescents’ extrinsic IER during parent-adolescent conflict discussions (Main et al., 2016) found that older (vs. younger) adolescents responded to their mothers’ affect with more frequent affect-improving extrinsic IER. Taken together, these studies provide preliminary evidence that typically developing youths increasingly improve in their use of extrinsic IER over time.

Gaps in the Literature

Questions remain regarding extrinsic IER and depression during adolescence. First, research on parent-child extrinsic IER has typically focused on either the person providing the regulation (the regulator) or the target, but not both (cf., Main et al., 2016). Further, studies on adolescents have focused almost exclusively on child outcomes; however, IER and depression development are bidirectional, transactional processes (Lougheed et al., 2020; Zaki & Williams, 2013). Thus, adopting a dyadic perspective when researching IER in parent-adolescent dyads is greatly needed.

Second, most studies of extrinsic IER in adolescence have used cross-sectional designs (cf., Schwartz et al., 2018). Moreover, to our knowledge, no studies have examined the associations between extrinsic IER and depression using daily assessments. Structured daily measurement techniques, such as daily diaries, are empirically validated to examine social-emotional processes (Russel & Gajos, 2020). Daily diaries reduce retrospective reporting bias and sampling noise by utilizing multiple assessment points (Trull & Ebner-Priemer, 2020). More importantly, they allow for uncovering fine-grained daily processes that contribute to psychopathology by tracking daily fluctuations in behaviors and symptoms (Russel & Gajos, 2020; Trull & Ebner-Priemer, 2020). Characterizing daily processes may also be particularly useful when designing interventions (Gadassi-Polack et al., 2021a).

Finally, most of the literature on extrinsic IER in parenting has focused on the regulation of negative but not positive affect (Breaux et al., 2022); although, the latter plays a key role in mental health, facilitates social adjustment and connectedness, and enhances recovery from stress (Fredrickson, 2004; Gilbert, 2012; Waugh, 2020). Moreover, anhedonia, defined as difficulties in experiencing pleasure, is one of the hallmark symptoms of depression (APA, 2013). Theoretical accounts of depression have tied anhedonia to difficulties in up-regulating and maintaining positive affect (Joormann & Vanderlind, 2014), and studies have found that depression in general, and anhedonia in particular, are positively associated with ineffective regulation of positive emotions (Werner-Seidler et al., 2013; Vanderlind et al., 2022). To our knowledge, no studies have examined how extrinsic IER of positive affect is associated with depression as the outcome.

The Present Study

The current study addresses the gaps in the literature by utilizing data from the Parent-Adolescent Interpersonal emotion Regulation Study (PAIRS), a 28-day dyadic daily-diary study. The current investigation examines how parents’ and adolescents’ extrinsic IER, in response to both negative and positive affect, is associated with targets’ and regulators’ depressive symptoms. To capture dyadic processes, we analyze the data using the Actor-Partner Interdependence Model framework (APIM; Kenny et al., 2020). The design, sample size, hypotheses, and analytic plan were pre-registered at https://osf.io/c7h98.

Pre-Registered Hypotheses

We hypothesized that:

1. Parents would regulate their children’s affect more frequently than children would regulate their parents’ affect.

2. Extrinsic IER strategy use would be related to target and regulator depressive symptoms. More specifically, we hypothesized that:

   • 2a.Greater use of affect-worsening IER would be associated with higher levels of partner and actor (i.e., regulator and target) depressive symptoms.
   • 2b.Greater use of affect-improving IER would be associated with lower levels of partner and actor (i.e., regulator and target) depressive symptoms.
   • 2c.The effects of parental IER on adolescents would be stronger than the effect of adolescent IER on parents.

Because prior studies found both within-person (longitudinal/experimental; e.g., Schwartz et al., 2018) as well as between-person (cross-sectional; e.g., Breaux et al., 2022) associations between IER and depression, we expected all hypotheses to be confirmed on both within-person and between-person levels. Additionally, considering prior findings showing that age and gender are associated with depression development in adolescence (e.g., Gadassi-Polack et al., 2021c; Rapee et al., 2019; Salk et al., 2016) and that child age is associated with parental and adolescent IER use (Main et al., 2016; McKee et al., 2022), all analyses were repeated with adolescent age and gender as moderators. For brevity’s sake, we report results of significant moderation in supplementary materials only.

Method

The current study uses data from the Parent-Adolescent Interpersonal emotion Regulation Study (PAIRS), a larger investigation of emotional processes in parent-adolescent dyads. Only measures relevant to the current study are described. ​​The study was approved by the Yale University Institutional Review Board (Human Investigation Committee #2000022492).

Participants

Parent-adolescent dyads were recruited via advertisements inviting parents and adolescents to participate in a study examining how they react to each other’s emotions. Advertisements were posted on social networks (e.g., Facebook) and online platforms for participant recruitment (e.g., ResearchMatch) and were emailed to families who had participated in our previous studies. To be eligible to participate, the adolescent had to be 12 to 18 years old and reside with the participating parent for at least half the study period. Both dyad members had to be fluent and literate in English, reside in the Eastern United States, and have daily access to a device connected to the internet. Dyads were excluded from analyses if either dyad member completed fewer than 14 of 28 diary entries. Additionally, given that our study was fully remote, we included procedures to ensure that our participants met recruitment criteria (e.g., residing in the United States; see Data Quality section in the Supplementary Online Materials for more details).

Participants received $15 each for the completion of a background session and an additional $70 if they completed at least 90% of the surveys, $50 if they completed 60–90%, and $10 if they completed fewer than 60%. Participants were compensated via Venmo payments or Amazon gift cards.

One hundred and fifteen dyads met our data quality criteria and started the daily diary. Of those, three dyads (2.6%) were excluded because a dyad member completed fewer than 14 daily diaries. Our final sample consisted of 112 adolescent-parent dyads (i.e., 112 adolescents and 108 parents, as parents were allowed to participate consecutively with different children). Most parents (90.4% of the dyads) reported living with their children 100% of the time. Adolescent ages ranged from 12 to 18 years old (M = 14.13, SD = 1.74, Median = 14), and parent ages ranged from 33 to 64 years old (M = 43.99, SD = 6.03, Median = 43). Sixty-one adolescents (54.5%) identified as female, 49 (43.8%) identified as male, and two (1.8%) identified as “other.” Ninety-six parents (88.9%) identified as female and 12 (11.1%) identified as male. Participants were largely Caucasian (64.3% of the adolescents; 75% of the parents) and non-Hispanic (92% of the adolescents; 91.7% of the parents). Most parents (75.9%) were married. Annual household income was high relative to the US median (2.8% reported < 20k USD; 8.3% reported 20–45k USD; 27.8% reported 45–100k; 39.8% reported 100–200k USD; 21.3% reported > 200k USD). Dyads were residents of the following US states: CT (32%), OH (17.9%), NY (11.6%), FL (5.4%), MD (4.5%), VA (4.5%), PA (4.5%), GA (4.5%), NJ (3.6%), MA (3.6%), MI (2.7%), NC (2.7%), WDC (1.8%), and DE (0.9%). Overall, parents completed a mean of 26.63 diary entries (SD = 2.37, Median = 27, Range = 19–32), and adolescents completed a mean of 25.27 diary entries (SD = 3.45, Median = 26, Range = 14–31); this resulted in 5,442 data points.

Procedure

Data for this study were collected between July 2022 and May 2023. The study started with an onboarding Zoom session with the parent and adolescent. A research assistant explained study procedures and reimbursement structure and reviewed all the daily diaries to ensure comprehension. If both parent and adolescent expressed interest in participating, they provided separate email addresses and chose a time at which they wanted to receive the diary (approximately an hour prior to regular bedtime). Then, participants were sent links to the background survey, including consent and assent forms, demographic questionnaires, and additional measures not used in the current investigation. After both dyad members completed the background survey, they received daily emails with links to the diary every evening for 28 consecutive days. All surveys were completed on a secure website (Qualtrics). Participants were instructed to complete the survey before going to bed each night, ideally at the same time as their dyad member, but independently. The link expired after 14 hours. To enhance compliance, participants received a weekly phone call or text message (based on their preference) from a research assistant to check whether they had any technical difficulties, and a weekly email with information about their completion rate and anticipated compensation. Additionally, participants were contacted if they missed entries for two or more consecutive days. In cases in which a participant missed a diary entry due to technical issues or another reason we deemed justified (e.g., illness), the diary period was extended by a day for both dyad members; this is why the number of completed diary entries could reach more than 28.

Sample Size

A power analysis using the R package EMAtools (Kleiman, 2017) suggested that a sample of 100 individuals who completed at least 75% of the daily diary entries would allow us to detect small effects (d = 0.2) with a power of β = 0.8. While each dyad includes two individuals, we anticipated that some effects may be evident for only one dyad member or only at the dyad level; therefore, we aimed to enroll a total of 100 dyads. We continued to onboard dyads until 100 dyads completed at least 60% of the diary entries.

Transparency and Openness

Study procedures, hypotheses, sample size, and analyses were pre-registered. The pre-registration can be found at https://osf.io/c7h98. Daily diary materials, data, and code pertaining to the current publication can be found at https://osf.io/4823s/.

Measures

Extrinsic IER - Negative Affect

Participants were asked to recall their dyad partner’s worst mood during the day. Participants were then asked to rate on 5-point scales, ranging from not at all to almost all of the time, the degree to which they had used different extrinsic IER strategies in response to their dyad partner’s worst mood. Items were adopted from validated questionnaires examining IER or ER, including Emotions as a Child (EC; Magai & O’Neal, 1997), Difficulties in Emotion Regulation Scale – Positive (DERS-positive; Weiss et al., 2015), Parental Assistance in Child Emotion Regulation (PACER; Cohodes et al., 2022), Coping with Children’s Negative Affect Scale-revised (CCNES-revised; Waslin et al., 2023), and Alexithymia Questionnaire for Children (AQC; Rieffe et al., 2006). Each IER strategy was assessed using the highest-loading item that was fit for both parent and adolescent daily use.

The following IER strategies were assessed: comfort, neglect (EC; Magai & O’Neal, 1997), non-acceptance (DERS-positive; Weiss et al., 2015), problem-solving, positive reappraisal, acceptance, distraction, co-rumination, behavioral disengagement, avoidance, encouragement/venting, expressive suppression (PACER; Cohodes et al., 2022), distress, minimizing, identification, information seeking, help in understanding the situation, self-regulation (CCNES-revised; Waslin et al., 2023), and labeling (reversed item from the AQC; Rieffe et al., 2006). In addition, parents (but not adolescents) were asked about using punishment (CCNES-revised; Waslin et al., 2023). Table S7 presents the mean, standard deviation, and range of use for each strategy across the whole diary period. Table S8 presents the definition for each strategy and the item used to assess it.

To enhance reliability and decrease the number of analyses, we grouped IER strategies as affect-improving (comfort, problem-solving, positive reappraisal, acceptance, distraction, behavioral disengagement, avoidance, encouragement/venting, information seeking, help in understanding the situation, and labeling) or affect-worsening (neglect, non-acceptance, co-rumination, expressive suppression, minimizing, identification, self-regulation). Grouping was done according to our pre-registered plan with the following exceptions: (1) distress was not included as it does not meet any IER definition; (2) punishment was not included as it was assessed only for parents; (3) identification was erroneously pre-registered as affect-improving for both positive and negative affect, whereas it should have been classified as affect-improving for positive affect, but affect-worsening for negative affect (Niven et al., 2009); (4) though not originally assigned as affect-improving or affect-worsening, labeling was assigned as affect-improving (Zaki & Williams, 2013), and self-regulation (defined as non-confrontational withdrawal from the target to allow them space) was assigned as affect-worsening (Niven et al., 2009). We calculated the between- and within-participant reliabilities using procedures outlined by Shrout and Lane (2012). The between-subject reliability coefficient (R1f) is the expected between-subject reliability coefficient for a single typical day. The within-subject reliability (Rc) coefficient is the expected within-subject reliability of change within individuals over the daily-diary entries. For negative-affect-improving IER strategies, the between-person reliabilities were .92 for parents and .94 for adolescents, and within-person reliabilities were .85 for parents and .81 for adolescents. For negative-affect-worsening, the between-person reliabilities were .67 for parents and .74 for adolescents, and within-person reliabilities were .46 for parents and .59 for adolescents. These reliabilities are considered between fair and substantial for within-individual measures (Nezlek, 2017). Within-person analyses used the average daily score; between-person analyses used the average of averages across the whole diary period.

Extrinsic IER - Positive Affect

Participants were asked to recall their dyad partners’ best mood during the day. Participants were then asked to rate on 5-point scales, ranging from not at all to almost all of the time, the degree to which they had used different extrinsic IER strategies in response to their dyad partner’s best mood. Items were adopted from well-validated questionnaires examining IER or ER, including Co-Dampening Co-Enhancing Questionnaire (Bastin et al., 2018), Responses to Positive Affect Questionnaires for Children (Bijttebier et al., 2012), CCNES-revised (Waslin et al., 2023), EC (Magai & O’Neal, 1997), DERS-positive (Weiss et al., 2015), PACER (Cohodes et al., 2022), and AQC (Rieffe et al., 2006). Due to the dearth of measures focusing on IER of positive affect, we also adapted items from measures of IER of negative affect. Each IER strategy was assessed using the highest-loading item that was fit for both parent and adolescent daily use.

The following strategies were assessed: co-dampening and co-enhancing (Co-Dampening Co-Enhancing Questionnaire; Bastin et al., 2018), other-focused positive rumination (Responses to Positive Affect Questionnaire for Children; Bijttebier et al., 2012), non-acceptance (DERS-positive; Weiss et al., 2015), distress, minimizing, information seeking, help in understanding the situation, identification (CCNES-revised; Waslin et al., 2023), neglect (EC; Magai & O’Neal, 1997), encouraging/venting, expressive suppression (PACER; Cohodes et al., 2022), and labeling (AQC; Rieffe et al., 2006). Two strategies (bragging and affection) were added based on discussions between R.G., who is a clinical psychologist working with adolescents, and E.J.G. Table S7 presents the mean, standard deviation, and range of use for each strategy across the whole diary period. Table S8 presents the definition for each strategy and the item used to assess it.

We grouped IER strategies as affect-improving (co-enhancing, other-focused positive rumination, information seeking, help in understanding the situation, identification, encouraging/venting, labeling, affection, bragging) or affect-worsening (co-dampening, non-acceptance, minimizing, neglect, expressive suppression). Grouping was done according to our pre-registered plan with the following exceptions: (1) distress was not included as it does not meet any IER definition; (2) though not originally assigned as affect-improving or affect-worsening, labeling was assigned as affect-improving (Zaki & Williams, 2013). For positive-affect-improving IER, the between-person (R1f) reliabilities were .91 for parents and .90 for adolescents, and within-person reliabilities were .70 for parents and .67 for adolescents. For positive-affect-worsening, the between-person reliabilities were .58 for parents and .70 for adolescents, and the within-person (Rc) reliabilities were .24 for parents and .43 for adolescents. These reliabilities are considered between slight to substantial for within-individual measures (Nezlek, 2017). Within-person analyses used the average daily score; between-person analyses used the average of averages across the whole diary period.

Depressive Symptoms

Parent and adolescent baseline depressive symptoms were measured using the child version of the Short Mood and Feelings Questionnaire (SMFQ; Angold et al., 1995). The SMFQ is a measure in which respondents are presented with 13 statements that reflect various aspects of depression (e.g., I feel miserable or unhappy, I find it hard to think properly or concentrate). To adapt the scale for daily diary use, instead of asking participants to report the presence of the symptoms in the last two weeks, we asked them to rate how they felt at the moment of filling out the diary entry. The scale for rating each statement was, as in the original version, either true, somewhat true, or not true. Scores range from 0 to 26, with higher scores reflecting more depressive symptoms. Studies assessing the psychometric properties of the child version of the SMFQ have found high internal consistency and criterion validity when used in adolescents and young adults (e.g., Angold et al., 1995; Eyre et al., 2021). The between-person reliabilities (R1f) were .85 for parents and .94 for adolescents. The within-person (Rc) reliabilities were .73 for parents and .82 for adolescents. These reliabilities are considered substantial for within-individual measures (Nezlek, 2017). Within-person analyses used the sum daily score; between-person analyses used the average of sums across the whole diary period.

Data Analytic Strategy

To examine the first hypothesis, regarding differences in the frequency of IER use between parents and adolescents, we conducted a series of paired sample t-tests comparing person-level averages across the entire diary period.

The second hypothesis was tested using a series of multilevel regression models that simultaneously examined how regulator and target IER predict target depressive symptoms. More specifically, data were analyzed using the nlme package (Pinheiro et al., 2014) of the statistical programming software RStudio (R Core Team, 2013). Our data have a multilevel structure (days nested within persons, and persons nested within dyads; Bolger & Laurenceau, 2013). Therefore, we used multilevel regression models comprising two levels (a within-individual level and a between-individual level). To address the non-independence inherent to dyadic data, we used the Actor-Partner Interdependence Model (APIM, Kenny et al. 2020). APIM is a data-analytic approach designed specifically to test dyadic effects by simultaneously estimating actor effects (i.e., the effects of the actor’s independent variable scores [e.g., parents’ extrinsic IER] on their own dependent variable score [e.g., parents’ depressive symptoms]), as well as partner effects (i.e., the effects of the other person’s variable scores [e.g., adolescents’ extrinsic IER] on the actor’s dependent variable score [e.g., parent’s depressive symptoms]). In the present manuscript, we refer to actor effects as target effects, and to partner effects as regulator effects.

Within-person-level predictors were centered around person means to make the interpretation of intercepts clearer and to allow testing interaction effects; between-person-level predictors were grand-mean centered. To reduce the concern of reverse causation, within-person (i.e., day-level) analyses included the lagged score (i.e., the previous day’s score) of the outcome variable. To tease apart between-person- and within-person-level effects, our within-person-level analyses also included the person’s mean score of the predictor IER (averaged across the entire study period). Importantly, multilevel models can accommodate missing and non-balanced data.

We ran a series of such models in which depressive symptoms were predicted by both dyad members’ IER. The model used to assess between-person-level results included a random intercept for dyads, as follows¹:

Y_ij (Mean depressive symptoms for person j in dyad i) =

β₀ +

β₁ * (target mean IER use) +

β₂ * (regulator mean IER use) +

β₃ * (Role²) +

β₄ * (Role * target mean IER use) +

β₅ * (Role * regulator mean IER use) +

b_i (random effect for dyad) +

r _ij

The model used to assess within-person-level results included random intercepts and random slopes on the dyad and between-person levels, as follows:

Y_ijk (Depressive symptoms on day k for person j in dyad i) =

(β₀+ b_0i + b_0ij) +

(β₁ + b_1i + b_1ij) * (lagged depressive symptoms [day k-1, mean-centered]) +

β₂ * (parental current depression) +

(β₃+ b_3i + b_3ij) * (target IER on day k [mean-centered]) +

(β₄+ b_4i + b_4ij) * (regulator IER on day k [mean-centered]) +

β₅ * (Role) +

β₆ * (target IER on day k * Role) +

β₇ * (regulator IER on day k * Role) +

β₈ * (mean depressive symptoms) +

β₉ * (mean target IER) +

β₁₀ * (mean regulator IER) +

ε_ijk

To estimate partial effect sizes for specific coefficients, we calculated partial f² values (Selya et al., 2012). f² for each predictor was calculated by first calculating R² for the original model (R²_Orig). Then, we removed the key predictor from the model and calculated R² again (R²_Removed). Finally, we calculated f² as (R²_Orig - R²_Removed) / (1- R²_Orig). Note that in some rare cases, this results in a negative effect size, generally when variables account for so little variance that removing them increases model R² by an infinitesimal amount. This never occurred with any of the significant effects discussed.

We ran all analyses with and without two moderators strongly associated with depressive symptoms: age and gender (Salk et al., 2016). For brevity’s sake, the results are presented without them, and significant moderation models are reported in the supplementary materials.

Results

Descriptive Statistics

Table 1 presents the mean of IER strategy implementation and depressive symptoms separately for parents and adolescents, as well as correlations between the variables. As can be seen in Table 1, parents’ IER strategy use and adolescents’ IER strategy use were positively and significantly correlated.

Table 1.

Means and SDs of IER and depressive symptoms for parents and adolescents, their correlations, and paired-sample t-tests examining the differences between them (N_dyads = 112)

	1	2	3	4	5	6
1. Negative-affect-improving1	.22 *	.65***	.79***	.43***	.19*	.22*
2. Negative-affect-worsening	.62***	.37 ***	.48***	.66***	.23*	.25*
3. Positive-affect-improving	.89***	.50***	.32 ***	.41***	.21*	.07
4. Positive-affect-worsening	.48***	.76***	.39***	.36 ***	.22*	.22*
5. Depressive symptoms (regulator)	.11	.44***	−.01	.48***	−.01	--
6. Depressive symptoms (target)	−.02	.09	−.04	.14	--	---
Parents: Mean (SD)	1.03 (0.75)	0.51 (0.35)	1.08 (0.72)	0.15 (0.22)	1.85 (2.16)	--
Adolescents: Mean (SD)	0.63 (0.70)	0.50 (0.43)	0.80 (0.66)	0.32 (0.37)	4.24 (4.70)	--
T	4.74***	0.19	3.73***	-5.05***	-4.87***	--
Cohen’s d	0.45	0.02	0.35	0.48	0.46	--

¹Correlations for parents are presented above the diagonal; correlations for adolescents are presented below the diagonal. The diagonal presents parent-adolescent correlations.

^*p < .05;

^**p < .01,

^***p < .001

Hypothesis 1: Parents would regulate adolescent affect more frequently than adolescents would regulate parent affect

As can be seen in Table 1, the results partially support our hypothesis. Parents used positive- and negative-affect-improving strategies more frequently than their children. However, adolescents used more positive-affect-worsening strategies compared to their parents. No difference was found in the use of negative-affect-worsening strategies.

Hypothesis 2: IER would be associated with regulator and target depressive symptoms

Results for sub-hypotheses are elaborated below. Please note that estimates for adolescents are presented in the Tables, whereas estimates for parents are presented in the text.

Hypothesis 2a: Greater use of affect-worsening IER would be associated with higher levels of regulator and target depressive symptoms

Negative-Affect-Worsening IER Strategies

Between-Person Level.

As seen on the right side of Table 2b, adolescents’ higher depressive symptoms were predicted by their own use of affect-worsening IER strategies, but not by their parents’ affect-worsening IER strategies. Parents’ depressive symptoms were not predicted from parents’ (β = 1.39, SE = 0.98, t(107) = 1.42, p = .159, 95% CI [−0.55, 3.34], f² = .009) or adolescents’ (β = 0.04, SE = 0.80, t(107) = 0.05, p = .957, 95% CI [−1.55, 1.63], f² < .001) use of negative-affect-worsening IER. Thus, adolescents who used more negative-affect-worsening IER strategies had higher depressive symptoms.

Table 2.

Predicting adolescents’ depressive symptoms from extrinsic IER of negative and positive affect: Between-person-level APIM results (N_dyads = 112)

	Extrinsic IER of negative affect
	(a) Affect-improving						(b) Affect-worsening
	β	SE	df	t	CI	f2	β	SE	df	t	CI	f2
Intercept	4.06	0.37	111	10.90***	3.32, 4.81	.165	4.25	0.32	111	13.41***	3.62, 4.88	.337
Role1	−2.38	0.54	107	−4.43***	−3.45, −1.31	.094	−2.41	0.46	107	−5.20***	−3.32, −1.49	.132
Target’s IER	0.42	0.50	107	0.84	−0.57, 1.41	.003	4.39	0.80	107	5.47 ***	2.80, 5.98	.138
Regulator’s IER	1.28	0.46	107	2.76 **	0.36, 2.21	.035	1.38	0.98	107	1.41	−0.56, 3.33	.009
Role x Target’s IER	0.18	0.68	107	0.27	−1.16, 1.53	< .001	−2.99	1.22	107	−2.39 *	−5.48, −0.52	.026
Role x Regulator’s IER	−1.50	0.68	107	−2.21 *	−2.85, −0.16	.022	−1.34	1.26	107	−1.07	−3.82, 1.14	.005
		Extrinsic IER of positive affect
	(c) Affect-improving						(d) Affect-worsening
	β	SE	df	t	CI	f2	β	SE	df	t	CI	f2
Intercept	4.14	0.37	111	11.28***	3.41, 4.87	.113	3.82	0.36	111	10.72***	3.12, 4.53	.264
Role	−2.46	0.52	107	−4.68***	−3.50, −1.42	.103	−1.84	0.52	107	−3.50***	−2.88, −0.80	.061
Target’s IER	−0.19	0.56	107	−0.34	−1.29, 0.91	.001	5.87	0.90	107	6.51 ***	4.08, 7.66	.196
Regulator’s IER	0.49	0.51	107	0.98	−0.51, 1.50	.004	1.01	1.54	107	0.65	−2.04, 4.06	.002
Role x Target’s IER	0.95	0.75	107	1.27	−0.54, 2.44	.007	−3.88	1.76	107	−2.20 *	−7.37, −0.39	.022
Role x Regulator’s IER	−0.91	0.75	107	−1.21	−2.40, 0.58	.007	−0.63	1.76	107	−0.36	−4.12, 2.86	.001

¹Role is coded as “0” for adolescent and “1” for parent. Therefore, default estimates in the table are those in which adolescents are the targets, and the outcome predicted is theirs.

^*p < .05;

^**p < .01,

^***p < .001

Within-Person Level.

Table 3b presents within-person-level results for negative-affect-worsening IER strategies. As predicted, adolescents experienced increases in depressive symptoms on days in which they used more negative-affect-worsening IER strategies compared to what they usually do. However, adolescents’ depressive symptoms did not change with their parents’ use of negative-affect-worsening IER. Parents experienced increases in depressive symptoms on days in which they used more negative-affect-worsening IER compared to what they usually do (β = 0.50, SE = 0.13, t(4476) = 3.93, p < .001, 95% CI [0.25, 0.74], f² = .002). However, parents’ depressive symptoms did not change with their adolescents’ use of negative-affect-worsening IER (β = 0.06, SE = 0.08, t(4476) = 0.82, p = .409, 95% CI [−0.09, 0.22], f² < .001). Thus, on days in which parents and adolescents used more negative-affect-worsening IER strategies, they themselves experienced more depressive symptoms.

Table 3.

Predicting adolescents’ depressive symptoms from extrinsic IER of negative affect: Within-person-level APIM results (df_person = 108, df_day = 4476)

	(a) Affect-improving					(b) Affect-worsening
	β	SE	T	CI	f2	β	SE	t	CI	f2
Intercept	−0.05	0.08	−0.71	−0.20, 0.10	<.001	−0.03	0.06	−0.44	−0.15, 0.10	<.001
Role1	−0.02	0.07	−0.26	−0.61, 0.12	<.001	−0.03	0.06	−0.59	−0.15, 0.08	<.001
Lagged depression	0.16	0.03	5.68***	0.10, 0.21	.076	0.27	0.03	9.34***	0.21, 0.33	.096
Mean depression	0.99	0.01	105.51***	0.97, 1.01	2.883	0.99	0.01	116.46***	0.97, 1.01	2.742
Mean Target IER	0.02	0.04	0.39	−0.06, 0.09	<.001	0.01	0.08	0.16	−0.14, 0.16	<.001
Mean Regulator IER	0.03	0.04	0.70	−0.05, 0.11	<.001	0.01	0.07	0.18	−0.12, 0.15	<.001
Target’s daily IER	0.01	0.17	0.07	−0.32, 0.34	<.001	1.08	0.15	7.18 ***	0.78, 1.37	.028
Regulator’s daily IER	0.35	0.15	2.34 *	0.06, 0.65	.002	0.34	0.22	1.57	−0.08, 0.77	.002
Role x Target’s daily IER	0.23	0.16	1.47	−0.08, 0.54	<.001	−0.60	0.17	−3.44 ***	−0.94, −0.26	.007
Role x Regulator’s daily IER	−0.34	0.16	−2.17 *	−0.66, −0.03	<.001	−0.30	0.24	−1.26	−0.76, 0.17	<.001

¹Role is coded as “0” for adolescent and “1” for parent. Therefore, default estimates in the table are those in which adolescents are the targets, and the outcome predicted is theirs.

^*p < .05;

^**p < .01,

^***p < .001

Positive-Affect-Worsening IER Strategies

Between-Person Level.

As seen on the right side of Table 2d, adolescents’ depressive symptoms were positively associated with their own use of positive-affect-worsening IER but were not associated with their parents’ positive-affect-worsening IER. Parents’ depressive symptoms were not associated with parents’ (β = 1.98, SE = 1.54, t(107) = 1.29, p = .200, 95% CI [−1.07, 5.04], f² = .008) or adolescents’ (β = 0.38, SE = 0.90, t(107) = 0.42, p = .677, 95% CI [−1.41, 2.16], f² = .001) use of positive-affect-worsening IER. Thus, adolescents who used more positive-affect-worsening IER strategies had more depressive symptoms.

Within-Person Level.

Table 4b presents within-person-level results for positive-affect-worsening IER strategies. As predicted, adolescents experienced increases in depressive symptoms on days in which they used more positive-affect-worsening IER compared to what they usually do. However, adolescents’ depressive symptoms did not change with their parents’ use of positive-affect-worsening IER. Parents experienced increases in depressive symptoms on days in which they used more positive-affect-worsening IER compared to what they usually do (β = 0.74, SE = 0.32, t(4477) = 2.29, p = .021, 95% CI [0.11, 1.37], f² < 0). However, parents’ depressive symptoms did not change with their adolescents’ use of positive-affect-worsening IER (β = −0.02, SE = 0.10, t(4476) = −0.22, p = .826, 95% CI [−0.23, 0.18], f² < 0). Thus, on days in which parents and adolescents used more positive-affect-worsening IER strategies, they themselves experienced more depressive symptoms.

Table 4.

Predicting adolescents’ depressive symptoms from extrinsic IER of positive affect: Within-person-level APIM results (df_person = 108, df_day = 4477)

	(a) Affect-improving					(b) Affect-worsening
	β	SE	T	CI	f2	β	SE	t	CI	f2
Intercept	−0.05	0.07	−0.76	−0.19, 0.08	<.001	−0.03	0.08	−0.39	−0.18, 0.12	<.001
Role1	−0.02	0.06	−0.42	−0.15, 0.09	<.001	−0.03	0.07	−0.40	−0.18, 0.12	<.001
Lagged depression	0.29	0.03	9.09***	0.22, 0.35	.1	0.13	0.02	5.74***	0.08, 0.17	.068
Mean depression	0.99	0.01	120.94***	0.97, 1.01	2.957	0.99	0.01	91.40***	0.97, 1.01	2.495
Mean Target IER	0.02	0.04	0.46	−0.06, 0.09	<.001	0.04	0.13	0.30	−0.22, 0.29	<.001
Mean Regulator IER	0.01	0.04	0.24	−0.07, 0.09	<.001	0.01	0.10	0.09	−0.19, 0.21	<.001
Target’s daily IER	−0.31	0.14	−2.13 *	−0.60, −0.02	.002	0.82	0.22	3.69 ***	0.39, 1.26	.014
Regulator’s daily IER	0.02	0.20	0.12	−0.37, 0.42	<.001	0.27	0.28	0.96	−0.28, 0.81	<.001
Role x Target’s daily IER	0.35	0.16	2.13 *	0.03, 0.67	.001	−0.30	0.30	−0.98	−0.90, 0.30	.001
Role x Regulator’s daily IER	−0.13	0.23	−0.56	−0.58, 0.33	<.001	−0.26	0.30	−0.87	−0.84, 0.32	<.001

¹Role is coded as “0” for adolescent and “1” for parent. Therefore, default estimates in the table are those in which adolescents are the targets, and the outcome predicted is theirs.

^*p < .05;

^**p < .01,

^***p < .001

Hypothesis 2b: Greater use of affect-improving IER would be associated with lower levels of regulator and target depressive symptoms

Negative-Affect-Improving IER Strategies

Between-Person Level.

As seen on the left side of Table 2a, adolescents’ depressive symptoms were positively associated with their parents’, but not their own, higher use of negative-affect-improving IER. Parents’ depressive symptoms were not associated with their own (β = 0.60, SE = 0.46, t(107) = 1.29, p = .198, 95% CI [−0.32, 1.52], f² = .008), or their adolescents’ (β = −0.22, SE = 0.50, t(107) = −0.44, p = .660, 95% CI [−1.21, 0.77], f² = .001) use of more negative-affect-improving IER. Thus, parents who used more affect-improving IER in response to their children’s negative affect had adolescent children with more depressive symptoms, contrary to our prediction.

Within-Person Level.

Table 3a presents within-person-level results for negative-affect-improving IER strategies. Contrary to our predictions, adolescents experienced increases in depressive symptoms on days in which their parents used more positive-affect-improving IER strategies compared to what they usually do. However, adolescents’ depressive symptoms were unrelated to their own IER use. Parents experienced increases in depressive symptoms on days in which they used negative-affect-improving IER compared to what they usually do (β = 0.22, SE = 0.06, t(4476) = 3.52, p < .001, 95% CI [0.10, 0.34] , f² = .002), but did not change with their adolescents’ use of affect-improving IER strategies (β = 0.01, SE = 0.08, t(4476) = 0.12, p = .905, 95% CI [−0.15, 0.17] , f² < 0). Thus, on days in which parents used more negative-affect-improving IER strategies, they and their children experienced more depressive symptoms.

Positive-Affect-Improving IER Strategies Predicting Depressive Symptoms

Between-Person Level.

As seen on the left side of Table 2c, neither adolescents’ nor parents’ positive-affect-improving IER strategy use predicted adolescent or parent depressive symptoms (parent IER effect on parent depressive symptoms β = 0.76, SE = 0.51, t(107) = 1.50, p = .136, CI [−0.24, 1.76] , f² = .01; adolescent IER effect on parent depressive symptoms β = −0.42, SE = 0.56, t(107) = −0.75, p = .456, 95% CI [−1.52, 0.68] , f² = .002). Thus, the use of positive-affect-improving IER strategies was not linked to depression.

Within-Person Level.

Table 4a presents within-person-level results for positive-affect-improving IER strategies. As predicted, adolescents experienced decreases in depressive symptoms on days in which they used affect-improving IER strategies more than they usually do. Adolescents’ depressive symptoms did not change with parents’ use of positive-affect-improving IER strategies. Parents’ depressive symptoms were not associated with their own (β = 0.02, SE = 0.08, t(4477) = 0.27, p = .784, 95% CI [−0.13, 0.18], f² < 0) or their adolescent’s (β = −0.05, SE = 0.09, t(4477) = −0.60, p = .549, 95% CI [−0.24, 0.13], f² < .001) use of affect-improving IER. Thus, on days in which adolescents used more positive-affect-improving IER strategies, they themselves experienced fewer depressive symptoms.

Hypothesis 2c: The effects of parental IER on adolescents would be stronger than the effect of adolescent IER on parents

To test the hypothesis that parental IER would have a stronger effect compared to adolescent IER on the dyad partner’s depressive symptoms, we examined the interaction effects between relationship role and regulator effects in all models run for hypotheses 2a and 2b. As can be seen in Tables 2a and 3a, the interaction between role and regulator negative-affect-improving IER was significant and negative at both between- and within-person levels. This shows that, as expected, parents’ effects on their adolescents were larger than adolescents’ effects on their parents. Importantly, although no other interactions with regulator IER were significant, none of the models showed that adolescents’ IER was associated with their parents’ depressive symptoms – only parents’ IER was associated with adolescents’ depressive symptoms.

Finally, as can be seen in Tables 2b and 3b, the significant role X target IER interaction indicates that the effect of adolescents’ negative-affect-worsening IER on adolescents’ depressive symptoms was significantly smaller than parents’ negative-affect-worsening IER’s effect on parents’ depressive symptoms, at both between- and within-person levels. Additionally, as can be seen in Table 2d, the significant role X target IER interactions shows that adolescents’ association between positive-affect-worsening IER and adolescents’ depressive symptoms was stronger than parents’ positive-affect-worsening IER and parents’ depressive symptoms.

Moderation by Age and Gender of the Adolescent and Additional Analyses

We examined age and gender differences in all study variables as well as age and gender moderation of all models. Because there were few significant results, we elaborate on them in the supplementary materials. Additionally, although we did not pre-register these analyses, we ran two additional model types for completeness. First, we ran models using lagged extrinsic IER as predictors. Only one effect was significant; therefore, we report these analyses in the supplementary materials only (Tables S1 and S2). Second, because we were surprised by the small number of regulator effects on the target, we ran all models examining the association between regulator IER and target depression only (as regulator and target IER were highly correlated), instead of the full APIM model. A few more regulator effects emerged in this (vs. the APIM) analysis. See supplementary materials Tables S3–S5 for the results of these models.

Discussion

The current study investigated whether extrinsic interpersonal emotion regulation (IER) is associated with depressive symptoms in parent-adolescent dyads. Using data from the Parent-Adolescent Interpersonal emotion Regulation Study (PAIRS), we explored whether parent and adolescent extrinsic IER are associated with the other person’s depressive symptoms (i.e., regulator effects on target) as well as with one’s own depressive symptoms (i.e., regulator effects on self). We examined these questions separately for negative and positive affect and for affect-worsening and affect-improving IER strategies. Our results show that extrinsic IER is associated with the regulator’s and target’s depressive symptoms, and that the nature of this association is different depending on the valence of the affect regulated (positive vs. negative), the type of strategy used (affect-improving vs. affect-worsening), the relational role (parent vs. adolescent), and whether you are the regulator or the target. Five major findings emerge from our analyses.

(1). Evidence for the Costs of Affect-Worsening IER is Robust

As predicted, we found that the use of more affect-worsening IER strategies was associated with higher levels of depressive symptoms. Specifically, adolescents who used more (vs. less) affect-worsening IER strategies in response to their parents’ negative and positive affect were more depressed on average; these effects were medium-sized. Similarly, on days in which adolescents and parents used negative- and positive-affect-worsening IER strategies more than they usually do, their own depressive symptoms increased; these effects were small for adolescents and small to negligible for parents. These results are consistent with prior findings that affect-worsening IER strategies are related to adverse effects in the regulator (e.g., Hale et al., 2023; Niven et al., 2012).

We extend prior research in several ways. First, unlike most existing literature, which focuses on negative affect, we present findings pertaining to the regulation of both negative and positive affect. Examining the regulation of positive affect is extremely important considering that deficits in positive affect play a central role in the emergence and maintenance of depression (Gilbert et al., 2012). Second, the present study is one of the first to examine extrinsic IER in adolescents, particularly in the context of the parent-adolescent dyads (cf., Main et al., 2016). Despite adolescents’ increasing independence from parents, our results show that parents’ behaviors (i.e., IER), as well as adolescents’ behaviors towards their parents, play a role in adolescent depression. Interestingly, the association between affect-worsening IER and one’s own depression were found both on the between- and within-person levels for adolescents, but only on the within-person level for parents. Indeed, effect sizes for target effects of affect-worsening IER were almost always lower for parents. This suggests that the costs of affect-worsening IER are both larger and accumulate for adolescents, whereas parents (at least in a community sample) can “shake off” the negative effects of affect-worsening IER.

An important question is why the use of affect-worsening strategies was related to the regulators’ depressive symptoms. Theoretical models of IER suggest that when another person’s affect worsens, the regulator may be impacted through the process of emotion contagion (Zaki & Williams, 2013). Another explanation is that the mere perception of changing another’s affect can change how we feel about ourselves (Zaki & Williams, 2013). For example, a person may feel guilty for making a loved one feel worse. However, as we did not inquire about participants’ regulation goals or the perceived effectiveness of their IER, we cannot determine whether they were aware that these strategies were affect-worsening. Indeed, prior research on co-rumination (which is considered affect-worsening IER) shows that people implement it because they believe it is supportive (Rose et al., 2007).

(2). Negative-Affect-Improving IER Seems to Backfire

In contrast to our hypotheses, we found that parents’ higher use of extrinsic negative-affect-improving IER strategies was associated with higher levels of adolescent depressive symptoms at the between- and within-person levels. Moreover, our within-person-level analysis found that on days in which parents used more negative-affect-improving IER, parents themselves were more depressed. It should be noted that the effect sizes for these effects are small on the between-person level and negligible at the within-person level, suggesting that even when accounting for their adverse effects, affect-improving strategies are at least better than affect-worsening ones. Even so, these results were surprising, as literature typically ties parents’ negative-affect-improving strategies to better child outcomes (e.g., Waslin et al., 2023; though cf., Viana et al., 2016). It is worth noting, however, that our finding that using more negative-affect-improving IER is related to parents’ higher depression is consistent with literature showing that continuously responding to another’s distress may lead to extensive strain on the regulator, including heightened depressive symptoms (e.g., Ma et al., 2022; Mottaghi et al., 2020; for a review, see Cohen & Arbel, 2020).

As between-person-level results are aggregated over days, it is possible that this positive association emerges because parents try harder to cheer up children who are more depressed. However, our within-person-level results go against this interpretation. First, while parents might also try to cheer up depressed children using positive-affect-improving IER, the use of such strategies by parents was not tied to adolescent depression, reducing the likelihood of this interpretation. Within-person models controlling for adolescent depression on the previous day suggested that the use of these strategies is not caused by adolescent depression on previous days. However, controlling for lagged outcomes cannot preclude the possibility that adolescent depression or other confounding factors (i.e., third variables) that occurred on that specific day could have causally influenced IER strategies (Rohrer & Murayama, 2023). Still, we would expect such effects to also increase positive-affect-improving IER, which again was not found. Thus, though unexpected, these results seem robust. Taken together, these findings suggest that attempts to down-regulate negative affect may backfire; however, future research using experimental designs is needed to determine causality.

There are several explanations for the adverse effects of negative-affect-improving IER strategies. First, it is possible that the mere focus on adolescents’ negative affect makes it more salient, thereby exacerbating adolescents’ depression. Second, it is possible that our findings diverge from previous literature because of our unique design. PAIRS is the first study to examine associations between parental extrinsic IER and adolescent depression using a daily-diary design. Perhaps daily processes differ from long-term consequences, even if averaged over a month; thus, we suggest that although affect-improving IER strategies were problematic in the short-term, their long-term effects may go in an opposite direction. A multiple-time-scale design (e.g., Gadassi-Polack et al., 2021c; Ram & Diel, 2014) is needed to examine this explanation. Finally, as adolescence is a time in which striving for autonomy is central, it is possible that regulation attempts were perceived as impinging on that developmental goal.

(3). Adolescents Benefit From Their Own Positive-Affect-Improving IER

As predicted and previously found (Breaux et al., 2022), there was a negative association between positive-affect-improving IER strategies and depressive symptoms. Interestingly, this association was found only for adolescents, and only at the within-person level. These findings are in line with previous literature suggesting that regulators benefit from regulating others as it boosts their self-esteem when feeling helpful (Cohen & Arbel, 2020), and extends this finding from adults to adolescents. It should be noted that these beneficial effects were small and did not seem to accumulate over time. Still, this is an interesting direction for further investigation, as most interventions focus on parents’ IER (e.g., Kehoe et al., 2014), but these findings suggest that perhaps adolescents’ IER should be targeted.

We did not find that parents benefit from their own or their children’s IER. In fact, we found that parents’ attempts to regulate their children predict their own increased daily depressive symptoms. These findings highlight the potential emotional burden of parenting (Vertsberger et al., 2022), an important topic that has not received enough attention. However, we do not conclude that parents necessarily only suffer from regulating their children. The effect sizes were small, and effects were limited to within-person levels, suggesting parents’ depressive symptoms do not accumulate over time. Moreover, it is important to consider that IER is associated with relational – and not only affective – outcomes. Indeed, prior research in adults shows that IER is associated with increased feelings of intimacy (Horn et al., 2019). Thus, it is quite likely that although parents’ IER was not associated with better outcomes for parental depressive symptoms, parents’ and their children’s IER may have contributed to feeling better regarding the parent-child relationship. Future studies examining relational outcomes of IER are needed.

(4). Use of IER is More Consistently Associated with Regulator (vs. Target) Outcomes

Our use of the APIM framework seemed to uncover that the associations between IER and outcome were stronger for the regulator – more than for the target of regulation. Thus, regulator effects were significant in more cases, and even when both effects were significant, effect sizes were larger for regulators’ effects on themselves (vs. on targets). There are several possible explanations for this finding. First, each dyad member reported the IER they provided but not the IER they received. It is possible that discrepancies in perception and report played an important role; for example, it is likely that although parents reported using IER, the adolescents did not notice that behavior. Indeed, dyadic studies often show such discrepancies (e.g., Sened et al., 2017). Future studies are needed to explore this explanation. Another explanation for the dearth of regulator effects (i.e., the effects of the regulator on the target) is the collinearity between target and regulator IER. In line with this explanation, additional analyses were conducted with only regulator IER as predictors, which revealed additional regulator effects. This finding highlights the importance of using the APIM framework, as it suggests that the associations previously found between parental IER and adolescent depression may have been mediated by adolescent IER. Additional research is needed to clarify this point.

Taken together, our results highlight the complexity of extrinsic IER. Theoretical models have claimed that extrinsic IER’s effects on the regulator depend on multiple factors (e.g., the characteristics of the regulator; Cohen & Arbel, 2020). Our results further elaborate on this model as we simultaneously examine associations between IER and depressive symptoms of both the regulator and the target. We show that the valence of the affect regulated (positive vs. negative), type of strategy (affect-worsening vs. affect-improving), the relational role of both the regulator and target (parent vs. adolescent), and whether you are the regulator or the regulated all moderate the impact of extrinsic IER. This complex picture is highly reminiscent of recent literature on intrinsic (i.e., intrapersonal) emotion regulation that argues against categorizing strategies as “good” or “bad” and suggests that the result of emotion regulation depends on multiple factors (Sheppes, 2020). The current investigation is an important first step in delineating moderators of extrinsic IER.

(5). Parents (vs. Adolescents) Use More Affect-Improving and Less Affect-Worsening IER

To our knowledge, the present study is the first to compare parents’ and adolescents’ extrinsic IER strategy use. We expected parents to regulate their children’s affect more frequently than adolescents regulate their parents’ affect. However, our results uncovered a more nuanced picture. Parents used more extrinsic IER strategies compared to their children – but only affect-improving ones. In fact, adolescents used more positive-affect-worsening IER strategies compared to their parents, and there were no differences between parents’ and adolescents’ use of negative-affect-worsening IER strategies. One interpretation, in line with prior research showing that the motivation and ability to use extrinsic IER improves with age (e.g., López-Pérez & Pacella, 2021), is that parents are more skilled in extrinsic IER compared to their children. Relatedly, our prediction that parents would affect adolescents more than the other way around was partly supported. Specifically, regulator effects on targets were found only when parents were the regulators. Taken together, we deduce that parents may be more motivated to regulate their children than vice versa because they feel responsible for their children’s well-being. Future research is needed to examine these interpretations and perhaps examine them in other types of hierarchical relationships (e.g., patient-therapist, employee-boss).

Strengths, Limitations, and Future Directions

The present study is the first to examine the associations between extrinsic IER and depressive symptoms using a dyadic perspective. Considering that extrinsic IER is an essentially dyadic process (Zaki & Williams, 2013) and that depression development is a highly transactional process (Felton et al., 2021), our results address an important gap in the literature and shed unique light on how parents’ and adolescents’ use of extrinsic IER is related to adolescent and parent depression. Relatedly, this is one of the first studies to examine extrinsic IER from early to late adolescence (cf., Main et al., 2016), as prior investigations examined either earlier or later periods in development. Considering that adolescence is a time of dramatic social-emotional change, our findings shed unique light on this developmental period. Finally, our use of an intensive longitudinal design increases the ecological validity of our findings and, therefore, their clinical relevance.

The current research has some limitations that should be acknowledged. First, we rely on a community sample, thus limiting the generalizability of our results to clinical samples. Future studies should examine these effects in clinical or high-risk samples. Second, although daily diaries have much higher temporal resolution than cross-sectional and typical longitudinal studies, this design still limited our ability to detect effects that unravel even more quickly (e.g., within a single day). Future studies with higher temporal resolutions are needed. Third, as we did not use an experimental design, our ability to deduce causality is limited. Although day-level analyses control for reverse-causality and reduce the risk of third-variable explanations, future studies that manipulate extrinsic IER in parent-adolescent dyads (e.g., intervention studies) are needed to provide evidence of causality. Fourth, our classification of IER strategies as affect-improving or affect-worsening was based on typology from Niven et al. (2009). However, that typology was based on a single assessment and may not apply to less sensitive longitudinal measures. Moreover, given the criticism of classifying strategies as “good” or “bad” (Sheppes, 2020), as well as the complexity revealed by the present study’s findings, future studies should consider alternative typologies. Indeed, some of the within-person reliabilities were low, which may have limited our ability to detect significant effects. Finally, our analyses use mixed linear models, which have been the model of choice for years for analyzing dyadic diary data (Bolger & Laurenceau, 2013). In recent years, researchers have developed ways to analyze such data using dynamic structural equation models (Savord et al., 2023), which allow for various advantages, including better treatment of missing data and modeling of latent variables. Future studies should consider using such models as the body of research on their use for modelling dyadic time series data becomes consolidated.

Clinical Implications

Our finding that the use of extrinsic IER strategies has costs for the regulating parent should be recognized by clinicians. Helping parents deal with the emotional burden of parenting could reduce parental burnout and improve mental health outcomes for both parent and child. Additionally, even though adolescence is a time of increased independence from parents, our results show that parental behaviors, and specifically how parents regulate their children’s affect, are still associated with adolescents’ depressive symptoms. Interestingly, we found that when adolescents try to regulate their parents’ affect, their own depressive symptoms change – for better or worse, depending on the valence of the affect regulated and the types of IER strategies employed. Prior literature had mostly warned against situations in which children attempt to care for their parents emotionally (Miller, 1981) and tied this type of parent-child “role-reversal” with parental unavailability or psychopathology (e.g., Florange et al., 2019). Our findings extend these theories by showing that these costs may be evident even in a community sample. Importantly, we also challenge these theories, as we show that at times, adolescents may also benefit from trying to regulate their parents’ affect (e.g., when trying to up-regulate parents’ positive affect).

Summary

The present study is an innovative investigation on how parents and adolescents regulate each other, and how those regulatory efforts are associated with both parent and adolescent depressive symptoms. To examine this question, we used data from PAIRS, a unique dyadic daily diary study of parents and adolescents. Our results suggest that both parents’ and adolescents’ extrinsic IER are associated with adolescents’ depressive symptoms, but only parents’ extrinsic IER is associated with parents’ depressive symptoms. Specifically, the use of positive- and negative-affect-worsening IER strategies were associated with higher depressive symptoms in the regulator (regardless of whether the regulator is the parent or adolescent), whereas the use of more negative-affect-improving IER and less positive-affect-improving IER was associated with more adolescent depression. These results greatly advance existing literature, highlight the benefits of a dyadic perspective in the study of both depression and IER, and show that the relational role of the regulator and target, the valence of the affect regulated, and the type of strategy used all moderate the association between IER and depression.