Links Between Cumulative Risk Factors and Child Temperament in Early School Age Children

Document Type


Publication Date



Developmental scientists have become increasingly interested in the relationship between cumulative demographic risk and developmental outcomes. Risk has been defined as “a process that serves to increase the chances of experiencing a negative outcome in one or several domains of functioning…” (Popp, Spinrad, & Smith, 2008). Cumulative risk models are often preferred over single risk models because individual risk factors such as poverty and single parenthood are so highly correlated. Although researchers have demonstrated strong associations between cumulative risk and a variety of child outcomes, to our knowledge only Popp et al. have investigated links with child temperament, with a specific focus on infancy. In the present study we investigate links between cumulative and single risk indices and child temperament in 4- to 6-year-olds. Unlike other studies, we also consider rural status as an possible risk indicator.

Data were collected in two types of setting: a university-affiliated child-care facility (N = 33, about 52% girls) and a group of rural, county-funded preschools and kindergartens (N = 21, about 62% girls). Mean age across the two samples was 4.57 years (SD = 1.11 years). A cumulative risk index was created by summing across eight risk indicators based on 1) income, 2) marital status, 3) ethnicity/race, 4) family size, 5) maternal education, 6) maternal age at birth, 7) maternal occupational status, and 8) rurality status. Risk factors were dichotomized (1 vs 0) based on whether the family met a specific risk criterion (Table 1). Temperament was measured via mother report using the Child Behavior Questionnaire, which produced three overarching temperament scores: surgency, negative affectivity, and effortful control.

In terms of cumulative risk scores, 16 (30%) of the mothers had zero risk indicators, 14 (26%) had one, 8 (15%) had two, 9 (17%) had three, 4 (8%) had four, and 2 (4%) had five. No cumulative risk score exceeded five. Mean cumulative risk was 1.64 (SD = 1.51). As shown in Table 2, greater cumulative risk was associated with higher scores on surgency and negative affectivity but not effortful control. The most strongly associated individual risk factors were household income and rurality status, which were also strongly related to one other [r(53) = .61, p = .000]. Regression analyses revealed that rurality accounted for unique variance over and above income in both surgency (R2 = .20, p = .000) and negative affect (R2 = .42, p = .000), but not vice versa.

These results support the contention that cumulative demographic risk is linked to at least two superdimensions of temperament in early school age, wherein a driving factor appears to be a child’s rurality status. Moreover, the valence of these associations is consistent with the notion that greater demographic risk may lead to negative temperament outcomes. Both negative affectivity and surgency (at least to the extent that surgency indexes activity level and impulsive behavior) are characteristics that many would regard as contributing to temperamental difficulty. This link is notable because many researchers regard temperamental difficulty as a risk indicator for negative developmental outcomes in its own right.

Fifty-six children (26 boys) visited the lab at M = 18.3 months (SD = 0.43 months). The Early Childhood Behavior Questionnaire (ECBQ; Putnam et al., 2006) superdimension of effortful control was used as a surrogate measure of early executive function. To measure child activity level, we used the mother-reported activity level subdimension from the ECBQ, and also coded mother-child free play periods to quantify children’s predilection to use physical activity in the service of social or cognitive objectives, such as grasping a spoon and extending the arm outwards to feed a baby doll, which we termed sociocognitive activity. To measure sociocognitive activity we used a modified version of Tamis-LeMonda and Bornstein’s (1990) play competence scale wherein each instance of sociocognitive activity was noted and summed for a total score of sociocognitive activity level (See Table 1). Finally, to gauge maternal encouragement, a modified version of the Dyadic Parent Child Interaction Coding System (DPICS; Eyeberg, Nelson, Duke, & Boggs, 2005) was used to identify maternal commands, praise, questions, physical involvement, talking, touching, and scaffolding behaviors during mother-child free play sessions.

Zero-order correlations revealed a significant negative relationship between mother-reported activity level and child executive function (r = -.42, p < 0.01), replicating previous findings. However, correlations between sociocognitive activity and executive function, while positive, was not significant. We conducted moderation analyses separately for each maternal encouragement variable, and found that a higher amount of maternal questioning during play corresponded to a positive association between sociocognitive activity and executive functioning (moderator = 1.00, p < 0.05). These findings partially support our hypotheses and suggest that the ways in which caregivers direct and train activity during play through questioning strategies may also direct and train cognitive functioning. However, further research is needed to support these claims. These results also point toward issues with the measurement of activity level, as our two measures of activity produced significantly different correlations with executive functioning (z = -3.4, p < 0.01). Future research in the area of motor development as it pertains to cognitive functioning should investigate and develop a standard measure of motor activity that is capable of capturing not only simple milestone achievement and intensity levels, but also the amount of sociocognitive engagement during physical activity.


Baltimore, MD

This document is currently not available here.