Table 4.

Description of Planned Analyses^a

Aims	Description	Planned Analyses	Post Hoc Analyses
Aim 1: To evaluate a power mobility intervention to promote developmental, activity, and participation outcomes of young children aged 12–36 mo who have CP	Descriptive statistics will be calculated for developmental, activity, and participation outcomes Data will be assessed for normality and symmetry. If normality assumptions are met, we will use parametric statistics to examine between- and within-subject effects. If assumptions are not met, we will use the nonparametric equivalent of the planned analyses, including Friedman test for ANOVA and multiple Wilcoxon tests for MANOVAs	Separate 2 (group: 1, 2) × 3 (time: T0, T1, T2) repeated measures ANOVAs will be calculated to determine between- and within-subject changes on mean scores of developmental outcomes (each Bayley subscale) A 2 (group: 1, 2) × 3 (time: T0, T1, T2) repeated measures MANOVA will be calculated to determine between- and within-subject changes on mean scores activity and participation outcomes (YC-PEM, CEDL) Stratification into low- and high-use groups will occur based on mean device use data We will use these stratified groups to run our analyses to determine if device use is related to developmental, activity, and participation outcomes. Separate 2 (group: low use, high use) × 3 (time: T0, T1, T2) repeated measures MANOVAs will be calculated to determine between- and within-subject changes on mean scores of developmental, activity, and participation outcomes Visual analysis will be used to determine if clinically significant differences exist	Bonferroni correction will be used for the planned analyses to control for the type I error rate and determine where the significant differences exist
Aim 2: To compare the use patterns (frequency, duration, environment) of 2 different power mobility options	Descriptive statistics will be calculated for automated device use tracking data, including 95% CIs, for number of driving sessions, duration (min) of driving sessions, and total duration (min) of driving sessions, stratified by device type Variable distributions will be examined to: (1) identify appropriate response variable distributions, (2) screen for outliers, and (3) characterize patterns of data missingness. If assumptions for parametric statistics are met, we will use parametric statistics to examine within-subject effects. If assumptions for parametric statistics are not met, we will use the nonparametric equivalent of the planned analyses, including multiple Wilcoxon tests for MANOVAs Descriptive statistics will be calculated for caregiver diaries and automated device use tracking data to describe home and community use trends for both devices (mean distance traveled, speed, number of mobility bouts, number of controller activations, etc), stratified by device type	A 2 (group: 1, 2) × 2 (time: T1, T2) repeated measures MANOVA will be calculated to determine within-subject changes on the dependent measures of the mean number of driving sessions, mean duration (min) of driving sessions, and summed total duration (min) of device use	Bonferroni correction will be used for the planned analyses to control for the type I error rate and determine where the significant differences exist

^a ANOVA = analysis of variance; Bayley = Bayley Scales of Infant Development; CEDL = Child Engagement in Daily Life; CP = cerebral palsy; MANOVA = multivariate analysis of variance; YC-PEM = Young Children’s Participation & Environment Measure.