It’s a balancing act: The ratio of maladaptive (vs. all) emotion regulation strategies is associated with depression

Abstract

Research on emotion regulation (ER) has increasingly recognized that people use multiple strategies simultaneously, often referred to as ER repertoire. Prior research found that ER repertoire is associated with psychopathology, but results have been mixed. Indeed, research from recent years suggests that it is the quality of ERs, but than their quantity, that needs to be considered. Based on the combination of the literatures on ER repertoire, polyregulation, and ER flexibility, we propose a novel metric: the ratio of using putatively maladaptive (vs. all) ER strategies. Using this metric, we examine (1) maladaptive ER ratio changes during the transition to adolescence, a developmental period in which the prevalence of depression sharply increases, and (2) how changes in maladaptive ER ratio are associated with depressive symptoms. One-hundred and thirty-nine youths (baseline age: 8–15) reported ER strategies and depression daily for 21 days. One year later, 115 completed another 28-day daily-diary (N_assessments= 5,631). Our results show that almost all youth use at least some maladaptive ERs. Importantly, maladaptive ER ratio decreases over a year of adolescence for most youths. Conversely, an increased maladaptive ER ratio predicted depression increases on the daily and on the yearly level. These results shed light on typical and atypical development of ER flexibility and emphasize the need to consider the balance between ERs in relation to psychopathology.

Depression is a highly prevalent psychiatric illness, with tremendous costs to individuals and society (Greenberg et al., 2015). Rates of depression increase dramatically during adolescence, with the incidence of depression rising steeply around ages 14–15 (Salk et al., 2016). Adolescent-onset of depression is associated with increased risk of suicide and a more severe illness course (Thapar et al., 2012). Thus, it is important to identify risk factors before the escalation of depression (Gonçalves et al., 2019). Emotion regulation (ER) – the processes that influence the frequency, intensity, duration, and expression of emotional experiences (Gross, 2014) – can serve as risk, maintenance, or protective factors for depression (Schäfer et al., 2017; van den Huevel et al., 2020).

The transition to adolescence brings with it increases in academic and social challenges that coincide with an imbalance between increased emotionality and protracted development of regulatory circuits (Casey et al., 2019). Furthermore, it is also a time in which children and adolescents (henceforth: youth) are expected to become more independent in their ER, in contrast to infancy and childhood, when regulation largely relies on caregivers (Eisenberg et al., 2010). Indeed, it is hypothesized that the increase in ER demands is related to the higher rates of depression during adolescence (Rapee et al., 2019). There is substantial literature tying ER to depression in youth (Schäfer et al., 2017). Most of the research focused on specific ER strategies and categorized them as “adaptive” or “maladaptive.” Adaptive strategies are considered effective in regulating affect (i.e., decrease negative affect and increase or sustain positive affect), whereas maladaptive strategies are considered ineffective (i.e., sustain or increase negative affect and decrease positive affect). Research on youth shows that maladaptive ERs are related to increased risk for depression (e.g., rumination, dampening; Gentzler et al. 2013; Nelis et al. 2016; Young et al., 2019), whereas adaptive ERs are usually found to be related to decreased risk for depression (e.g., problem-solving, savoring; McMahon & Naragon-Gainey, 2019). However, in the past few years, there has been much criticism on the categorization of ER strategies as inherently “adaptive” or “maladaptive,” and instead started focusing on ER flexibility – i.e., the ability to implement ER strategies in a way that is synchronized with situational demands (Aldao et al., 2015; Bonnano & Burton, 2013).

Moreover, research on ER has increasly recognized that ER strategies do not work in isolation – that individuals, in fact, have a repertoire of ER strategies, and they often use them simultaneously. Recent models of ER refer to this phenomenon as emotion polyregulation – i.e., the observation that ER strategies are often used simultaneously or in close succession (Ford et al., 2019; Gadassi Polack et al., 2021b). ER repertoire is defined as the range of different ER strategies that an individual utilizes across situations (Grommisch et al., 2019). Similarly, emotion polyregulation is the range of strategies used during a specific emotional episode (Ladis et al., 2023). Research has shown that individuals differ in the size of their ER repertoire as well as in their repertoire components. These differences are related to their risk for psychopathology, including depression (e.g., Grommisch et al., 2019; Ladis et al., 2023; Lougheed & Hollenstein, 2012; Southward & Cheavens, 2020; van den Huevel., 2020; Wen et al., 2021).

ER repertoire has been operationalized using different methods. Some studies simply summed up the number of ER strategies a person uses (e.g., Aldao & Nolen-Hoeksema, 2013; Bonanno & Burton, 2013; Heiy & Cheavens, 2014; Ladis et al., 2023; Quiñones-Camacho & Davis, 2018). The disadvantage of this method is that it treats all ER strategies as interchangeable. Indeed, research based on this operationalization has led to mixed results, with some studies reporting a positive association between ER repertoire size and mental health (Quiñones-Camacho & Davis, 2018), others reporting a negative association (Aldao & Nolen-Hoeksema, 2013), and some reporting no association (Heiy & Cheavens, 2014; Ladis et al., 2023). Relatedly, a study that examines ER diversity (which is similar to ER repertoire but also takes into account the frequency of individual strategy use; Wen et al., 2021), found that those with current and remitted depression have larger ER diversity compared to healthy controls. However, in an in-depth examination, they found that, in fact, those with current and remitted depression (vs. controls) had more maladaptive ER diversity and less adaptive ER diversity (Wen et al., 2021).

Driven by the realization that it is not only the quantity, but also the quality that matters, another set of studies used person-centered approaches, such as latent profile analysis (LPA), to characterize how individuals cluster according to their typical use of ER strategies. Although there are variations, most of these studies identified (at least) the following four groups of individuals: (1) “high regulators” – individuals who frequently use both maladaptive and adaptive ER strategies, (2) “low regulators” – individuals who infrequently use ER strategies of either type, (3) “adaptive regulators” – individuals who frequently use adaptive strategies, and infrequently use maladaptive strategies, and (4) “maladaptive regulators” – individuals who frequently use maladaptive strategies, and infrequently use adaptive strategies. The most consistent finding in this literature is that membership in the “maladaptive regulators” group is related to worse psychological well-being, whereas membership in the “adaptive regulators” group is related to the best psychological well-being (Chesney & Gordon, 2017; Grommisch et al., 2019; Lougheed & Hollenstein, 2012; van den Huevel et al., 2020). Similar results were obtained by a study applying a multilevel exploratory factor analysis, which found that it is not the number of ERs that were endorsed that mattered, but their quality (Southward & Cheavens, 2020). Specifically, Southward and Cheavens (2020) found that more reliance on adaptive engagement and less reliance on aversive cognitive strategies (categories that largely overlap with “adaptive” and “maladaptive”) is associated with improved mood in college students. Together with findings that show that the interaction between adaptive and maladaptive ERs is associated with psychopathology (e.g., Aldao & Nolen-Hoeksema, 2012; McMahon & Naragon-Gainey, 2018), it can be concluded that the balance between maladaptive and adaptive ER strategy use may be critical.

Using LPA to assess ER repertoire quality has some limitations. First, using LPA requires relatively large samples (i.e., N>250; Tein et al., 2013) – and even then, specific profiles sometimes contain a small number of participants (e.g., Chesney & Gordon, 2017), which makes it harder to draw conclusions that are generalizable. Second, the different profile groups are not identical across studies, as they are reliant on the specific sample examined. Similarly, though informative, multilevel exploratory factor analysis is highly sample-dependent. Thus, based on conclusions from the literature on ER repertoire, polyregulation, and ER diversity, we propose a novel aspect of ER flexibility: the ratio of using putatively maladaptive (vs. all) ER strategies. We base our metric on findings that it is not the number of ERs endorsed, but the relative balance between maladaptive and adaptive ERs that is most consistently associated with psychopathology (Chesney & Gordon, 2017; Grommisch et al., 2019; Lougheed & Hollenstein, 2012; van den Huevel et al., 2020; Wen et al., 2021). This operationalization is further supported by studies that examined interactive effects between ER strategies and found that the use of a maladaptive strategy (e.g., rumination) can moderate the effectiveness of adaptive ER strategies (e.g., reappraisal; Aldao & Nolen-Hoeksema, 2012; McMahon & Naragon-Gainey, 2018). Therefore, we computed an index of the ratio of maladaptive ER strategies used relative to all strategies endorsed, thus capturing the dynamic interplay between maladaptive and all other ER strategies. We chose to focus on maladaptive (and not adaptive) ER ratio because findings that strategies categorized as typically “maladaptive,” that are related to adverse consequences (e.g., Bean et al., 2022; Cavicchioli et al., 2023; Watkins & Roberts, 2020; cf., Ciarocco et al., 2010), are more robust than findings regarding the positive consequences of “adaptive” ER strategies (e.g., reappraisal – McMahon & Naragon-Gainey 2018; problem-solving – Gadassi Polack et al., 2021b; distraction – Wolgast & Lundh, 2017).

The research on ER flexibility has additional gaps. For example, it is unclear how ER flexibility changes over development. Examining the development of ER flexibility during adolescence is incredibly important because, during adolescence, individuals increasingly learn, and are expected, to regulate their own emotions, and are hypothesized to complete their “ER toolbox” during this period. Studies suggest that the ability to effectively use ER strategies increases across development. For example, it has been found that the use of adaptive regulation strategies increases (Silvers et al., 2012; Zimmermann & Iwanski, 2014) and the use of maladaptive strategies decreases (Gullone et al., 2010; cf., Cracco et al., 2017). Furthermore, the neurocircuitry underlying ER matures and becomes more specialized during adolescence (Gaussi Moreira et al., 2021). The only study that examined developmental aspects of ER repertoire quality (which is close to our maladaptive ER ratio), showed that younger (vs. older) adolescents were more likely to be included in the “low regulators” or “maladaptive regulators” group (van den Heuvel et al., 2020), thus indicating that ER improves with development. However, since the study was cross-sectional, individual differences may have been confounded with developmental effects. The current study used a longitudinal design to address this limitation.

Another gap in the literature on ER flexibility is that it is largely based on retrospective self-reports (e.g., Wen et al., 2021), with very few studies using intensive longitudinal designs (e.g., Grommisch et al., 2019; Southward & Cheavens, 2020). Intensive longitudinal designs, which involve repeated measurements over short periods of time, are highly recommended for studying psychopathology in developing samples, particularly depression, as they are less influenced by memory biases and require a lower degree of self-awareness (Russel & Gajos, 2020). In the context of ER, intensive longitudinal designs have been shown to have higher validity compared to retrospective self-reports (Koval et al., 2023; McMahon & Naragon-Gainey, 2020). In the context of ER flexibility, intensive longitudinal designs are ideal as they allow for assessment across many situations (vs. lab-based studies; Grommisch et al., 2019).

A third gap in the literature on ER flexibility is that this literature largely ignored the regulation of positive (vs. negative) emotions (cf., Southward & Cheavens, 2020). This is a significant gap considering that difficulties in maintaining positive emotions is one of the hallmark symptoms of depression (APA, 2013), and the regulation of positive emotions is particularly important during adolescence (Forbes et al., 2021; Young et al., 2019). Only a few of the studies examining ER flexibility included strategies aimed at regulating positive emotions, and none, to our knowledge, have done so in youth. The present study will be the first to do so.

The Present study

The present study examines an operationalization of a novel aspect of ER flexibility: the ratio of maladaptive vs. all ER strategies (henceforth, maladaptive ER ratio). Using a unique multiple- time-scale design (Ram & Diehl, 2014), in which two waves of an intensive longitudinal design were collected one year apart, we examine the associations of this maladaptive ER ratio with depressive symptoms. The unique design allows us to examine both short-term (within-day) and long-term (across-year) associations. Including two waves of an intensive longitudinal design is particularly beneficial during adolescence, as a longitudinal design is the optimal way to examine developmental processes. An additional benefit of this design comes from the fact that while Wave 1 examined typical everyday life, Wave 2 focused on a stressful time, around the first onset of COVID-19 and subsequent closures. As regulation is particularly needed during stressful times, having the second data wave is particularly informative and acts as a type of “experiment in nature,” which is important in intensive longitudinal designs (Kalokerinos et al., 2019). In the context of learning about the development of psychopathology, our design is optimized to examine processes of psychopathology development because it focuses on a high-risk developmental period and a time of increased stress. We examined the following hypotheses:

1. Feasibility.

Considering work on polyregulation across development, we hypothesized that most youths would use more than one emotion regulation strategy over the diary period. To provide evidence of the feasibility of the maladaptive ER ratio metric, we examined whether it can be calculated on the day-level and on the person-level. (1a) Evidence supporting day-level feasibility is provided by examining (i) for how many days (out of all diary entries) the maladaptive ER ratio can be calculated, and (ii) for how many days its value is larger than zero (1b) Evidence supporting person-level feasibility is provided by examining (i) for how many participants a maladaptive ER ratio metric can be calculated at Wave 1 and at Wave 2, and (ii) for how many participants the maladaptive ER ratio is larger than zero. Feasibility is achieved if for most participants and on most days, maladaptive ER ratio can be calculated (indicating that most individuals, on most days, use some ERs of any kind) and is larger than zero (indicating that most individuals, on most days, use some maladaptive ERs). As there are no comparable studies that we are aware of, we do not have a-priori hypotheses.

2. Development of ER flexibility.

Prior studies on the development of emotion regulation suggest that younger (vs. older) youth rely more strongly on maladaptive (vs. adaptive) emotion regulation strategies. Therefore, we hypothesize that (2a) cross-sectionally, maladaptive ER ratio will be higher for younger (vs. older) youth at Wave 1 and at Wave 2; and (2b) longitudinally, we hypothesize finding within-person decreases in the maladaptive ER ratio.

3. Clinical validation and utility.

To assess the validity and clinical utility of our new index, we examined its association with depressive symptoms. Cross-sectionally, (3a) we hypothesized that youth with clinical levels of depressive symptoms would use a higher maladaptive ER ratio compared to youth with non-clinical levels of depressive symptoms both at Wave 1 and at Wave 2. Longitudinally, (3b) we hypothesize that on days in which youth use a higher maladaptive ER ratio, they would experience increases in depressive symptoms, and (3c) we hypothesize that youth who have increased maladaptive ER ratio after a year would experience increases in their depressive symptoms. Considering vast literature showing increases in depressive symptoms are related to being female and older, we examine whether biological sex and age moderate the associations between maladaptive ER ratio and depressive symptoms.

Method

The current study is part of a larger investigation of children’s and adolescents’ emotions and social experiences (Deng et al., 2021; Dworschak et al., 2023; Gadassi Polack 2021a,b,c); only relevant measures are described.

Transparency and openness.

We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study. This report was not pre-registered. Data and R code can be obtained from the first author. All procedures have been approved by Yale University Institutional Review Board.

Participants and procedure

Wave 1:

Data were collected between 1/31/2019 and 9/23/2019. One hundred forty-eight youths were recruited via flyers, on Craigslist, and on Facebook. Advertisements invited youths 9–15 years old to participate in a 21-day diary study about emotions and social experiences. The only inclusion criteria were age and daily access to a device connected to the internet. Participants came to the lab for an initial visit with a legal guardian. A research assistant reviewed the daily diary questionnaire with each youth and the parent to ensure that all questions were clear and acceptable. Youths signed assent forms and their parents signed consent forms. Then, youths completed a practice survey and a demographics questionnaire on a lab computer. Every evening, approximately an hour before their regular bedtime, participants received a link via email to the daily survey, which they completed on a secure website (Qualtrics). The link expired after 14 hours. Participants who completed 60% of surveys received $40; participants who completed at least 90% received $60. Participants who completed less than 60% received $10 for participation.

One hundred and thirty-nine youths (94% of the original sample; 73 Girls; M_AGE=11.89, SD=2.14) completed at least 13 diary entries. Age range of the final sample was 8–15 since we allowed 8-year-old youths who would turn 9 during the diary period to participate. Participants completed an average of 19.08 (SD=2.14) of the surveys. The sample was largely White (N=97, 69.8%) and non-Hispanic (N=130, 93.5%). All youths resided in the New Haven, Connecticut, USA area. Information regarding income, education, or socioeconomic status was not collected.

Wave 2:

Data were collected between 3/30/2020 and 6/8/2020. Data collection was conducted during the acute early stage of the COVID-19 pandemic; two weeks after schools closed in Connecticut and a few days after the “stay at home” and mandatory face–covering orders were issued. According to the CT Department of Public Health (https://data.ct.gov/Health-and-Human-Services/COVID-19-Tests-Cases-Hospitalizations-and-Deaths-S/rf3k-f8fg), during data collection, there were 44,179 confirmed cases of COVID-19 in CT, between 293–1,972 hospitalizations per day, and 4,097 deaths. In that same timeframe, there were 1,961,781 confirmed cases and 111,774 deaths reported across the U.S. Vaccines for COVID-19 were at the preliminary stages of development at the time of data collection. Over 95% of our participants did not have in-person school at Wave 2 (for more details see Gadassi-Polack et al., 2021c).

Participants who completed Wave 1 of the study and indicated that they were interested in participating in additional studies were invited to participate in this follow-up. Interested participants were invited to an online Zoom session with the parent and the child, during which time youths and parents gave assent and consent, respectively, and received explanations about the diary. Then, youths filled out a demographic questionnaire as well as some additional questionnaires not used in the current report. Subsequently, every evening for 28 days, participants received a link via email to the daily diary survey. Participants were instructed to complete the survey before going to bed. The link expired after 16 hours. Participants who completed at least 60% of the surveys received $50; Participants who completed at least 90% of the surveys received $70; those who completed less than 60% received $10 for participation. One hundred and fifteen out of the 117 participants who started Wave 2 completed at least 13 diary entries (62 girls; M_AGE=12.70, SD=2.12). Participants completed an average of 25.9 (SD=3.28) of the surveys. The sample was largely White (N=84, 73%) and non-Hispanic (N=111, 96.5%). Information regarding their cultural/geographic background (apart from residing in Connecticut during the first wave of the study), income, education, or socioeconomic status was not collected. Table S1 in supplementary materials presents demographic characteristics for both waves of data.

Power analysis.

Sample size was determined with a power analysis conducted using PASS software (https://www.ncss.com/software/pass), based on data from the first 18 participants and adjusting for intra-class correlations. According to the power analysis, we needed a sample of 120 participants assuming 80% power and a two-sided α of 0.05 to detect all slopes of ER strategies predicting depressive symptoms (slopes beta values ranged from 0.43–1.29 in the preliminary data). We continued with participant recruitment until 120 participants completed at least 60% of the daily diary. To offset attrition, we increased the diary period in Wave 2 to 28 days (vs. 21 in Wave 1).

Measures

Regulation of negative emotions.

To assess regulation of negative emotions, we adopted items from the Children’s Response Style Questionnaire (CRSQ; Abela et al., 2000). The CRSQ is a self-report questionnaire used to assess three strategies for regulating negative emotions: problem-solving (e.g., “I talked it out with someone I think can help me feel better”), distraction (e.g., “I did something I enjoy”), and rumination (e.g., “I thought about: “I’m ruining everything”). Past research using the CRSQ has reported high levels of internal consistency (e.g., Abela et al., 2007, 2009), test-retest reliability (Abela et al., 2007), and predictive validity (Abela et al., 2002, 2004, 2007, 2009).

In Wave 1, we used two items from each scale to assess each strategy; in Wave 2, we used three items to assess each strategy with the goal of increasing reliability. Items were rated on 5-point scales, ranging from 0 (“not at all”) to 4 (“almost all of the time”). Instructions and items were adapted for daily-diary use by asking participants in each daily-diary entry to report the extent to which they had engaged in these ER strategies since the last evening and adding a response option of not using that strategy at all during that period.

We calculated the between- and within-subject reliabilities using procedures outlined in Shrout and Lane (2012). For a given measure, the between-subject reliability coefficient is the expected between-subject reliability estimate for a single typical day. The within-subject reliability coefficient is the expected within-subject reliability of change within individuals over the daily diary entries. The between-person and within-person reliabilities in Wave 1 were .68 and .60 for rumination, .69 and .54 for distraction, and .77 and .45 for problem-solving; in Wave 2, between-person and within-person reliabilities were .74 and .61 for rumination, .77 and .54 for distraction, and .85 and .70 for problem-solving. These reliabilities are considered acceptable for within-individual measures (Nezlek, 2017).

Regulation of positive emotions.

Regulation of positive emotion was assessed using the Responses to Positive Affect Questionnaire for Children (RPA-C; Bijttebier et al., 2012). The RPA-C is a self-report questionnaire assessing three strategies for regulating positive emotions: Emotion-focused positive rumination (e.g., “Notice how you feel full of energy”), Self-focused positive rumination (e.g., “Think ‘I am the best I could be’”), and Dampening (e.g., “Think ‘I don’t deserve this’”). The RPA-C has shown good internal consistency and test-retest reliability, as well as predictive validity (Bijttebier et al., 2012).

In Wave 1, we used three items to assess emotion-focused positive rumination and two items to assess each of the other scales; In Wave 2, we added two items so that each of the subscales was assessed using three items. Items were rated on 5-point scales, ranging from 0 (“not at all”) to 4 (“almost all of the time”). Instructions and items were identical to those for regulation of negative emotions. The between-person and within-person reliabilities in Wave 1 were .84 and .62 for emotion-focused positive rumination, .82 and .46 for self-focused positive rumination, and .66 and .41 for dampening, and in Wave 2 .82 and .56 for emotion-focused positive rumination, .82 and .59 for self-focused positive rumination, and .76 and .59 for dampening. These reliabilities are considered acceptable for within-individual measures (Nezlek, 2017).

Depressive Symptoms.

To assess depressive symptoms, we used the Children’s Depression Inventory – short version (CDI-S; Kovacs, 1985). The CDI-S is a self-report measure consisting of 10 items used to assess severity of depressive symptoms. The short form is similar to the full measure in its specificity and sensitivity to screen for depression in children (Allgaier et al., 2012). Each item consists of three sentences representing different degrees of symptom severity (from 0 to 2), from which the participant needs to choose the ones that describe them best. Instructions were adapted for use of a daily diary by asking participants to choose from each group of sentences the sentence that describes them best in the moment of answering the survey. Previous studies have found that the CDI-S has acceptable to good internal-consistency reliability, as well as convergent and divergent validity (Ahlen & Ghaderi, 2017; Allgaier et al., 2012).

The between-person and within-person reliabilities were .91 and .75 in Wave 1, and .92 and .74 in Wave 2. These reliabilities are considered good for within-individual measures (Nezlek, 2017). In Wave 1, the mean of depressive symptoms was 2.71 (SD=3.05, range 0–14.94). In Wave 2, the mean of depressive symptoms was 3.90 (SD=3.77, range 0–18.96). The score suggested as the clinical cutoff for the short version is ≥3 (Allgaier et al., 2012); thus, 35.3% of our participants in Wave 1 and 49.6% in Wave 2 met this criterion.

Ratio of Maladaptive vs. All ERs

We computed the maladaptive ER ratio as follows:

$$\begin{gathered} \text{Maladaptive}\text{ER}\text{ratio}=(\text{rumination}+\text{dampening})/(\text{rumination}+\text{problem} \\ \text{solving}+\text{distraction}+\text{dampening}+\text{emotion-focused}\text{positive}\text{rumination+self-focused}\text{positive}\text{rumination}) \end{gathered}$$

Thus, on the day level, we summed, for each participant, the frequency of use of rumination and dampening, and divided it by the sum of frequency of use of all ER strategies for that day. On the person level, we summed, for each participant, the average frequency of use of rumination and dampening across the diary period and divided it by the sum of average frequency of use of all ER strategies across the diary period. In cases in which no ER was used at all, maladaptive ER ratio could not be calculated and resulted in a missing value. On days it was calculated, it could range from 0 (if neither rumination nor dampening were used, but one or more of the other ERs were used to some degree) to 1 (if only maladaptive ERs were used). It should be noted that the same ratio score could have resulted from different patterns of ER use. For example, a person could get a 0 in the maladaptive ER ratio if they had not used rumination or dampening at all throughout the diary, used only problem-solving (to any degree) and no other ERs, or if they used any combination of ERs that are not dampening or rumination. Conversely, a value of 1 could be obtained by using only rumination, only dampening, or both.

Validation of the maladaptive ER ratio compared to previous analytic methods.

To assess the validity of the novel maladaptive ER ratio index, we compared it to a classification by groups of regulators that emerged in prior research on ER in adolescents (van den Heuvel, 2020), namely: low regulators, high regulators, adaptive regulators, and maladaptive regulators.

Person level.

We categorized our participants to each of the regulation groups: (a) low regulators had lower-than-group-mean levels of all types of ER strategies. (b) high regulators had higher-than-group-mean levels of all types of ER strategies. (c) adaptive regulators had lower than group-mean maladaptive ERs (including rumination and dampening) and higher than group-mean adaptive ERs (including distraction, problem-solving, emotion-focused positive rumination, and self-focused positive rumination). (d) maladaptive regulators had higher than group-mean maladaptive ERs and lower than group-mean adaptive ERs. Table S3 describes the distribution of profile group membership for each of the data waves. We examined if the regulation groups were different in the maladaptive ER ratio index using two one-way ANOVAs, with ER-profile group membership as the independent variable, and maladaptive ER ratio in each data wave as the dependent variable. As can be seen in Table S4, both ANOVAs revealed that, supporting the validity of the maladaptive ER ratio indices, regulation groups had significantly different maladaptive ER ratios: in both data waves, adaptive regulators had the lowest maladaptive ER ratio and maladaptive regulators had the highest maladaptive ER ratio, whereas low and high regulators were in-between. In Wave 1, low regulators had a similar maladaptive ER ratio to high regulators, but in Wave 2, they had a lower maladaptive ER ratio compared to high regulators and similar to the adaptive regulators’ ratio.

Day level.

We categorized days into four day-types, along the lines of the person-level profile categorization: (a) low regulation: days in which participants used all ER strategies less than their person-average, (b) high regulation: days in which participants used all ER strategies more than their person-average, (c) adaptive regulation: days in which participants used maladaptive strategies less than their own person-average, and adaptive strategies more than their own person-average. (d) maladaptive regulation: days in which participants used adaptive strategies less than their person-average and maladaptive strategies more than their person-average. We ran a multilevel model with data wave, day categorization, and their interaction as the independent variables, and daily maladaptive ER ratio as the dependent variable. As can be seen in Table S5, supporting the validity of the maladaptive ER ratio index, day type significantly predicted maladaptive ER ratio. Similar to person-level categorization, adaptive regulation days (average maladaptive ER ratio=.15) were characterized by low maladaptive ER ratio, whereas maladaptive regulation days (average maladaptive ER ratio=.37) were characterized by high maladaptive ER ratio, and low regulation days (maladaptive ER ratio =0.23) and high regulation days (maladaptive ER ratio=.28) were intermediate; differences between all day types were significant (all ps<.001).

Data Analytic Plan

Feasibility.

To examine the first hypothesis, that most youths use a combination of maladaptive and adaptive strategies to regulate their emotions, we (a) conducted descriptive statistics to show the number of days the maladaptive ER ratio could be calculated as well as the frequency in which it was larger than zero; and (b) descriptively examined for how many participants the maladaptive ER ratio could be calculated in each data wave, as well as the number of participants for which the maladaptive ER ratio was larger than zero.

Development of maladaptive ER ratio.

To examine how maladaptive ER ratio changes over development we will examine (a) cross sectionally: Pearson correlations between age and maladaptive ER ratio separately for each data wave (b) longitudinally: a mixed-model ANOVA with wave as the within-participant factor, gender and age between-participant factors, and maladaptive ER ratio as the dependent variable. For completeness, we will also examine what how many participants show negative/positive change in maladaptive ER ratio over a year. Finally, to increase interpretability, we examined whether specific ER strategies contributed to changes in maladaptive ER ratio using a series of paired-sample t-tests. All analyses examining development were conducted on participants who completed 13 entries or more on both data waves (N=112; see Gadassi Polack et al., 2021c for more details).

Clinical validation and utility.

To examine hypothesis 3a, that maladaptive ER ratio would be different for youths who are above (vs. below) the clinical cutoff for depression, we first dichotomized the depressive symptoms variable, separating youths into a group of those below the clinical cutoff (mean sum CDI-S across the diary period <3) and those above it (≥3). Then, for each wave, we conducted a one-way ANOVA to examine if youth in the two depression groups had different maladaptive ER ratio levels.

To examine hypothesis 3b, we conducted multilevel modeling with the nlme package (Pinheiro et al., 2014) of the statistical programming software R Studio (R Core Team, 2013). Level 1 was the day level and Level 2 was the person level. We centered the day-level predictors at the person-mean to make the interpretation of intercepts clearer and to separate Level 1 and Level 2 effects (see Zhang et al., 2009). We used a lag-1 auto-regressive structure across the daily errors. As covariates, we entered into the analyses (1) the lagged person-mean-centered outcome score (i.e., the previous day’s outcome variable) and (2) the person’s mean outcome score (averaged across the entire diary period). Thus, the outcome becomes a residualized change score. These models examine the bi-directional association between depressive symptoms and maladaptive ER ratio. To examine differences between waves, we entered wave (dummy coded such that 0=Wave 1 and 1=Wave 2) and its interactions with the predictor into the model.

Level 1 equation:

$$\begin{gathered} {\mathrm{Y}}_{\mathrm{j}\mathrm{k}}\left(\mathrm{D}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{y}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{o}\mathrm{m}\mathrm{s}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{y}k\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{o}\mathrm{n}j\right)={\beta }_{0\mathrm{j}}+ \\ {\beta }_{1j}\mathrm{*}(\mathrm{l}\mathrm{a}\mathrm{g}\mathrm{g}\mathrm{e}\mathrm{d}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{o}\mathrm{n}-\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}-\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{d}\mathrm{d}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{y}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{o}\mathrm{m}\mathrm{s}[\mathrm{d}\mathrm{a}\mathrm{y}\mathrm{k}-1\left]\right)+ \\ {\beta }_{2\mathrm{j}}\mathrm{*}(\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{o}\mathrm{n}-\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}-\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{d}\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{a}\mathrm{d}\mathrm{a}\mathrm{p}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{E}\mathrm{R}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{y}\left.\mathrm{k}\right)+ \\ +{\epsilon }_{j\mathrm{k}} \end{gathered}$$

Level 2 equation:

$$\begin{gathered} {\beta }_{0\mathrm{j}}={\gamma }_{00}+{\gamma }_{01}\mathrm{*}(\text{Wave})+{\gamma }_{02}\mathrm{*}(\text{mean}\text{depressive}\text{symptoms})+{\gamma }_{03}\mathrm{*}(\text{mean}\text{maladaptive}\text{ER} \\ \text{ratio})+{\gamma }_{04}\mathrm{*}(\text{Age})+{\gamma }_{05}\mathrm{*}(\text{Gender})+{\mathrm{u}}_0 \\ {\beta }_{2\mathrm{j}}={\gamma }_{20}+{\gamma }_{21}\mathrm{*}\text{Wave} \end{gathered}$$

For example, to examine if daily maladaptive ER ratio predicted daily variations in depressive symptoms and to test whether these associations were moderated by data wave, we ran a model in which wave, daily maladaptive ER ratio, and the interaction between daily maladaptive ER ratio and wave were the predictors. In addition, we entered yesterday’s depressive symptoms into the model, along with the individual’s mean level of depressive symptoms. Including lagged depressive symptoms means that any effect we find for daily maladaptive ER ratio would not include variance that is due to yesterday’s depressive symptoms and its effects on the daily maladaptive ER ratio (or directly on today’s depressive symptoms). We also entered (3) the person’s mean score of the predictors (in the same example, this meant entering an individual’s mean level of maladaptive ER ratio). Including the person-mean variables allows estimation of both person-level and day-level effects (Bolger & Laurenceau, 2013) and also allows to rule out static spurious “third variables” as alternative explanations.

To examine hypothesis 3c, that the maladaptive ER ratio changes would predict depressive symptoms a year later, we conducted multiple regression analysis in which change in maladaptive ER ratio from Wave 1 to Wave 2 predicted depression in Wave 2 while controlling for depression at Wave 1.

As reported elsewhere (Gadassi-Polack et al., 2021c), being a girl (vs. boy) and older was significantly associated with higher levels of depressive symptoms in this dataset. Therefore, we examined gender and age as moderators in all models testing hypothesis 3; only significant interactions are reported. In addition, Table S2 in the supplementary materials presents a correlation between participants’ age and all research variables.

Results

1. Feasibility

Table 1 presents means, ranges, and SDs of maladaptive ER ratio and of the separate emotion regulation strategies. Distribution of the maladaptive ER ratio index can be seen in Figures S1 and S2; skewness and kurtosis values suggested normal distribution at both data waves (skewness=0.31 at Wave 1 and 0.33 at Wave 2; kurtosis=2.00 at Wave 1 and −0.08 at Wave 2; Hare et al., 2010).

Table 1.

Depressive symptoms and emotion regulation (ER) strategy use in Wave 1 and Wave 2

		Wave 1 (N=139)				Wave 2 (N=115)
	Mean (SD)	Median	Range	N scored 0	Mean (SD)	Median	Range	N scored 0
Depressive symptoms	2.71 (3.05)	1.19	0–14.94	9	3.90 (3.77)	2.86	0–18.96	3
ER NA 1
Rumination	0.81 (0.55)	0.73	0–3.08	1	0.66 (0.67)	0.40	0.74-2.77	0
Distraction	1.22 (0.96)	0.85	0–3.83	4	1.14 (0.83)	1.14	0–3.32	3
Problem Solving	0.96 (0.86)	0.69	0–3.58	8	0.86 (0.84)	0.61	0–3.59	6
ER PA
Dampening	1.40 (0.58)	1.29	0–3.43	1	0.92 (0.74)	0.82	0–2.89	3
Emotion-focused positive rumination	2.09 (0.85)	2.00	0.35–3.88	0	1.70 (0.88)	1.82	0.01-3.70	0
Self-focused positive rumination	1.90 (0.90)	1.88	0.31–4	0	1.35 (0.90)	1.33	0–3.76	1
Maladaptive ER ratio	0.28 (0.09)	0.28	0.01–0.57	0	0.22 (0.12)	0.23	0–0.60	1

¹ER=emotion regulation; NA=negative affect; PA=positive affect

Demonstrating feasibility, (1a) maladaptive ER ratio could be calculated for 5,228 of the 5,631 days included in the study (92.8%). The value of maladaptive ER ratio was larger than zero on 5,055 days (89.8%). (1b) As can be seen in Table 1, maladaptive ER ratio could be calculated for all participants in both data waves. One-hundred and thirty-nine participants in Wave 1 (100%) and 114 (99%) participants in Wave 2 had maladaptive ER ratio larger than zero.

2. Development of maladaptive ER ratio

1. Contrary to our hypothesis, there was a significant and positive correlation between age and maladaptive ER ratio at Wave 1 (r[112]=.22, p=.022), and a non-significant correlation at Wave 2 (r[112]=.11, p=.229).

2. As predicted and can be seen in Table 1, our longitudinal analysis showed that over time, there was a significant decrease in maladaptive ER ratio (F[1, 109]=36.69, p<.0001, η²=0.25). The main effects of gender and age, as well as their interactions with time, were not significant (all ps>.076). For completeness, we examined how many participants experienced this decrease; descriptive statistics revealed that for 78 participants (69.6% of the sample) maladaptive ER ratio decreased over time.

To better understand whether changes in specific strategies led to this overall decrease, we examined whether specific ER strategies contributed to reductions using a series of paired-sample t-tests. As can be seen in Table 1, the frequency of using rumination (t[111]=2.22, p=.029, Cohen’s d=0.21), dampening (t[111]=8.12, p<.001, Cohen’s d=0.77), self-focused positive rumination (t[111]=8.10, p<.001, Cohen’s d=0.76), and emotion-focused positive rumination (t[111]=5.38, p<.001, Cohen’s d=0.51), significantly decreased from Wave 1 to Wave 2. There were no significant changes in the use of problem-solving (t[111]=0.39, p=.699), or distraction (t[111]=−0.18, p=.862).

3. Clinical validation and utility

1. To examine whether participants with clinical (vs. not clinical) levels of depression had a higher maladaptive ER ratio, we conducted two one-way ANOVAs (one for each data wave), with depression levels, gender, and age as between-subject factors, and maladaptive ER ratio as the dependent variable. As predicted, youths who had clinical levels of depressive symptoms had significantly higher maladaptive ER ratio (Mean ratio=0.34, SD=0.08 at Wave 1; Mean ratio=0.28, SD=0.11 at Wave 2) compared to the low depression group (Mean ratio=0.24, SD=0.07 at Wave 1; Mean ratio=0.16, SD=0.10 at Wave 2) at Wave 1 (F[1, 132]=36.56, p<.001, η²=.22) and at Wave 2 (F[1, 105]=27.94, p<.001, η²=.21). Gender, age, and their interactions did not have significant effects (all ps≥.073). See Table S6 in the supplementary materials for a parallel regression analysis examining the cross-sectional association between maladaptive ER ratio and depressive symptoms.

2. Table 2 presents the results of the multilevel model examining whether daily changes in maladaptive ER ratio predicted daily variations in depressive symptoms while adjusting for previous-day depressive symptoms, and whether this association was moderated by gender, age, and data wave. As can be seen in Table 2, on days in which participants’ maladaptive ER ratio was higher than their own average, they experienced increases in depressive symptoms. The significant interaction between daily maladaptive ER ratio and data wave indicates that this association was significantly smaller in size in Wave 2, although it was still significant (β=3.77, SE=0.88, t=4.29, p<.0001, for Wave 1; β=2.63, SE=0.83, t=3.18, p=.002, for Wave 2). Figure 1 depicts the significant interaction. As can be seen in Table 2, the effects of age, mean maladaptive ER ratio, and the interactions of daily maladaptive ER ratio with gender and age were not significant.

3. Table 3 presents the results of the regression analyses predicting depression at Wave 2 from changes in maladaptive ER ratio from Wave 1 to Wave 2, grand-mean-centered age, gender, and their interactions with maladaptive ER ratio change, while adjusting for depression at Wave 1. The model was significant (F[6, 105] = 25.04, p<.001, R²=.56), explaining 56% of the variance in change in depressive symptom levels. As can be seen Table 3, as predicted, increases in maladaptive ER ratio predicted higher levels of depression at Wave 2. Higher levels of depression at Wave 1 also significantly predicted higher levels of depression at Wave 2. Age, gender, and their interactions, did not significantly contribute to the model.

Table 2.

Predicting daily depressive symptoms from daily changes in maladaptive ER ratio.

	β	SE	df	t	p-value	95% CI
Intercept	−0.01	0.05	4486	−0.19	.849	−0.10,0.08
Age1(γ04)	−0.002	0.004	4486	−0.54	.589	−0.01,0.006
Gender (γ05)	−0.04	0.02	4486	−2.05	0.04	−0.08,−0.002
Wave (γ01)	0.004	0.05	4486	0.08	0.939	−0.10,0.10
Lagged Depression (β1j)	0.21	0.02	4486	9.06	<.0001	0.16,0.25
Mean Depression (γ02)	1.02	0.004	4486	250.31	<.0001	1.01,1.02
Mean maladaptive ER ratio (γ03)	−0.03	0.13	4486	−0.20	0.845	−0.29,0.24
Daily maladaptive ER ratio (β2j)	3.76	0.88	4486	4.29	<.0001	2.04,5.49
Daily maladaptive ER ratio X Age	0.43	0.26	4486	1.66	.097	−0.08,0.93
Daily maladaptive ER ratio X Gender	0.65	1.05	4486	0.62	.533	−1.40,2.71
Daily maladaptive ER ratio X Wave	−1.13	0.56	4486	−2.02	.044	−2.23,−0.03

¹β are the beta coefficients in the multilevel models; Age is grand-mean centered and daily ratio ER is person-mean centered; Gender is coded 0 for male and 1 for female; Wave is coded 0 for Wave 1 and 1 for Wave 2; ER= emotion regulation

Figure 1.

Daily maladaptive ER ratio predicting depressive symptoms

Table 3.

Predicting depression at Wave 2 from changes in maladaptive ER ratio over one year.

	B	SE	t	p−value	CI 95%
Intercept	1.48	0.38	3.91	<.001	0.73–2.24
Age	0.07	0.24	0.28	.783	−0.42–0.55
Gender	0.16	0.52	0.30	.762	−0.87–1.19
Depression Wave 1	0.84	0.08	9.92	<.001	0.67–1.00
Maladaptive ER1 ratio change	1.06	0.40	2.63	.010	0.26–1.86
Age * maladaptive ER ratio change	−0.13	0.26	−0.49	.623	−0.65–0.39
Gender* maladaptive ER ratio change	−0.16	0.50	−0.33	.742	−1.15–0.82

¹ER=emotion regulation

Discussion

The present study used an index to assess an aspect of ER flexibility: the ratio of maladaptive ER strategies out of all ER strategies used, and examined the feasibility, developmental course, and clinical utility of this novel index. This novel index – examining the balance between maladaptive and adaptive strategies – is a direct continuation of recent advances in the field of ER stressing the need to focus on emotion polyregulation (Ford et al., 2019), as it captures the simultaneous impact of multiple ERs. Moreover, this novel index is in line with recent theoretical advances on ER flexibility that argue that we should investigate ERs not as inherently “maladaptive” or “adaptive” (Aldao et al., 2015; Bonanno & Burton, 2013). Indeed, this maladaptive ER ratio captures the dynamic interplay between maladaptive and all ER strategies thus focusing on how the balance between ERs impacts depressive symptoms. Importantly, we implemented a unique multi-time-scale design (Ram & Diehl, 2014) examining two waves of intensive longitudinal data, one year apart, during a critical period for the development of ER and psychopathology, the transition to adolescence (Rapee et al., 2019).

As predicted, using a maladaptive ER ratio index was highly effective in capturing youths’ ER flexibility, and demonstrated that almost all of the youths used some amount of maladaptive ERs almost every day both at Wave 1 and at Wave 2. Moreover, supporting the validity of the maladaptive ER ratio and its clinical utility, our results show that youths who suffer from clinical levels (vs. youths with non-clinical levels) of depressive symptoms use a higher proportion (10–12% more) of maladaptive ER strategies when regulating their emotions. This finding was robust: it replicated in both data waves, and it was not moderated by gender or age. These results – especially the observation that everyone uses a mixture of ERs, and at least some degree of “maladaptive” strategies – demonstrate the importance of examining multiple ER strategies at the same time, as recent studies have increasingly done (Chesney & Gordon, 2017; Grommisch et al., 2019; Ladis et al., 2023; Lougheed & Hollenstein, 2012; Southward, & Cheavens, 2020; van den Huevel et al., 2020; Wen et al., 2021).

The current study extends previous research on ER repertoire and flexibility in several ways. First, we offer to examine a novel aspect of ER flexibility and operationalize it as the ratio of maladaptive ERs vs. all ERs. Second, while prior work focused mostly on one-time self-report or single lab visits (e.g., Wen et al., 2021), we utilized a 2-wave intensive longitudinal design, allowing us to examine both within- and between-person processes as well as developmental changes. Third, we examined youths during a sensitive time in development, the transition to puberty, a period that is likely to shed light on processes of developmental psychopathology. Finally, our maladaptive ER ratio included strategies for the regulation of both positive and negative emotions, whereas the majority of ER repertoire research focused on the regulation of negative emotions only (cf., Southward, & Cheavens, 2020).

The development of the maladaptive ER ratio

The present study is the first to examine how ER flexibility changes over development. As predicted, there was a decrease in the maladaptive ER ratio after a period of a year. This is in line with other studies on ER in youths, which generally show an increase in using adaptive ERs (Silvers et al., 2012; Zimmermann & Iwanski, 2014) and a decrease in the use of maladaptive strategies across development (Gullone et al., 2010; c.f., Cracco et al., 2017), as well as with neuroimaging studies of ER in youth (Gaussi-Moreira et al., 2021). The fact that our one-year follow-up was conducted during a population-level stressor complicates interpretation, as we cannot conclude with certainty whether the changes were due to development or due to the stressor. The literature on stress-related changes in ER in youth shows that stress is related to increases in ER difficulties (e.g., Jenness et al., 2020); thus, this literature would predict increases in maladaptive ER ratio because of the stress. However, as we found decreases in maladaptive ER ratio, we believe that the changes we found are more likely related to developmental processes (rather than the impact of COVID-induced-stress). Future studies are needed to elucidate the developmental trajectory of maladaptive ER ratio in the absence of severe stressors such as COVID-19.

Changes in the balance between maladaptive and all ER strategies predicts changes in depressive symptoms.

Our study was specifically designed to capture processes implicated in the co-development of ER and depression in youth. In Wave 1, we assessed youths who were on the cusp of the emergence of depression (i.e., 15 years old or younger). Wave 2, which was conducted a year later, helped capture the developmental period during which there are substantial increases in depressive symptoms (Salk et al., 2016). The fact that the second wave also co-occurred with a significant stressor – the onset of COVID-19 lockdowns – acted as an “experiment in nature” and increased the likelihood of capturing significant increases in depressive symptoms due to stress (Gadassi-Polack et al., 2021c; Kalokerinos et al., 2019).

Further demonstrating the clinical utility of this novel index, and as predicted, maladaptive ER ratio was significantly associated with youths’ depressive symptoms on the day level. Daily increases in maladaptive ER ratio were associated with daily increases in depressive symptoms. This finding echoed also on a larger timescale: increases in maladaptive ER ratio over a year predicted increases in depressive symptoms over a year. Considering that on both timescales we reduced the risk for reverse causality by adjusting prior time-point levels of depressive symptoms (yesterday’s or last year’s), and that our day-level analyses reduced the risk of third-variable explanations by adjusting for person-level predictors and outcomes, increases confidence in our results.

Interestingly, our results that daily increases in maladaptive ER ratio are associated with daily increases in depressive symptoms were moderated by data wave, such that this association was weaker during Wave 2. Considering that Wave 2 coincided with a population-level stressor, these results were surprising, as we expected that ERs would be more significantly associated with mental health status during a stressful time. It is possible that the characteristics of this unique stressor, that brought with it multiple unprecedented changes to youths’ lives (Gadassi-Polack et al., 2021c; Magson et al., 2021), rendered ER strategies themselves less central to youth mental health. For example, as more youth were home with family members, it may be that in addition to using ER to self-regulate, they relied more on their family for regulation. Another explanation may be related to developmental processes – it is possible that as our sample aged a year, other social, cognitive, or hormonal factors started playing a more important role in daily variations in depressive symptoms. For example, it is possible that advances in pubertal development increased the impact of hormonal fluctuations on depressive symptoms, thus reducing the relative influence of ER. Future studies are needed to further understand under which conditions (e.g., social isolation, stress) and developmental periods this balance matters more, and when it might matter less.

Strengths, Limitations, and Future Directions

The current investigation has several strengths. First, we examined an important risk factor during a time of increased depression risk (Rapee et al., 2019) and developed a metric that can be easily applied to clinical settings, hopefully increasing the impact of our findings. Second, our use of daily diaries increased the ecological validity of our findings, which is particularly important considering the limitations of using retrospective self-report to assess ER (Koval et al., 2023; McMahon & Naragon-Gainey, 2020). Third – and most importantly – our maladaptive ER ratio captures a novel aspect of ER flexibility and thus makes a significant theoretical contribution.

The current investigation has some limitations that should be acknowledged. First, despite the longitudinal design, our ability to deduce causality is limited since our design is not experimental. Future studies manipulating the maladaptive ER ratio are needed to better examine its causal role in depressive symptoms. Second, our sample was not clinically diagnosed with depression, and we relied on self-report questionnaires to assess symptom levels. Relatedly, we used the clinical cutoff to help deduce the clinical significance of our findings by dichotomizing a continuous variable, which may have resulted in a loss of information. Future studies using clinically diagnosed samples are needed. Third, our study contains an inherent confound between data waves, as any difference between them may be explained either by the stressor evident in the second wave only, or in naturally occurring developmental processes. Conducting additional follow-ups on this sample after the pandemic subsides may help disentangle developmental from stress-related effects. Fourth, although we were one of the first to examine ER of positive and negative emotions (e.g., Southward, & Cheavens, 2020; Verhees et al., 2021), we examined only six ER strategies overall. Future research is needed to ensure our results generalize when a wider range of strategies is examined. Finally, our study did not assess to what degree youth were aware of their emotions, a factor that is known to predict emotion regulation use (Riley et al., 2019). Future studies assessing emotional awareness and its associations with the maladaptive ER ratio are needed.

Clinical implications

The current study has important clinical implications. First and foremost, our findings suggest that it is not how much you regulate, but rather it is the balance between strategies that is important for mental health. This suggests that instead of focusing on reducing the use of maladaptive strategies, increasing the use of other ERs (that are not maladaptive) would be a more fruitful solution. Moreover, our finding that most youths, most of the time, use some degree of maladaptive strategies to regulate their emotions should have implications for psychoeducation, as it can help normalize the use of “maladaptive” strategies, as they may not be so harmful if balanced. This is an important direction as classifying a client’s behavior as “maladaptive” may increase their self-criticism, thus contributing to their depression.

Summary

The results of the present study show that an overwhelming majority of youths use at least some maladaptive ERs in addition to other strategies to regulate their emotions on a daily basis. The vast majority of youths showed a decrease in the proportion of putatively maladaptive ER (vs. all ER) over the course of development. Conversely, increases in the proportion of maladaptive ER strategies was associated with increases in depressive symptoms on the day and year level. Thus, the present study is an important first step in understanding how the development of ER flexibility is associated with typical and atypical development during a developmentally sensitive period of time – the transition to adolescence.

Highlights.

• We examine the development of emotion regulation (ER) flexibility in adolescence

• Most adolescents used fewer maladaptive (vs. all) ER after a year

• Increases in the ratio of maladaptive (vs. all) ERs was associated with depression

• Findings replicated on the within-day and between year levels