Positive Parenting, Parenting Stress, and Child Self-Regulation Patterns Differ Across Maternal Demographic Risk

Abstract

We examined relations between positive parenting, parenting stress, and children’s regulatory abilities across infancy and early childhood. First-time mothers and their infants (N = 682) were recruited prenatally. Mothers belonged to one of three potential demographic risk groups: adolescent mothers (AM), adult low-educated mothers (LEM), and adult high-educated mothers (HEM). Parent–child interactions were observed in the home and scored for positive parenting (at 4, 8, and 18 months) and child self-regulation (at 4, 8, 18, and 30 months). Parenting stress was measured by questionnaire at 6, 12, and 24 months. A multigroup cross-lagged panel model indicated different patterns of effects for adolescent and adult mothers: for AM, positive parenting and parenting stress were related early in infancy, but no relations emerged with regulation until 18 months. For LEM, parenting related to regulation only at later ages, and for HEM, positive parenting, parenting stress, and regulation were associated early in the infant’s life but not later. Differing patterns of influence indicate, perhaps, the importance of considering demographic risk and timing of parenting in childhood in relation to child regulatory outcomes.

Effective regulation is a hallmark of early socioemotional development (Kopp, 1982) and relates to later cognitive, social, and academic outcomes (Eisenberg et al., 2005). Self-regulation is a multidimensional construct comprising cognitive (executive function and attention), emotional (emotion regulation and effortful control), and behavioral domains. Self-regulation includes a wide range of skills aimed toward motivating or modifying behavior in order to achieve a goal (Cole et al., 2019; Gagne et al., 2021; Kopp, 1982; McClelland & Cameron, 2012; McClelland et al., 2007; Nigg, 2017). Children learn effective regulatory strategies in large part through repeated interactions with caregivers, who guide children’s emotional development through the use of positive parenting practices that incorporate sensitivity, warmth, and responsiveness (Baker et al., 2011; Eisenberg et al., 1998; Morris et al., 2007). Infants use distraction and self-soothing to regulate, but also rely on caregivers to aid in effective regulation by making bids for their caregiver’s attention to elicit soothing behaviors (Planalp & Braungart-Rieker, 2015). As infants age into toddlerhood, they develop the cognitive and motor skills necessary for more independent forms of regulation (Montroy et al., 2016), and by the time children are about 3 years old, they are expected to be able to follow the rules and regulate their behaviors and emotions in a social context; this regulation is observable in many everyday situations, for example, at the grocery store with caregivers or in a doctor’s office waiting room. Thus, infancy and toddlerhood provide a salient developmental period during which to study processes important in emerging regulation. In this work, we define self-regulation as children’s ability to diffuse negativity and engage in socially appropriate emotions and behaviors.

Parenting Processes Impacting Regulation

The parent–child relationship is integral to the development of healthy and effective self-regulation. Early in infancy, there are individual differences in how and when children learn to regulate their behaviors and emotions depending on caregiver and environmental characteristics (Baker, 2018; Fay-Stammbach et al., 2014). Children of parents who are more responsive to infant-needs regulate more effectively from 3 to 14 months (Braungart-Rieker et al., 2014). On the other hand, higher parenting stress negatively relates to the regulatory abilities of 3-year-old children (DeGangi et al., 2000). In this work, we examine how and when mothers’ positive parenting and parenting stress might impact a child’s developing self-regulation, and how these patterns may differ depending on maternal demographic risk.

Positive Parenting

Positive parenting is a relatively broad dimension that reflects a mother’s warmth, contingent responsiveness toward her child’s affective and physical cues and needs, affective expression, vocal positivity and praise, and physical affection toward the child (Landry et al., 1997, 2006). Responsive mothers react to their children’s needs during dyadic, everyday interactions using multiple types of behaviors: speaking to the infant, engaging the infant in shared enjoyment, or encouraging the infant to attend to something in the environment (Bornstein & Manian, 2013). Evidence consistently connects positive parenting to multiple indicators of child self-regulation (Baker, 2018; Fay-Stammbach et al., 2014). Mothers who are more sensitive with infants from 3 to 7 months have more regulated infants at 12–14 months (Braungart-Rieker et al., 2014); mothers who express more warmth at 20 months have toddlers who were more regulated at 34 months of age (Jennings et al., 2008); and maternal positive parenting at 30 months predicted regulation (in the form of effortful control) at 42 months (Spinrad et al., 2012). In a sample of German kindergartners, maternal warmth, but not response to distress, was related to behavior regulation during a snack delay task, whereas contingent response to distress was specifically related to children’s rule-following (von Suchodoletz et al., 2011). Additionally, the timing of mother’s positive parenting in a child’s life may be important in the development of regulation, with effective parenting in early infancy having downstream effects on child outcomes. In a study examining mother–child dyads at 12, 24, 40, and 54 months of age, early and concurrent responsiveness had a direct effect on children’s later social skills (Steelman et al., 2002). Of note, responsiveness was measured only at 12 and 54 months, not at 24 or 40 months of age. In sum, positive parenting during the first years of life, a period when infants rely on parents for help in learning how to engage with the world, plays an important role on how children learn to regulate, which relates to a variety of child outcomes.

Parenting Stress

Parenting stress incorporates the task demands of parenting, parents’ behavior and well-being, the parent–child relationship, and the child’s adjustment (Deater-Deckard, 1998). While most parents show normative levels of stress associated with daily experience of parenting, others experience more extreme stress that impacts parenting and family functioning. High parenting stress is associated with increased child behavior problems (Crnic et al., 2005), but there are relatively few studies that relate parenting stress to child regulation, specifically. In early school-aged children, parenting stress, but not positive parenting practices, was associated with poor regulation (Mathis & Bierman, 2015). In the only study we found that specifically related early parenting stress to infant regulation, Riva Crugnola et al. (2016) did not find a significant relation between parenting stress and regulation at 3 months of age. However, they did not extend their work to older ages even though cumulative stress can build over time and increase risk for poor parenting and child functioning at later ages (Crnic et al., 2005). Feldman et al. (2004) found that higher levels of parenting stress when infants were 3 months of age was associated with lower levels of maternal sensitivity, which in turn interfered with infant cognitive development at 12 months (Feldman et al., 2004). Though these findings are not specific to self-regulation in early childhood, we build on these two studies by examining relations between maternal positive parenting behaviors, parenting stress, and the development of self-regulation through the toddler years. As self-regulation is impacted by the parent–child relationship (Braungart & Stifter, 1991; Calkins & Hill, 2007), parenting stress likely relates to the child’s developing regulatory abilities.

Alternatively, there are likely bidirectional relations between parenting stress and child regulation, such that parents of children who are more difficult may experience higher stressors. Specifically, 12-month-old infants’ difficult temperament relates to higher maternal parenting stress (Molfese et al., 2010), and mothers of difficult preschoolers report higher stress (Coplan et al., 2003). Thus, we examine how parenting might predict child regulation, as well as how child self-regulation relates to subsequent positive parenting and parenting stress. In addition, parenting stress may also indirectly impact children’s regulation through proximal parenting processes. Limited work indicates that, though there are links between stress and other parenting practices, this does not necessarily impact child regulation directly (Anthony et al., 2005; Crnic et al., 2005; Mathis & Bierman, 2015). Of note, the limited work examining indirect effects of parenting stress on child regulation studies toddler and preschool-aged children, but not infants.

Maternal Demographic Risk

Demographic risk has been operationalized in numerous ways, including but not limited to measuring aspects of income, education, marital status, or minority status. In general, mothers at higher demographic risk (lower education, income, or social support) are less prepared for parenting more generally, and less likely to develop the skills needed for positive parenting in particular (Lanzi et al., 2009; Mollborn & Morningstar, 2009). Two groups of mothers that tend to be at higher demographic risk are adolescent mothers (AM; Wakschlag et al., 2000) and mothers with lower educational attainment (LEM; Liaw & Brooks-Gunn, 1994); these mothers often have access to fewer resources than high-educated adult mothers (HEM; Coley & Chase-Lansdale, 1998; Savio Beers & Hollo, 2009). Though results vary, AM and LEM are at heightened risk for parenting stress, financial strain, and problems with mental health (e.g., depression; Huang et al., 2014; Whitson et al., 2011). Both AM and LEM exhibit lower positive parenting and harsher parenting practices with their children when compared to parents who are older (Coley & Chase-Lansdale, 1998) or have higher education (Richman et al., 1992). In addition, children in families with more limited resources may be at higher risk for behavioral problems (Whitman et al., 2001) and poorer regulatory skills ( Raver, 2004). In one study examining risk at child age 18 months and regulation and parenting at 30 months, maternal demographic risk, a composite of maternal age, education, income, and number of children, predicted both of the later behaviors directly (Taylor et al., 2013). Here, we operationalize demographic risk using mothers’ age and educational attainment by comparing patterns of parenting behaviors in AM, LEM, and HEM.

The Present Study

Positive parenting helps infants learn how to better regulate distress, and within contexts of high stress, infant’s self-regulation is linked to mothers’ demographic risk (Feldman et al., 2004). Previous work has highlighted the importance of early positive parenting and later child outcomes (Steelman et al., 2002; Wakschlag & Hans, 1999). In addition, AM and LEM can be at greater risk for parenting stress, mental health issues, and poorer parenting practices. Yet there is a dearth of research that links these concepts over time in early childhood, and specifically examines how maternal context impacts the development of children’s self-regulation. Previous work examining sex differences in child regulation is mixed; two meta-analyses indicate that girls are generally rated higher in regulation than boys from an early age (Chaplin & Aldao, 2013; Else-Quest et al., 2006). Other work finds no sex differences in child regulation using both parent report and observational scoring of infants and preschoolers (Cole et al., 1996; Planalp et al., 2019). In addition, previous research has not found that mothers’ positive parenting impacts boys and girls differently (Jennings et al., 2008; Planalp et al., 2019; Smith et al., 2006; von Suchodoletz et al., 2011). The goal of this work is to elucidate pathways through which positive parenting, parenting stress, and child regulation vary across three groups of first-time mothers: mothers who were adolescents at the time of their child’s birth, low-educated adult mothers (who differ from AM only in age), and a smaller comparison group of educated, adult mothers (who differ from low-educated adult mothers only in education level). Our hypotheses are as follows:

1. In general, higher positive parenting will relate to higher child regulation and increased parenting stress will relate to lower positive parenting and lower child regulation.

2. We expect that girls’ ability to regulate may be rated higher than boys’, but do not expect sex differences in how regulation is associated with parenting behaviors.

3. We anticipate that AM and LEM will experience higher levels of stress and lower positive parenting than HEM. Using HEM as a comparison group, we will explore whether maternal age (by comparing to AM) or maternal education (by comparing to LEM) differentially relates to parenting behaviors. We anticipate that AM, as they are likely less prepared for parenthood and have less parenting knowledge, will have higher parenting stress and poorer parenting skills.

4. Patterns through which parenting impacts child regulation will vary by group; as AM and HEM may experience more stressful contexts, we anticipate that relations between parenting stress and child self-regulation will also be stronger in these groups.

Method

Participants

First-time mothers were recruited as part of the Parenting for the First Time Project, a prospective 3-year longitudinal study designed to understand neglectful parenting among mother–child dyads (Borkowski et al., 2012). Mothers were recruited from 2002 to 2005 in four cities across the United States: South Bend, Indiana; Kansas City, Kansas and Missouri; Washington, DC; and Birmingham, Alabama. The sample included 682 mother–infant dyads (48.2 % male infants) who participated in at least one timepoint of data collection. Mother age ranged from 15 to 35 years; monthly household income ranged from $415 to more than $15,000; mothers reported on their own race, with 64.5% of the sample reporting themselves to be African American, 18.8 % European American, 15.2% Hispanic/Latina, and 1.4% of other ethnicities. More detail on sample demographics is presented in Table 1.

Table 1.

Demographic Characteristics of the Sample

Demographic	Total sample	Adolescent mothers	Low-educated mothers	High-educated mothers
N	682	396	169	117
Infant sex	330 females	186 females	80 females	64 females
	329 males	199 males	79 males	51 males
Mother age in years (M and SD)	21.23 (5.09)	17.50 (1.13)	25.48 (3.04)	27.88 (3.94)
Average yearly income at infant age 6 monthsa	$22,185	$15,818	$16,842	$40,352
Marital status (% by group)
Single	61.1%	72.2%	57.4%	29.1%
Married	16.0%	2.3%	19.5%	57.3%
With partner	21.4%	24.2%	20.1%	13.7%
Other	.5%	0%	2.4%	0%
Educational attainment (% by grade completed)
Less than 8th grade	2.5%	2.0%	5.3%	0%
Less than 12th grade	42.4%	64.4%	20.2%	0%
Completed 12th grade	24.2%	26.3%	35.5%	.9%
Some college or vocational school	12.4%	1.1%	28.5%	4.3%
4-year college degree	14.1%	.5%	5.3%	72.6%
Postgraduate degree	3.8%	0%	0%	22.2%
Mother race/ethnicity (% by group)
European American	18.8%	13.9%	15.4%	40.2%
African American	64.5%	69.9%	65.1%	45.3%
Hispanic/Latina	15.2%	14.5%	19.0%	12.8%
Other	1.4%	1.8%	.6%	1.8%

Note.

^aThere is significant missing data (almost 50%) on reported income. All other demographic data is mostly complete, though some participants did not report all information so not all rows/columns total 100% or N = 682. Unless otherwise noted, demographics are reported at the time of the infant’s birth.

Procedure

Mothers were recruited through primary care facilities during the third trimester of pregnancy. Mothers were interviewed prenatally, and mother–infant dyads were followed at infant age 4, 6, 8, 12, 18, 24, 30, and 36 months. The prenatal and 6-month visits were typically completed in a laboratory setting and included surveys and interviews; visits at 12, 24, and 36 months added in child developmental testing. Visits at 4, 8, 18, and 30 months occurred in a home setting and included a short survey, an interview about the home environment, and observed mother–child interaction. We use available longitudinal data from the 4- to 30-month home visits and questionnaire data at 6, 12, and 24 months.¹ The home interviews at 4, 8, 18, and 30 months typically took 1.5–2 hr and included observations of mother–child interaction (Landry et al., 1997) and interview and observation of the home environment (HOME; Caldwell & Bradley, 1984). Mothers also were interviewed and responded to questionnaires about family history and structure, parenting perspectives, stress, activities, and support. The study was conducted with Institutional Review Board approval from each research site; analyses presented in this work were not preregistered. Data are available upon request from the first author.

Measures

Self-Regulation

Infant and child regulation was scored when infants were 4, 8, 18, and 30 months of age. During home visits, mothers were instructed to carry out their typical daily activity within proximity of their infant for a total of 30 min. Each 30-min session was broken into a 2-min warm up and four 5-min observation periods, allowing for breaks between each period for the interviewer to record notes. During each 5-min period, several dimensions of child behavior were scored (Landry et al., 1997); in the current project we focus on the infant’s self-regulation. Raters were instructed to observe the infant’s behavior, make notes, and were given 7 “anchors” or exemplars describing the child’s actions during each scoring epoch. Anchors included: hypersensitivity to the environment, negative affect, adjustment to change, low frustration tolerance (indicated by both fussiness and lack of persistence), ability to make wants known in appropriate ways (words or gestures instead of whining or throwing objects), needing a lot of help from mothers to organize behaviors, and ability to sustain attention on objects/environment. Notably, because infants’ motor and cognitive skills as well as engagement with the environment change over time, examples for regulated behavior were age-appropriate. For example, exemplars for hypersensitivity to the environment, ability to not be overwhelmed by negative emotions, and frustration tolerance were the same at all scoring ages. The 4- and 8-month behavior exemplars differed from the 18- and 30-month observations on the child’s ability to make wants known in an appropriate way as well as their ability to focus attention as these behaviors change with age (using words instead of crying). Therefore, scores are reflective of the child’s self-regulation within their age-appropriate developmental period, but still on the same scaling with overall levels of regulation ranging from “unregulated” (1) to “always regulated” (5) on each scale for each 5-min interval. The scores were averaged across 5-min intervals to create one score for self-regulation at each time point (4, 8, 18, and 30 months). A local master coder was trained by Susan Landry’s master coder, who then trained interviewers on the study team to score child and parent behaviors in vivo. Coders were not able to score independently until they reached an 80% agreement with the master coder using both video-taped and live observations. Inter-rater agreement at 4, 8, 18, and 30 months, on average, ranged from 80% to 84% agreement. The master coder also took part in home visits to code live with research assistants in vivo every 6 months to maintain reliability and avoid coder drift. At 4 and 8 months, 23.2% of the visits were double-coded and at 18 and 30 months, 24.6% of the visits were double-coded. Reliability for observed behaviors was calculated using a weighted kappa; for child self-regulation, κ = .82 for the 4- and 8- month visits and κ = .82 for the 18- and 32- month visits.

Maternal Positive Parenting

A Positive Parenting composite was created using four observed measures of maternal parenting: Maternal Warmth subscale from the Infants/Toddlers Home Observation for Measurement of the Environment (HOME; Caldwell & Bradley, 1984) and three items from the Landry Scoring (mother’s display of positive affect, warmth/sensitivity, and flexible responsiveness; Landry et al., 1997). The HOME measurement consists of a total of 45 items rated as either present or absent. Ratings are based on observation of the child in the home environment throughout the home visit, as well as an interview with the mother. Similar to the Landry, a master coder was trained by Robert Bradley, and local coders had to reach 90% reliability with the master coder before independent scoring. Higher scores reflect a more enriching environment for the child. The original scoring for the HOME offers a variety of subscales; for the purposes of this study, we used one of the conceptually designed subscales created by analyzing the data across four large data sets: maternal warmth (Fuligni et al., 2004; Linver et al., 2004). For the Landry scoring, parental behavior was observed during the same time as the child behavior using the same procedures described above for child self-regulation. Display of positive affect was quantified by counting the number of times a mother smiled at her child. The amount of maternal warmth/sensitivity displayed was scored on a 5-point scale ranging from 1 “almost never” to 5 “almost always,” with exemplars specifying, for example, engagement, proximity, enthusiasm, affection, acceptance, concern, and praise. Maternal flexible responsiveness was rated on a 5-point scale ranging from 1 “almost completely uninvolved” to 5 “almost always displays,” with exemplars or descriptions for patience, prompt and sensitive response to both affective and nonaffective cues, attention/ interest, or involvement. Similar to regulation described above, more than 80% reliability with interviewers and a master coder was achieved in both video-taped and live observations. Inter-rater agreement at 4, 8, and 18 months for positive affect, warmth, and maternal responsiveness ranged from 94% to 96% agreement. Like the child self-regulation in vivo coding, 23.2% of the visits at 4 and 8 months were double-coded and at 18 and 30 months, 24.6% of the visits were double-coded. Reliability for parenting from the Landry scoring system was κ = .65 for the 4- and 8-month visits and κ = .61 for the 18- and 30-month visits. Though these reliability estimates are low, Landry items were combined with maternal warmth from the HOME. All four items reflecting maternal positive parenting were z-transformed to account for different scoring used by the HOME and the Landry systems, and an average was computed. Overall internal consistency of the maternal positive parenting scale was α = .81 at 4 months of age, α = .84 at 8 months of age, and α = .83 at 18 months of age.

Maternal Parenting Stress

Mothers completed the Parenting Stress Index—Short From (PSI; Abidin, 1990) at 6, 12, and 24 months of age. The 36-item form has three subscales (Difficult Child, Parental Distress, and Parent–Child Dysfunctional Interaction) and a Total Stress score. The Total Stress score was used in analyses. Items were rated on a 5-point scale with higher scores indicating more stress. Internal consistency was very high for the overall stress of this sample (α range from .90 to .94 across time point).

Maternal Demographic Risk Group

During recruitment, mothers were delineated into one of the three potential demographic risk groups: (a) adolescent mother (AM; <19 years old at the time of the infant’s birth; N = 396), (b) adult low-educated mother (LEM; > 21 years old at the time of the infant’s birth with less than 2 years of college education; N = 169), and (c) adult high-educated mother (HEM; >21 years old at the time of the infants birth with at least 2 years of college education; N = 117). Extant previous work includes income as a source of demographic risk (Carneiro et al., 2013; Davis-Kean, 2005; Taylor et al., 2013). Of the 682 first-time mothers in our study, less than half reported on income when the infant was 6 months old (n = 316), with only slight increases in percent reporting for subsequent time points. Therefore, we are not able to accurately include income as an indicator of maternal demographic risk. We did, however, have complete data on the maternal risk groups (age and education) as defined by the recruitment strategy. Therefore, based on age and educational attainment, the AM group represents the relatively higher risk mothers whereas the HEM group represents the relatively lower risk mothers.

Data Analysis Plan

The first step in our analyses was to use analysis of variance (ANOVA) to examine differences in child self-regulation, positive parenting, and parenting stress by maternal demographic risk group. We also examined main effects for sex differences in child self-regulation to determine how child sex may influence more complex models. Though ANOVAs identify group differences, they do not elucidate associations between variables. To answer the question of “how and when” positive parenting and parenting stress relate to child self-regulation, we used structural equation modeling (SEM) to examine patterns of relations between our variables of interest. SEM is a robust method for conducting path analysis with multiple variables across multiple time points. A specific type of SEM model, the cross-lagged panel model, is used to identify reciprocal effects of two or more variables on each other using multiple repeated measures of the same variables. We ran three sets of models: a full model with no grouping variables, a model that included mother demographic risk group, and a model that added child sex comparisons. We also included all possible indirect paths linking parenting to child regulation.

Results

Results are organized into three sections. First, basic descriptive statistics are presented and interpreted. Second, we compare study variables across infant sex, time, and mother resource group. Lastly, we test the full path model for each mother resource group. Correlations within and across each variable for the entire sample are in Table 2. Child regulation was significantly correlated at 4 and 8 months, and again between 18 and 30 months. There was no significant relation between 8 and 18 months. Parenting stress was significantly correlated across time, with rs ranging from .53 to .66 (all p values <.0001). Positive parenting was also significantly correlated across time, with rs ranging from .39 to .61 (all p values <.0001). Correlations indicate that child regulation at 18 and 30 months was inversely related to parenting stress at 12 and 24 months, such that children who were less regulated at 18 months had mothers reporting higher parenting stress at 24 months; children who were less regulated at 30 months had mothers reporting higher parenting stress at 12 months. Children who were more regulated at 18 and 30 months had mothers who exhibited more positive parenting behaviors at each age, with significant rs ranging from .17 to .28 (all p values <.001).

Table 2.

Correlations Between Variables and Infant Sex Differences Across Total Sample

Variable or statistic	1	2	3	4	5	6	7	8	9	10
1. 4-Month child self-regulation	1
2. 8-Month child self-regulation	.53***	1
3. 18-Month child self-regulation	0.06	−0.01	1
4. 30-Month child self-regulation	0.06	0.01	.22***	1
5. 4-Month positive parenting	0.004	−0.06	0.06	.19**	1
6. 8-Month positive parenting	−0.05	0	0.04	.22***	.60***	1
7. 18-Month positive parenting	−0.04	−0.07	.17**	.27***	.49***	.61***	1
8. 6-Month parenting stress	0.1	−0.05	−.19**	−0.1	−.22***	−.34***	−.21**	1
9. 12-Month parenting stress	0.003	0.01	−.20**	−.15*	−.17**	−.20**	−.14*	.66***	1
10. 24-month parenting stress	0.03	−0.01	0.06	−0.15	−0.07	−.20*	−.20*	.58***	.67***	1
Total N	463	399	368	341	465	398	372	333	344	196
Total M (SD)	4.47 (0.84)	4.61 (0.77)	4.48 (0.65)	4.62 (0.62)	0 (0.80)	0 (0.82)	0 (0.81)	69.31 (16.81)	68.58 (17.64)	68.84 (21.17)
Female/male N	230/233	205/194	179/189	168/173	231/234	205/193	181/191	174/159	179/165	110/86
Female M (SD)	4.52 (.75)	4.63 (.68)	4.58 (.61)	4.73 (.51)	.04 (.77)	.09 (.80)	.14 (.74)	66.82 (15.26)	66.04 (17.44)	68.02 (21.05)
Male M (SD)	4.42 (.93)	4.58 (.86)	4.38 (.67)	4.52 (.51)	−.04 (.83)	−.08 (.83)	−.13 (.86)	72.04 (18.01)	71.33 (17.50)	69.90 (21.41)
t-value (df)	1.24 (461)	.76 (397)	3.04 (366)	3.19 (339)	1.16 (463)	2.11 (396)	3.27 (370)	−2.87 (331)	−2.81 (342)	−.61 (194)
p value	.217	.45	.002	.002	.247	.036	.001	.004	.005	.539

Note. Positive parenting is measured with a composite of z-scores, so means are 0; t denotes the difference between female and male children; significant differences are also bolded.

^*p < .05.

^**p < .001

^***p < .0001.

Child Sex Differences

We used independent samples t-tests to compare each variable for boys and girls across the entire sample. Full results are in Table 2. Mothers of boys reported higher parenting stress at 6 and 12 months, but not 24 months, and mothers were more positive with girls than boys at 8 and 18 months. For child regulatory behaviors, girls were significantly more regulated at 18 and 30 months of age (girls’ means of 4.58 and 4.73 compared to boys’ means of 4.38 and 4.52, respectively).

Mother Risk Group Differences

Multigroup comparisons used ANOVA to compare each variable of interest across maternal demographic risk groups (AM, LEM, and HEM). Results indicated that child regulatory behaviors did not differ across maternal risk group until 30 months of age, when children of HEM had significantly higher regulation than children of AM or LEM. AM reported significantly higher parenting stress than LEM and HEM at 6 months, but at 12 and 24 months, the AM and LEM both reported higher parenting stress than the HEM. Finally, AM were consistently lower in positive parenting than either LEM or HEM (See Table 3).

Table 3.

Group Differences in Each Variable of Interest by Mother Demographic Risk Group Within and Across Time-Point

Variable	N AM/LEM/HEM	Total M (SD)	AM M (SD)	LEM M (SD)	HEM M (SD)	Group comparisons	Group differences F (df)	p value
4-month child self-regulation	264/109/90	4.47 (.84)	4.50 (.75)	4.56 (.84)	4.38 (1.08)	AM = LEM = HEM	.65 (2, 460)	.520
8-Month child self-regulation	224/92/83	4.61 (.77)	4.62 (.65)	4.54 (1.00)	4.64 (.81)	AM = LEM = HEM	.53 (2, 396)	.587
18-Month child self-regulation	208/89/71	4.47 (.65)	4.44 (.65)	4.47 (.72)	4.60 (.54)	AM = LEM = HEM	1.56 (2, 365)	.211
30-Month child self-regulation	181/88/72	4.62 (.62)	4.53 (.72)a	4.66 (.57)	4.79 (.33)b	AM = LEM < HEM	4.68 (2, 338)	.010
4-Month positive parenting	226/109/90	.0012 (.80)	−.18 (.81)a	.15 (.71)b	.34 (.71)b	AM < LEM = HEM	18.33 (2, 462)	.000
8-Month positive parenting	224/91/83	.0039 (.82)	−.22 (.83)a	.16 (.74)b	.43 (.66)b	AM < LEM = HEM	23.77 (2, 395)	.000
18-Month positive parenting	21190/71	.0038 (.81)	−.20 (.80)a	.10 (.76)b	.48 (.70)c	AM < LEM < HEM	21.71 (2, 369)	.000
6-Month parenting stress	193/70/70	69.31 (16.81)	74.07 (16.47)a	65.21 (16.24)b	60.29 (13.35)b	AM > LEM = HEM	22.49 (2, 330)	.000
12-Month parenting stress	187/80/77	68.58 (17.64)	71.14 (17.72)a	71.73 (17.72)ab	59.08 (14.63)b	AM = LEM > HEM	15.64 (2, 341)	.000
24-Month parenting stress	95/42/59	68.84 (21.17)	73.86 (23.00)a	68.50 (15.90)ab	61.00 (19.16)b	AM = LEM > HEM	7.15 (2, 193)	.001

Note. Significant differences are also bolded. AM = adolescent mother; LEM = adult low-educated mother; HEM = adult high-educated mother

^abcsuperscripts indicate group differences.

Cross-Lagged Panel Models Relating Parenting to Child Self-Regulation

We used MPlus Version 7.2 (Muthen & Muthen, 1997) to examine SEM relating each variable across time. All variables were examined for normality prior to conducting path models: child regulation did not meet normality assumptions, so analyses used the MLR estimator in MPlus, a maximum likelihood estimation that computes robust standard errors using a sandwich estimator. This more appropriately accounts for nonnormality as well as missing data (Yuan & Bentler, 1998). We use a slight variation of a classic cross-lagged panel model as our time points for measurement are not the same for all variables: parenting stress is measured at 6, 12, and 24 months; parenting measured at 4, 8, and 18 months; and regulation measured at 4, 8, 18, and 30 months. In the specified models, autoregressive paths (those within each variable across time) were allowed to vary and each path was estimated (no within or cross-variable or time constraints) for each maternal risk group. MPlus provides absolute fit statistics: the comparative fit index (CFI; Bentler, 1990) and the root mean square error of approximation (RMSEA; Hu & Bentler, 1999). A CFI above .90 and RMSEA below .08 indicate that the data fit the specified model.

Determination of the "Best-Fitting" Model

First, we estimated a model constraining all estimates to be equal across group (i.e., a single group model). The model did not fit the data (CFI = .81, RMSEA = .07). Next, we ran the same model removing equality constraints across groups which fit the data well (CFI = .92, RMSEA = .05). A likelihood ratio test comparing the simple model to the model with mother risk as a grouping variable indicated that the single group model fit the data significantly worse than the multigroup model, χ²(155) = 290.98, p < .001, indicating that patterns of relations differed across AM, LEM, and HEM groups. Lastly, we included both mother risk group and child sex as grouping variables. This model also did not fit the data (CFI = .81, RMSEA = .08). Model comparisons indicate that the path model without child sex as a grouping variable fit the data significantly better than the model that did include sex, χ²(117) = 201.68, p < .008. However, as child sex differentiated mean levels of child regulation at 18 and 30 months, we included child sex as a covariate in the final model.² Model fit indices for the final group model controlling for child sex were χ²(75) = 133.20, p < .001, CFI = .92, RMSEA = .06. The full model is illustrated in Figure 1.

Figure 1.

Full Pathway Model and Significant Paths for Adolescent Mothers, Adult Low-Educated Mothers, and Adult High-Educated Mothers; Estimates Are Standardized Path Estimates

Comparing Patterns Across Mother Resource Groups

Full results of pathway estimates and significance are in Table 4 and Figure 1. Some significant pathways were consistent across groups: child sex was unrelated to regulation in the full model; parenting stress and positive parenting were consistent across time for AM, LEM, and HEM; and infant regulation was significantly related between 4 and 8 months, but not across other time points for all mothers. For AM, parenting stress and positive parenting were related early in infancy, with no significant pathways relating to infant regulation. Twelve-month parenting stress was inversely related to 18-month regulation, such that mothers reporting higher parenting stress had children with lower regulation. At 18 months, mothers who exhibited more positive parenting had children with higher regulation at both 18 and 30 months of age. Results for LEM did not elucidate any significant pathways early in infancy, but a similar pattern to AM emerged, such that 18-month positive parenting was positively related to child regulation at 18 and 30 months of age. Results for HEM show that parenting stress at 6 months was inversely related to positive parenting at 8 months, and higher positive parenting at 4 months was inversely related to infant regulation at 8 months.

Table 4.

Standardized Beta Estimates for Full Path Model by Maternal Demographic Risk Group

	Parameter estimate
Pathway	AM	LEM	HEM
Child regulation 4m → male	−.05	−.11	−.06
Child regulation 8m → male	.06	−.01	−.02
Child regulation 18m → male	−.11	−.06	−.14
Child regulation 30m → male	−.10	−.17	−.10
Child regulation 4m → child regulation 8m	.34**	.55***	.67***
Child regulation 8m → child regulation 18m	−.03	.08	.03
Child regulation 18m → child regulation 30m	.15	.01	.22
Positive parenting 4m → positive parenting 8m	.53***	.57***	.56***
Positive parenting 8m → positive parenting 18m	.54***	.50***	.71***
Parenting stress 6m → parenting stress 12m	.71***	.54***	.68***
Parenting stress 12m → parenting stress 24m	.56***	.71***	.76***
Parenting stress 6m → positive parenting 8m	−.20**	−.04	−.20*
Parenting stress 12m → positive parenting 18m	−.01	−.18	−.04
Positive parenting 4m → parenting stress 6m	−.18*	−.16	−.08
Positive parenting 8m → parenting stress 12m	.06	−.17	.01
Positive parenting 18m → parenting stress 24m	.11	−.16	−.15
Positive parenting 4m → child regulation 4m	.04	−.04	.03
Positive parenting 8m → child regulation 8m	.09	.02	−.02
Positive parenting 18m → child regulation 18m	.19**	.33**	.20
Child regulation 4m → parenting stress 6m	−.06	−.03	.07
Child regulation 8m → parenting stress 12m	.08	.01	.05
Child regulation 18m → parenting stress 24m	−.03	−.07	.01
Positive parenting 4m → child regulation 8m	−.03	−.01	−.12*
Positive parenting 8m → child regulation 18m	−.01	.30	.17
Positive parenting 18m → child regulation 30m	.18*	.40***	.06
Parenting stress 6m → child regulation 8m	−.13	.04	−.01
Parenting stress 12m → child regulation 18m	−.18*	−.02	.06
Parenting stress 24m → child regulation 30m	−.19	.08	.17

Note. AM = adolescent mother, LEM = adult low-educated mother, HEM = adult high-educated mother.

^*p < .05.

^**p < .001.

^***p < .0001.

Indirect Paths Linking Parenting and Child Self-Regulation

As it is possible that parenting stress impacts children indirectly through positive parenting, we tested all possible indirect effects for each mother risk group as well. The only significant pathway that emerged included each autoregressive path from positive parenting at 4 months, to 8 months, to 18 months predicting child regulation at 30 months for both AM and LEM, and child regulation at 18 months for LEM only. No indirect paths were significant for HEM.

Discussion

We examined patterns of relations between maternal positive parenting, parenting stress, and child self-regulation in early childhood. By comparing relations across three different maternal risk groups, we add to the literature on contextual factors that may impact parenting and child development. Here, girls regulated more effectively than boys at 18 and 30 months, but not earlier in infancy. The sample behaviors described in scoring of regulation at 18 and 30 months captured developmentally important behaviors such as attentional focusing and environmental exploration/sensitivity that are not as apparent as in early infancy and may emerge earlier in girls than boys. Indeed, sex differences in age-related expressions of emotion (Chaplin & Aldao, 2013; Else-Quest et al., 2006), regulation (Moilanen et al., 2010), and motor/activity (Junaid & Fellowes, 2006) processes are well-documented. Therefore, the identified differences in regulation that emerge as (a) motor skills become more fine-tuned and regulated for girls, (b) anger becomes more expressive in boys, and (c) the socialization of emotion becomes gendered as a result of parental and societal interpretations (Chaplin & Aldao, 2013) is in line with previous literature. In addition, though our gender-difference findings are similar to previous work, we did not find stability in child self-regulation across time. The scoring system used in this work reflects self-regulation more broadly and includes aspects of emotional, cognitive, and behavioral regulation. Though recent syntheses indicate the overlapping nature of multiple regulatory constructs (effortful control, executive function, inhibitory control; Gagne et al., 2021), future studies might include more fine-tuned measures that highlight the dynamic processes involved in self-regulation.

In addition, we do not have a measure of temperament available to explore how a child’s underlying emotional disposition may alter emerging regulatory skills. As Cole et al. (2019) recently proposed, new methods assessing the dominant (emotional) reaction in addition to the more top–down behavioral and cognitive regulatory responses are necessary to truly understand which individual children possess higher regulatory skill, and which simply do not experience the dominant reaction to begin with. This would be particularly important in understanding self-regulation as a developmental construct, whereby regulation is learned as children mature through early childhood.

Parenting Stress and Child Regulation

Similar to previous work, there were group differences in mean levels of mothers’ parenting behaviors and stress (Huang et al., 2014), with varying degrees of significant relations with child regulation across time. Path models indicated that for AM, parenting stress and positive parenting were inversely related in early infancy. This is a time of transition, when adverse postpartum effects can be heightened (Riva Crugnola et al., 2016). In addition, it was only in the AM group that higher parenting stress at 12 months related to lower child self-regulation at 18 months. Mothers reporting higher stress may be more openly negative with their infant (Crnic et al., 2005), damaging the parent–child relationship and preventing the infant from effective self-regulation. We did not find that child regulation predicted subsequent maternal stress, but it is possible that limitations inherent with an autoregressive model, such as shared variance across time, may have obscured such findings.

Interestingly, parenting stress did not adversely impact child self-regulation for LEM or in HEM. To our knowledge, no previous work has directly compared parenting in AM and LEM thereby examining age independent of education or other risk factors as a determinant of parenting. Yet, previous work examining younger first-time mothers indicates that these mothers are more rejecting or use unsupportive parenting practices in both observed and self-reports (Trentacosta et al., 2008). A separate study indicates that AM are at increased risk for depression when compared to adult mothers (Lanzi et al., 2009). Thus, perhaps it is maternal age and not education that adversely impacts mothers’ parenting behaviors and through them, child regulation. Age may contribute to deficits in knowledge or spillover relating to parenting and child development. Further, motherhood is deemed off-time in adolescence but not in adulthood; thus, societal and cultural attitudes may contribute to stress and challenges in AM that adult mothers do not have to face. Of note, there are likely causal relations that are not specifically related to age. Many other factors contribute to adolescent pregnancy such as depression or engagement in higher risk behaviors and drug use (Coley & Chase-Lansdale, 1998). Future work would benefit from considering socioemotional and sociocultural factors other than simply age to examine patterns of parent–child interactions across risk contexts.

Positive Parenting and Child Regulation

Positive parenting is related to child self-regulation at different ages for different maternal risk groups. For HEM, infants who experienced more positive parenting at 4 months were actually less well-regulated at 8 months. Particularly at early ages, infants rely on their caregivers to aid in regulation, for example, infants elicit cooing and soothing behaviors from mothers to down-regulate negative affect (Kopp, 1982). Infants of HEM mothers who are more in tune with their infants’ needs (warm and responsive) may have infants who initially express negativity as a form of communication with their mothers about such needs. These infants are not necessarily less well-regulated, but more expressive when their mother is available. It is also possible that these infants have not yet learned how to regulate, as they rely on their mothers to do so. As mothers and infants were scored on parenting and regulation behaviors during the same observational period, it is difficult to tease apart whether the mother–infant coregulation strategies observed indicate increased communication or lack of regulation. Future work would benefit from multiple measures of parenting and child behaviors.

On the other hand, positive parenting impacted child self-regulation in AM and LEM mothers only at older ages, such that higher positive parenting at 18 months was related to more effective child regulation at 18 and 30 months. The same pattern was not found for HEM even though positive parenting was the highest in this group and child regulation was the highest at 30 months. Previous work in 2–4 year olds shows stronger child regulation across time in families with more positive parenting support (Moilanen et al., 2010). Though we were not able to test this specifically, AM and LEM first-time mothers may not be prepared or educated on parenting practices or expectations during their child’s first months; they may not have the skills necessary to understand how they need to adjust with their developing child. In a recent study examining the effect of a parenting intervention on child regulation from ages 3 to 6, children of mothers who reported more positive parenting were higher in regulation, but also showed improvements in regulation over time when mothers underwent a parenting sensitivity training (Speidel et al., 2020). As all mothers in our sample were first-time mothers, we were able to control for previous experience in childrearing, yet it is possible that mothers’ parenting improves over time as they learn more about their child’s needs. Parenting behaviors of lower risk mothers (e.g., older and with a higher education) may not be affected by maternal stress (Cooper et al., 2009) or child difficult behaviors (i.e., dysregulation; Augustine & Crosnoe, 2010). Therefore, in higher risk families, both mother and child are not protected from other potential adverse life factors. Of note, we use a composite positive parenting variable and did not differentiate between maternal warmth and contingent responsiveness, even though prior research found that maternal warmth may be uniquely related to child regulatory outcomes (von Suchodoletz et al., 2011; Wakschlag & Hans, 1999). Future work may wish to delineate between the two types of mother behaviors. In addition, we scored mother and child behaviors during the same at-home 30-min dyadic interaction. We acknowledge that one interviewer scoring both behaviors is a limitation. However, using in vivo coding, we are likely better able to tap into a more naturalistic interpretation or observation of children’s regulatory abilities and mothers’ behaviors in everyday life without the intrusion of cameras.

In this work, indirect effects of parenting stress or positive parenting on child regulation were not significant; only those that included longitudinal autoregressive pathways predicted later child regulation. Stress may impact children directly or through mechanisms which include more proximal parenting practices, such as warmth and responsiveness. Mothers who experience stress may have less capacity for positivity, particularly toward a child in distress. Future work may, with the aid of our findings, propose more specific mechanisms through which child regulation develops across multiple parenting domains and environmental contexts. In addition, without direct measures of mothers’ own regulatory patterns or skills, we are not able to draw conclusions suggesting that children learn to mimic regulatory behaviors through direct observation of the same. It is possible that mothers who are more positive or report lower stress model more effective regulatory strategies themselves. In this case, children learn regulation through direct observational processes, and not indirectly through mothers’ parenting.

Limitations With Using Age and Education as Indicators of Demographic Risk

Maternal demographic risk is measured differently across studies. Income or economic resources are included in much of the existing literature (Lengua et al., 2007; Osborne et al., 2012), but constructs in addition to economic resources change across studies. Here, we were unable to include income as an indicator of risk or resources as more than 50% of mothers did not report an annual household income. As such, our resource groups relied heavily on maternal age and education. In this work, group comparisons and patterns of relations between positive parenting and stress were not consistent; AM did not always report higher stress and exhibit lower positive parenting than LEM; HEM were not always the least stressed and the most positive parents. For 12- and 24-month parenting stress, the LEM group reported similar levels to the AM group, whereas for 4- and 8-month positive parenting, the LEM group was more similar to the HEM group. This implies that something other than simply age and educational attainment may be important when comparing parenting behaviors across different mothers. Indeed, mothers rely on a variety of resources, including but not limited to higher education, housing and financial stability, and social support to aid them in effective childrearing (Thomson et al., 1994). Parenting stress and financial stress are often associated with having fewer resources which can influence caregiver behaviors (Conger et al., 2002; Whitson et al., 2011). Alternatively, access to higher quality or quantity of resources can enhance both mother and child well-being.

We used predefined cutoffs of age and education to delineate between maternal risk groups; younger mothers who are less well-educated may be less prepared for parenting or less knowledgeable surrounding parenting topics (Davis-Kean, 2005). When mothers are unsure of how to raise their children and experience pressure due to low contextual resources, they are especially at risk for parenting stress and negative interactions with their children. Further, we acknowledge that there may be other equally effective or valid ways to identify maternal risk groups (including measures of income or social support). It is possible that the lack of variation among our HEM mothers prevented us from finding significant patterns of relations between parenting and child self-regulation in the HEM group. However, we did find group differences not only in the means of positive parenting, parenting stress, and child self-regulation using age and education, but also in how behaviors related to each other over time. In addition, our study is limited to observed parenting behaviors during a short interaction and mother-reported stress. Notably, our coding reliabilities for positive parenting and child self-regulation were relatively low. However, as we scored these behaviors in the home, we believe that even though the reliability was low, the mean of the four positive parenting items accurately reflects how mothers engage with their children in everyday situations. Future research would benefit from in vivo scoring that, potentially, has more stringent indicators of reliability. Nonetheless, it is important to consider multiple parenting behaviors and their influence on child development (i.e., Belsky’s Process Model of Parenting, Belsky, 1984); children not only experience their mothers’ stress or her positive parenting independent of each other but develop in an environment which includes a combination of these plus other factors that may be unmeasured here. Future work could consider how multiple factors impact child self-regulation.

Conclusions

We demonstrate differing relations between parenting stress, positive parenting, and child self-regulation across three maternal risk groups: AM, LEM, and HEM. In general, and similar to previous research (Feldman et al., 2004; Landry et al., 2006; Steelman et al., 2002), simple relations among variables indicated that mothers who were higher in stress were also lower in positive-parenting and had less well-regulated children. The timing of these relations, however, differed as a function of maternal age and education, and it is possible that it is maternal age and not low educational risk that confers higher risk for adverse child outcomes. We do not wish to overinterpret these results by saying definitively that younger and/or less-educated mothers are ideal candidates for intervention programs which assist in reducing parenting stress or improving positive parenting. Indeed, there are potentially many overlapping, complex variables that alter a mothers’ engagement with her child that we were not able to examine here. Instead, we add to a growing literature on how specific resources, in this case maternal age and education, impact the parent–child relationship and child regulatory abilities. Results highlight the need to understand parental and environmental context in relation to child development.