Developmental Trajectories of Maternal Sensitivity across the First Year of Life: Relations among Emotion Competence and Dyadic Reciprocity

SYNOPSIS

Objective.

Maternal sensitivity is a commonly used construct to capture the quality of mother-child interactions, but inconsistencies in conceptualizing and defining maternal sensitivity limit understanding of how sensitive caregiving may be associated with child development. The purposes of this study are to (1) examine and compare the developmental trajectories of individual maternal sensitivity behaviors to that of a global index of sensitivity across the first year of infant life and (2) determine whether differences in trajectories of sensitivity are meaningful for infant emotion competence and dyadic reciprocity at 12 months.

Design.

A total of 322 low-income, Mexican American mothers and infants were observed during a free play task at 3, 4.5, 6, and 12 months. Observations were coded for 11 distinct behaviors known to compose maternal sensitivity. At 12 months, mother-infant interactions were also coded for dyadic reciprocity, and mothers reported on infant emotion competence.

Results.

Latent growth models indicated that individual sensitivity behaviors differed from the global index of sensitivity with respect to initial levels and slopes, with increasing (e.g., vocal appropriateness), decreasing (e.g., touch), and stable (e.g., elaboration) trajectories. The individual and global indices of sensitivity differed in prediction of emotion competence and dyadic reciprocity. Trajectories of global and individual indices of maternal sensitivity operated similarly in predicting dyadic reciprocity, with the exception of consistency of style. In contrast, the global index of sensitivity was unrelated to emotion competence, and only initial levels of positive affect emerged as significant predictors of emotion competence.

Conclusions.

The findings offer a more nuanced understanding of maternal sensitivity and suggest that component aspects of maternal sensitivity uniquely contribute to child and family competencies.

INTRODUCTION

Maternal sensitivity is a developmental construct that is central to current conceptualizations of the quality of early parenting behavior. Early sensitive interactions between infants and their mothers lay an important foundation for the infant’s development of socioemotional skills and healthy relationship schemas (Ainsworth, Blehar, Waters, & Wall, 1978). Indeed, maternal sensitivity has been linked to fewer externalizing behaviors (Leerkes, Blankson, & O’Brien, 2009), better cognitive and language development (Landry, Smith, & Swank, 2006), and decreased physiological stress responding (Enlow et al., 2014) as well as other emerging maternal behaviors related to children’s competence, such as less harsh discipline (Joosen, Mesman, Bakermans-Kranenburg, & Van IJzendoorn, 2012). Although maternal sensitivity is often considered a coherent global construct in its own right (Lohaus, Keller, Ball, Voelker, & Elben, 2004; Murray, Fiori-Cowley, Hooper, & Cooper, 1996), it is also understood as a set of many different behaviors and characteristics that help mothers respond in a nurturant way to children’s cues (Feldman, 2010; Leckman et al., 2004). Whether the component behaviors and qualities that compose maternal sensitivity cohere and operate similarly to the more global construct, however, is not well understood (but see Bohr, Putnick, Lee, & Bornstein, 2018).

Conceptualizing Maternal Sensitivity

Framed in the context of attachment theory, maternal sensitivity refers to a mother’s ability to notice, interpret, and promptly respond to her infant’s signals with behaviors appropriate to the infant’s needs (Ainsworth, Bell, & Stayton, 1974). From an evolutionary perspective, sensitive and responsive caregiving enhances an infant’s well-being because the infant’s needs for survival are met (e.g., food, safety, and comfort; Mesman et al., 2018). Western conceptualizations of maternal sensitivity have evolved to include aspects of mother-infant interaction quality, such as warmth and affection. The definition of the maternal sensitivity construct varies widely across the field, with some descriptions emphasizing synchronicity of mother-infant interactions (Isabella & Belsky, 1991), whereas others include the specific caregiving behaviors that may characterize sensitivity (e.g., positive affect; Feldman, 2012). This lack of uniformity further complicates an understanding of the role of early sensitive caregiving in child development.

Darling and Steinberg (1993) provided a useful framework for characterizing maternal sensitivity by differentiating parenting style from parenting practices, where parenting style provides the context in which specific parenting practices, or behaviors defined by specific socialization goals, are expressed. Maternal sensitivity can be considered a parenting style, and the specific behaviors that characterize sensitivity (e.g., gaze, touch) can be thought of as parenting practices. Further, Dallaire and Weinraub (2005) asserted that it is conceivable that parenting styles, such as maternal sensitivity, demonstrate stability over time, whereas specific parenting practices may not.

The Developmental Trajectory of Maternal Sensitivity

Maternal sensitivity consistently demonstrates continuity and stability (Bornstein, Putnick, & Esposito, 2017) during the infancy period (Isabella, 1993; Spangler, Schieche, Ilg, Maier, & Ackermann, 1994; Feldman, Greenbaum, Mayes, & Elrich, 1997; Dallaire & Weinraub, 2005; Ferber, Feldman, & Makhoul, 2008), but there is evidence that maternal sensitivity varies across time in form and levels of component behaviors that underlie the broader construct. For example, affectionate touch has been shown to significantly decrease during infancy (Tronick, 1995; Ferber et al., 2008), perhaps due to infants’ increased mobility (e.g., crawling, walking) and desire for autonomy. Given that maternal sensitivity is relatively stable across infancy, it follows that other sensitive behaviors are increasing while touch is decreasing to maintain that stability. It is likely that sensitive mothers adjust their specific behaviors to meet their infants’ rapidly changing needs and developmental capabilities across the first year of life (Bigelow et al., 2010). For example, Jahromi, Putnam, and Stifter (2004) reported that during inoculation stressor tasks mothers were more likely to respond with affectionate touch to infant distress when infants were 2 months old but emphasized vocalization and distraction at 6 months. This change may reflect mothers’ sensitivity to developmental shifts in their infants’ cognitive development, as an infant’s ability to sustain attention improves dramatically between 3 and 6 months of age (Colombo, 2001).

Although global sensitivity may demonstrate relative stability across time, the specific factors that compose the global construct may vary in their function and operation, with different factors exerting influence at different points in time in ways that support overall sensitivity. Indeed, mothers modify their caregiving behaviors to meet the changing needs of their rapidly developing infants, but too few studies have examined how multiple, specific sensitive caregiving behaviors change across time in systematic ways in the first year of life. An exception is Bornstein, Tamis-LeMonda, Hahn, and Haynes (2008), who examined general versus specific maternal responsiveness—a construct that overlaps conceptually with maternal sensitivity—to infant bids for their mother’s attention. They reported that in most cases, mothers demonstrated a similar increase in general responsiveness from infant age 10 to 21 months. However, the ways in which mothers responded to their infants differed with respect to initial levels and trajectories, with some behaviors decreasing (e.g., exploratory prompts) and others increasing (e.g., imitation). What remains unclear is whether component behaviors of sensitivity change over time in a way that is meaningful for children’s socioemotional functioning as well as for the quality of future dyadic interactions between mothers and their children.

Maternal Sensitivity and Later Competencies

Maternal sensitivity has been linked to a variety of critical developmental competencies and processes, including the emergence of children’s emotion competence and the quality of the mother-child relationship as indexed by reciprocity in dyadic interaction.

Emotion competence.

Early sensitive care has been consistently connected to later socioemotional functioning. Mothers who acknowledge their children’s emotions and support a full range of emotion expressions typically have infants with advanced emotion regulation strategies (Shaffer, Suveg, Thomassin, & Bradbury, 2012), more prosocial behavior (Roberts, 1999), and more competent peer relationships (Denham, 1997; Garner & Estep, 2001; Garner, Jones, & Miner, 1994). Similarly, mothers who are prompt and appropriate in their responses to infant distress at 6 months had more emotionally competent 2 year olds (Leerkes et al., 2009). In contrast, lower levels of warmth and support for appropriate emotion expressions have been associated with difficulties in emotion regulation and internalizing and externalizing problems (Eisenberg, Fabes, Schaller, Carlo, & Miller, 1991; Eisenberg, Cumberland, & Spinrad, 1998; Garner & Estep, 2001). Nonetheless, it remains unclear whether other component behaviors commonly associated with maternal sensitivity (e.g., affectionate touch, consistent gaze, vocalizations) similarly contribute to infant’s socioemotional development.

Dyadic reciprocity.

Whereas maternal sensitivity emphasizes a mother’s ability to appropriately respond to infant’s cues, dyadic reciprocity refers to the “dance-like” fluency and rhythm created when infants begin to play a more active role in the mother-child interaction. Although similar to synchrony, dyadic reciprocity focuses on the dynamic nature of the relationship instead of the harmony of each individual’s behaviors or affective states, examining the way infants and mothers shape each other’s behaviors in a reciprocal manner. As infants develop the capacity to initiate social interactions, dyadic reciprocity increases substantially between ages 3 and 24 months (Feldman, 2010). Reciprocal interactions are thought to foster the development of individual identity (Dunham & Dunham, 1995), self-regulation, and empathy (Feldman, 2007). Furthermore, mother-infant reciprocity during infancy has been linked to lower rates of depression and better adjustment in adolescence (Feldman, 2010). Although there is little research directly linking maternal sensitivity to dyadic reciprocity, maternal sensitivity may be a prerequisite for dyadic synchrony (Skuban, Shaw, Gardner, Supplee, & Nichols, 2006). The evidence is further limited when considering how individual factors that compose sensitivity contribute to dyadic interactions. For example, there is support for a connection between early affectionate touch and later dyadic reciprocity (Ferber et al., 2008). In a study comparing mothers who used kangaroo care (i.e., increased skin-to-skin touch) compared to a control group, those who used kangaroo care were significantly more sensitive and had higher dyadic reciprocity (Feldman, Weller, Sirota, & Eidelman, 2003). Appropriate and coordinated maternal vocalizations have also been linked to reciprocal interactions. Specifically, the adaptation of caregivers’ vocalizations was related to movement by the infant, resulting in a back-and-forth interaction marked by movement by the infant and coordinated vocalizations by the mother (Condon & Sander, 1974). Finally, there is evidence that attunement or responsivity to infant behavior leads to reciprocal interactions (Ross, Allan, & Lollis, 1988).

The Present Study

The first aim of the current study was to examine and compare developmental trajectories of maternal sensitivity as a global construct and as a range of specific individual behaviors that have been proposed to compose sensitivity across the first year of infancy. It was hypothesized that trajectories would differ among global maternal sensitivity—defined as the composite of multiple sensitive maternal behaviors—and individual sensitive behaviors that compose the global composite. Specifically, it was expected that some indices of sensitivity would significantly increase (e.g., vocal appropriateness), some would remain stable (e.g., the global index of sensitivity), and some would significantly decline (e.g., touch). The second aim was to examine whether trajectories of the individual behaviors and the global index of maternal sensitivity would differentially predict infant emotion competence and dyadic reciprocity at 12 months. It was predicted that the developmental trajectory of the global index of maternal sensitivity would significantly and positively predict both emotion competence and dyadic reciprocity at 12 months, but there would be differences in prediction among the individual sensitive behavior trajectories.

METHOD

Participants

Participants included 322 mothers and their infants (53.7% female). Eligibility criteria for mothers included (1) self-identification as Mexican or Mexican American, (2) fluency in Spanish or English, (3) age 18 years or older, (4) low-income status (eligibility for Medicaid or Federal Emergency Services coverage or self-reported family income below $25,000), and (5) anticipated delivery of a singlet baby with no prenatal evidence of serious health or developmental problems. The majority of mothers (86.1%) were born in Mexico, 54.1% of whom came to the United States before age 17. Of the mothers who immigrated to the United States, mothers varied in how long they had lived in the United States, ranging from zero to 32 years (M = 11.89, SD = 5.98). The majority (82.0%) spoke Spanish as their primary language. Mothers ranged in age from 18 to 42 years old (M = 27.86, SD = 6.48). Mothers most often reported that they were unmarried but living with a romantic partner (45.7%) and supporting a family of four (including themselves) on a total household income of $10,001–$15,000 per year (27.7%). The vast majority (94.2%) reported receiving some type of public assistance. More than half (59.1%) of mothers in this sample did not complete high school, and 83.6% indicated that they were unemployed at the time of recruitment.

Procedures

Mothers were recruited from community prenatal clinics in the southwestern United States serving low-income families. Informed consent and demographic information were obtained during a prenatal home visit in the participant’s preferred language. Additional home visits were completed when infants were 1.5, 3, 4.5, 6, and 9 months old. To minimize participant burden, the study followed a “planned missingness” design (Enders, 2010; Little & Rhemtulla, 2013) for the home visits. Using this method, all dyads participated in the prenatal and 1.5-month home visit but were randomly assigned by a computer program to complete two of the three remaining home visits. This design allowed for a reduction in participant burden but retained statistical power and produced data missing completely at random (Enders, 2010). During the home visits, bilingual interviewers traveled to participants’ homes and videorecorded mothers interacting with their infants during five specified tasks. At 12 months, mothers and their children traveled to a university campus to complete similar tasks in a laboratory setting. Interviewers also surveyed parents regarding children’s socioemotional development during this laboratory visit. All questions were administered in participants’ language of choice and were read aloud due to variability in literacy. Participants were compensated monetarily for data collection at each time point. The university’s Institutional Review Board approved all study procedures. Note that for the present study, only data from the prenatal, 1.5-, 3-, 4.5-, 6-, and 12-month visits were used.

Mother-infant interaction tasks.

During the 3-, 4.5-, and 6-month home visits, mothers and infants participated in a free play task (5 min), which constituted a “warm up” context for the structured tasks that followed (i.e., arm restraint, soothing, teaching, peekaboo). Only observations during the free play task during the home visits were used in the current study. During the free play task, mothers were provided with a small basket of toys and objects and asked to play with their infants as they normally would when alone. At the 12-month laboratory visit, the mother-infant dyads participated in an unstructured free play task (5 min) followed by a clean-up task (2 – 5 min). In addition, mothers blew bubbles (3 min) for their infant to pop and completed four separate teaching tasks (4 min each) in which they were asked to engage their infants in activities intended to elicit mild frustration in both the infant and mother (e.g., have the infant make a block tower, build a puzzle).

Coding of interaction tasks.

Videorecorded mother-infant interactions during the 3-, 4.5-, and 6-month home visits and the 12-month laboratory visit were coded using the Coding Interactive Behavior protocol (CIB; Feldman, 1998) for maternal sensitivity and dyadic reciprocity. CIB is a global rating system for parent-child interactions that has been validated across a variety of social and cultural contexts (Feldman & Masalha, 2007). Undergraduate research assistants coded mother-infant interactions for 42 parent, infant, and dyadic behaviors. Each behavior was rated on a scale of 1 (a little) to 5 (a lot), and half-point scores could be given (e.g., 2.5). Coders were trained in pairs of two by a graduate-level, master coder until they met reliability criteria (i.e., at least 85% agreement within 1 scale point with the master coder) and met weekly with the master coder to reduce coding drift. Each research assistant in a coding team separately coded video observations and then came to consensus when disagreement was present. The same coding teams coded all 42 behavior categories. Twenty percent of all videos were master coded.

Measures

Maternal sensitivity.

Coded ratings of 11 individual parent behaviors typically associated with maternal sensitivity during a free play task were made at each data collection period. These behaviors included: acknowledging, consistency of style, elaborating, enthusiasm, gaze, positive affect, appropriate range of affect, resourcefulness, supportive presence, affectionate touch, and vocal appropriateness. Interrater reliability across all sensitivity behaviors was good (3 months: ICC = .76; 4.5 months: ICC = .65; 6 months: ICC = .67; 12 months: ICC = .62). The maternal sensitivity composite (i.e., the global index of sensitivity) was created by averaging the maternal behaviors within each time point. Internal consistency for the composite score was excellent each time point (3 months: α = .89; 4.5 months: α = .87; 6 months: α = .88; 12 months: α = .84).

Acknowledging.

Behaviors associated with acknowledging include vocalizations, gaze, facial expressions, and body movements that indicate that the mother is aware and receptive of the child’s cues. For example, the infant yawns and the mother asks, “Are you tired?” Scores for this scale are contingent on infant social signals and refer to infant-leads-parent-responds interactions.

Consistency of style.

Mothers are consistent during interactions with the infant, and their movement, behavior, and expression appear predictable to the infant. No abrupt changes are observed in the mother’s level of attention or involvement during the interaction.

Elaborating.

Mothers expand and elaborate their imitated behaviors or expressions by adding variations and increasing the complexity of the infant’s social signal. For example, the infant may vocalize “ba,” and the mother responds with “ba ba ba” in an excited voice, adding, “You are really telling me how much you like this game.”

Enthusiasm.

Mothers demonstrate genuine enthusiasm during the interaction when they are involved, display positive affect, and demonstrate clear signs that they enjoy interacting with their infant.

Gaze.

Maternal gaze is defined as any instance in which the mother focuses her gaze/attention on the infant or on an object of joint activity.

Positive affect.

Expressions of positive affect are characterized by warmth and emotional openness during the interaction. Maternal positive affect is expressed by relaxed body posture, warm tone of voice, frequent smiling or laughter, and happy facial expressions.

Appropriate range of affect.

Mothers were assessed as to whether they expressed a full range of emotional behaviors and flexibly shift affective states in accordance with the infant’s activity and emotional state. During infancy, it is important to consider whether the mother’s affect is predictable and comfortable versus shocking, unnatural, or uncomfortable.

Resourcefulness.

Mothers who are resourceful are flexible and creative in managing changes in their infant’s affect, behavior, fussiness, or level of interest. They are also able to maintain their infant’s interest in creative and flexible ways

Supportive presence.

This construct addresses the degree to which the mother’s presence provides a “secure base” for the infant with respect to warmth, closeness, and mutuality (Bowlby, 1988). Maternal responses must be appropriate, receptive, and provide an external regulatory structure for the infant’s activities and emotions. Supportive presence may be observed in maternal affect, verbalizations, touch, gaze, or physical proximity (e.g., holding the infant in the mother’s arms).

Affectionate touch.

Maternal affectionate touch refers to the mother touching her infant affectionately, often, and spontaneously. Examples of affectionate touch include kissing, hugging, and caressing. Affectionate touch is distinct from instrumental touch (e.g., dressing, feeding, diaper changing), physical manipulation (e.g., playing with the infant’s fingers), or unintentional touch (e.g., accidentally brushing against the infant).

Vocal appropriateness.

Mother’s use of appropriate tone of voice and level of repetition was assessed based on the infant’s developmental ability. For infants up to 6 months of age, vocal appropriateness is defined by “motherese” (i.e., high-pitched, repetitive vocalizations) and beyond the first year, it is defined by clear, warm, expressive, and variable speech during interactions.

Emotion competence.

Child socioemotional developmental competencies were assessed at 12 months using the Brief Infant-Toddler Social and Emotional Assessment (BITSEA; Briggs-Gowan & Carter, 2006), a standardized 42-item parent report questionnaire. The BITSEA includes a Problem Scale (31 items) and a Competence Scale (11 items). The Competence Scale was used in all study analyses. Example items include “looks for you [or other parent] when upset”; “follows rules”; and “hugs or feeds dolls or stuffed animals.” Parents rated these items as not true/rarely (0), somewhat true/sometimes (1), or very true/often (2). Lower total scores on the Competence scale indicate a lower level of competence. The BITSEA has been established as a reliable and valid measure for the evaluation of social-emotional competence in children ages 12 to 36 months (α = .66; Briggs-Gowan, Carter, Irwin, Wachtel, & Cicchetti, 2004). Internal reliability was lower than expected in the current study sample (α = .53). Hungerford, Garcia, and Bagner (2015) evaluated the psychometric properties of the BITSEA in a predominantly low-income, Latin American sample of mothers and their 12- to 15-month-old infants, similar to the current study sample. They reported that the reliability of the Competence Scale was also low (α = .57) but still a valid measure of emotion competence in infants.

Dyadic reciprocity.

Coded ratings of dyadic reciprocity during each of the 12-month interaction tasks were also obtained. Reciprocity during interactions refers to “give and take” interactions between mother and infant. Highly reciprocal interactions are synchronous and mutual, exhibiting both partners’ ability to read the other’s cues and respond appropriately and timely. They involve participation from both mother and infant, interpreting and responding to each other’s cues appropriately, and frequent bouts back and forth. Scores were averaged across tasks to create a single composite. Interrater reliability across all tasks was very good (ICC = .73).

Covariates.

During the prenatal home visit, mothers were asked to report their age, number of years in the United States, and years of education. During the 1.5-month home visit, mothers reported the infant’s sex (0 = male, 1 = female).

Analytic Plan to Address Study Aims

Our first aim was to describe the individual developmental trajectories of specific behaviors associated with maternal sensitivity. To address this aim, a no growth (i.e., intercept-only) and linear growth model were fitted and compared for the global index of sensitivity and for each sensitive maternal behavior. The factor loadings of the intercept factor were fixed to 1 for each model, whereas the slope factor loadings were fixed to [0 0 0 0] for the no growth model and [0 1.5 3 9] to account for unequal spacing between time points for the linear growth models. All models were specified and estimated using Mplus 8.1 (Muthén & Muthén, 1998–2017). Missing data were handled with full information maximum likelihood estimation. Model fit was evaluated based on multiple indices, including the chi-square test, the comparative fit index (CFI; Bentler, 1990), the standardized root mean square residual (SRMR), and the root mean squared error of approximation (RMSEA; Browne & Cudeck, 1993). Note that an adjusted CFI, which uses the chi-square statistic from the intercept-only (i.e., no growth) model as the null model (Widaman & Thompson, 2003), was used. Good model fit is indicated by a nonsignificant chi-square value, and values of greater than 0.95 on CFI, less than 0.08 on the SRMR, and less than 0.05 on the RMSEA (Hu & Bentler, 1999). Our second aim was to examine prediction of dyadic reciprocity and infant’s emotion competence at 12 months from trajectories of maternal sensitivity and individual trajectories of sensitivity behaviors. Once linear growth models for the global and individual indices of maternal sensitivity were identified, dyadic reciprocity and infant emotion competence were regressed on the intercept and slope of each specific index of sensitivity (individual and global) in a series of separate models.

RESULTS

Descriptive Statistics and Preliminary Analyses

Table 1 presents descriptive statistics for all study variables. Notably, mean values varied widely among global and individual indices of maternal sensitivity. The majority of means for sensitive behaviors were greater than 4 across all four time points (i.e., consistency of style, enthusiasm, gaze, positive affect, appropriate range of affect, resourcefulness, and supportive presence). Given that the CIB scores maternal behavior on a Likert scale of 1 to 5, mothers frequently displayed these sensitive behaviors, on average, during the infancy period. Others were lower, such as touch (Ms range = 1.43–2.73, SDs range = .72–1.09) and elaboration (Ms range = 1.21–1.37, SDs range = .42–63). The means of global sensitivity, however, ranged from 3.68 to 3.95 across all time points. Observed means are plotted in Figure 1. The majority of variables were within conventional benchmarks for skew and kurtosis (upper and lower bounds of 2 and 7, respectively; Chou & Bentler, 1995; Curran, West, & Finch, 1996), with the exception of elaboration (positively skewed at the 3-, 4.5-, and 6-month time points). Notably, consistency of style, enthusiasm, positive affect, appropriate range of affect, supportive presence, and vocal appropriateness were also negatively skewed, but only at the 12-month time point. Six participants were missing on all study variables and were not included in subsequent analyses.

TABLE 1.

Descriptive Statistics for all Study Variables (N = 316)

Variable	3 months				4.5 months				6 months				12 months
	n	Range	M(SD)	Skewness/ Kurtosis	n	Range	M(SD)	Skewness/ Kurtosis	n	Range	M(SD)	Skewness/ Kurtosis	n	Range	M(SD)	Skewness/ Kurtosis
Sens	198	1.45–4.59	3.69(.57)	−.87/.90	197	1.91–4.68	3.77(48)	−.79/1.33	199	2.00–4.45	3.68(.52)	−.83/.45	189	1.82–4.64	3.95(.40)	−1.95/5.70
Ack	198	1.5–5.0	3.92(.67)	−.69/.69	198	1.50–5.0	4.06(.62)	−.67/.91	199	1.0–5.0	3.72(.72)	−.84/1.15	189	2.0–5.0	3.45(.73)	−.25/−.68
Con	198	2.0–5.0	4.31(.56)	−.58/.59	197	3.0–5.0	4.47(.49)	−.54/−.47	199	2.5–5.0	4.45(.67)	−.90/−.18	189	2.5–5.0	4.79(.48)	−2.49/6.10
Elab	197	1.0–4.0	1.34(.63)	2.10/4.10	197	1.0–4.0	1.29(.54)	2.24/6.04	199	1.0–3.5	1.21(.45)	2.62/7.94	189	1.0–3.0	1.37(.52)	1.32/1.03
Enth	197	1.0–5.0	4.11(.89)	−1.11/.93	197	2.0–5.0	4.09(.71)	−.67/.53	199	1.5–5.0	4.11(.86)	−.86/.30	189	1.0–5.0	4.61(.73)	−2.30/5.72
Gaze	198	1.5–5.0	4.06(.66)	−1.04/1.64	197	2.0–5.0	4.23(.63)	−1.02/1.34	199	2.5–5.0	4.39(.61)	−.83/.12	189	2.0–5.0	4.49(.59)	−1.04/.83
Pos	198	1.0–5.0	4.15(.86)	−1.35/1.73	197	1.0–5.0	4.20(.76)	−1.09/1.66	199	2.0–5.0	4.32(.72)	−.94/.61	189	1.5–5.0	4.74(.60)	−2.99/10.09
Rang	198	1.0–5.0	4.09(.77)	−1.04/1.36	197	1.5–5.0	4.20(.66)	−1.03/2.05	199	1.5–5.0	4.16(.78)	−.72/.06	189	2.0/5.0	4.71(.58)	−2.33/5.73
Res	198	1.5–5.0	4.00(.82)	−.72/−.17	197	1.5–5.0	4.19(.67)	−1.00/1.92	199	1.0–5.0	4.20(.84)	−1.07/1.08	189	1.5–5.0	4.54(.70)	−1.74/2.97
Sup	198	1.5–5.0	4.01(.74)	−.63/.13	197	1.5–5.0	4.23(.70)	−1.11/1.84	199	1.5–5.0	4.19(.78)	−.85/.43	189	1.5–5.0	4.72(.56)	−2.69/8.93
Tch	197	1.0–5.0	2.73(1.09)	.04/−1.11	197	1.0–5.0	2.45(1.04)	.30/−.78	199	1.0–4.5	1.88(.86)	.82/−.03	189	1.0–4.0	1.43(.72)	1.76/2.34
Voc	192	1.0–5.0	3.87(1.27)	−1.25/.38	195	1.0–5.0	4.03(1.05)	−1.45/1.83	197	1.0–5.0	3.81(1.13)	−1.07/.49	189	1.0–5.0	4.55(.79)	−2.26/5.30
Rec													189	2.07–3.80	3.66(.53)	−.21/−.13
Comp													195	7.0–22.0	14.56(2.66)	.09/.17

Note. Sens = global index of sensitivity; ack = acknowledging; con = consistency; elab = elaboration; enth = enthusiasm; gaze = gaze; pos = positive affect; rang = appropriate range of affect; res = resourcefulness; sup = supportiveness; tch = affectionate touch; voc = vocal appropriateness; rec = dyadic reciprocity; comp = infant emotion competence.

Figure 1.

Plotted means of observed CIB scores and maternal sensitivity composite.

With respect to covariates (i.e., maternal age, years in the United States, maternal education, and child sex), only child sex was significantly related to emotion competence, with female infants demonstrating higher emotion competence (M = 15.16, SD = 2.72) than male infants (M = 13.79, SD = 2.40), t(192) = −3.67, p < .001. Zero-order correlations among sensitive behaviors across time points are presented in Table 2. The global index of sensitivity was positively correlated with each consecutive time point (i.e., at 3 and 4.5 months, 4.5 and 6 months, and 6 and 12 months). Only enthusiasm, positive affect, appropriate range of affect, affectionate touch, and vocal appropriateness demonstrated a similar pattern across all four time points. Table 3 presents bivariate correlations among maternal sensitivity, dyadic reciprocity, and emotion competence. Dyadic reciprocity was uncorrelated with maternal sensitivity and all sensitivity behaviors at 3 months. Maternal sensitivity, appropriate range of affect, and vocal appropriateness were positively correlated with dyadic reciprocity across the 4.5- and 6-month time points. All correlations among dyadic reciprocity and sensitive maternal behaviors were significant, with the exception of consistency of style and affectionate touch, at 12 months. Emotion competence was uncorrelated with maternal sensitivity and all sensitive behaviors at 3 and 4.5 months. Emotion competence was positively correlated with enthusiasm, positive affect, and supportive presence at 6 months. At 12 months, emotion competence was positively associated with the global index of sensitivity, acknowledging, and vocal appropriateness.

TABLE 2.

Zero-order Correlations among Sensitive Maternal Behaviors Across Time Points (N = 316)

	3 months	4.5 months	6 months	12 months
Sens (3mo)	-
Sens (4.5 mo)	.44***	-
Sens (6 mo)	.18	.30**	-
Sens (12 mo)	−.01	.32***	.28**	-
Ack (3mo)	.74***
Ack (4.5 mo)	.04	.77***
Ack (6 mo)	−.11	.14	.71***
Ack (12 mo)	−.03	.09	.14	.63***
Con (3mo)	.53***
Con (4.5 mo)	−.06	.52***
Con (6 mo)	.02	.24*	.60***
Con (12 mo)	−.15	.009	.09	.42***
Elab (3mo)	.39***
Elab (4.5 mo)	.18	.39***
Elab (6 mo)	.09	−.001	.21***
Elab (12 mo)	−.16	.08	.03	.37***
Enth (3mo)	.85***
Enth (4.5 mo)	.48***	.87***
Enth (6 mo)	.23*	.27**	.91***
Enth (12 mo)	.17	.34***	.19*	.82***
Gaze (3mo)	.60***
Gaze (4.5 mo)	.37***	.51***
Gaze (6 mo)	.09	.20	.60***
Gaze (12 mo)	−.08	.02	.05	.59***
Pos (3mo)	.83***
Pos (4.5 mo)	.53***	.80***
Pos (6 mo)	.18	.35***	.85***
Pos (12 mo)	.12	.23**	.34***	.82***
Rang (3mo)	.90***
Rang (4.5 mo)	.32**	.90***
Rang (6 mo)	.09	.23*	.86***
Rang (12 mo)	.00	.24**	.23**	.76***
Res (3mo)	.80***
Res (4.5 mo)	.10	.80***
Res (6 mo)	.12	.17	.86***
Res (12 mo)	−.20*	.19*	.16	.72***
Sup (3mo)	.82***
Sup (4.5 mo)	.20	.84***
Sup (6 mo)	.12	.10	.86***
Sup (12 mo)	−.07	.15†	.22*	.70***
Tch (3mo)	.41***
Tch (4.5 mo)	.22*	.33***
Tch (6 mo)	.19	.20*	.18*
Tch (12 mo)	.15	.04	.19*	.27***
Voc (3mo)	.76***
Voc (4.5 mo)	.43***	.72***
Voc (6 mo)	.27*	.47***	.69***
Voc (12 mo)	.02	.20*	.31***	.77***

Note. Sens = global index of sensitivity; ack = acknowledging; con = consistency; elab = elaboration; enth = enthusiasm; gaze = gaze; pos = positive affect; rang = appropriate range of affect; res = resourcefulness; sup = supportiveness; tch = affectionate touch; voc = vocal appropriateness. Bivariate correlations presented below the diagonal are correlations for each variable across all four time points. Correlations presented on the diagonal and in bold are bivariate correlations among the global index of sensitivity and the specific indicator at each time point.

^*p < .05.

^**p < .01.

^***p < .001.

TABLE 3.

Zero-order Correlations among All Maternal Sensitivity Variables at 3, 4.5, 6, and 12 Months; and Dyadic Reciprocity and Emotion Competence at 12 Months (N = 316)

Variable	Dyadic Reciprocity					Emotion Competence
	3 months	4.5 months	6 months	12 months		3 months	4.5 months	6 months	12 months
Sens	−.07	.22*	.16	.47***		.11	.04	.17	.15*
Ack	.01	.15	.14	.39***		.16	−.06	.03	.21**
Con	.07	.02	.05	.11		.17	.06	.11	.05
Elab	−.03	.12	−.03	.15*		.05	.05	.13	.14†
Enth	.01	.20*	.10	.34***		.08	.07	.18*	.10
Gaze	−.03	.002	.01	.29***		.09	.09	.03	.02
Pos	−.04	.12	.15	.31***		.05	.11	.18*	.09
Rang	−.08	.25**	.19*	.31***		.11	.07	.14	.14†
Res	−.06	.11	.11	.47***		.16	−.01	.16	.03
Sup	−.12	.20*	.05	.35***		.15	−.03	.19*	−.06
Tch	−.05	.13	.14	.14		−.05	−.02	.12	.11
Voc	−.12	.18*	.15	.29***		−.01	−.02	.03	.16*

Note. Sens = maternal sensitivity composite score; ack = acknowledging; con = consistency; elab = elaboration; enth = enthusiasm; gaze = gaze; pos = positive affect; rang = appropriate range of affect; res = resourcefulness; sup = supportiveness; tch = affectionate touch; voc = vocal appropriateness; rec = dyadic reciprocity; comp = infant emotion competence.

^*p < .05.

^**p < .01.

^***p < .001.

Linear Growth Models of Maternal Sensitivity and Individual Sensitive Behaviors

To account for the nonnormal distribution of some of the variables—and based on the missing at random (MAR) assumption—all models were specified using a full information maximum likelihood estimator (ESTIMATOR=MLR). Model-estimated means based on the linear growth models for the global and individual indices of maternal sensitivity are plotted in Figure 2. A full interpretation of the fit indices and parameters for the global index of maternal sensitivity linear growth model is presented in this section to serve as a model for the interpretation of all estimated models. Model fit, intercept, and slope results for the component behaviors models are presented in the following sections for concision. The linear growth model for the global index of sensitivity fit the data well based on some fit indices, χ²(N = 314; df = 5) = 11.99, p = .06; SRMR = .06, but less well on others (CFI = .88; RMSEA = .06). The means and variances for the intercept and linear slope from the unstandardized solution were: Intercept_sens (M = 3.68, p < .001; s² = .12, p < .001) and Slope_sens (M = .03, p < .001; s² = .003, p = .009). On average, mothers varied in their initial levels of maternal sensitivity. Mothers increased somewhat in sensitivity across the infancy period, but also varied in their rate of change. The intercept and slope correlation was negative (standardized estimate = −.58, p < .001), indicating that higher levels of sensitivity at 3 months were associated with a lower rate of change in overall sensitivity across the first year of life. Conversely, lower levels of sensitivity at 3 months corresponded to a higher rate of change (i.e., a steep increase) in sensitivity from 3 to 12 months.

Figure 2.

Model-estimated means based on linear growth models.

Model fit.

Based on all fit indices, the linear growth model fit well for consistency of style, positive affect, appropriate range of affect, resourcefulness, and supportive presence. A linear growth model demonstrated appropriate fit for enthusiasm based on some fit indices, χ²(N = 314; df = 5) = 10.51, p = .06; SRMR = .06, but not others (CFI = 0.90, RMSEA = 0.06). Linear growth modeling was not appropriate for the other sensitivity behaviors: acknowledging, elaboration, gaze, touch, and vocal appropriateness. Nonlinear models (i.e., quadratic) were tested but did not fit the data well.

Intercepts.

The intercepts for the majority of sensitive behaviors were greater than 4 (Ms range = 3.80–4.34, p < .001). Two behaviors—touch and elaboration—had much lower intercepts (M = 2.54, p < .001 and M = 1.26, p < .001, respectively), which could account for the lower global sensitivity intercept (M = 3.68, p < .001). The majority of sensitive behaviors had a significant intercept variance (s²s range = .06-.70), indicating that mothers varied on their initial levels of a specific behavior. The intercept variance, however, was nonsignificant for acknowledging (M = 4.00, p < .001; s² = .004, p = .93) and consistency of style (M = 4.34, p < .001; s² = .03, p = .32), suggesting that there are no individual differences in mothers’ levels of acknowledging behavior, consistency, or predictability when infants are 3 months old.

Growth.

The majority of component behaviors followed a similar positive, linear trend to the global sensitivity index (Ms range = .05-.08, p < .001). However, two individual behaviors—acknowledging (M = −.06, p < .001) and touch (M = −.13, p < .001)—decreased across the first year of infancy. The linear slopes of many of the individual sensitive behaviors (i.e., consistency of style, elaboration, positive affect, appropriate range of affect, resourcefulness, supportive presence, and vocal appropriateness) had a significant variance, suggesting that there were individual differences in the rate of change in each behavior across the first year of life. Another behavior, elaboration, did not vary across time (M = .01, p = .08). See Table 4 for a summary of results of all growth analyses.

TABLE 4.

Summary of Fit Statistics, Means, and Variances for all Maternal Sensitivity Linear Growth Models

		χ2						Means			Random Effects (variances & correlations)
Model	n		df	CFIa	RMSEA(CI)	SRMR		I	S		I	S	I w S
Sensb	314	11.99	6	.88	.06(.00, .10)	.05		3.68***	.03***		.12***	.003**	−.58***
Ack	314	21.10***	5	.80	.10(.06, .15)	.12		4.00***	−.06***		.004	.003	−.37
Con	314	9.26	5	.95	.05(.00, .10)	.08		4.34***	.05***		.03	.003*	−.69***
Elab	314	12.89*	5	.34	.07(.02, .12)	.07		1.26***	.01		.06*	.003*	−.79***
Enth	314	10.51	5	.90	.06(.00, .11)	.06		4.02***	.06***		.27***	.003	−.59**
Gaze	314	11.76*	5	.87	.07(.01, .12)	.11		4.16***	.04***		.15***	.003	−.87***
Pos	314	3.78	5	1.00	.00(.00, .07)	.07		4.11***	.07***		.31***	.006***	−.62***
Rang	314	8.39	5	.96	.05(.00, .10)	.05		4.06***	.07***		.16**	.005*	−.52***
Resb	314	4.96	6	1.00	.00(.00, .07)	.06		4.06***	.05***		.12*	.009***	−.59***
Supb	314	4.17	6	1.00	.00(.00, .06)	.04		4.04***	.08***		.11*	.006***	−.56***
Tch	314	33.62***	5	.85	.14(.09, .18)	.09		2.54***	−.13***		.23**	.002	−.77
Vocb	312	17.38**	6	.87	.08(.04, .12)	.08		3.80***	.08***		.70***	.02*	−.71***

Note. Sens = maternal sensitivity composite score; ack = acknowledging; con = consistency; elab = elaboration; enth = enthusiasm; gaze = gaze; pos = positive affect; rang = appropriate range of affect; res = resourcefulness; sup = supportiveness; tch = affectionate touch; voc = vocal appropriateness. I = Intercept; S = Slope (unstandardized coefficients reported); I w S = correlation between Intercept and Slope (standardized coefficient reported).

^aAdjusted CFI reported uses the chi-square statistic from the intercept-only (i.e., no growth) model as the null model (Widaman & Thompson, 2003).

^bDuring the initial specification, an error message indicated that the residual covariance matrix was not positive definite. Closer inspection revealed that the residual variance at the 12-month time point was negative, nonsignificant, and close to 0. According to Muthén (2007), it is appropriate to fix a residual variance to zero provided the value is small, negative, and nonsignificant. The variance at the 12-month time point was fixed to zero when the model was respecified, after which the models ran without error.

^*p < .05.

^**p < .01.

^***p < .001.

Predicting Dyadic Reciprocity and Emotion Competence

Before connecting trajectories of maternal sensitivity to child and dyadic outcomes, correlations among covariates (including maternal age, education, and years in the United States) and the Intercept and Slope factors for each model were examined to determine whether covariates should be included in subsequent models. Dyadic reciprocity and infant emotion competence at 12 months were then regressed on the intercept and slope of the global sensitivity index. Separate models were also estimated for consistency of style, positive affect, appropriate range of affect, resourcefulness, and supportive presence, as these behaviors demonstrated good fit for linear growth curve modeling in previous analyses. See Table 5 for a summary of all model results.

TABLE 5.

Latent Growth Curve Analysis of Initial Maternal Sensivity (at 3 months) and Trajectories of Maternal Sensitivity across the First Year of Life to Dyadic Reciprocity and Emotion Competence at 12 months

									Dyadic Reciprocity		Emotion Competence
Variable	n	χ2	df	CFI	RMSEA(CI)	SRMR			β	SE	β	SE
Sens	321	42.25**	21	.81	.06(.03, .08)	.08		Intercept	.33**	.11	.14	.15
								Slope	.39***	.08	.08	.10
Con	315	15.37	14	.91	.02(.00, .06)	.06		Intercept	.27	.31	.49	.36
								Slope	.25	.21	.28	.28
Pos	321	17.52	19	1.00	.00(.00, .05)	.05		Intercept	.36*	.17	.23*	.11
								Slope	.44*	.19	.07	.10
Rang	321	22.71	19	.94	.03(.00, .06)	.06		Intercept	.40**	.14	.24	.13
								Slope	.41*	.16	.14	.11
Res	321	16.56	20	1.00	.00(.00, .04)	.05		Intercept	.33*	.1	.27	.17
								Slope	.59***	.08	.09	.09
Sup	321	29.86	25	.91	.03(.00, .05)	.06		Intercept	.27*	.14	.23	.16
								Slope	.42***	.09	−.06	.10

Note. Sens = maternal sensitivity composite score; con = consistency; pos = positive affect; rang = appropriate range of affect; res = resourcefulness; sup = supportiveness. Emotion competence was regressed on child sex in each model. Based on significant correlations among covariates and Intercept and Slope factors, the models for sens, pos, rang, and res were estimated controlling for maternal age on the Intercept and Slope factors. The model for sup was estimated controlling for maternal age and education on the Intercept and Slope factors. The con model did not include any covariates on the Intercept and Slope Factors. Standardized (β) estimates are reported.

^*p < .05.

^**p < .01.

^***p < .001.

Both the maternal sensitivity intercept (β = .33, SE = .11, p = .002) and slope (β = .39, SE = .08, p < .001) positively predicted dyadic reciprocity. Mothers with higher sensitivity at 3 months had a higher frequency of engaging in “give and take” with their infants during structured interactions at 12 months. Of the individual sensitive maternal behaviors, all but the consistency of style intercept predicted dyadic reciprocity (.27 < β < .40). The slope of maternal sensitivity was also related to dyadic reciprocity, wherein mothers with higher rates of change in sensitivity had a higher frequency of engaging in “give and take” interactions with their infants at 12 months. Similarly, mother’s rate of change in all individual components of maternal sensitivity predicted dyadic reciprocity (.41 < β < .59), except for consistency of style.

The maternal sensitivity intercept was unrelated to emotion competence (β = .14, SE = .15, p = .35). With regard to the sensitivity components, only the intercept of positive affect (β = .23, SE = .11, p = .03) was related to infant emotion competence, indicating that children with mothers who expressed more positive affect at 3 months had higher emotion competence at 12 months. Mother’s initial levels of appropriate range of affect (β = .24, SE = .13, p = .06) were marginally associated with infants’ later emotion competence. Notably, none of the slopes of the global or individual sensitive maternal behaviors examined predicted infant emotion competence.

DISCUSSION

The reliance on maternal sensitivity as a construct in the absence of a consensus of its conceptualization and operationalization has posed a challenge for the field of child development, as it has been difficult to come to a richer understanding of the complexity inherent in its representation. This study deconstructed maternal sensitivity, compared trajectories of the component behaviors to the trajectory of the global construct, and examined contributions of individual component behaviors to later infant and dyadic competencies. Overall, findings suggested differences in the trajectories of mothers’ sensitive caregiving behaviors across the first year of infant life compared to the trajectory of the global construct, and these differences were meaningful for infant’s emotion competence and dyadic reciprocity at 12 months.

Developmental Trajectories of Maternal Sensitivity

We anticipated that mothers would vary in initial levels of sensitivity across both global and individual indices. That is, some indices would be higher or lower than others in relative values reflective of sensitive responding. These expectations were met in some cases but not others. On average, mothers were moderately sensitive on the global index when infants were 3 months old, with some individual sensitive behaviors fluctuating higher (e.g., positive affect, acknowledging) and some lower (i.e., touch, elaboration) than the global factor. It is likely that the relatively lower instances of touch and elaboration in this sample substantially lowered the global index of sensitivity. Although there is some evidence that Mexican-origin mothers may be more likely to touch their infants during interactions compared to Dominican and African American mothers (Tamis-LeMonda et al., 2012), touch was a low frequency behavior in the current study sample. It could be that, although Mexican culture typically emphasizes touch in parenting practices with infants, the measure used to evaluate affectionate touch did not adequately capture typical touch practices in this study sample. Observational coding of touch emphasized distal parenting practices—which describes face-to-face contact and object stimulation—compared to proximal parenting practices—characterized by body contact and stimulation. Hispanic cultures may be higher in proximal parenting practices compared to European American groups (Keller et al., 2009). Similarly, verbal interactions—which are included in the definition of elaboration—are less common among low-income, less educated families (Hoff, Laursen, & Tardif, 2002), which may account for the low instances of elaboration in this study sample. Other research studies using CIB to measure maternal sensitivity that did not include these two constructs in the overall composite reported much higher sample means for maternal sensitivity (e.g., Kim, Mayes, Feldman, Leckman, & Swain, 2013). It is possible that, by including different behaviors and constructs in composites and latent variables of maternal sensitivity, maternal sensitivity is conceptually distinct across studies.

Although maternal sensitivity is equally valued across cultures (Emmen, Malda, Mesman, Ekmekci, & Van IJzendoorn, 2012), cultural and socioeconomic factors may impact how sensitivity is defined and expressed in specific contexts (Bornstein, 1995). It is also possible that certain behaviors (such as elaboration in the current study sample) are less pertinent to understanding maternal sensitivity in specific cultural contexts and should not be given equal weight in measuring sensitivity in this context compared to other contexts. Likely there are other behaviors that are crucial to maternal sensitivity in certain contexts that are not yet studied, as measurement of maternal sensitivity stems from research among primarily Western, middle-class populations (Mesman & Emmen, 2013; Mesman et al., 2018). It is equally important in any case to understand the distinct components that make up sensitive caregiving responses in diverse socioeconomic and cultural contexts.

We also expected that mothers would vary in their rates of change in both global and individual indices of sensitivity. More specifically, we anticipated that some trajectories would increase, whereas others would decrease or remain stable across the first year of life. Global sensitivity increased slightly from 3 to 12 months. With regard to individual indices, most followed a similar positive trend as the global construct, although touch and acknowledging decreased. Although levels of acknowledging were higher than the global index of sensitivity at 3 months, acknowledging was actually lower than the global index at 12 months. The instability of acknowledging supports the hypothesis that, although global maternal sensitivity is relatively stable across the first year, its component behaviors vary in level and change across time. In other words, maternal sensitivity is distinct at 3 and 12 months, as sensitive mothers adapt their behavioral responses to meet the changing needs of their infant. Feldman (2012) has argued that certain behaviors—such as acknowledging—are fundamental components of maternal sensitivity regardless of developmental stage, whereas the importance of other behaviors may depend on developmental stage (e.g., touch). Nonetheless, behaviors such as acknowledging that are foundational and meaningful to sensitivity may still vary in their absolute amount across different stages of development. It may be that absolute levels of acknowledging decrease because other sensitive behaviors (e.g., supportive presence, vocal appropriateness) are increasing at a greater rate, and it is these aspects of maternal sensitivity that are more salient at 12 months.

Prediction of Dyadic Reciprocity and Emotion Competence from Sensitivity Trajectories

We anticipated that the global index of maternal sensitivity would be linked to both emotion competence and dyadic reciprocity at 12 months, but that differences in prediction among the individual sensitive behavior trajectories would emerge. Higher initial levels of maternal global sensitivity at 3 months and a higher rate of change in sensitivity were both associated with higher dyadic reciprocity at 12 months. With respect to the individual sensitivity behaviors, associations among dyadic reciprocity and initial levels and rates of change for positive affect, appropriate range of affect, resourcefulness, and supportive presence followed similar positive trends as the global index of maternal sensitivity. Consistent with our findings regarding positive affect, Cohn and Tronick (1987) reported that maternal positive affect elicited infant attention during dyadic interactions at ages 6 and 9 months, but not 3 months. It is likely that higher rates of change are associated with the accumulation of more high-quality experiences (i.e., more interactions characterized by positive affect) across the infancy period, which may be more meaningful for dyadic reciprocity. Infants are rapidly becoming more competent social partners as they approach their first birthday, and mothers who are able to become more sensitive (e.g., recognize cues, respond appropriately) as their infant develops are likewise able to become better social partners for their infants.

In contrast to dyadic reciprocity, only mother’s initial sensitivity levels were linked to infant emotion competence. Similarly, only the initial levels for positive affect and appropriate range of affect mattered for infant’s emotion competence. Mother’s expression of emotion plays a crucial role in the development of emotion competence by providing infants with a sense of their own emotion states, helping them understand emotions, and by communicating social rules about emotion expression (Denham, 1989). Mothers who are initially high in expression of emotion are likely to continue to be high in positivity and use a full range of emotion expressions appropriate to the situation to support their infant’s emotion learning.

Consistency of style only contributed to dyadic reciprocity as part of the global maternal sensitivity index, not as an individual construct, and it was entirely unrelated to emotion competence. Broadly, Feldman (2012) conceptualized maternal sensitivity as a higher-order construct composed of components that capture the specific behaviors (i.e., content) associated with sensitive caregiving and the quality of those behaviors (i.e., form). Content refers to the specific relational behaviors associated with sensitive caregiving (e.g., positive affect, vocalizations), whereas form characterizes more fundamental, organizational aspects of sensitivity, such as synchronicity. Feldman proposed that it is likely the content of maternal sensitivity (i.e., the specific component behaviors) rather than the form (i.e., the organizational features of the construct) that changes across development. Consistency of style may constitute a form component and therefore represent a more organizational and foundational aspect of sensitivity, whereas the other behaviors (e.g., positive affect) described fall better under the content category. Supportive presence and resourcefulness could also be construed as form behaviors, as they characterize structural components of the quality of sensitivity. It may be that content behaviors are more proximally important than form behaviors for emotion competence, whereas both form and content behaviors matter for dyadic reciprocity.

Although the global and individual indices of maternal sensitivity operate similarly in predicting dyadic reciprocity, the individual component indices of sensitivity still differed in how they predicted the quality of dyadic interactions. In other words, the strength of the associations differed among individual indices and the global construct of maternal sensitivity. In contrast, the global index of sensitivity was unrelated to emotion competence at 12 months, but component behaviors (i.e., positive affect) of sensitivity emerged as significant predictors. These differences suggest that, even though the global index of sensitivity is a composite composed of the component behaviors individually examined, the individual indices of sensitivity contribute uniquely to understanding developmental competencies. The findings from the current study provide evidence that maternal sensitivity does not support development in a systematic way, with certain aspects of sensitivity mattering more for specific competencies.

Limitations

Despite its many strengths, this study was not without limitations. The linear growth models did not fit all sensitivity behaviors equally well. Indeed, we determined fit based on SEM fit indices, which may not be equally sensitive to all sources of misfit for latent growth curve analysis (Wu, West, & Taylor, 2009; Wu & West, 2010) and should be interpreted with caution. It is possible that some models estimated in the current study (e.g., enthusiasm, vocal appropriateness) that did not meet benchmarks for fit indices were still appropriate models for the data. It is also possible that some behaviors (e.g., acknowledging) might have been better modeled using nonlinear growth trajectories, as other studies have found support for nonlinear trajectories in maternal sensitivity in infancy and toddlerhood (e.g., Feldman & Eidelman, 2009). Although the longitudinal design was a strength of the current study, the ability to examine nonlinear trajectories of global and individual indices of maternal sensitivity was limited by the availability of only four time points. Additional time points, particularly at 9 months, would contribute to capturing more complex, nonlinear developmental trajectories.

Another limitation was the low reliability of the Emotion Competence Scale (α = .53). Although there is limited psychometric work evidencing the use of the BITSEA with low-income, Hispanic populations, the work that has been done reported similarly low reliability yet provided evidence for internal consistency, test-retest reliability, discriminative validity, and predictive validity in the BITSEA (Hungerford et al., 2015). Even in a sample of primarily European American, middle class infants and toddlers, internal consistency for emotion competence using the BITSEA is expected to be lower because many of the items may not necessarily co-occur (Briggs-Gowan et al., 2004).

Future Directions

There are several directions future studies should pursue. First, future studies should model individual trajectories for different caregiving behaviors beyond the first year of life. It is necessary to understand the full developmental course of individual sensitive behaviors to understand when specific behaviors may become more or less important for development. Second, future studies should examine trajectories of individual sensitivity behaviors in relation to infant factors (e.g., behavioral cues, temperament) to better explain why some behaviors increase or decrease over time. Certain infant characteristics, such as irritability, have been shown to differentiate the growth trajectories of maternal caregiving behaviors, such as visual contact, soothing, and responsiveness (Van den Boom & Hoeksma, 1994). There is also a need to connect change in maternal caregiving behavior to other sociodemographic and maternal characteristics or factors. For example, low maternal depression has been connected to higher maternal sensitivity that increases across childhood, whereas mothers experiencing higher rates of depression tend to exhibit lower levels of maternal sensitivity and decreases across time (Campbell, Matestic, Von Stauffenberg, Mohan, & Kirchner, 2007). Third, future studies should examine trajectories of specific sensitivity behaviors in multiple cultural contexts to identify which forms and levels of behaviors are more salient in specific ethnic groups across development. It is equally important to understand how different environmental conditions (e.g., poverty) may intersect with culture to affect the expression of sensitive caregiving behavior. It is likely that there are important developmental changes in sensitive maternal caregiving behavior that occur across various interactional contexts. The current study examined maternal behaviors in the context of unstructured free play tasks, but it is possible that the developmental trajectories of these behaviors may vary in other contexts (e.g., teaching, discipline). Studies such as these would go far to informing strength-based prevention and early intervention programs that target parenting behavior in multiple interactional, cultural, and socioeconomic contexts as a mechanism of change.

IMPLICATIONS FOR THEORY

The current study is the first to use person-centered approaches to concurrently test differences among trajectories of maternal sensitivity and trajectories of the component sensitive behaviors across the infancy period. Other studies (e.g. Ferber et al., 2008) that have previously examined developmental change in individual maternal behaviors across time (e.g., touch) were limited by cross-sectional designs and cannot account for intra-individual differences across time. The use of longitudinal growth modeling in the current study allowed us to examine mothers’ individual, developmental trajectories of multiple sensitive behaviors relative to one another. It is necessary to concurrently evaluate trajectories of sensitive behaviors to understand the developmental timing of how change in one behavior may be related to another (i.e., as one behavior increases, another decreases).

The results from this study also shed light on how maternal sensitivity may impact children’s development by connecting specific behaviors that compose the broader construct to different outcomes. Although maternal sensitivity is consistently identified as an important early caregiving experience for children’s optimal development, less is known about how specific sensitive behaviors (e.g., supportive presence versus positive affect) may matter for children’s development. The findings from the current study demonstrate not only that different sensitivity behaviors matter for different outcomes, but there were also differences in the influence of initial levels compared to rate of change of sensitivity behaviors in relation to the quality of parent-child interactions and infant socioemotional competence. Because the component parts of sensitivity do not equally contribute to different dyadic and child outcomes, it is critical to consider which specific aspects of sensitivity may be important for selected outcomes in future studies, and this focus should be reflected in the operationalization of sensitivity.

Finally, the findings from the current study add to a growing body of literature that seeks to explicate the role of culture in maternal sensitivity. Although some (e.g. Mesman et al. 2018; Mesman, 2018), posit that maternal sensitivity is ubiquitous, and the characteristics, levels, and forms of responsivity may vary across cultures; others argue that sensitive caregiving can be conceptually distinct across cultural contexts (Tamis-LeMonda & Baumwell, 2011; Keller et al., 2018). Given that culture influences notions of optimal parenting, parental expectations, and goals for the socialization of children (Feldman & Masalha, 2007; Tamis-LeMonda, Wang, Koutsouvanou, & Albright, 2002), there is often a substantial lack of consideration for cultural differentiation in measurement of underlying constructs. Indeed, definitions of sensitivity and the value of maternal sensitivity are considered to be bound by culture (Crittenden & Claussen, 2000), and several studies have identified cultural differences in the specific behaviors that compose maternal sensitivity in mother-infant interactions (Kärtner, Keller, & Yovsi, 2010; Keller, Otto, Lamm, Yovsi, & Kärtner, 2008). Furthermore, parenting behaviors are thought to have differential effects on children from diverse ethnic backgrounds (e.g., Deater-Deckard & Dodge, 1997), suggesting that cultural context should be considered when evaluating the impact of maternal behaviors on child and dyadic outcomes. Examining trajectories of maternal caregiving behavior in multiple sociodemographic populations will lend deeper understanding of which caregiving behaviors are universal versus culturally bound and whether these cultural differences in sensitive caregiving are meaningful for children’s developmental outcomes.

In sum, the current study contributes to a deeper, more nuanced understanding of the maternal sensitivity construct. The developmental trajectories of individual sensitivity behaviors were found to differ from the global index of sensitivity across the first year of infant life, and these differences were meaningful for infant’s emotion competence and dyadic reciprocity when infants were 12 months old. The findings from the current study underscore the need for future studies to explicate which specific aspects of maternal sensitivity differentially predict to a variety of later child and family competencies, particularly in variety of cultural and socioeconomic contexts.