Several developmental assessments have been widely studied in the ASD population and found to be psychometrically sound ways of measuring skills of preschoolers with autism. Construct, convergent and divergent validity of the Mullen was similar in ASD and nonspectrum samples (Swineford, Guthrie, & Thurm, 2015). The Bayley-III has also been identified as a useful developmental assessment to evaluate young children with ASD (Long, Gurka, & Blackman, 2011; Torras-Mana et al., 2016), as convergent validity of the Bayley-III with cognitive assessment measures has been found for those with ASD (Torras-Mana et al., 2016). The PEP-3 has also been found to be a useful indicator of cognitive and language functioning in young children with ASD (De Giacomo et al., 2016), as well as of adaptive functioning (Fu et al., 2012). However, Fulton and D’Entremont (2013) warned that PEP-3 scores varied from other developmental measures’ scores in their study, underscoring the need to include multiple developmental measures to obtain the most accurate picture of an individual’s abilities.
Moreover, some developmental tests are predictive of later cognitive test scores for preschoolers with ASD. For example, Dempsey et al. (2018) found that the Cognitive domain of the M-P-R showed a moderate but positive correlation with later WISC-IV scores. Mullen and DAS-II scores have also been found to be highly correlated, which suggests they measure a similar construct; however, mean differences in scores varied depending on the child’s overall cognitive ability (Farmer, Golden, & Thurm, 2015).
Long and colleagues (2014) found that language skills of preschoolers with ASD were significantly lower than their typically developing peers’, but fewer differences were found between overall cognitive skills of children with and without ASD, as measured on the Bayley-III. Similarly, Dempsey et al. (2018) found that preschoolers with ASD had higher general cognitive scores than receptive language scores on the M-P-R.
Developmental assessments are not designed to measure symptoms of ASD, but ASD symptoms can and do affect performance on developmental instruments. For example, higher autism severity scores were associated with total scores of the BDI-2, as well as predicted scores on all BDI-2 domains (Goldin, Matson, Beighley, & Jang, 2014). The ASQ-3, particularly scores on the Communication Domain, can also alert clinicians to the need for autism-specific follow-up evaluation (Hardy, Haisley, Manning, & Fein, 2015). However, Williams and colleagues (2018) found that the ASQ-3 used in conjunction with the ASQ:SE is more likely to accurately identify subtle early concerns that may indicate ASD.
Change-over-time among young children with ASD has also been investigated with developmental assessments. In a 2014 study, Williams and colleagues found that participants with lower overall developmental scores demonstrated greater gains in their BDI-2 scores than those with average scores at baseline. The authors hypothesized that some of the children who initially showed low scores exhibited regression toward the mean upon subsequent testing; alternately children with greater deficits at the outset may have received more intense and/or variety of services, thus affecting DQ. Regardless of the explanation for this observed effect, there is wide variability in development among young children and clinicians should be mindful of the various reasons that developmental scores may change over time. This is also applicable to comparing child performance on developmental tests during the preschool years to their performance on cognitive tests in the school-age years. For example, even though Mullen and DAS-II scores are highly correlated, DAS-II scores were consistently higher in a study conducted by Farmer, Golden, and Thurm (2015); the researchers underscored that initial evaluations using the Mullen with follow-ups on the DAS-II may produce score “changes” that give the false impression of considerable improvement over time. Use of PEP-3 raw scores and developmental ages may be quite useful for tracking change-over-time, including responsiveness to intervention or treatment effects (Chen et al., 2011).
Some research has suggested that developmental assessments may indicate the need for ASD evaluation in children, or serve as a kind of screening instrument. In a large sample (n = 1,668) of 17-36 month-old children, Sipes and colleagues (2011) identified BDI-2 cut-off scores that optimized sensitivity and specificity if this instrument was used for ASD screening purposes. One study with the M-P-R concluded that this test was able to identify developmental delay but not differentiate autism specifically (Peters, 2014).
As in other domains of evaluation, multifaceted data collection is important in developmental assessment. Ulvand (2014) concluded that direct assessment with the Mullen, as well as indirect assessment through both parent and teacher reports, yielded valid measurement of the language functioning of preschoolers with autism.