The multi-group model (Figure 2) exhibited good fit [χ2 (df)= 612.99(482); p<.001; χ2/df ratio = 1.27; CFI= .94; RMSEA=.04]. Among H/I children (Figure 2A), F1 and F2 neuropsychological functioning were associated with F2 and F3 ADHD severity (β=−.22 and −.23 respectively; p<.05), respectively. Elevated ADHD severity at F2 was associated with subsequently worse neuropsychological functioning at F3 (β= −.16; p<.05). Lower BL SES was significantly associated with poorer neuropsychological functioning and with more severe ADHD severity at F1. All paths leading-up to the time-point explained 7% (F1), 62% (F2), and 83% (F3) of the variance in neuropsychological functioning. At F2 and F3, this included the variance explained by neuropsychological functioning at F1 and F2 respectively. Similarly, paths leading up to ADHD severity explained 4% (F1), 37% (F2), and 42% (F3) of the variance in ADHD severity. At F2 and F3, this included the variance explained by ADHD severity at F1 and F2 respectively. Among TD children (Figure 2B), the only cross-lagged path that attained significance was that greater ADHD severity at F2 was associated with subsequently worse neuropsychological functioning at F3 (β= −.29; p<.01).
