TABLE 2

Child Outcome Measures and Linear Mixed Model Results

Child Outcome MeasuresLinear Mixed Model Results
Individual-ESIGroup-ESITime (Across Condition)Time × Condition
MSEΔ95% CIMSEΔ95% CIpHedges’s gCLESpHedges’s gCLES
Communication and Symbolic Behavior Scales
 Social Composite59.43.822.315.5 to 29.151.93.912.45.5 to 19.3<.0010.760.70.040.480.63
 Speech Compositea1.40.10.70.5 to 0.91.30.10.80.6 to 1.0<.0011.270.82.810.050.51
 Speech Compositeb42.13.833.724.9 to 42.634.04.127.518.4 to 36.6
 Symbolic Composite52.63.628.921.5 to 36.348.63.827.019.4 to 34.6<.0011.280.82.720.130.54
Autism Diagnostic Observation Schedule
 Social Affect10.90.7−2.6−3.8 to −1.412.30.7−2.1−3.4 to −0.9<.001−0.510.36.610.150.54
 Restricted, Repetitive Behavior3.40.30.4−0.2 to 1.03.60.30.70.1 to 1.4.010.380.61.430.170.55
Vineland Adaptive Behavior Scales
 Communication90.22.211.87.7 to 15.982.92.33.1−1.0 to 7.3<.0010.500.64.0040.690.69
 Daily Living91.01.74.20.2 to 8.185.01.8−2.5−6.5 to 1.5.560.070.52.020.580.66
 Socialization84.11.5−0.9−4.1 to 2.581.21.5−5.6−9.0 to −2.3.008−0.310.41.040.660.68
 Motor88.81.6−5.5−8.8 to −2.287.61.6−8.4−11.8 to −5.0<.001−0.690.31.220.340.59
 Adaptive Behavior Composite86.21.62.2−1.0 to 5.581.71.6−3.3−6.7 to 0.0.65−0.140.46.020.720.69
Mullen Scales of Early Learning
 Visual Reception42.12.2−0.1−4.2 to 4.039.22.2−1.2−5.4 to 2.9.65−0.040.49.700.030.51
 Fine Motor37.22.0−9.0−13.1 to −5.037.72.1−4.7−8.8 to −0.7<.001−0.570.34.140.410.61
 Receptive Languagea1.60.00.10.1 to 0.21.50.00.00.0 to 0.1<.0010.410.61.0080.580.66
 Receptive Languageb39.52.210.16.2 to 14.134.12.32.8−1.2 to 6.8
 Expressive Languagea1.60.00.10.1 to 0.21.50.00.10.0 to 0.1<.0010.540.65.610.180.55
 Expressive Languageb39.02.29.04.7 to 13.336.22.27.53.3 to 11.8
 Early Learning Composite81.73.35.90.2 to 11.676.23.42.1−3.7 to −7.9.060.200.56.360.130.54
  • F values for linear mixed model results can be found in the Supplemental Table 3. Estimated marginal means are presented. CI, confidence interval; CLES, Common Language Effect Size; Δ, mean change from baseline to end of intervention; CLES, which is derived from the g effect, represents the likelihood that a randomly selected child from individual-ESI will show greater improvement than a randomly selected child from group-ESI. —, not applicable.

  • a Estimated marginal means, SEs, and linear mixed model results using log transformed variables.

  • b Estimated marginal means and SEs presented using nontransformed variables.