Associations between Children’s Intelligence and Academic Achievement: The Role of Sleep

Summary

Sleep problems (long wake episodes, low sleep efficiency) were examined as moderators of the relation between children’s intelligence and academic achievement. The sample was comprised of 280 children (55% boys; 63% European Americans, 37% African Americans; M age = 10.40 years, SD = .65). Sleep was assessed through seven consecutive nights of actigraphy. Children’s performance on standardized tests of intelligence (Brief Intellectual Ability index of the Woodcock-Johnson III) and academic achievement (Alabama Reading and Math Test) were obtained. Age, sex, ethnicity, income-to-needs ratio, single parent status, zBMI, chronic illness, and pubertal development were controlled in analyses. Higher intelligence was strongly associated with higher academic achievement across a wide range of sleep quality. However, the association between intelligence and academic achievement was slightly attenuated among children with more long wake episodes or lower sleep efficiency compared to children with higher-quality sleep.

Introduction

Intelligence is a construct generally associated with the capacity to learn. Academic achievement is knowledge acquired through instruction. Intelligence is the best individual predictor of academic achievement (Deary, Strand, Smith, & Fernandes, 2007). However, intelligence is not the sole determinant of achievement, and children may over- or underachieve relative to their estimated ability (e.g., depending on sustained attention; Steinmayr, Ziegler, & Trauble, 2010). We considered sleep problems as a possible source of academic underachievement relative to intelligence. We examined whether the strength of the association between intelligence and academic achievement depends on the quality of children’s sleep.

Sleep supports neurocognitive functioning (Dewald, Meijer, Oort, Kerkhof, & Bogels, 2010). Poor sleep may undermine neural development as well as day-to-day emotion regulation and executive functions including processing speed and working memory, and may thereby affect academic achievement (Dewald et al., 2010). Sleep problems may also prevent consolidation of information learned at school into long-term memory (Wilhelm, Prehn-Kristensen, & Born, 2012).

The present study constitutes the first examination of sleep problems as moderators of the association between intelligence and academic achievement. Two patterns of moderation effects seemed plausible. Sleep problems might exacerbate the association between lower intelligence and lower achievement, such that children with relatively low intelligence and poor sleep would have lower achievement than children with relatively low intelligence and good sleep. Alternatively, sleep problems may attenuate the association between higher intelligence and higher achievement. In this case, children with relatively low intelligence would exhibit similarly low levels of achievement regardless of their sleep, whereas children with relatively high intelligence and poor sleep would have lower achievement compared to children with relatively high intelligence and good sleep.

Method

Participants

Fourth- and fifth-graders (N = 282) from public schools in the Southeastern United States were between the ages of 8 and 10 years at recruitment with no diagnosed learning disability or sleep disorder (Auburn University Sleep Study); data collected in 2010–2011. We excluded two children without any data on the primary study variables. Thus, the sample included 280 children (55.0% boys; M age = 10.40 years, SD = .65), including 63% European- and 37% African-Americans. Income-to-needs ratio (annual family income/federal poverty threshold; U.S. Department of Commerce) ranged from .27 to 4.10 (≤ 1 = living in poverty, ≥ 3 = middle class), with 27.9% of children from single-parent households. Children were rated by mothers as prepubertal on average (M = 1.74, SD = .54; range = 1 (prepubertal) to 5 (postpubertal); Petersen et al., 1988), and 21.8% had a chronic illness (e.g., asthma; 0 = no, 1 = yes). Height and weight were assessed in our lab and used to calculate zBMI with 68% of participants considered to be of healthy weight.

Procedures

The study was approved by the university’s institutional review board; written consent and assent were obtained from parents and children, respectively. Sleep data were collected during the regular school year. At home, children wore actigraphs on their non-dominant wrist for seven consecutive nights; parents completed sleep diaries which were used to cross-validate actigraphy data. On average, children visited the laboratory 4 days (SD = 13) after the last night of actigraphy and were administered the Woodcock-Johnson III Test of Cognitive Abilities (Woodcock et al., 2001). Achievement data were obtained from schools.

Measures

Intelligence

Intelligence was assessed with the well-established Brief Intellectual Ability (BIA) index of the Woodcock-Johnson III (WJ-III; Woodcock, McGrew, & Mather, 2001). The BIA is comprised of three tests: Verbal Comprehension, Concept Formation, and Visual Matching which are indicators of crystallized intelligence, fluid and inductive reasoning, and processing speed, respectively.

Sleep

Octagonal Basic Motionlogger (Ambulatory Monitoring Inc., Ardsley, NY, USA) measured motion in 1-minute epochs using zero crossing mode. The analysis software package (Action W2, 2000; Ambulatory Monitoring Inc.) utilized the Sadeh algorithm (Sadeh et al., 1994). Thirty four children had four nights or less of valid actigraphy data and their sleep data were excluded from analyses. Two common sleep quality parameters were derived: Long wake episodes (number of wake episodes ≥ 5 minutes) and sleep efficiency (% of epochs scored as sleep between sleep onset and offset). Sleep minutes were examined in initial analyses but yielded null effects. High night-to-night stability of the sleep parameters across the week was observed (long wake episodes α = .87, sleep efficiency α = .89).

Academic achievement

The Alabama Reading and Math Test (ARMT; Montgomery, AL: Alabama State Department of Education) is a criterion-referenced test based on Alabama’s academic standards in reading and mathematics. The ARMT includes items from the Stanford Achievement Test (Stanford 10; Pearson) that match the Alabama (AL) state standards in reading and math as well as items that were developed to cover AL reading and math content standards. Performance on the ARMT ranges from Level I (does not meet academic content standards) to Level IV (exceeds academic content standards); 60% of participants in this study scored within Level IV, which is representative of the percentage of students scoring within Level IV in reading and math in the school system where the students were enrolled at the time of data collection.

Plan of Analysis

To reduce outlier effects, values of study variables > 3 SDs from the mean were recoded as the highest or lowest observed value within 3 SDs. Regression analyses were conducted in Amos (Arbuckle, 2012) and used full information maximum likelihood estimation to handle missing data. Control variables (age, sex, ethnicity, income-to-needs, single parent status, zBMI, chronic illness, and puberty) were entered, followed by intelligence and sleep, and then the interaction between intelligence and sleep. Significant interactions were plotted at high (+1 SD) and low (−1 SD) levels of intelligence and sleep.

Results

See Table 1 for descriptive statistics and correlations. Intelligence was positively correlated with achievement. Long wake episodes and sleep efficiency were negatively correlated. Regression results presented in Table 2 indicate that after accounting for the effects of the control variables (not shown), intelligence predicted higher achievement. Sleep moderated the association between intelligence and academic achievement. Simple slopes analyses revealed that higher intelligence was associated with higher achievement among children who exhibited fewer and more long wake episodes (B = 2.69, SE = .20, p < .001; B = 1.92, SE = .20, p < .001, respectively) and among children who exhibited higher and lower sleep efficiency (B = 2.55, SE = .14, p < .001; B = 2.06, SE = .14, p < .001, respectively). However, the association between intelligence and achievement was slightly weaker among children with more long wake episodes or lower sleep efficiency (see Fig. 1 for an illustrative plot).

Table 1.

Descriptive Statistics and Correlations among Primary Study Variables

	1	2	3	4	5	6	7	8	9	10	11	12
1. Age	-
2. Sex	.04	-
3. Race/Ethnicity	−.05	−.04	−
4. Income-to-needs	−.05	.10	−.39***	-
5. Single parent status	−.07	−.01	.38***	−.36***	-
6. zBMI	−.07	.02	.14*	−.13	.15*	-
7. Chronic illness	.14*	.11	.03	−.07	.09	−.04	-
8. Puberty	.26***	−.42***	.19**	−.06	.03	.23***	.04	-
9. Intelligence	−.13*	−.02	−.29**	.25***	−.19**	.04	−.17**	−.05	-
10. Long wake episodes	.00	.11	−.10	.01	.01	.11	.10	−.03	-.11	-
11. Sleep efficiency	−.04	−.12	.08	.03	.00	−.12	−.13*	.01	.15*	−.95***	-
12. Achievement scores	.18*	.06	−.14*	.18*	−.12	.05	−.15*	.03	.77***	−.03	.05	-
Mean (SD)	124.76 (7.81)	-	-	1.62 (.97)	-	.63 (1.18)	-	1.74 (.54)	98.90 (13.67)	3.41 (2.16)	88.68 (6.51)	660.16 (41.27)

Note. Age measured in months; sex (0 = girl, 1 = boy); race (0 =European American, 1 = African American); single parent status (0 = no, 1 = yes); chronic illness (0 = no, 1 = yes); zBMI = standardized body mass index.

^*p < .05.

^***p < .001.

Table 2.

Independent and Interactive Associations among Intelligence, Sleep Disruptions, and Academic Achievement

	Academic achievement
	B (SE)	β	R2
Long wake episodes model
Step 2			64.4%
Intelligence	2.31 (.14)***	.77
Long wake episodes	−.09 (.87)	−.01
Step 3			65.6%
Intelligence x Long wake episodes	−.17 (.06)**	−.13
Sleep efficiency model
Step 2			64.5%
Intelligence	2.31 (.14)***	.77
Sleep efficiency	.02 (.29)	.00
Step 3			65.0%
Intelligence x Sleep efficiency	.04 (.02)*	.09

Note. Controls were entered in Step 1 (not presented) and accounted for 11% of the variance in academic achievement. Age (B = 1.12, SE = .38, β = .21, p < .01) and chronic illness (B = −15.41, SE = 6.73, β = −.15, p < .05), but not sex, ethnicity, income-to-needs, single parent status, zBMI, or puberty, predicted achievement.

^*p < .05.

^**p < .01.

^***p < .001.

Figure 1.

Intelligence predicting achievement at lower and higher levels of long wake episodes.

Discussion

Results suggest that children with relatively high intelligence may not reach their academic achievement potential when they experience sleep problems. Higher intelligence was strongly associated with higher achievement across a wide range of sleep quality. However, this association was slightly attenuated among children with more long wake episodes or lower sleep efficiency compared to children with higher-quality sleep. Although this study provides new evidence that children with relatively high intelligence may underachieve in the context of sleep disruptions, it should be noted that interaction effects were very small in magnitude and require further study before conclusions can be drawn.