3.1. COMPARISON Results
Table 1 shows descriptive results of the sample from the COMPARISON study. Among the intervention group, 55.3% were female and 69.9% were Hispanic/Latino. In the comparison group, 59.8% were female and 72.8% were Hispanic/Latino. Of note, the percent of children in the intervention group that were in the obese or overweight percentile category (>85th percentile) was 41.5%. Among the comparison group, 47.6% of children had a BMI percentile in the obese or overweight category.
PMA 12-Skills. The repeated measures ANOVA showed that after controlling for age, sex, race/ethnicity and BMI z-score, participation in the intervention was statistically significant between the intervention and comparison groups (F = 83.451, df = 1,
p < 0.001,
= 0.555) with regards to the 12-skill composite scores. The interaction of time and treatment group confirmed that increases in skills was not due to maturation alone (F = 20.662,
df = 1,
p < 0.001,
= 0.236). Overall, children in the intervention group had a PMA composite score mean change of 4.029 compared to 1.029 in the comparison group. For more detailed analyses, we considered specifically examining the impact on those children who were not competent or able to reach the pre-Kindergarten standard at pre-test. As such, children in the intervention group had a PMA composite score mean change of 7.225 and significant change in ten skills, and those in the comparison group had a mean change of 1.892 and significant change in only six skills pre and post-test (
Table 2).
Balance (timed). With regards to balance, we first considered all of the children in the COMPARISON study. The children in the comparison group did not have a statistically significant change in balance, while the children in the intervention group did increase significantly from a mean of 5.32 s pre-intervention to 6.42 s post-test (
p = 0.028) (
Table 3). Again, we wanted to examine the impact on children who were not able to meet the state pre-K standard of balancing for 5 s at pre-test. The number of children who switched from balancing less than 5 s (did not meet the standard) to greater than 5 s (met the standard), did not increase significantly in the comparison group. However, the number of children did increase significantly in the intervention group (
p = 0.007) (
Table 3).
Boxing (crossing the midline). With regards to the entire COMPARISON study, the intervention group had a greater percentage of those who were competent in crossing the midline post-intervention than the comparison group (69%, 31%, respectively). With regards to children who were unable to complete the skill at pretest, crossing the midline improved significantly for both the intervention (pre-test 70 children to post-test 21; p < 0.001) and comparison groups (pre-test 105 children to post-test 53; p < 0.001).
Hops (number in 15 s). The number of hops on one foot in 15 s increased significantly from a mean of 24.25 (SD = 11.65) hops pre intervention to 28.49 (SD = 10.53) hops post-test in the intervention group (t = −3.545, df = 112, p = 0.001). The mean number of hops did not significantly increase in the comparison group (p = 0.064).
3.2. SCALE-UP Results
Table 4 shows descriptive results of the sample from the SCALE-UP study. The mean age of the children was 4.16 years, and 51.1% were female. The majority was Hispanic/Latino (54.1%). The mean BMI percentile was the 66th percentile and 34.7% were obese or overweight, which is slightly higher than Nevada’s obese/overweight preschool population (32%) [
41]. In the SCALE-UP results, we consider first pre-intervention differences only. Then we consider post-intervention differences. In our post hoc analyses we use Wald tests to compare 12-skill, boxing, balance and hops scores pre- and post-intervention both with and without controlling for age. Lastly, Wald tests were run between boys and girls comparing the difference in the four scores from pre- to post-intervention. A Bonferroni adjustment was applied to maintain a familywise Type I error rate of α
fw = 0.05. With this adjustment, the slope (hierarchical linear model) or chi-square (Wald test) was viewed as statistically significant only if the
p-value was less than α
adj = 0.0125.
3.2.1. Pre-Intervention Differences
Pre-intervention scores were investigated whereby children were considered nested within one of the 28 ECE sites. For each task (sum 12-skills score, boxing score, number of hops and balance time), the unconditional model (intercept-only) was run to determine the proportion of variance explained by site alone. Then demographics (sex, age, ethnicity, and BMI z-scores) were added to the model.
For the sum 12-skills score, the ICC showed that 9.56% (p = 0.010) of the total variation was accounted for by site. When the demographics were added to the model, age was significant (b = 1.269, se = 0.201, p < 0.001), ethnicity was significant (b = −1.442, se = 0.309, p < 0.001), and the ICC decreased to 4.94% (p = 0.027). These slopes imply that for each year increase in age, children complete more than one additional skill on average, and Hispanic children on average complete nearly one and a half fewer skills.
For the boxing score, the ICC showed that 1.96% (p = 0.246) of the total variation was accounted for by the site. When demographics were added to the model, age was significant (b = 0.067, se = 0.015, p < 0.001), sex was significant (b = 0.065, se = 0.017, p < 0.001), and the ICC decreased to 1.05% (p = 0.367). These slopes confirm that older children and boys perform significantly higher on boxing prior to the intervention.
For the number of hops, the ICC showed that 6.06% (p = 0.014) of the total variation was accounted for by site. When the demographics were added to the model, sex was not significant (b = 1.370, se = 0.586, p = 0.020), age was significant (b = 4.933, se = 0.690, p < 0.001), BMI was significant (b = −0.692, se = 0.260, p = 0.008), and the ICC decreased to 6% (p = 0.05). Overall, these slopes show that boys on average complete more hops, and children with higher BMI complete fewer hops on average. Interestingly, for each year older, children on average complete approximately five more hops.
For the timed balance, the ICC showed that 4.52% (p = 0.085) of the total variation was accounted for by site. When demographics were added to the model, age was significant (b = 1.974, se = 0.342, p < 0.001), ethnicity was significant (b = −1.113, se = 0.314, p < 0.001), BMI was not significant (b = −0.14, se = 0.061, p = 0.022) and the ICC decreased to 0.84% (p = 0.140). These slopes show that on average for each year older, children balance for nearly two more seconds. Conversely, Hispanic children and children with a higher BMI present on average balance for fewer seconds.
3.2.2. Post-Intervention Differences
Post-intervention scores were investigated whereby children were considered nested within one of the 28 ECE sites. For each task (sum 12-skills score, boxing score, number of hops and balance time), the unconditional model (intercept-only) was run to determine the proportion of variance explained by site alone. Then demographics (sex, age, ethnicity and BMI z-score) and the pre-intervention score were added to the model.
For the sum 12-skills score, the ICC showed 16.52% (p = 0.007) of the total variation was accounted for by site. When the demographics and the pre-12-skills score were added to the model, the pre-12-skills sum score was significant (b = 0.517, se = 0.026, p < 0.001), age was significant (b = 0.608, se = 0.117, p < 0.001), ethnicity was significant (b = −0.758, se = 0.161, p < 0.001) and the ICC decreased to 5.38% (p = 0.013). These slopes show that on average a child gained half of a skill by completing the intervention. In addition, older children continued to score higher on average, whereas Hispanic children continued to score lower on average.
For the post-intervention boxing score, the ICC showed that 3.74% (p = 0.209) of the total variation was accounted for by site. When demographics and the pre-boxing score were added to the model, the pre-boxing score was significant (b = 0.425, se = 0.037, p < 0.001), age was significant (b = 0.155, se = 0.027, p < 0.001), sex was significant (b = 0.053, se = 0.020, p = 0.010), ethnicity was not significant (b = −0.051, se = 0.022, p = 0.021), and the ICC decreased to 1.10% (p = 0.242). These slopes show that on average children increased their boxing score by nearly half a point. However, on average, older children, boys, and non-Hispanic children continued to score significantly higher in boxing.
For the number of hops, the ICC showed that 5.73% (p = 0.032) of the total variation was accounted for by site. When demographics and the pre-hops score were added to the model, the pre-hop score was significant (b = 0.454, se = 0.021, p < 0.001), age was significant (b = 2.208, se = 0.618, p < 0.001), sex was significant (b = 1.528, se = 0.317, p < 0.001), and the ICC decreased to 1.55% (p = 0.101). These slopes confirm that on average children score nearly half a hop more after the intervention. However, older children and boys continue to complete significantly more hops than younger children and girls, respectively.
For the timed balance, the ICC showed that 2.21% of the total variation was accounted for by site. When demographics and the pre-balance time were added to the model, the pre-balance time was significant (b = 0.473, se = 0.036, p < 0.001), age was significant (b = 2.084, se = 0.487, p < 0.001), and the ICC decreased to 2.08% (p =.131). These slopes show that on average children significantly increased their balance time with the intervention; however, older children on average continued to balance longer than younger children.
3.2.3. Post hoc Wald Test Comparisons
Finally, Wald tests were run by simply comparing the means of the pre-intervention and post-intervention scores for each task.
Table 5 summarizes the results. On average, scores for all four tasks significantly increased after the intervention: sum 12-skills score (
= 1520.39,
df = 1,
p < 0.001), boxing (
= 611.95,
df = 1,
p < 0.001), number of hops (
= 323.58,
df = 1,
p < 0.001) and time balancing (
= 129.23,
df = 1,
p < 0.001).
However, since age was consistently significant for all tasks, Wald tests were run with age as a covariate.
Table 6 shows the results when controlling for age. When controlling for age pre-intervention and post-intervention scores were significantly different for the sum 12-skills score (
= 31.02,
df = 1,
p < 0.001), boxing (
= 11.64,
df = 1,
p = 0.001) and the number of hops (
= 12.21,
df = 1,
p = 0.001).
Lastly, Wald tests were run between boys and girls comparing the difference in scores from pre- to post-intervention. No significant differences were detected for any task. Consequently, the intervention does not seem to benefit boys or girls more.