Temporal Associations Between Parents’ Daily Reports of Media Motivations, Infant Affect, and Parenting Behavior

Abstract

Past research links media use to infant negative affect, and parents report using media to cope. However, trait-level measures mask within-person effects that reflect regulatory and relational mechanisms more directly. In this study, 401 predominantly White (70.32%), college-educated (88.28%) parents (75.31% mothers) of 12- to 24-month-olds (M_age=16.43, 51.37% boys) completed daily diaries reporting the frequency of parent and child behaviors and affective states for 21 consecutive days. Temporal network modeling was used to examine associations among parents’ daily reports of media motivations, infant affect, and parenting behavior. On days when infants were fussier than usual, parents reported more yelling and more media use to occupy children, regulate children, and regulate parents’ own emotions the same day, and more media to occupy children the next day. The reverse was not true: Media use did not predict negative affect the next day. On days when parents reported more frequent media use to relax alone and to regulate children than usual, they also reported more comforting the next day, while more frequent media to connect with children one day predicted more infant positive affect the next day. Again, the reverse was not true. Results differed little as a function of infant age. Thus, temporal effects suggest media use may be an adaptive short-term strategy across the second year of life. Critically, within-person effects were not found at the between-person level, underscoring the importance of examining fluctuations within individuals to identify mechanisms and potential intervention targets.

Parents, scholars, and medical practitioners alike have a vested interest in understanding how digital media use affects child development, especially given many households exceed screen time recommendations for infants and young children (Lammers et al., 2022; McArthur et al., 2022). Additionally, there is mounting evidence that more screen time, particularly during infancy, is associated with a range of negative outcomes for young children (Mallawaarachchi et al., 2024). Importantly, however, most studies are cross-sectional, and findings from longitudinal studies are mixed, leaving directionality unclear (Gordon-Hacker & Gueron-Sela, 2020; Radesky et al., 2022). To better understand how media use operates in daily family life, scholars have called for research that considers the family context surrounding child media use (Nabi & Krcmar, 2016; Reich et al., 2025) and that goes beyond global estimates of screen time that make assumptions about the stability of media use patterns, ignoring dynamic processes (Barr et al., 2024).

Parents are especially influential in shaping media exposure during infancy, a period characterized by rapid fluctuations in children’s emotional states (Aktar & Pérez-Edgar, 2020). Parental decision making around media use has changed rapidly over the past decade due to the increasing availability of mobile devices, streaming platforms (e.g., YouTube), and infant-directed media (Mann et al., 2025). As such, an important contextual factor is grounded in parental motivations for media use, or parents’ perceived affordances of digital media as they relate to parenting goals (e.g., taking a break away from children, calming a fussy infant, relaxing together). A clearer understanding of parents’ media motivations, including how such motivations vary within individuals and in relation to parenting experiences, will help isolate underlying mechanisms that can guide evidence-based recommendations and illuminate potential targets for intervention. To meet this need, we examined parents’ daily reports of media motivations as they relate to infant affect and parenting behavior. We used a temporal network modeling approach to characterize both same-day and next-day associations to reveal the direction of short-term (daily) temporal effects.

Infant Affect and Parent Co-regulation

Parents monitor infant emotion carefully, especially during the first few years of development, to respond to their infants’ rapidly fluctuating mood states (Aktar & Pérez-Edgar, 2020). In turn, infants learn directly from their caregivers’ responses how to recognize and understand emotions, and develop the ability to understand others’ emotions. For infants one to two years old, this developmental process is dependent on caregivers’ verbal and physical emotional input in response to infant affect (LoBue & Ogren, 2022; Tan et al., 2022).

The reverse is also true: Child affect shapes parenting behaviors, such as responsiveness and engagement (Lee, 2013). Child negative affect may prompt parents to comfort and soothe infants as parents attempt to calm their child or redirect attention; alternatively, child negative affect may overwhelm parents, thereby reducing the quantity or quality of parent-child interactions. This may ultimately contribute to parents engaging in less favorable behaviors, such as yelling at or withdrawing from their children (Fabes et al., 2002; Pereira et al., 2022). Such punitive actions are often spurred by a parent’s desire for their child’s negative emotional state to end, or, at minimum, for their child’s outward expression of the emotion to be tempered (Fabes et al., 2002). A child’s emotional state may thus prompt the parent to either (a) self-soothe to manage the consequences of a child’s negative affect, or (b) assist the child in regulating their emotions.

When navigating an undesirable emotional state, parents may employ a number of strategies, some of which are more adaptive than others (Rodriguez & Shaffer, 2021). Some parenting behaviors (e.g., soothing and comforting) support child self-regulatory control, while other strategies (e.g., yelling) have the opposite effect (Pereira et al., 2022). Notably, some parenting strategies—whether adaptive or maladaptive—can be supplemented by external tools such as digital media (Bentley et al., 2016; Wolfers & Schneider, 2021). In the present study, we examine the extent to which different ways of using digital media with and around infants may be adaptive or maladaptive, particularly as it relates to parenting behavior and infants’ affective states. Next, we review research on parents’ motivations to use media for regulating themselves and their young children.

Regulatory Motivations for Media Use

The current work is motivated by the Dynamic, Relational, and Ecological Approach to Media Effects Research (DREAMER) (Barr et al., 2024). This framework represents media use as both a cause and consequence of real-time emotional, behavioral, and relational states that interact dynamically over relatively short time scales (e.g., hours or days) and accumulate over time to influence developmental cascades that unfold over longer time scales (e.g., months or years). This dynamic family media ecology is centered on parents’ media motivations, or the ways in which parents’ own media use and their provision of media to their children can serve to meet immediate parenting goals.

Of particular relevance to the current study are parents’ motivations to use media to alter their own or their children’s emotional and behavioral states. Several studies have demonstrated parents’ use of media for their own stress management and emotion regulation (Wolfers & Schneider, 2021). For example, parents may watch videos to escape or distract themselves, look up parenting resources, or connect with others to receive social support (Suh et al., 2024; Torres et al., 2021; Wolfers, 2021). Parents may similarly offer entertaining media to distract their children, regulate their children’s emotions, or connect by watching media together (Coyne et al., 2022; Radesky et al., 2016, 2022; Suh et al., 2024).

Notably, some uses of media for regulation are characterized by individual use—either the child viewing media by themselves, or the parent using their own device for emotion management—or by co-use, where joint engagement is the mechanism by which the media exposure aids in regulation (Ewin et al., 2021; Wolf & Tomasello, 2020). Regardless of the specific behavior, parents whose children are expressing negative affect may find media to be an accessible tool for aiding emotion regulation (Bentley et al., 2016; Coyne et al., 2022).

Although parents report using media as a coping mechanism (Wolfers & Schneider, 2021), parents are rarely asked what their motivations are at the time of media use, particularly in the context of parenting (Nikken, 2019). Such motivations include parents’ own media use to regulate their emotions or to relax by themselves, as well as parents’ provision of media to their children so parents can take a break or manage challenging child behaviors (Suh et al., 2024). Additionally, while some studies have examined young children’s social-emotional skills in relation to the frequency of using media to calm distressed children (Coyne et al., 2021; Radesky et al., 2016), the efficacy of different media motivations are not well understood. As such, recent advances in conceptualizing and measuring parents’ media motivations inform the current study by characterizing different regulatory uses of media for both parents and children.

Temporal Dynamics

Critically, past research on parents’ media motivations, particularly as they relate to parent and child emotions and behavior, has often relied on single-time-point measures that capture global patterns. For example, in one study with low-income families, parent reports of using mobile technology to calm upset toddlers or to keep peace and quiet in the home were associated with greater socio-emotional difficulties in toddlers (Radesky et al., 2016). However, emotional states and caregiver behavior fluctuate rapidly in infancy. Due to these fluctuations, some emotion theorists have applied dynamic systems theory, which considers in-the-moment interactions between infant mood states and caregiver responses to externally regulate affect (Cole et al., 2017). In line with dynamic systems theory and the DREAMER framework, regulatory and relational processes in the family media ecology likely unfold across relatively short time scales, which may differ from those that unfold across longer time-scales (Barr et al., 2024). As such, the current research is informed by literature examining day-to-day fluctuations in emotion and emotion expression (Kerr et al., 2021; Kuppens et al., 2010; Pérez-Edgar, 2019). In contrast to temperament, emotions operate on a smaller timescale, as many discrete emotion displays, such as infant fussing, are transient and context-dependent responses to internal and external factors (Pérez-Edgar, 2019). As emotions are not static across time, a dynamic approach is necessary to best understand the interaction between a child’s media ecology and their affective and behavioral states.

Overview of the Current Study

The current study is motivated by the DREAMER framework (Barr et al., 2024), which posits that: (1) media use is driven by parents’ media motivations; (2) such motivations are related to both regulatory and relational processes (e.g., to calm children, to take a break from parenting, to connect by using media together); and (3) these associations fluctuate within families across relatively short time scales. To test this framework, we examined parents’ daily reports of media motivations, infant affect, and parenting behavior over a 21-day period. We used temporal network modeling to examine same-day (contemporaneous) and next-day (temporal) associations among key variables, as well as mean-level associations among variables (between-subject).

In this study, we focused on parents of 12- to-24 month-olds because this is a period of rapid development in infants’ emotion regulation (Aktar & Pérez-Edgar, 2020; Pérez-Edgar, 2019). This period is characterized by rapid fluctuations in infants’ emotional states and parents’ responses, allowing us to examine day-to-day fluctuations within dyads. This is also a period when media use is guided by parents, and parents report using media to regulate infants’ emotional and behavioral states, yet little is known about parents’ motivations to use media in this age range (Coyne et al., 2021; Gordon-Hacker & Gueron-Sela, 2020; Radesky et al., 2016). This is particularly important given that current public health messaging advises against media use for children under age two years (AAP Council on Communications and Media, 2016; World Health Organization, 2019), but accumulating evidence suggests that most infants are exposed to screen media (Lammers et al., 2022; Mann et al., 2025; McArthur et al., 2022).

Prior research demonstrates associations between parent reports of infants’ trait-level negative affect and global estimates of infants’ screen time, and such associations are thought to reflect parents’ use of screen media as a tool for calming their children down or for giving parents a break away from their children (Aishworiya et al., 2022; Radesky et al., 2014). As such, we expected that parents who reported more infant negative affect in general would also report more frequent infant media use to occupy and to regulate infants in general (between-subjects effect), and that higher-than-average infant negative affect on a given day would be associated with higher-than-average infant media use to occupy and to regulate infants that same day (within-subjects contemporaneous effect). Whether this resulted in next-day temporal associations (i.e., more negative affect today predicting more media use tomorrow, or vice-versa) was an open research question. Similarly, the degree to which infant affect (negative or positive) related to parents’ motivations for their own media use was an open research question.

There is little prior research to guide directional hypotheses about associations between media motivations and parenting behavior. However, the pattern of observed temporal effects may reveal the degree to which media use is an adaptive strategy. For instance, if parents’ own media use to self-regulate is an effective coping strategy, we would expect it to predict an increase in positive parenting behavior (comforting/soothing) and a decrease in less optimal parenting behavior (yelling) the next day. Given rapid development in infants’ emotion regulation across the age range studied, we also conducted exploratory analyses to determine the degree to which results from our main analysis varied as a function of infant age.

Methods

Participants and Recruitment

Participants were drawn from a larger longitudinal study examining parenting experiences and family media use among 528 parents of an infant 12 to 24 months old at the time of enrollment. The study was approved by the Institutional Review Board at Georgetown University (#MOD00014994). The target sample size in the larger study was derived from a priori power analysis to ensure power at or above .80 to detect effects related to the original study aims and modeling approaches. The current study, which addresses different research questions using novel modeling approaches, includes a subsample of participants who met eligibility criteria (described later). Power for our statistical approach, temporal network modeling, is complex and depends on the number of variables, timepoints, and participants. However, a review of studies using these approaches reported a median number of participants of 66 and a mean of 119.6 (SD = 210.7) (Blanchard et al., 2023).

Data were collected online from parents across the US between February 2023 and July 2024. Participants were recruited through online advertisements (e.g., social media, website banners), postings in online parent support groups, a platform for hosting online studies for children and families, emails to parents on existing lab contact lists, flyers and emails to parents at a university lab preschool, and a listserv mailing from a university parent resource office. Inclusion criteria for parents included being 18 years or older and having at least one child who was 12- to-24 months old at the time of enrollment.

The current study included 401 parents who completed at least 15 daily diaries. Parents’ self-identified gender was woman (75.56%), man (23.44%), gender expansive/nonconforming (0.50%), or not reported (0.50%). Parent age averaged 33.42 years (SD = 4.24). Children (51.37% boys, 47.13% girls, 0.25% gender neutral, 1.25% not reported) were on average 16.43 months old (SD = 3.46). The sample represented mostly highly educated parents, with most reporting a 4-year college degree (35.91%) or an advanced degree (52.37%). Household income was also generally high, with most parents (71.32%) reporting an annual household income at or above $80,000, which is approximately equivalent to the US median household income in 2023 (Guzman & Kollar, 2024). Parent race was White (70.32%), Black or African American (11.72%), Asian or Pacific Islander (10.22%), American Indian or Alaska Native (0.75%), another or mixed race (4.49%), or not reported (2.49%). Additionally, 13.72% of parents self-identified as Hispanic or Latiné. See Supplemental Table S1 for a complete report of demographic information and comparison between this subsample and the full sample.

Procedure

Parents who were interested in the study completed a screening form, and eligible parents were contacted to schedule an initial interview with a research assistant who described the general study protocol, answered questions, and obtained written informed consent. Parents then completed a survey packet that included demographic information used to describe the sample in the current study. The following day, parents began receiving short surveys throughout a 21-day intensive longitudinal burst.

The current study uses data collected in the daily diaries that were sent to parents at the end of each day for 21 days. The link for each survey was distributed via text message at 8PM in parents’ local time using SurveySignal (Hofmann & Patel, 2015), an online messaging service. Automated reminders were sent as needed via text message after one hour and again after two hours if the survey link was not yet clicked on. Parents had a six-hour window (8PM to 2AM) to complete the survey before it was deactivated. The median survey duration was 3.75 minutes (IQR = 2.63 – 5.65 minutes). At the end of 21 days, parents received $3 for each completed survey (up to $63 total).

Measures

The surveys were the same each day and asked about the frequency of parent and child behaviors and emotional states throughout the day. Parents rated the frequency of each item on an 11-point slider scale ranging from not at all (0) to very frequently (10).

Child Affect

Parents reported the frequency of five child affective/behavioral states following the prompt, “Over the course of the day today, how often was [child’s name] [description of affective/behavioral state]?” Two items represented positive affective states (i.e., “happy or laughing”, “content or calm”) and three items represented negative affective states (i.e., “fussy, clinging, or crying”, “whimpering or whining”, “angry, mad, or having tantrums”). We created composite variables representing infant positive and negative affect by averaging parents’ responses to items in each category. Multilevel reliabilities are reported here as omega coefficients (Geldhof et al., 2014). The positive affect composite had a within-person omega coefficient of .63 (95% CI [.61, .64]) and a between-person coefficient of .72 (95% CI [.67, .78]). The negative affect composite had a within-person omega coefficient of .70 (95% CI [.69, .71]) and a between-person omega coefficient of .87 (95% CI [.84, .89]). These reliabilities are considered moderate to substantial based on guidelines for between-person (Shrout, 1998) and within-person measures (Brose et al., 2020; Nezlek, 2017).

Parenting Behavior

Parents reported on the frequency of parenting behaviors in general (not necessarily toward the focal child) following the prompt, “Over the course of the day today, how often did you [description of parenting behavior]?” For the current study, we selected two items that best captured different responses to children’s distress or dysregulation: “yell at your child/children” and “comfort or soothe your child/children.”

Motivations for Media Use

During each daily survey, parents indicated whether they (the parent) and their infant each used screen media (e.g., TV, videos, apps, video games) that day. On days when media use was reported, parents also rated the frequency of motivations for their own or their child’s media use; else frequency for each motivation was inferred to be 0 (“not at all”). When reporting motivations for their own media use, parents responded to the prompt: “Thinking about your whole day, how often did you (the parent) personally use screen media [to meet a stated goal]?” Similarly, when reporting on motivations for their child’s media use, parents responded to the prompt, “Today, how often did you use screen media for [child’s name] [to meet a stated goal]?” The media motivation items were adapted from validated scales measuring parents’ motivations for both parent and child screen media use in general (i.e., not on a specific day), and we reduced our data by averaging items that corresponded to each factor in the prior research (Suh et al., 2024). This resulted in five composite measures representing parents’ regulatory and relational motivations for their own and their child’s media use. We describe each composite measure next.

Parent regulate.

Four items captured parents’ own media use to regulate their own emotional states. These items asked about the frequency of using media on a given day to (1) calm themselves down when upset or stressed, (2) avoid or reduce conflict with others, (3) mentally “check out” or escape when the day was overwhelming, and (4) help them fall asleep or stay asleep. Reliability was moderate at the within-person level, 𝜔_W = .62 (95% CI [.61, .64]), and substantial at the between-person level, 𝜔_B = .93 (95% CI [.92, .94]).

Child regulate.

Eight items captured child media use for the purpose of regulating the child’s behaviors or emotional states. These items asked about the frequency of using media on a given day to (1) help the child calm down when they were upset or showing big emotions, (2) stop the child from moving around too much when they were too active or hyper, (3) help the child sit still or focus, (4) keep the child occupied in public places (e.g., doctor’s office, restaurant), (5) prevent the child from getting overwhelmed or upset in a new situation, (6) stop the child from begging for media, (7) avoid or reduce conflict, and (8) reward the child for good behavior. Reliability was substantial at both the within- and between-person levels, 𝜔_B = .83 (95% CI [.82, .83]) and 𝜔_B = .98 (95% CI [.98, .98]).

Parent relax.

Three items captured parents’ own media use to relax or unwind by themselves. These items asked about the frequency of using media on a given day to (1) laugh or be entertained, (2) alleviate boredom, and (3) take a break and relax by themself. Reliability was moderate at the within-person level, 𝜔_W = .63 (95% CI [.62, .65]), and substantial at the between-person level, 𝜔_B = .83 (95% CI [.80, .86]).

Child occupy.

Three items captured child media use to entertain or to keep the child busy. These items asked about the frequency of child media use on a given day to (1) let the parent get things done (e.g., work, chores), (2) take a break and relax by themself, and (3) mentally “check out” or escape when the day was overwhelming. Reliability was moderate at the within-person level, 𝜔_W = .65 (95% CI [.64, .67]), and substantial at the between-person level, 𝜔_B = .93 (95% CI [.91, .94]).

Connect.

The last three items captured both parent and child media use to connect or spend time together. These items asked about the frequency of (1) child media use to laugh or be entertained, as well as both (2) parent and (3) child media use for the purpose of relaxing or unwinding together. Reliability was moderate at the within-person level, 𝜔_W = .73 (95% CI [.72, .74]), and substantial at the between-person level, 𝜔_B = .95 (95% CI [.94, .96]).

Analytic Approach

Main Analysis

To test our hypotheses, we conducted temporal network model analyses using the mlVAR (Epskamp et al., 2024) and qgraph (Epskamp et al., 2012) packages in R to estimate and visualize the temporal network models, respectively. We generated temporal network models via a two-step multilevel vector autoregressive (mlVAR) approach (Epskamp et al., 2018), yielding three network structures estimating how well each variable predicts all others on the same day (contemporaneous) and the next day (temporal), as well as a between-subjects network demonstrating mean-level associations between all variables. Predictors were within-person centered in the mlVAR model, following the default in the mlVAR package, and the temporal and contemporaneous results can thus be interpreted as deviations from a person’s mean (e.g., on days when parents reported more X than usual, they also tended to report more Y than usual the next day).

Best practices for mlVAR analysis recommend having at least 20 data points to reduce bias in analyses; however, temporal network models have been conducted with as few as seven possible timepoints, and several studies have employed models with 10–20 observations per person (Blanchard et al., 2023). Across our full sample (N = 542), participants completed an average 16.02 diaries (SD = 6.09) out of 21 possible days, for an average response rate of 76.29%. In the full sample, 74% of participants had at least 15 observations, whereas only 36% had 20 or more. To avoid bias resulting from reducing our sample to only 36% of the total participants, we chose to conduct models with the 74% of participants who had at least 15 observations (n = 401). Because 15 is still under the recommended number of observations for temporal models, we also ran sensitivity analyses with the reduced sample of participants (n = 196) with at least 20 observations.

We first checked our data for normality and stationarity as recommended by other researchers (Blanchard et al., 2023). Kolmogorov-Smirnov tests and visual inspection of the residual plots (Aalbers et al., 2019; Blanchard et al., 2023) revealed that most variables were not normally distributed. However, there is little information on how to best deal with non-normality in multilevel VAR models (Blanchard et al., 2023), with some researchers suggesting that it is not critical or that it may mainly impact power to detect weaker associations (McNeish & Hamaker, 2020; Veenman et al., 2024). We also checked for stationarity violations using the Kwiatkowski-Phillips-Schmidt-Shin unit root test (KPSS) (Kwiatkowski et al., 1992), which verified that the variables in our data showed stable variance across time (Blanchard et al., 2023; Jordan et al., 2020).

To estimate the three networks, we modeled the media motivations, parenting behavior, and child affect variables using a multilevel vector autoregressive (VAR) approach. To estimate how well each variable predicts all other variables at the next timepoint (temporal network), the VAR approach regresses each variable at time t on itself and on all other variables at time t-1 (Epskamp et al., 2018). We also accounted for the dependency of timepoints within subjects by estimating the VAR model in a multilevel framework. The visual representation of the temporal network shows all significant associations (p < .05) between variables at time t and t-1 using arrows, controlling for all other associations.

To generate the contemporaneous network, the model regresses the residuals of the multilevel VAR model on all other residuals from that same timepoint. This network visualizes how variables are related within the same timepoint, after controlling for all other contemporaneous associations and temporal associations. The contemporaneous network, therefore, is akin to a partial-correlation network.

Finally, the model uses participants’ means to estimate a between-subjects network, which shows mean-level associations controlling for all other associations and collapsing across time. The between-subjects network is most similar to a partial-correlation cross-sectional network, which is similar to between-subjects cross-sectional associations that have been reported in prior work on media motivations and media use in relation to parent and child characteristics.

Moderator Analyses

Given rapid development in infants’ emotion regulation from 12 to 24 months (Aktar & Pérez-Edgar, 2020; Pérez-Edgar, 2019), we conducted exploratory analyses to determine whether the results from our main analysis varied by infant age. The mlVAR approach estimates separate networks for each participant. We regressed these individual network effects on infant age at the time of enrollment. For parsimony, we focused on within-person effects that (1) were significant in our main analysis and (2) addressed our main research questions (i.e., association between media motivations and either infant affect or parenting behavior).

Sensitivity Analyses

To confirm the robustness of our main analysis, we conducted two sets of sensitivity analyses. First, we generated another set of network analyses limiting the sample to only the 196 participants who had 20 or more observations, given that 20 is the recommended minimum number of observations for mlVAR in most research (Blanchard et al., 2023, p. 20; Epskamp et al., 2024; Jordan et al., 2020). We then compared the overall pattern of results to the original models including participants with 15 or more responses. The demographic distribution of this sub-sample was similar to those with less than 20 timepoints. See Supplemental Table S1 for a complete report of demographic information and comparisons across subsamples.

Second, we conducted additional sensitivity analyses for the contemporaneous network in light of ongoing debate about how best to interpret this network. Some propose that the contemporaneous network likely captures processes that occur more quickly than the lag interval in the data, in this case daily (Epskamp et al., 2018). Others, however, suggest that there are too many factors at play leading to correlated residuals that make it difficult to interpret results from this network as correlations within the same timepoint (Bringmann et al., 2024). As such, we also examined same-day associations between our variables using multilevel structural equation modeling (MSEM) in MPlus (Muthén & Muthén, 2023) using the same sample of 401 participants from our main analyses (i.e., those with 15 or more daily diaries).

MSEM models the associations between variables at the same timepoint using a multilevel framework to account for dependency in observations, but without controlling for the time-lagged, or temporal, associations. These additional models, therefore, confirm whether the findings revealed in the contemporaneous network were robust across different analytic approaches. One limitation of this approach is that while multiple dependent variables can be specified in MSEM, it does require that the “directionality” of associations be hypothesized or modeled by specifying independent (exogenous) and dependent (endogenous) variables. To account for this, we ran two MSEM models: one model where the parenting behavior and child affect variables were regressed on all five media motivations, and a second model where all five media motivations were regressed on the parenting behavior and child affect variables. MPlus estimates covariances for all independent (exogenous) variables by default, and we specified covariances for all dependent variables as well. The results of these models were then compared to the significant associations that emerged in the contemporaneous network in our main analysis.

Transparency and Openness

We report how we determined our sample size, all data exclusions, and all measures used in this study. All data, analysis code, and daily diary questions are available on Open Science Framework (Kirkorian et al., 2024). Reliability and network analyses were conducted using R, version 4.4.1 (R Core Team, 2024) and the packages multilevelTools, version 0.1.1 (Wiley, 2020), mlVAR, version 0.5.2 (Epskamp et al., 2024), and qgraph, version 1.9.8 (Epskamp et al., 2012). Sensitivity analyses using multilevel structural equation modeling were conducted using Mplus, version 8.11 (Muthén & Muthén, 2023). This study’s design and its analyses were not pre-registered.

Results

Descriptive Statistics

Sample-level average intra-individual means, standard deviations, minimums, and maximums and intraclass correlation coefficients (ICC) for each variable are reported in Table 1. In general, parents reported more frequent child positive affect than negative affect. Yelling was infrequently reported by most participants, while comfort and soothing was common. Media use for parents to relax was the most frequently reported motivation, while media use to regulate children’s emotions was the least frequently reported. However, there was variability both between and within participants. ICCs reflect the amount of variance explained by group membership, in this case the between-person variance. For example, with an ICC of .49, 49% of the variance in daily reports of child negative affect occurs at the between-person level, with the other 51% of the variance occurring within individuals (i.e., across time). Bivariate correlations among intraindividual means and with infant age can be found in Supplemental Table S2.

Table 1.

Intraindividual Descriptive Statistics and Intraclass Correlations for Each Item

Item	iiMean	iiSD	iiMin	iiMax	ICC
Child Affect
Negative affect	2.07	1.08	0.57	4.51	0.49
Positive affect	6.66	1.36	3.71	8.68	0.45
Parenting Behavior
Yelling	0.81	0.67	0.12	2.42	0.62
Comforting/soothing	5.08	1.71	2.06	8.10	0.51
Media Motivations
Regulate Parent	0.99	0.68	0.20	2.53	0.71
Regulate Child	0.61	0.42	0.17	1.57	0.77
Relax Alone	2.84	1.37	0.73	5.56	0.58
Occupy Child	0.99	0.76	0.19	2.75	0.67
Connect	1.43	0.99	0.30	3.56	0.68

Note.All variables were measured on an 11-point frequency scale ranging from 0 (not at all) to 10 (very frequently). iiMean = intraindividual mean; iiSD = intraindividual standard deviation; iiMin = intraindividual minimum; iiMax = intraindividual maximum; ICC = intraclass correlation coefficient. Intraindividual descriptive statistics represent the sample-level average across individuals (e.g., average across all intraindividual means).

Main Analyses: Temporal Network Model

Results from the temporal network model are shown in Figure 1. In network analyses, “nodes” represent variables and “edges” represent associations, or the links between nodes (Epskamp et al., 2018). Figure 1 depicts all significant edges in each of the three networks. A full report of all edge values can be found in Supplemental Tables S3-S5.

Figure 1. Significant Edges in the Main Network Analysis.

Note. CNegAff = child negative affect; CPosAff = child positive affect; ParReg = parents’ own media use to self-regulate; ChReg = child media use to regulate child behaviors and emotional responses; ParRelax = parents’ own media use to relax alone; ChOcc = child media use to occupy the child; PCConn = media use to connect or relax together; ParYell = yelling at the child; ParComf = comforting or soothing the child. The figure only shows significant edge values given p < .05. Solid blue lines represent positive edge values. Dashed red lines represent negative edge values. Line thickness represents the magnitude of significant effects in the temporal network (range 0.03–0.15), contemporaneous network (range .03-.32), and between-subjects network (range .13-.49). See Supplemental Tables S3-S5 for a full report of all edge values.

Temporal Network

Results from the temporal network represent how each variable on a given day predicts another variable (or itself) the next day, accounting for all other associations. The model indicated that most nodes (variables) temporally predicted themselves (i.e., autocorrelated), except for child regulate (see Figure 1), and all significant edges reflected positive associations. While many edges emerged as significant, here we focus primarily on the associations related to our research questions, namely associations involving media use. No bidirectional associations or feedback loops emerged between nodes for media use and either parenting or child affect.

For single direction edges, however, media use both for regulating the child (ChReg) and for helping the parent relax (ParRelax) on a given day predicted more parent comfort and soothing (ParComf) the next day. That is, on days when parents reported more frequent media use than usual for child regulation or parent relaxation, they also reported more frequent comforting and soothing than usual the next day. Additionally, more frequent media use to connect on a given day (PCConn) predicted more frequent positive child affect (CPosAff) the next day compared to that child’s average. In contrast, more infant negative affect (CNegAff) than usual one day predicted more media use to occupy children (ChOcc) the next day.

Several significant edges also emerged among the five media factors. Media use to occupy children (ChOcc) on a given day predicted media use to regulate both the parent (ParReg) and the child (ChReg) the next day. Greater media use for parent relaxation (ParRelax) on one day predicted greater media use to regulate the parent (ParReg) and occupy the child (ChOcc) the next day. All significant edge values are visualized in Figure 1, and a full report of edge values is provided in Supplementary Table S3.

Contemporaneous Network

Results from the contemporaneous network represent partial correlations within the same day, controlling for all other associations (see Figure 1). This network revealed that all significant edge values from the temporal network were also significant in the contemporaneous network, with only one exception: Media to regulate the child (ChReg) was not related to parent comforting (ParComf) within the same day. Several additional connections emerged in this network as well. As with the temporal network, more frequent media use to connect (PCConn) than usual on a given day was associated with more child positive affect (CPosAff), and was also associated with more parent comforting (ParComf) than usual on the same day. Media use to occupy (ChOcc) was still associated with child negative affect (CNegAff), and was also associated with parent yelling (ParYell). Media use for parent relaxation (ParRelax) was still associated with parent comforting (ParComf) the same day, and was also associated with child positive affect (CPosAff). Although media use to regulate children (ChReg) was not associated with parent comfort (ParComf) as in the temporal network, it was positively associated with child negative affect (CNegAff). Finally, when parents reported more media use than usual to regulate parents (ParReg), they also reported more frequent yelling (ParYell) and child negative affect (ChNegAff) than usual the same day. Each of the media factors were also strongly connected with two or three other media factors within the same day. See Figure 1 and Supplemental Table S4 for edge values.

Between-Subjects Network

The between-subjects network revealed mean-level associations between variables, controlling for other variables, which is equivalent to partial correlations among the person-level means of each variable. In this network, none of the key associations (i.e., edges between media motivations and either parenting behavior or child affect) that emerged earlier in the temporal network were significant. Further, this network revealed more negative associations than the other two networks, in which edges were primarily positive. For example, parents who tended to use more media to regulate their own emotions in general (ParReg) also reported more frequent yelling at their children (ParYell), relative to other parents. Parents who used more media to regulate their children (ChReg) were also more likely to report yelling at their children (ParYell), but less likely to report child negative affect (ChNegAff). Parents who tended to report using media to relax (ParRelax) also tended to report more child positive (CPosAff) and negative affect (CNegAff), but less yelling (ParYell) compared to other parents. Finally, parents who used more media to connect (PCConn) also reported more frequent comforting of their children (ParComf).

Again in the between-subjects network, most of the media factors were associated with each other, suggesting that parents who use media often do so for many reasons. However, parents who reported using media more often to connect (PCConn) were less likely to report using media to regulate their own emotions (ParReg) compared to other parents. Further, parents who used media to relax (ParRelax) were less likely to say they used media to regulate their children’s emotions (ChReg) than other parents. See Figure 1 and Supplemental Table S5 for edge values.

Moderator Analyses

Results from the moderator analyses are reported in Supplemental Tables S6 and S7. These analyses focused on within-person effects that were significant in the main analysis and that involved a media motivation and another variable of interest (i.e., infant affect or parenting behavior). Of the thirteen effects examined (four temporal, nine contemporaneous), only one significantly varied by infant age. The main analysis indicated that more frequent media use than usual for regulating the child (ChReg) on a given day predicted more parent comfort and soothing (ParComf) than usual the next day. Regressing this association on infant age revealed that this temporal association was stronger among parents of older infants (B = 0.001, SE < 0.001, p = .014, R-squared = .012). Infant age did not significantly predict any other within-person effects.

Sensitivity Analyses

Sensitivity Analyses 1: Network Models with 20+ Observations

Results from this sensitivity analysis are visualized in Supplemental Figure S1 and reported in Supplemental Tables S8-S10. The overall pattern of results was quite consistent between the main model (401 participants with 15+ observations) and the sensitivity model (196 participants with 20+ observations). In the temporal network, 15 of 20 significant effects remained significant at p < .05, and another four were in the same direction and of similar magnitude with p < .10. Only the temporal link from parent-child media to connect (PCConn) to positive affect (CPosAff) was reduced in magnitude with a p-value > .10 in the sensitivity model. For the contemporaneous network, only three of 22 significant links in the main model were not significant in the sensitivity model, including two that involved media use: the link between between parent-child media to connect and child positive affect (PCConn and CPosAff) and that between child media to occupy and child negative affect (ChOcc and CNegAff). For the between-subjects network, only two of 20 significant links in the main model were not significant in at least one direction in the sensitivity model, including one that involved media use: the link between child media to regulate and child negative affect (ChReg and CNegAff).

Sensitivity Analyses 2: MSEM to confirm contemporaneous network

The overall pattern of results was consistent between the contemporaneous network in the main model and both MSEM sensitivity models (media predicting affect/parenting and affect/parenting predicting media). See Supplemental Tables S11 and S12 for a full report of MSEM results. For two associations only, the association was significant in one MSEM model but not the other (i.e., positive affect (CPosAff) predicted comfort (ParComf) but not the reverse; media for parent regulation (ParReg) predicted yelling (ParYell) but not the reverse). For the non-significant associations, however, the p-values were still low at .052 (ParReg on ParYell) and .066 (ParComf on CPosAff), which still supports the direction and overall pattern of our contemporaneous network findings.

Discussion

In this study, we examined associations among parents’ daily reports of media motivations, child affect, and parenting behavior among parents of infants 12–24 months old. As such, we observed temporal dynamics in regulatory processes during a developmental period that is characterized by children’s rapidly fluctuating mood states, reliance on caregivers for external regulation of affect, and rapid development of emotion regulation skills (Aktar & Pérez-Edgar, 2020; LoBue & Ogren, 2022; Pérez-Edgar, 2019; Tan et al., 2022). We used temporal network modeling to examine same-day (contemporaneous) and next-day (temporal) associations among key variables within dyads, as well as mean-level associations between dyads (between-subject). Of particular interest were parents’ regulatory and relational motivations for media use (e.g., to calm themselves or their children, occupy children, connect with children). The findings build on prior research by examining the temporal dynamics of parents’ media motivations, as well as how these dynamics relate to parent-reported fluctuations in child affect and parenting behavior.

Several key findings emerged from our analyses. First, the pattern of results for within-person effects (contemporaneous and temporal networks) differed markedly from the pattern of results for the between-person network, illustrating how global measures of typical behavior may mask fluctuations within dyads. Second, correlations between child age and intra-individual means on key variables were small to negligible, and our within-person results were mostly consistent across the age range studied, further underscoring the importance of daily fluctuations during the second year of life. Third, results from the temporal network revealed next-day associations that provide some insight into directionality (e.g., degree to which changes in child negative affect precede versus follow changes in media use). Fourth, the results varied as a function of parents’ media motivations. We discuss these patterns and their implications for parents’ adaptive coping and children’s emotion development in the sections that follow.

Infants’ Negative Affect Drives Parents’ Media Use for Coping

On days when parents reported that their children were fussier than usual, they also reported more frequent yelling at their children and more frequent media use to occupy their children, to regulate their children’s behaviors and emotions, and to regulate their own emotions on the same day. Critically, the temporal effects suggest child negative affect drives later media use, not the reverse. Specifically, higher-than-usual negative affect on a given day predicted higher-than-usual media use to occupy children the following day, which in turn predicted higher-than-usual media use to regulate both parent and child emotions the day after that. The reverse was not true: None of the media motivations predicted child negative affect the following day.

Our findings are consistent with growing evidence that parents use media as a coping mechanism (Wolfers & Schneider, 2021), and with survey studies showing that parenting stress among parents of young children is associated with higher child screen time cross-sectionally (Brauchli, Sticca, et al., 2024; Shin et al., 2021) and longitudinally, especially for children with higher externalizing behaviors (McDaniel & Radesky, 2020). Our findings are also consistent with studies examining longitudinal associations among children’s trait-level temperament and parent reports of media use. For example, higher trait levels of negative affect at 9 months predicted higher screen time at 24 months (Radesky et al., 2014). However, our findings depart from previous research insofar as the temporal link between infant negative affect and media use was not bidirectional in our study. For example, results from prior research show that higher regulatory media use at 18 months longitudinally predicted higher child negative affect at 24 months, albeit only for children with lower levels of negative affect at 18 months (Gordon-Hacker & Gueron-Sela, 2020; see also Radesky et al., 2022 for similar findings in preschoolers). Together, these findings suggest that temporal effects may differ by timescale. For instance, media use may be an effective tool for quickly calming infants in the short term (i.e., the same day or next day), but cumulative long-term effects may still be negative by displacing opportunities for children to develop their own self-regulatory skills. There may be developmental differences in these associations as well, with more associations emerging among older children (Vasconcellos et al., 2025).

Counter to our predictions, our within-person effects (both same-day and next-day) between child negative affect and media motivations did not emerge in the between-subjects network. That is, parents who reported more negative affect than other parents in general were no more or less likely to report media use to regulate their own emotions or to occupy children. In fact, parents who reported more frequent child negative affect also reported less frequent use of media to regulate children’s behaviors and emotional responses than other parents. This was unexpected given prior research linking toddlers’ trait-level negative affect to global estimates of screen time (Brauchli, Edelsbrunner, et al., 2024; Coyne et al., 2021; Radesky et al., 2014). This departure from past research may be due to differences in developmental sensitivity or trajectories, given that our sample, with a mean age of about 16 months (range 12–24 months), is younger on average than samples in most prior research with infants and toddlers. Perhaps associations between media use and social-emotional skills may be less evident among younger children who are still reliant on caregivers for external emotion regulation, or perhaps such associations accumulate over time, becoming more evident among older children who increasingly rely on self-regulatory skills.

Alternatively, the difference across studies may be due to measurement. For example, trait-level measures of infant temperament capture a wider range of behaviors than those included in our daily diaries. On the other hand, most prior studies rely on a single question asking about the use of media to calm children, whereas our composite measure of regulatory media use averaged across eight items (e.g., reduce conflict, help the child sit still or focus, occupy the child in public spaces, prevent the child from getting overwhelmed in new situations).

It is also possible that averaging across parents’ daily reports captures something qualitatively different from trait-level or global assessments. Global self-report measures require that respondents generalize across variable experiences or generate retrospective reports of discrete events, both of which can be impacted by recall biases and current mood (Schwarz, 2007; Thomas & Diener, 1990). Thus, intensive longitudinal designs that capture parent reports across multiple days offer more precise estimates of parents’ media motivations and how they relate to parenting behavior and child affect concurrently and over time.

Is Media Use an Adaptive Strategy for Parents?

While it is clear parents report using media as a coping strategy, it is less clear whether this use of media is adaptive or maladaptive. Here we consider whether the frequency of media use predicted parenting behavior. As discussed earlier, we found that higher-than-average infant negative affect on a given day predicted higher-than-average media use to occupy children the next day, which in turn predicted higher-than-average media use to regulate infant behavior the day after that. Going one step further, we found that increases in media to regulate infants on a given day in turn predicted increases in comfort and soothing the next day. These findings may provide some evidence of an adaptive strategy, with media use representing a welcome reprieve for distressed parents, allowing them to experience more positive interactions with their infant the next day. Alternatively, this finding could be interpreted as a compensatory mechanism whereby parents compensate for media use one day by actively engaging in more positive interactions with infants the next day (Linder et al., 2022). Regardless of how they are framed, our results illustrate that higher-than-average media use to regulate the child one day predicted higher-than-average positive parenting the next day, particularly among parents of older infants. Notably, this general pattern was robust in sensitivity analyses. As such, our findings suggest mixed effects of regulatory media use when combined with past research, which highlights the importance of using an intensive longitudinal design to understand the potential short-term effects of different media motivations.

Unlike comforting, there were no significant temporal links between yelling and media use. Parents who reported more frequent media use than usual to regulate their own emotions and to occupy their children also reported more frequent yelling than usual on the same day, but not the next day. In the absence of temporal effects, directionality is unclear. The absence of a temporal effect may be due to the low frequency of parent-reported yelling in this study. Alternatively, if parents use media as a strategy to avoid yelling at their children (e.g., by soothing fussy infants or giving parents a break to calm down), such effects could be fleeting; they may not be robust enough to decrease the frequency of yelling on the following day. Thus, the degree to which media use may be considered an adaptive strategy for parents likely depends on the specific indicator and timescale studied. For example, different findings might emerge for indicators of parents’ own subjective well-being and perceptions of media use as an effective coping mechanism to manage parenting challenges (Wolfers et al., 2023). Indeed, we have reported elsewhere that parents’ daily reports of parental burnout relate to their media motivations the same day and, in some cases, the next day (Kerr et al., 2025).

Of particular interest in the current study was parents’ use of media to connect or relax with their child. Previous research has linked parent-child joint media use with positive outcomes for young children. For example, joint media use appears to buffer against negative effects of screen time on language development while maximizing potential educational value (Dore et al., 2020; Ewin et al., 2021; Madigan et al., 2020). Our findings suggest there may be relational and regulatory benefits as well, at least in the short term. On days when parents reported higher-than-average media use to connect or relax with their children, they also reported higher-than-average infant positive affect and parent comforting the same day. Temporal effects suggest media use may be driving these associations, insofar as media to connect one day predicted positive affect the next day, and media use to relax one day predicted more comforting the next day (but not the reverse). Further research is needed to replicate this effect over a longer period of time given these temporal links were not significant in our sensitivity analysis using a substantially reduced sample with more timepoints.

Together, the temporal findings offer some evidence that media use may be an adaptive strategy for parents, at least in the short term. More media to regulate children predicted an increased frequency of comfort and soothing the next day, particularly for older infants. However, there was not a compensatory decrease in the frequency of yelling, perhaps due to the low frequency of parent-reported yelling in general. Similarly, more media use to connect and relax alone on a given day predicted more positive affect and positive parenting the next day, respectively. While there was some preliminary evidence that media use served as an adaptive short-term strategy, there was no clear evidence that media use was a maladaptive strategy (e.g., predicting less comforting or more yelling the next day).

Implications

Prior research shows that parents report using media as a regulatory tool to respond to rapid mood fluctuations during this developmental period (Coyne et al., 2021; Radesky et al., 2016; Suh et al., 2024), but the degree and direction of effects on infants’ emotional development remain unclear and understudied. By focusing on parents’ media motivations, and examining day-to-day fluctuations in these motivations as they relate to infants’ affective states and parents’ responses, we can better identify potential mechanisms that link media use to child outcomes (e.g., via parenting behavior). These dynamic factors may represent modifiable targets for intervention.

For example, our findings replicate and extend prior research, illustrating how child negative affect is driving media use, but not the reverse, at least across relatively short timescales. We did not find evidence that media use predicted infant negative affect the next day, yet prior studies examining much longer timescales suggest screen time can predict more child emotional and behavioral dysregulation over much longer timescales (Aishworiya et al., 2022; Gordon-Hacker & Gueron-Sela, 2020; Mallawaarachchi et al., 2024). Moreover, research with general adult populations suggests that adults’ own media use is an ineffective or maladaptive strategy for coping with negative emotions and stress (Greenwood & Long, 2009; Rasmussen et al., 2020). Together, the findings suggest media use may be an adaptive short-term strategy for parents in the moment, even while cumulative effects over the long term may be less positive. Testing these suppositions necessitates future research that captures these processes across both shorter (e.g., days) and longer (e.g., months) time scales simultaneously.

Our findings suggest that efforts to reduce children’s screen time consider parents’ media motivations, particularly as they relate to managing challenging child behaviors. For example, effective interventions might focus on providing parents with alternate strategies for regulating children. Simply recommending reductions in screen time may take away a seemingly effective parent coping strategy without providing a feasible alternative (Torres et al., 2021). Removing an accessible coping tool from parents, even if not the most adaptive in the long-term, may have unintended consequences for children in the short term, given that parents’ negative affect and stress both consistently predict harsh parenting behavior (Deater-Deckard, 1998; Rueger et al., 2011). Instead, prompting parents to reflect on their motivations for media use may help them shift to more adaptive strategies, such as reframing media use as an opportunity to connect with children or to relax and unwind, rather than to disconnect or escape.

Limitations and Future Directions

The current study has several strengths that advance research on this topic, including a focus on a development period that is understudied and characterized by rapid fluctuations in infant affect, the assessment of parents’ regulatory and relational media motivations in the context of such fluctuations, an intensive longitudinal design with a large sample size that revealed within-person effects, and a temporal network modeling approach that provided insights about directionality. That said, the study also has limitations that temper its implications and motivate future research directions.

First, the number of timepoints is relatively low for temporal network models (Blanchard et al., 2023). This required excluding several participants with insufficient data, substantially reducing our sample size (although there were few differences in demographic distributions among those who were excluded versus included). The relatively low number of timepoints also limited our ability to detect temporal effects because of less variability within (i.e., day to day) than between individuals. Future studies with more timepoints would allow for higher inclusion rates and more capacity for detecting within-person temporal effects.

Second, the heterogeneity in our data set was limited in a few ways, including low variability for some measures and reports from just one parent per family. While all of our models converged, it is possible that low variability led to floor effects for some associations that might otherwise have emerged. Variability was particularly low for some of the media motivations (e.g., to regulate parents and children), perhaps given the young age of focal children in this study (McArthur et al., 2022; Suh et al., 2024). Similarly, only two parenting behaviors were included, and one (yelling) had low variability. A more comprehensive measure of parenting experience that captures a wider range of behaviors from multiple reporters is likely to yield more variability, as would measures of parents’ subjective well-being (e.g., parenting stress), which likely relates to the constructs under study here. In addition, although this study provides a proof of concept for how temporal models can reveal mechanisms, our sample was largely white and highly educated, reducing generalizability. As such, this work could be extended to include multiple reporters per family, a wider range of parenting responses, and parents with more diverse backgrounds to capture more variability in focal behaviors and to increase the generalizability of the findings.

Third, our daily diary method allows us to observe day-to-day fluctuations and test both same-day and next-day effects. However, it is possible a one-day lag is insufficient to capture some media dynamics. For example, to the extent media use is an adaptive emotion regulation strategy for parents in the short term (e.g., reducing the likelihood of yelling at children), its effects may dissipate more quickly than our design could detect. Our temporal network results illuminated directionality for some same-day associations (e.g., infant negative affect on a given day consistently predicted media use to occupy children on the same day as well as the next day). However, some temporal links were not robust in sensitivity analysis. Other same-day associations that emerged in the contemporaneous network were not significant in the temporal network, making directionality uncertain. Having more timepoints would likely result in more temporal links, as described earlier. In addition, it is possible that some effects are more fleeting than a one-day lag can capture. Future research could test temporal dynamics over even shorter timescales by increasing measurement density (e.g., experience sampling several times per day).

Finally, while our intensive longitudinal design allows us to examine short-term temporal dynamics, it does not allow us to test potential longer-term outcomes. This is critical, given short-term effects are not always replicated in the long term. A full understanding of media effects should take into account the potential opportunity costs and gains for different family members, and how they collectively influence longer-term developmental cascades (Barr et al., 2024). For example, media use may represent an adaptive strategy for parents that helps them self-regulate in the moment, but it may also displace opportunities for rich parent-child interactions in ways that can accumulate over time to predict worse child outcomes in the long term (McDaniel et al., 2025). As such, future research could adopt an intensive longitudinal burst design to reap the benefits of measuring at both short and long time scales.

Conclusion

This study used temporal network analysis to reveal the dynamic nature of everyday media use as it relates to infant mood states and parenting responses during the second year of life, a developmental period characterized by rapid fluctuations in emotional state. As a result, we observed more complexity in the relations among infant affect, media use, and parenting behavior than what was possible in past research. The overall pattern of temporal results suggests infants’ negative affect drives media use, which in turn drives parenting responses, but not the reverse. Observed patterns differed based on parents’ media motivations, with some motivations seeming to be more adaptive than others. Importantly, within-person results from same-day (contemporaneous) and next-day (temporal) networks revealed associations that were otherwise masked in between-person analyses. The fact that infant age was at most weakly correlated with variables of interest, and rarely moderated our main findings, underscores the importance of examining within-dyad processes that reflect the rapid fluctuations characteristic of the second year of life. By understanding what drives parents’ decisions about using media with and around their young children, we will be better able to identify underlying mechanisms and help support parents in building a healthy family media ecology.

Public Health Significance:

This study examined day-to-day fluctuations in parent-reported media use, infant mood, and parenting behavior. The findings show how parents use media to cope with parenting challenges (e.g., occupying children, helping parents and children calm down when they are upset, connecting with each other). In some cases, this appeared to be a short-term adaptive strategy because it predicted positive parenting on the same day and also the next day.