Temperament and Sleep Behaviors in Infants and Toddlers living in Low-Income Homes

Abstract

Temperament is a dynamic trait that can be shaped by maturity and environmental experiences. In this study, we sought to determine whether and the extent to which temperament was predicted by sleeping behaviors in an understudied sample of primarily Black and White infants and toddlers living in low-income homes (N = 150). Sleeping behaviors were assessed at 15–19 months of age with caregiver report of the Tayside Children’s Sleep Questionnaire. Temperament was examined as effortful control, negativity, and surgency with the Infant Behavior Questionnaire Very Short Form at 9–12 months of age and with the Early Childhood Behavior Questionnaire Short Form at 20–24 months of age. Covariates were maternal education, household income, and child sex and race. Continuous variables were standardized, then missing data from independent variables were multiply imputed in 20 datasets. Regression analyses showed that about 1 SD improvement in toddler sleep behaviors significantly predicted about 1/5 SD better toddler effortful control. However, sleep behaviors did not significantly predict toddler surgency or negative affect. This study shows that for a sample of infants and toddlers in low-income homes, how a child learns to regulate sleeping behaviors may influence the development of overall effortful control about six months later.

1.0. Introduction

The origins of many health and academic behaviors can be traced back to the sensitive period of early childhood (Moffitt et al., 2011) and early childhood temperament (Canals et al., 2011; Pesonen et al., 2008; Posner & Rothbart, 2018; Rothbart, 2007; Williams et al., 2017). The sensitive period of early childhood is when rapid neurophysiological adaptations occur in response to environmental experiences, guiding behavioral trajectories (Shonkoff, 2012). Generally, temperament is one’s disposition and behavior to their environment (Rothbart, 2011), which can change with maturity and environmental experiences (Rothbart et al., 2000). In this study, we examine whether early childhood temperament changes with an environmental experience in which young children spend over half of their day – sleeping. Specifically, we focus on behavioral sleeping problems, which are influenced by parental management and represent the most common clinical sleep disorders (Owens, 2020).

Behavioral sleeping problems have been linked to later adverse behaviors (Morales-Muñoz et al., 2020; Sadeh et al., 2015) and qualities related to temperament, such as internal and external reactivity (Gregory et al., 2005; Thunström, 2002), as well as structural brain changes in areas linked to attention and sociality (Isaiah et al., 2021). Behavioral sleeping problems may also have the most profound adverse developmental effects on children living in low socioeconomic status (SES) homes compared to those living with more SES advantage (El-Sheikh et al., 2010). Toddlers (1–2 years of age) may also be particularly vulnerable to adverse environmental effects on development, as they have some of the highest synaptic density or growth as compared to other age periods (Huttenlocher, 1979; Huttenlocher & Dabholkar, 1997). The quantity and developing nature of these neural circuits may be more susceptible to environmental experiences, influencing cognitive and behavioral trajectories (Holditch-Davis et al., 2005; Scher, 2005). However, whether sleep behaviors may change temperament has not been examined for toddlers living in low-SES homes. With this study, we hope to begin informing the nature of this relationship and identify a potential malleable target – sleeping behaviors – for parents and clinicians to better support the developmental outcomes of children living in low-income homes.

1.1. Temperament in Early Toddlerhood

Modern Western understandings of temperament blossomed in child development labs in the latter half of the 20^th century with researchers such as Thomas and Chess (Goldsmith et al., 1987; Thomas et al., 1963). Thomas and Chess classically classified children into “difficult” or “easy” temperaments: children with difficult temperament were generally described as those with high intensity regarding negative mood, low adaptability, low predictability, and withdrawal; and children with easy temperaments had contrasting qualities (Thomas et al., 1963). Early understandings of temperament also considered that temperament was relatively stable and unchanging. However, emerging scientific research and opinion suggests that temperament can change because young children undergo rapid developmental transformation, and early experiences may alter the expression of temperament (for a review, see Bornstein et al., 2019). Influenced by this work, Rothbart’s lab tested and expanded upon Thomas and Chess’s binary categories of temperament into more modern classifications based on age-related milestones of behavior in specific scenarios (Putnam & Stifter, 2008), which are widely used to measure temperament today.

Rothbart’s lab generally appraises early childhood temperament across three higher-order factors: 1) regulatory capacity/effortful control, 2) negative affect, and 3) surgency/positive affectivity/extraversion (Putnam et al., 2008). For conciseness, we hereafter refer to regulatory capacity/effortful control as effortful control and surgency/positive affectivity/extraversion as surgency. Effortful control comprises qualities of attention, low-intensity pleasure, and self-soothing behaviors (Rothbart et al., 2011), each of which children gain more independent “effortful” control of in toddlerhood and beyond (Kochanska et al., 2000; Kochanska et al., 1998; Putnam et al., 2008). Negative affect consists of qualities of fear, recovery from distress, frustration, sadness, discomfort, and fidgeting (Putnam et al., 2008). Surgency includes “surges” of reactivity, gross motor and vocal activity, high-intensity pleasure, positive anticipation, and sociability. Together, these three factors comprise individual differences in temperament and can individually or interactionally shape later behaviors (Putnam et al., 2008). For example, high levels of negative affect and effortful control may contribute to heightened vigilance and avoidant behaviors (Zhang et al., 2018), often seen in anxiety disorders. Additionally, low levels of effortful control and high levels of surgency and negative affect could lead to increased symptoms of attention-deficit/hyperactivity disorder (Martel, 2016).

While temperament has been extensively studied in the past 50 years, given the changes in understandings of temperament over the last half century, there is currently little research on the stability of Rothbart’s specific temperament profiles from infancy to toddlerhood. However, the overarching and growing body of temperament research shows that generally, early temperamental traits in the preschool years predict later outcomes across mental health, violence, school adjustment, and adult personality (Caspi et al., 1996; Dougherty et al., 2010; Guerin et al., 1997; Nelson et al., 1999). Yet, it is not known if these predictions are related to extensions of temperament or due to temperament triggering environmental changes – for example, sleep behaviors and associated parental management – that in turn foster changes in later temperament (Martin et al., 2020). As children’s behavior and capabilities related to temperament develop and fluctuate as a function of age and developmental period, researching adjacent developmental periods such as from infancy to toddlerhood, could provide more specific information about temperament stability. For example, general research shows that short-term temperament stability or consistency (i.e., from one to a few months) from infancy to toddlerhood is small to moderate, but from later childhood to adolescence is moderate to high (Bornstein et al., 2019; Guerin & Gottfried, 1994; Martin et al., 2020). This fluctuating stability may be likely due to the rapid developmental changes and capabilities from infancy to toddlerhood, compared to older school ages to early adolescence.

From two known studies with Rothbart’s measures of temperament in early childhood, Casalin et al. (2012) and Putnam et al. (2008) show that in samples of infants and toddlers from the perspective primarily White mothers of economic privilege, temperament constructs have small-moderate stability from infancy to toddlerhood. For example, from the transition of infancy to toddlerhood, when children gain more “effortful” control over their internal reactivity, there is moderate stability between the developmental periods (Casalin et al., 2012; Putnam et al., 2008). Casalin et al. (2012) and Putnam et al. (2008) also show that negative affect can have small-moderate and sometimes insignificant stability from infancy to toddlerhood, depending on the mother’s or father’s perspective. Maturational effects of negative affect are also evident, as other researchers show that fine-grained negative affect attributes of fear and anger can increase from infancy to toddlerhood, with steeper increases at younger ages and in those with less regulatory abilities (Braungart-Rieker et al., 2010). Showing similarly mixed results, some attributes of early surgency are stable across time, yet some research shows that this factor is influenced more by the environment than genetics (Plomin et al., 1993). Varied environment or genetic influences in temperament may explain why surgency also has varied stability in terms of significance and effect from infancy to early toddlerhood (for a review, see Sallquist et al., 2010). For example, Casalin et al. (2012) and Putnam et al. (2008) show that surgency can have small-moderate effects and varying significance of stability, based on the perspective of the mother or the father.

The varying stability of very early temperament based on the perspective of the mother or father is likely because mothers and fathers may have unique experiences with their children. These unique experiences determine the parent’s perspective of the child and consequently the environment that the parent then creates with their child (Mangelsdorf et al., 2000; Molfese et al., 2010). Notably, maternal report of child behavior is critical, as mothers are often the primary caregivers of children (e.g., Casalin et al., 2012). However, as parents differ less in the amount of time they spend caregiving, parents have more similar perceptions of infant and toddler temperament (Casalin et al., 2012). This study adds to the temperament literature by examining Rothbart’s higher-order constructs of temperament from infancy to toddlerhood from the missing perspective of Black or White mothers, who are often the primary caregivers of infants and toddlers living in low-income homes.

1.2. Sleep and Sleep Behaviors in Early Toddlerhood

In early infancy, sleep and circadian cycles undergo dramatic maturation (de Weerd & van den Bossche, 2003; Ednick et al., 2009). However, sleep in later infancy and toddlerhood has stabilized into more consistent sleep-wake cycle patterns (Acebo et al., 2005) and is more markedly influenced by the environment, such as parental management (Brescianini et al., 2011; Dionne et al., 2011; Fisher et al., 2012). For example, Dionne et al. (2011) showed that sleep consolidation at 6 months of age was primarily related to genetics (64% genetic, 36% unique environment), while sleep consolidation at 18 months of age was primarily due to shared environmental influences (58% shared environment, 42% unique environment). Environmental influences such as parental management of a child’s sleep can directly influence behavioral sleeping problems, which are the most common sleep problem in children (Owens, 2020). Behavioral sleeping problems are often called behavioral insomnia of childhood or disorders/difficulties initiating or maintaining sleep (DIMS; McGreavey et al., 2005). DIMS include toddler bedtime resistance, postponing sleep initiation, and recurrent or extended night awakenings that command parental assistance (Owens, 2020). Researchers show that early sleep problems such as DIMS are related to later behavioral problems (Ringli & Huber, 2011), particularly for younger children (Sadeh et al., 2002; Touchette et al., 2007) and those with low self-regulation abilities (Goodnight et al., 2007). Because parents play a crucial role in shaping sleeping behaviors for dependent toddlers, and mothers are overwhelmingly the primary caregiver of children during the night (National Sleep Foundation, 2004), we examine toddler sleeping behaviors from the mother’s perspective.

Prevalence estimates suggest that DIMS are common in early childhood. One report suggests that about 34% of young children overall may have problems with night awakenings (Weinraub et al., 2012). Other reports show that sleep problems may increase 50% from infancy to toddlerhood, with nearly 45% of toddlers having different sleeping problems as compared to when they were infants (Van Tassel, 1985; Weinraub et al., 2012). Toddlers specifically may have unique sleeping behavior problems, as they may have become accustomed to particular sleep routines as infants (e.g., co-sleeping), but their locomotor independence may challenge changes to their routine as they age. For example, toddlers may have difficulty self-soothing or have fears of missing out; their developing ability to challenge routines and climb out of cribs or beds can make it difficult for parents to train toddler sleep behaviors to ensure the toddler attains adequate sleep. Estimates suggest that if these DIMS are not treated, they can persist for about 40% of children (Meltzer et al., 2014; Zuckerman et al., 1987).

DIMS may also be higher for children living in low-SES homes. Reports show that 25% of toddlers living in low-income homes have sleep problems (n = 5,006 infants and toddlers; Sadeh et al., 2009), which may be higher than for children more generally (Acebo et al., 2005; National Sleep Foundation, 2014; Williamson & Mindell, 2019). These higher prevalence estimates can be problematic, as DIMS could significantly interrupt the sleep of toddlers living in low-income homes, who should be spending about half of their day sleeping (11–14 hours each day; Hirshkowitz et al., 2015).

1.3. Relations Between Temperament and Sleep

Sleep may shape neuronal plasticity, as evidence shows that sleep induces experience-dependent changes in synaptic circuitry during early and sensitive periods of development (Frank et al., 2001). This might have important implications for cortical regions involved in shaping experience-dependent temperament. Rothbart’s theory of temperament (Rothbart et al., 2000) posits that temperament can influence behavior in specific contexts with specific environmental stimuli and develops over time with maturation and interactions with the child’s environment. By this theoretical definition, we could surmise that temperament could influence sleep behaviors and that the child’s sleep behaviors could influence the development of later temperament. In other words, sleeping behaviors during critical periods of development may change the development of temperament. This concept is further supported by Dahl (1996), who shows in a theoretical paper that sleep regulatory systems are closely interlinked with qualities of temperament such as affect, arousal, and attention. Therefore, in this study, we address whether sleep behaviors during a sensitive period of development, toddlerhood, may affect later temperament.

Research has shown some support for the hypothesis that sleep may change temperament; however, there has been no study that has investigated how sleep may change temperament over time for toddlers living in low-income homes. Instead, research often shows one-way connections with sleep and different factors of temperament. For example, Sorondo and Reeb-Sutherland (2015) found that negative temperaments influence worse sleep and sleep behavior problems in a diverse sample of infants across their first year of life. In a national study, Weinraub et al. (2012) showed that compared to infants and toddlers with fewer sleeping difficulties, those with sleeping problems were more likely to have a difficult temperament at six months of age. Of note, harsh parenting did not differentiate infants with sleeping behavior problems (Weinraub et al., 2012). On the aggregate, this work indicates that higher negative affect and lower effortful control are related to worse sleep qualities.

In other work, Boddy et al. (2014) found that in a sample of mothers and infants and toddlers living primarily with SES disadvantage, higher infant negative affect and lower effortful control (e.g., regulatory capacity) predicted more sleep problems in toddlerhood. In a cross-sectional study on toddlers 30 months of age, Molfese et al. (2015) found that toddlers with better soothability or lower fear had less parent-reported sleep problems and less variability in nighttime sleep onset or duration. In a sample of children 2.5 to 6 years of age, Touchette et al. (2007) found that children with more sleep fragmentation had higher observed hyperactivity-impulsivity and lower effortful control. In another study of toddlers living in low-income homes, Bates et al. (2020) found that higher levels of physiologic arousal predicted worse changes in sleeping behaviors. Bernier et al. (2010) also found that despite SES and verbal abilities, toddlers with better nighttime sleep had better impulse control. Finally, Spruyt et al. (2008) found that at 11 months of age, increased overall sleep was associated with better effortful control, and at 12 months of age, lower overall diurnal sleep was associated with worse emotional regulation (Spruyt et al., 2008).

In studies on surgency and sleep, researchers have found mixed relations. First, De Marcas et al. (2015) found in an observational study that infants who were either hyporeactive or hyperreactive to stimuli had worse sleep quality. De Marcas et al. (2015) suspected that this curvilinear relationship was related to differential temperamental sensitivity to environmental stimuli. Finally, Molfese et al. (2015) found that more active toddlers had worse sleep as measured through actigraphy.

On the aggregate, the findings on the relations between temperament and sleep behaviors suggest that infants with temperaments rated as higher in negativity, lower in effortful control, and higher in surgency may have worse sleeping behaviors. One of the reasons for these findings may be related to arousal. For example, DeGangi (2017) suggested that some toddlers with sleep problems may have had a higher need for stimulation and arousal (e.g., surgency), leading to less ability to sleep continuously. Additionally, Morrell and Steele (2003) suggested that children with more difficult temperaments (e.g., higher negativity or less effortful control) may prompt more parental intervention, which could either lead to higher levels of arousal or interfere with the child’s development of self-soothing behaviors surrounding sleep. It may also be that children with more difficult temperaments prompt difficult parent-child nighttime interactions, in turn increasing child arousal surrounding bedtime. Finally, Sadeh (1994) suggested that sleep fragmentation (e.g., increased night awakenings) may impair a young child’s ability to regulate emotions and cognition, leading to hypervigilance (e.g., increased arousal). Sadeh (1994) also suggested that parental management of sleep behaviors, such as in bedtime-related routines, may also explain outcomes in sleep behavior problems and characteristics of difficult temperaments. In other words, there could be a linear relationship in which a) temperaments associated with increased arousal could lead to poor sleeping behaviors and that ultimately b) the poor sleeping behaviors could lead to changes in temperament.

1.4. The Current Study

Addressing the gap in the literature, the primary objective of this study was to examine whether and the extent to which sleeping behaviors may predict changes in temperament for a sample of infants and toddlers living in low-income homes. We also examined whether and the extent to which infant temperament predicted poor sleeping behaviors. We screened for sleeping behaviors as DIMS from the mother’s perspective, to capture potential parenting perspectives and toddler sleeping behaviors that may be difficult to capture with more invasive observational methods. Based on prior research and theory, we hypothesized that a) temperament related to heightened arousal would predict later worse DIMS and b) when controlling for previous temperament, worse DIMS would predict later temperamental characteristics related to heightened arousal. In other words, we mainly hypothesized that poorer sleep behaviors would later result in lower levels of effortful control, higher levels of negative affect, and higher levels of surgency. Based on prior research and theory, we understood that temperament has some stability in early childhood, and that sleeping behaviors likely do not fully explain the relationship between previous and later temperament. Rather, sleeping behaviors may predict changes in temperament. With this study, we hope to identify a potential malleable target, sleep behaviors, for parents to help children develop optimal temperament behaviors to support healthy academic and health behaviors later in life.

2.0. Methods

2.1. Design, Participants, and Procedure

Here we report determination of our sample size, measures, and statistical analyses for the study. The study was an observational, longitudinal, correlational design. Data are from a longitudinal birth cohort study, the Kids in Columbus Study (Salsberry et al., 2016). In brief, participants primarily included mothers and their children, who were recruited at or near the time of the focal child’s birth from Women, Infant, Children (WIC) clinics. Participants were recruited following a robust quota sampling method to represent the race and ethnicity of families living at or below the Federal Poverty Level in Franklin County, Ohio (Salsberry et al., 2016). Eligibility criteria for the study were: mothers were 18 years of age or older, eligible for WIC services, planned to live in Franklin County for the duration of the 5-year study, provided consent for themselves and their child, and the child had no known severe medical problems at or near birth. The study was approved by the local university Institutional Review Board (IRB). Demographics of the larger dataset are described elsewhere (Jiang et al., 2020).

The analytic sample of the present study included 150 mother-toddler dyads who provided data for the outcome measures regarding temperament at toddlerhood, when children were between 20–24 months of age (September 2016–January 2018). In over 95% of cases, mothers reported on child temperament and sleep. As shown in Table 1, the sample demographics show that participants were primarily living with socioeconomic disadvantage – at or near poverty level. That is, over 75% of dyads lived in homes earning less than $30,000 per year. For reference, 100% of the Federal Poverty Level in 2017 was $24,600 for a family of four (United States Department of Health and Human Services, 2017), yet families need an income of at least 200% of the Federal Poverty Level to afford basic needs such as housing and food (Jiang et al., 2017). About half of the mothers had no college education. About half of the children were described by the caregivers as White, and about half were described as Black. About 54% of the toddlers were female.

Table 1.

Demographics and Descriptive Statistics of Key Variables of the Analytic Sample (N = 150)

Variable	N	%	Min	Max	Mean	SD
Child age (in months)3	130		8.50	14.30	11.12	1.01
Child age (in months)4	134		14.10	19.90	16.99	1.76
Child age (in months)5	144		19.80	25.40	22.70	1.37
Mother age (in years)5	144		20.00	45.00	28.40	5.32
Maternal education attainment4	130
8th grade or less		2.3
Some high school, no diploma		10.8
High school graduate (diploma or GED)		34.6
Some college, no degree		36.9
Associate degree		5.4
Bachelor’s degree		9.2
Master’s degree		0.8
Annual income in USD4 (TP4)	132
< $10,000		32.6
$10,001–20,000		22.7
$20,001–30,000		21.2
$30,001-$50,000		15.9
>$50,000		7.6
Child race/ethnicity1,2
Black	149	54.4
White	149	46.3
Asian	149	2.7
Latino	148	10.1
Child sex female2,3	144	54.2
Toddler sleeping DIMS4	133		2.00	30.00	18.98	6.22
Infant effortful control3	128		2.08	7.00	5.32	0.84
Infant negative affect3	128		1.45	6.67	4.28	1.08
Infant surgency3	128		1.85	6.92	5.44	0.81
Toddler effortful control5	137		3.00	6.12	4.30	0.60
Toddler surgency5	136		3.33	6.27	4.95	0.64
Toddler negative affect5	135		1.80	6.09	3.90	0.87

Note. Superscripts indicate the time point of measurement (¹TP1, ²TP2, ³TP3, ⁴TP4, ⁵TP5). Temperament and sleep variables are unstandardized. Mothers could select more than one race/ethnicity.

2.2. Measures

This study used data collected during timepoints (TP) in infancy and toddlerhood when sleep and temperament measures were collected. These data were primarily gathered during three home visits: 1-year visit (TP 3: child age ~9–12 months), 1.5-year visit (TP 4: child age ~15–19 months), and 2-year visit (TP 5: child age ~20–24 months).

2.2.1. Difficulties Initiating or Maintaining Sleep (DIMS)

DIMS were represented with maternal report of the Tayside Children’s Sleep Questionnaire (McGreavey et al., 2005) at TP4 with mild adaptations for KICS (Bates et al., 2020). The Tayside is a clinical questionnaire on Disorders of Initiating or Maintaining Sleep, with strong reliability and validity for toddlers and preschoolers 1–5 years of age (McGreavey et al., 2005; Spruyt & Gozal, 2011). The survey has been cited in 16 articles in PubMed and over 25 articles on Google Scholar. Generally, mothers rated 10 items during an unspecified period on a 4-point scale with 0 = strongly agree and 3 = strongly disagree. After following reverse scoring rules, items were summed. Scores were then standardized with a mean of 0 and SD of 1. Higher scores indicate fewer difficulties initiating or maintaining sleep (i.e., higher scores indicate better sleep behaviors). Cronbach’s alpha reliability for this scale was .87.

2.2.2. Infant Temperament: Effortful Control, Negative Affect, and Surgency

Infant temperament was measured at TP3 with three subscales from the Infant Behavior Questionnaire-Revised Very Short Form (IBQ-VSF; Putnam et al., 2014; Putnam et al., 2010). The three scales measured were: effortful control, surgency, and negative affect. Caregivers were asked to evaluate how often their baby performed a behavior in the past week, on a frequency scale from 1 to 7 where 1 = Never and 7 = Always. The effortful control subscale has 12 questions on an infant’s ability to modulate and regulate behavior, duration of orienting, low-intensity pleasure, cuddliness, and soothability. The negative affect subscale has 12 questions on an infant’s overall sadness, distress to limitations, and fear. The surgency subscale has 13 questions on approach, vocal reactivity, high-intensity pleasure, smiling and laughter, activity level, and perceptual sensitivity. To score each subscale, questions are processed with reverse scoring rules, summed, then averaged. These subscale scores were then standardized to M = 0 (SD = 1) for analyses. Higher scores indicate more surgency behaviors, more negative affect, or better effortful control. Cronbach’s alpha reliability for the surgency subscale was .80, negative affect was .86, and effortful control subscale was .80.

2.2.3. Toddler Temperament: Effortful Control, Negative Affect, and Surgency

Toddler temperament was measured with select subscales from the Early Childhood Behavior Questionnaire Short Form (ECBQ-SF; Putnam et al., 2006; Putnam et al., 2010; Rothbart, 2006) collected during TP5 of KICS. Following guidelines from Putnam et al. (2008), each ECBQ subscale corresponded to a specific IBQ-VSF subscale. There were eight available ECBQ-SF subscales from KICS (indicated in italics, with correspondence to IBQ-VSF subscales in parentheses): activity level (surgency), impulsivity (surgency), sociability (surgency), frustration/distress to limitations (negative affect), shyness (negative affect), attention/duration of orienting (effortful control), inhibitory control (effortful control), and soothability (effortful control).

To evaluate toddler temperament within these subscales, mothers were asked to evaluate how often their toddlers performed a behavior in the past two weeks, on a frequency scale from 1 to 7 where 1 = Never and 7 = Always. Activity level included eight questions on gross motor activity of the child, including the frequency in which they engaged in movement. Impulsivity included four questions on how quickly the child engaged in a situation. Sociability included four questions on how often the child sought interactions with others. Frustration/distress to limitations included six questions on negative affect related to confinement or interruptions. Shyness included five questions on the child’s inhibited approach or discomfort during social situations. Attention/duration of orienting included six questions on the child’s ability to sustain attention. Inhibitory control included six questions on the child’s ability to suppress inappropriate behaviors. Soothability included five questions on a child’s reduction of distress when a caregiver uses soothing techniques.

The questions from each of these ECBQ-SF subscales were categorized to the respective surgency, negative affect, and effortful control IBQ-VSF subscales from Putnam et al. (2008), processed with reverse scoring rules, added, and then averaged to comprise ECBQ-SF subscales respectively entitled surgency, negative affect, and effortful control. The subscales were then standardized to M = 0 (SD = 1) to aid in practical interpretation of the analyses as the science on sleep and temperament builds. Higher scores indicate more surgency behaviors, more negative affect, or better effortful control, respectively. To maximize subscale reliability, one item with low reliability was removed from the surgency subscale (“When offered a choice of activities, how often did your child stop and think before deciding?”). Cronbach’s alpha reliability for the effortful control subscale was .70, surgency was .64, and negative affect was .76.

2.2.3. Covariates

Covariates included child race, child sex, maternal education, and household income. Child race was measured with one parent-reported question regarding whether the caregiver considered the child to have 1 = Black/African American race or 0 = no Black/African American race. This dichotomy was chosen because most children in the sample were Black or White, and there may be differences in how mothers in low-income homes in the Midwest report infant affect and temperament by race (Palmer et al., 2013). Child sex was scored 1 = female and 0 = male to account for potential differences in temperament or sleep by sex (Martin et al., 1997; Olino et al., 2013; Sorondo & Reeb-Sutherland, 2015). We also controlled for socioeconomic status (SES) within the sample by controlling for maternal education and household income to account for potential SES differences in child sleep (National Sleep Foundation, 2014; Sadeh et al., 2009; Williamson & Mindell, 2019). For maternal education, mothers were asked to report their highest category of education on a scale from 0 to 8, with 0 = 8^th grade or less and 8 = doctorate degree. Given the characteristics of the sample, we merged education into three categories: no high school diploma (13.1%), high school diploma (34.6%), or attended at least some college (52.3%). For household income, mothers were asked to report which of seven income categories, in $10,000 increments, represented their annual household income: from $0 to more than $60,000.

2.3. Analytic Overview

Statistics were performed in SPSS 27 and 28. We first examined descriptive statistics and correlations for preliminary analysis. We examined missing data (missing n in parentheses): child race (n = 1), child sex (n = 6), maternal education (n = 20), household income (n = 18), DIMS (n = 17), infant temperament subscales (n = 22), toddler effortful control (n = 13), toddler surgency (n = 14), and toddler negative affect (n = 15). For reference, if more than 10% of data were missing from a survey or subscale in KICS, the participant was given a missing score for that respective survey or subscale. For the missing data on quantitative variables, Little’s MCAR was not significant (χ² = 72.35, p = .43), indicating insufficient evidence against the hypothesis that the quantitative data were missing completely at random (Little’s MCAR is not used for categorical variables; Little, 1988). We then multiply imputed 20 datasets with fully conditional specification for dyads who participated in the toddler temperament assessment (n = 150) but had missing data on any variable. After imputation, we performed hierarchical multivariable linear regression with autoregressive models to interpret the hypotheses. We averaged R² and change in R² following guidelines from van Ginkel (2019). From a known sample size of 150 from the longitudinal cohort, we have a power of 0.81 to detect moderate omnibus effect sizes of f ≈ 0.33.

3.0. Results

3.1. Preliminary Results

Descriptive statistics of raw data are shown in Table 1. Correlations between all study variables with standardized measures and imputed data are shown in Table 2. To supplement the correlation tables, Supplemental Tables 1, 2, and 3 show the extent to which infant temperament predict toddler DIMS while controlling for covariates.

Table 2.

Correlations between all study variables (n = 150)

	1	2	3	4	5	6	7	8	9	10	11	12	13
1. Difficulties initiating or maintaining sleep	--
2. Infant effortful control	.33**	--
3. Infant negative affect	−.33**	.13	--
4. Infant surgency	.23*	.72**	.35**
5. Toddler effortful control	.27**	.25**	−.30**	.07	--
6. Toddler negative affect	−.24**	−.11	.31**	−.04	−.44**	--
7. Toddler surgency	.05	.06	−.09	.04	−.02	.19*	--
8. Mother not high school graduate	.01	−.15	−.04	−.13	.00	−.08	.04	--
9. Mother high school graduate	−.08	.02	−.10	.03	−.07	.18*	.13	−.18	--
10. Mother completed some college	.04	.05	.15	.05	.03	−.11	−.10	−.34**	−.68**	--
11. Household income	.13	.09	−.04	.09	.05	−.08	−.13	−.10	−.14	.19	--
12. Child Black	−.06	.06	.19*	.08	−.02	.14	.05	−.08	−.01	.09	−.25**	--
13. Child sex	.05	.03	−.05	−.04	.20*	.00	−.15	−.08	−.11	.14	−.06	−.04

Note. DIMS = difficulties initiating or maintaining sleep. Analysis performed with imputed data and standardized sleep and temperament variables

^*p < .05.

^**p < .01.

Based on correlational data, we found significant associations between race and income, sex and toddler effortful control, and race and infant negative affect. Infant effortful control and negative affect were respectively significantly, positively, and moderately associated with toddler effortful control (r = .25) and negative affect (r = .31). Infant surgency was not significantly associated with toddler surgency (r = .04).

Additional correlational data showed that toddler DIMS had directly significant and moderate associations with most aspects of infant and toddler temperament. Toddler DIMS were significantly, moderately, and positively associated with infant effortful control (r = .33), toddler effortful control (r = .27), and infant surgency (r = .23). In contrast, toddler DIMS were significantly, moderately, and negatively associated with negative affect in infancy (r = −.33) and in toddlerhood (r = −.24). There were negligible associations between toddler DIMS and toddler surgency (r = .05).

Supplemental Tables 1, 2, and 3 corresponded to the correlational data between infant temperament and toddler DIMS. When controlling for covariates (child race, child sex, maternal education, and household income), we found the following. A 1 SD increase in infant effortful control predicted about 1/3 SD better toddler DIMS, 1 SD higher infant negative affect predicted about 1/3 SD worse toddler DIMS, and 1 SD higher infant surgency predicted about 1/4 better toddler DIMS.

3.2. Do DIMS Predict Changes in Later Temperament?

We performed hierarchical regression to determine if DIMS predicted changes in temperament, by using autoregressive models controlling for infant temperament to predict toddler temperament from DIMS. We examined three stepwise models determining the effects of 1) covariates to predict toddler temperament; 2) covariates and infant temperament to predict toddler temperament; then 3) the effects of the full model including covariates, infant temperament, and toddler DIMS to predict toddler temperament. The results for toddler effortful control are in Table 3, toddler negative affect in Table 4, and toddler surgency in Table 5.

Table 3.

Pooled Hierarchical Regression Results Predicting Toddler Effortful Control^d (N = 150)

Variable	B	SE	t	p	95% CI for B		R 2	ΔR2
					LL	UL
Step 1: Covariates only							.06
Constant	−0.26	0.29	−0.90	.37	−0.82	0.31
Child race Black	0.02	0.18	0.11	.91	−0.34	0.38
Child sex female	0.42	0.19	2.23	.03	0.05	0.78
Household income	0.04	0.06	0.58	.56	−0.09	0.16
Mother not a high school graduatea	0.05	0.30	0.17	.86	−0.54	0.65
Mother high school graduatea	−0.05	0.22	−0.25	.81	−0.49	0.38
Step 2: Covariates + infant effortful control							.12	.06
Constant	−0.21	0.27	−0.78	.44	−0.75	0.33
Child race Black	−0.02	0.18	−0.10	.92	−0.37	0.34
Child sex female	0.40	0.18	2.23	.03	0.05	0.76
Household income	0.02	0.06	0.36	.72	−0.10	0.14
Mother not a high school graduatea	0.14	0.28	0.49	.63	−0.42	0.70
Mother high school graduatea	−0.07	0.21	−0.31	.75	−0.49	0.36
Infant effortful controlb,d	0.24	0.09	2.76	.01	0.07	0.41
Step 3: Covariates + infant effortful control + toddler DIMS							.15	.04
Constant	−0.20	0.26	−0.76	.45	−0.72	0.32
Child race Black	0.00	0.18	0.02	.98	−0.34	0.35
Child sex female	0.39	0.18	2.19	.03	0.04	0.74
Household income	0.01	0.06	0.22	.83	−0.10	0.13
Mother not a high school graduatea	0.11	0.28	0.38	.70	−0.44	0.66
Mother high school graduatea	−0.04	0.21	−0.20	.84	−0.46	0.38
Infant effortful controlb,d	0.17	0.09	1.88	.06	−0.01	0.35
Toddler DIMSb,c	0.19	0.09	2.02	.04	0.00	0.37

Note. CI = confidence interval; LL = lower limit; UL = upper limit; DIMS = difficulties initiating or maintaining sleep.

^aEducation referent group is at least some college.

^bStandardized variable.

^cHigher scores indicate better sleeping behaviors.

^dHigher scores indicate better effortful control.

Table 4.

Pooled Hierarchical Regression Results Predicting Toddler Negative Affect^d (N = 150)

Variable	B	SE	t	p	95% CI for B		R 2	ΔR2
					LL	UL
Step 1: Covariates only							.06
Constant	−0.29	0.26	−1.15	.25	−0.80	0.21
Child race Black	0.27	0.19	1.45	.15	−0.10	0.64
Child sex female	0.04	0.18	0.22	.83	−0.32	0.40
Household income	−0.01	0.06	−0.24	.81	−0.12	0.09
Mother not a high school graduatea	−0.10	0.27	−0.39	.70	−0.63	0.43
Mother high school graduatea	0.36	0.20	1.84	.07	−0.03	0.76
Step 2a: Covariates + infant negative affect							.16	.10
Constant	−0.30	0.24	−1.23	.22	−0.77	0.18
Child race Black	0.16	0.18	0.89	.37	−0.20	0.52
Child sex female	0.08	0.18	0.45	.65	−0.27	0.43
Household income	−0.01	0.05	−0.17	.86	−0.11	0.09
Mother not a high school graduatea	−0.05	0.26	−0.21	.83	−0.56	0.45
Mother high school graduatea	0.45	0.19	2.41	.02	0.08	0.82
Infant negative affectb,d	0.31	0.09	3.53	.00	0.14	0.49
Step 3: Covariates + infant negative affect + Toddler DIMS							.17	.02
Constant	−0.32	0.24	−1.33	.18	−0.78	0.15
Child race Black	0.17	0.18	0.93	.35	−0.19	0.53
Child sex female	0.09	0.17	0.50	.61	−0.25	0.43
Household income	0.00	0.05	−0.03	.98	−0.10	0.10
Mother not a high school graduatea	−0.05	0.25	−0.19	.85	−0.54	0.45
Mother high school graduatea	0.43	0.18	2.35	.02	0.07	0.79
Infant negative affectb,d	0.27	0.09	2.88	.00	0.09	0.46
Toddler DIMSb,c	−0.12	0.09	−1.34	.18	−0.31	0.06

Note. CI = confidence interval; LL = lower limit; UL = upper limit; DIMS = difficulties initiating or maintaining sleep.

^aEducation referent group is at least some college.

^bStandardized variable.

^cHigher scores indicate better sleeping behaviors.

^dHigher scores indicate higher negative affect.

Table 5.

Pooled Hierarchical Regression Results Predicting Toddler Surgency^d (N = 150)

Variable	B	SE	t	p	95% CI for B		R 2	ΔR2
					LL	UL
Step 1: Covariates only							0.06
Constant	0.12	0.28	0.44	.66	−0.43	0.68
Child race Black	0.04	0.19	0.20	.85	−0.33	0.40
Child sex female	−0.27	0.18	−1.48	.14	−0.63	0.09
Household income	−0.07	0.06	−1.13	.26	−0.19	0.05
Mother not a high school graduatea	0.09	0.26	0.33	.74	−0.42	0.59
Mother high school graduatea	0.22	0.23	0.94	.35	−0.24	0.67
Step 2: Covariates + infant surgency							.06	.00
Constant	0.13	0.28	0.45	.65	−0.43	0.68
Child race Black	0.03	0.19	0.17	.87	−0.34	0.40
Child sex female	−0.27	0.18	−1.46	.14	−0.63	0.09
Household income	−0.07	0.06	−1.15	.25	−0.19	0.05
Mother not a high school graduatea	0.10	0.26	0.38	.70	−0.42	0.62
Mother high school graduatea	0.22	0.23	0.94	.35	−0.24	0.67
Infant surgencyb,d	0.04	0.10	0.44	.66	−0.15	0.23
Step 3: Covariates + infant surgency + Toddler DIMS							.07	.01
Constant	0.14	0.28	0.48	.63	−0.42	0.70
Child race Black	0.04	0.18	0.21	.83	−0.32	0.40
Child sex female	−0.28	0.18	−1.51	.13	−0.64	0.08
Household income	−0.08	0.06	−1.23	.22	−0.19	0.04
Mother not a high school graduatea	0.08	0.27	0.31	.76	−0.45	0.62
Mother high school graduatea	0.22	0.23	0.98	.33	−0.23	0.67
Infant surgencyb,d	0.02	0.11	0.22	.82	−0.19	0.23
Toddler DIMSb,c	0.08	0.10	0.85	.40	−0.11	0.28

Note. CI = confidence interval; LL = lower limit; UL = upper limit; DIMS = difficulties initiating or maintaining sleep.

^aEducation referent group is at least some college.

^bStandardized variable.

^cHigher scores indicate better sleeping behaviors.

^dHigher scores indicate higher surgency.

Based on the results of the full model, we found partial support for our main hypothesis. Toddler DIMS significantly predicted nearly 1/5 SD higher change in toddler effortful control (p = .04), predicting nearly 33% more variance than the model with covariates and infant effortful control. However, toddler DIMS did not significantly predict changes in toddler negative affect nor toddler surgency. In sum, we found that lower toddler DIMS (i.e., better sleep behaviors) predicted nearly 1/5 SD better changes in toddler effortful control about six months later.

4.0. Discussion

In this study, we determined whether and the extent to which characteristics of sleeping behaviors in toddlerhood, DIMS, predicted changes in three higher-order factors of temperament for a sample of toddlers living in low-income homes. The primary findings showed mixed support for our hypotheses that sleep behaviors would predict arousal-related changes in temperament. Preliminary findings showed that infant temperament predicted about 6–11% of the variance in later toddler sleeping behaviors. When controlling for infant temperament, toddler DIMS predicted about 1/5 SD lower toddler effortful control. Additionally, while we found that better toddler DIMS were directly associated less toddler negative affect about six months later, we did not find that DIMS significantly predicted changes in toddler negative affect when controlling for demographics. We also did not find direct associations between toddler DIMS and later toddler surgency. This research highlights the need for targeted attention towards sleeping behaviors and effortful control in toddlerhood. We discuss the primary findings in turn.

This study helps researchers and clinicians understand how early effortful control may change from adverse toddler sleeping behaviors within the first to the second year of life. These findings align with those of Boddy et al. (2014), Molfese et al. (2015), Touchette et al. (2007), and Spruyt et al. (2008), who also found connections between lower effortful control and more sleeping problems in infancy and toddlerhood. In line with previous research and theory, these findings may be related to aspects of arousal (Bates et al., 2020; Dahl, 1996). For example, some of the reasons for this may be related to how parents manage the sleeping behaviors of children with varying temperaments. For example, children with less effortful control may have a more difficult time regulating sleep behaviors, which may prompt more parental intervention. Increased parental intervention around bedtime for children who have difficulty regulating themselves may lead to cascades of arousal around bedtime because children may also not learn to self-soothe (Morrell & Steele, 2003). On the other hand, toddlers with poorer effortful control surrounding bedtime need guidance to learn how to sleep. This pattern may disrupt sleep, leading to later problems with effortful control, underpinning later adverse internalizing or externalizing behaviors. While heightened arousal may be an evolutionarily protective mechanism to be aware of danger, in today’s structured society, arousal near bedtime may simply interrupt sleep and self-regulation abilities. Cascading effects of interrupted self-regulation abilities could lead to poorer adaptation to structured environments (e.g., preschool classrooms).

Yet, in low-income homes, there may be other environmental characteristics that may heighten arousal that we could not measure in this correlational design, which may explain our findings. For example, conditions of the neighborhood and home may be impairing the sleep of those in low-income environments; noise may be common in low-income, urban neighborhoods and has been linked to adverse sleep (Basner et al., 2011; Chambers et al., 2016) through its effects on arousal (Basner et al., 2017). Another potential explanation may be related to increased access to blue-light ambiance from screen media (e.g., mobile phones and tablets) before bedtime. Blue-light ambiance may also heighten arousal and adversely affect sleeping behaviors and the transition to sleep (Cain & Gradisar, 2010). Today, over 98% of children 0–8 years of age have access to a mobile device – compared to 75% in 2013 – and the largest increase in media use has been by those living in the lowest-income homes (Common Sense, 2017). Future research can continue investigating the relations among effortful control, sleep behaviors, and arousal in early childhood.

Surprisingly, we did not find that sleeping behaviors significantly predicted negative affect or surgency, in contrast to studies by Molfese et al. (2015) and Boddy et al. (2014). Null findings do not necessarily mean that there are no relations. One of the possible explanations for the null finding may be related to small effect sizes; however, in early childhood, the effects of adverse sleep may compound over time and lead to measurable effects later on. The null findings could also be related to our measurement of higher-order factors of temperament. Instead, finer-grained attributes of negative affect and surgency, respectively as frustration or hyperactivity, could be related to earlier poorer sleep behaviors.

Limitations

Although the present findings highlight that adverse sleep behaviors undermine later effortful control abilities, our reliance on maternal reports for these analyses is a limitation of our work. Indeed, maternal report of child temperament and sleep may have shared method variance, although our small-moderate correlation results between DIMS and temperament (r < |.33|) suggest that these constructs are independent. It may also be that maternal report of both temperament and sleep constructs could be tapping into similar behaviors, which is still an important area of research that this study helps uniquely elucidate. Yet, maternal report is a crucial first step for elucidating this emerging area of research for several reasons. First, maternal report of infant temperament and sleep behaviors may be a unique window into the infant’s microenvironment (Mangelsdorf et al., 2000). Further, research has shown that maternal appraisal of child behavior can overlap with observational measures (Rothbart & Bates, 2006) and reflects both subjective and objective components of the microenvironment (Bates and Bayles, 1984; Mebert, 1991). This may be because mothers, compared to other assessors or observers, have intimate familiarity with their child’s behavior across time, in numerous contexts (Mangelsdorf et al., 2000), and in situation-specific interactions involving the mother and the child (Achenbach et al., 1987). This is critical to highlight, as a national survey showed that 85% of mothers are the primary attendees to toddler needs during the night (National Sleep Foundation, 2004). Second, maternal perceptions have also been linked to later child outcomes (Molfese et al., 2010). It could be that maternal perception of infant temperament influences how she manages toddler sleep behaviors, which in turn could influence later toddler temperament. Indeed, clinicians are aware that screening for adverse sleep and temperament concerns often begins with maternal reports before more invasive testing. Given that characteristics of effortful control may shape self-regulation abilities and gradients of health and academic behaviors later in life (Moffitt et al., 2011), it may be crucial that researchers and clinicians consider more widespread screening of sleeping behaviors to protect optimal development of effortful control.

A second limitation of our study is that while our timeline of temperament measures is in line with previous research on primarily White mothers and infants living with socioeconomic advantage (Casalin et al., 2012; Putnam et al., 2008), we measured child sleep behaviors once with a 10-item maternal report questionnaire. We did not have access to objective sleep measures such as actigraphy or more refined child sleep measures such as sleep diaries. However, the Tayside Children’s Sleep Questionnaire uniquely captures caregiver reports of sleep behaviors, distinct from actigraphy. Nevertheless, a strength of this study is that we gathered this information in a longitudinal design from an understudied population of racially diverse children, from infancy to toddlerhood, living at or near poverty in the United States. As previously mentioned, maternal report from this population is missing in the literature and critically important to understanding more comprehensive relations between temperament and sleep in young children.

A third limitation of this study is that our temperament measures were slightly different from TP3 (very short form) to TP5 (short form). The slight differences between these measures limited our ability to capture nuances in the stability of temperament from TP3 to TP5, which may have explained why there was little variance between the surgency measures at two different time points. Nevertheless, the temperament measures are widely used and were linked to a clinical questionnaire for an understudied population, helping inform this growing area of research.

Implications for Future Research

As this is one of the first studies examining relationships between infant/toddler temperament and toddler sleep behaviors by maternal report, there are several avenues to strengthen this area of research. First, it would be important to move beyond maternal report of child temperament and sleep. Supplementing this research with unique variance from the father’s perspective of child development is a critical area often missing from infant behavior and development research. Additionally, leveraging observational accounts of infant and toddler sleep, such as with actigraphy, and more fine-grained measures of sleep behaviors, such as through sleep diaries for two weeks, would provide an important understanding beyond mother’s perspective of child sleep behaviors. Second, given that the period from infancy to toddlerhood is marked by rapid developmental change, future research should investigate more short-term repeated measures of temperament and sleep with a more robust sample size. Understanding short-term nuances between these relations might help uncover small effects between early childhood sleep and surgency or negative affect. Finally, researchers could consider randomized controlled trials in how adjusting sleep behaviors change later temperament, including for how long the effects may last.

Conclusion

In this research, we examined how toddler sleep behaviors predicted potential changes in later toddler temperament, controlling for infant temperament, in an understudied population of racially diverse infants living at or near poverty in the United States. While sleep behaviors failed to predict significant changes in toddler negative affect or surgency, we found that better toddler sleeping behaviors predicted about 1/5 SD better effortful control about six months later. Researchers and clinicians can expand upon this work with this understudied population by further investigating the critical importance of early sleeping behaviors on later outcomes, which could compound over time during the sensitive period of early child development.

Highlights.

• We determined whether toddler sleeping behaviors predicted later temperament

• We found that better toddler sleep behaviors predicted better effortful control

• Sleep behaviors did not significantly predict toddler surgency or negative affect