Introduction

Neurodevelopmental disorders consist of a group of disorders that occur early in key developmental periods (i.e., early childhood), impact life-long developmental processes (American Psychological Association [APA], 2013), and include autism spectrum disorder (ASD), intellectual disability (ID), and global developmental delay (GDD). ASD is characterized as deficits in social communication and interaction and the presence of restricted repetitive behaviors, and is estimated to occur in 1 in every 54 children in the United States (Manner et al., 2020). ASD is a heterogeneous disorder and autistic youth differ in symptom presentation and other related behaviors and skills such as cognitive ability and adaptive behavior (i.e., broad conceptual, social, and practical skills required to function in everyday life). For example, 31% of children with ASD are diagnosed with ID (Baio et al., 2018) and 5.3% are diagnosed with GDD (Zablotsky et al., 2015). The identification of ID requires deficits in cognitive abilities and deficits in adaptive behavior that are apparent from an early age (i.e., adaptive behavior deficits define the disorder), whereas GDD is applied to youth under the age of five who demonstrate delays in multiple cognitive domains and often have corresponding deficits in adaptive functioning (APA, 2013). Although specific diagnostic criteria for ASD, ID, and GDD differ, all diagnoses require that there be evidence of functional impairment as a result of symptom presentation; thus, adaptive behavior deficits are likely to be evident across all disorders.

Comprehensive Evaluations and Adaptive Behavior

Comprehensive evaluations of ASD, ID, and GDD require professionals to assess developmental domains using data from multiple sources to make diagnostic decisions and capture an individual’s present level of functioning (Hyman et al., 2020; Jordan et al., 2019; McDonald et al., 2016; Wilkinson, 2017). Youth with these suspected neurodevelopmental disorders (i.e., ASD, ID, GDD) may present with developmental differences across a wide range of skills. As such, best practice evaluations include comprehensive assessment across myriad domains including developmental history, cognition, language, academic achievement, social, emotional, and behavioral functioning, and adaptive behavior skills (Wilkinson, 2017). Obtaining reports on standardized measures about client behaviors and skills from informants who know the individual is a critical aspect of best practice evaluations for ASD, ID and GDD, and this is especially important regarding adaptive behavior skills (Floyd et al., 2015; Jordan et al., 2019; Stratis & Lecavalier, 2015).

Adaptive behavior refers broadly to three primary factors (i.e., conceptual, social, and practical skills) required to function in everyday life (Schalock et al., 2010). Conceptual skills include language (expressive and receptive), academic skills such as reading and writing, telling time, and understanding numbers. Social skills not only encompass interpersonal skills (e.g., going places with friends) but also safety related skills such as understanding when one is in an unsafe social situation or when one is not welcome in a social group. Practical skills include other common activities of daily living such as toileting, dressing, cooking, and following schedules. As noted, clinical data from measures of adaptive behavior are necessary when making differential diagnostic decisions (e.g., documented deficits in adaptive behavior is a requirement for a diagnosis of ID), but also essential in developing recommendations for subsequent intervention and ensuring that supports are available in both home and school settings as appropriate (Floyd et al., 2015; Jordan et al., 2019). Deficits in adaptive behavior are not an operational requirement for an ASD diagnosis (i.e., the Diagnostic and Statistical Manual for Mental Disorders-5 does not require deficits in adaptive behavior be documented using specific measurement); however, adaptive behavior difficulties are pervasive in ASD. Clients with ASD and average cognitive abilities have also demonstrated difficulty in adaptive behavior (Kanne et al., 2011). In other words, despite meeting benchmarks in cognitive abilities, autistic individuals have unexpected deficits in adaptive skills and still struggle with the basic skills needed to get through an entire day.

Deficits in adaptive behavior impede daily living skills (e.g., bathing, feeding, shopping) as well as attainment of independence and autonomy. Adaptive behavior data offers information about how an individual’s differences and diagnosis specifically impact their daily living and functioning. Data from measures of adaptive behavior can be used to contribute to the clients’ strengths and weaknesses profile and can help to determine long term goals and a prognosis on their ability of being independent (Saulnier & Klaiman, 2018). Specific skills can be targeted to increase the likelihood of independent living in the future, and it is not uncommon for adaptive skills to be prioritized when planning intervention.

Measurement of Adaptive Behavior

Without reliable data, an individual risks not receiving an accurate diagnosis and the targeted intervention needed to optimize their long-term outcomes. Adaptive behavior should be assessed using standardized measures—though observation is also often included in assessment—completed by individuals who have adequate knowledge of an individual’s functioning in various settings (Tasse et al., 2012). Ideally, data is collected from multiple raters because particular skills may be more salient in different settings (e.g., dressing is required in the home but may not be at school). As such, standardized rating forms may be completed by parents, teachers, or by the individual themselves (i.e., self-report). However, when multiple informants are involved, there may be differences in reporting, and, as a result, challenges of interpreting and reconciling discrepant data. Such discrepancies are possible whenever multiple informants are used and are not only found when measuring adaptive behaviors. For example, in a review of 49 studies examining the agreement between informants rating youth with autism emotional and behavioral concerns, there were significant differences across similar (parent/parent), different (e.g., teacher/parent), and all informants for externalizing behavior (r = 0.39–0.64), internalizing behavior (r = 0.31–0.66), social skills (r = 0.28–0.42), and total behavior (r = 0.36–0.64) (Stratis & Lecavalier, 2015). Moreover, in a study evaluating informant agreement on the Behavior Assessment System for Children 2nd Edition (BASC-2; Reynolds & Kamphaus, 2004), reports on internalizing and externalizing behaviors were consistent between informants, however, teachers reported their students as having higher adaptive skills (McDonald et al., 2016).

On measures specific to adaptive behaviors, such as the Adaptive Behavior Assessment System, Third Edition (ABAS-3; Harrison & Oakland, 2015) and Vineland Adaptive Behavior Scales-Second Edition (Vineland-II Sparrow et al., 2006), there is some evidence of discrepancies between parent and teacher ratings in autistic youth (Dickson, 2018; Jordan et al., 2019). Specifically, one study found that teachers rated youth with ASD as having more overall skills (i.e., higher scores on the general adaptive behavior composite) and more Practical skills, but found no differences in ratings between teachers and parents on the ABAS-3 Social and Conceptual domains (Jordan et al., 2019). While there were discrepancies in the composite adaptive behavior and Practical scores, their inclusion of intra-class correlation coefficients (ICCs) indicated moderate reliability between informants on the ABAS-3 composite, Conceptual, Social, and Practical scores. Dickson (2018) found that parent and teacher ratings were significantly different at two separate time points on the Vineland-II, with parents consistently reporting more skills than teachers. The previous studies were limited to samples of autistic youth with IQ scores higher than 70, meaning that these samples excluded youth with ID. However, there is some evidence that a pattern of higher scores on teacher ratings of adaptive behavior is evident in youth with lower cognitive abilities. For example, Ellison (2016) found that teachers rated 67 students with autism (IQ range = 38–128) as having more adaptive skills compared to their parents’ ratings (d = 0.42–0.78) on the BASC-2.

As described above, there is evidence of multiple informant discrepancies in samples of youth with ASD, but whether diagnoses and other individual factors influence informant discrepancies on adaptive behavior measures is largely unknown. It is critical that clinicians and other stakeholders relying on data from adaptive behavior measures understand what individual child factors might result in discrepant scores across raters so that the data is interpreted in meaningful ways. There are indications that youth characteristics may play a role in the degree of concordance between informants’ ratings on several measures of behavior (Stratis & Lecavalier, 2015). Meta-analytic research suggests that a clinical diagnosis of ASD and ID moderates the degree to which pairs of informants (i.e., child, parent, teacher) agree across several measures of internalizing and total behavior measures (Stratis & Lecavalier, 2015). Specifically, there may be higher agreement on total scores from measures evaluating broad-band externalizing and internalizing behaviors for both autistic youth and youth with ID, as well as on scales assessing internalizing behaviors for youth with ASD (Stratis & Lecavalier, 2015). Importantly, this meta-analysis did not include measures specific to adaptive behaviors.

Data from adaptive behavior measures certainly offer more insight to a child’s daily living skills, as well as provide more insight and context for many other related skills like communication, cognitive, and social skills. In some ways adaptive behavior skills are the functional outcome of these related skills, which are commonly targeted in intervention. For example, if a child is not able to communicate their wants and needs this may suggest that their speech and language functioning may be impacted. This then becomes an adaptive behavior concern when the child is unable to use verbal/nonverbal strategies to complete daily living skills (e.g., requesting a snack; telling an adult when they need to use the restroom). Moreover, engagement and maintenance of conversation is considered an important skill for both adaptive and social behavior, and example of functional contexts include social relationships (e.g., peers at school) and employment. Unsurprisingly, both speech/language and social skills are represented on many adaptive measures. Given what we know about adaptive behavior and related skills, adaptive behavior likely varies across child variables like cognitive ability, age, and language ability and these child variables have also been identified as predictors for adaptive behavior in autistic youth with varying cognitive abilities (e.g., Liss et al., 2001). Moreover, a handful of studies have specifically examined factors that may moderate concordance between respondents’ ratings on adaptive behavior measures with inconsistent results. Dickson found that child cognitive ability, as measured by the Mullen Scales of Early Learning; (Mullen; Mullen, 2001) and Differential Ability Scale, Second Edition (DAS-II; Elliot, 2007), predicted parent/teacher discrepancies on adaptive behavior (Vineland-II) and autism specific rating scales. Specifically, higher cognitive ability was related to decreased discrepancies in ratings of youths’ adaptive behavior, restricted repetitive behaviors, and social approach/withdrawal behaviors (Dickson et al., 2018). However, results from studies using the ABAS-3 found both child (i.e., age, IQ, language level, ASD symptomology) and family factors (i.e., parent education level) did not predict discrepancies in ratings (Hattier et al., 2013; Jordan et al., 2019; McDonald et al., 2016).

Current Study

Information about adaptive behavior is essential for best practice clinical evaluations and intervention planning for youth with ASD, ID, and GDD. Previous research has documented informant discrepancies on measures of emotional/behavioral functioning and measures of adaptive behavior (e.g., the ABAS-3). Understanding patterns of ratings across respondents is an important component of an evaluation as clinicians are often tasked with reconciling discrepant data when they are making important decisions about diagnosis and intervention. This line of research has established that use of multi-informant ratings provides meaningful information about youth behavior and discrepancies commonly represent differences across settings and context (De Los Reyes, 2011), which directly informs clinical treatment recommendations. Data obtained from multi-informants allows for specific recommendations to support skill acquisition in the settings the child is observed (i.e., schools, home). For example, adaptive behavior measures are commonly referenced for school-based services (e.g., individualized education plan). Furthermore, depending on an individual’s diagnosis, there may be particular patterns in respondent ratings that could help inform clinical decision making. However, research to date is inconsistent and generalizability of findings is limited by sample characteristics. For example, to date, study samples have primarily included youth with ASD and higher cognitive abilities. It is important to include youth with ID, because cognitive ability has been identified as a significant predictor of informant discrepancies on rating scales (Dickson et al., 2018). Moreover, there is limited research including younger children who are identified with GDD. ASD, ID and GDD are highly comorbid (Baio et al., 2018; Zablotsky et al., 2015). For this reason, more research is needed that captures adaptive presentation in autistic youth with comorbid diagnoses (e.g., ASD and ID) and different diagnoses that overlap in symptomology (e.g., GDD).

The current study includes a well-characterized clinical sample of youth with ASD, ID, and/or GDD to extend on research evaluating informant discrepancies on adaptive behavior rating scales to help providers and caregivers better support youth with adaptive behavior deficits. We did this by replicating a comprehensive, rigorous data analysis plan (Jordan et al., 2019; McDonald et al., 2016) that goes beyond use of single metric of informant agreement (i.e., correlations) by including: the examination of intra-class correlation coefficients (ICCs), between-group comparisons, and Bland–Altman plots and regressions (Bland & Altman, 1986). The current study extends on previous research with the inclusion of linear regressions to determine whether age, diagnoses, IQ, autism severity, and/or parent education predict concordance between parent and teacher ratings on the ABAS-3. The following research questions are addressed: (1) What is the level agreement between parent and teacher ratings on the ABAS-3? (2) Are discrepancies between informants associated with the child’s diagnosis? (3) What are the trends in the magnitude of discrepancies across the entire sample? and (4) Do client factors predict parent/teacher discrepancies on the ABAS-3?

Method

This cross-sectional research study was determined to be exempt and approved by the university Institutional Review Board. The research questions were addressed with data from 115 participants from a well-characterized clinical sample. Each participant was evaluated at an autism specialty clinic in the Midwest. The data was gathered for secondary analysis from an existing database housing clinical data for the specialty center. The participants, measures, and procedures are described below.

Participants

Youth who participated in diagnostic clinics at an interdisciplinary, specialty-care center in the Midwest and who received diagnoses of ASD, ID, or GDD were included in this study. Inclusion criteria also required complete parent and teacher reports on the ABAS-3 (Harrison & Oakland, 2015), The resulting sample consistent of 115 youth between 2 and 19 years of age (M = 9.01 years old; SD = 3.53 years). Participants were predominantly White (83%) and male (83%), and most families obtained specialized training and/or a partial college training. Most participants received an ASD diagnosis (n = 91), followed by comorbid ID and ASD (n = 13), GDD and ASD (n = 6), ID diagnosis (n = 3), and GDD diagnosis (n = 2). See Table 1 for descriptive statistics and details of the participants. In the current study the diagnosis(es) variable was dichotomized to address a gap in the literature related to informant discrepancies and to evaluate parent-teacher ratings for youth with and without lower cognitive functioning as a complete sample and separate subsamples. Participants with only ASD (i.e., without comorbid ID or GDD) represented one group (n = 91) and all other subsamples in the other (n = 24).

Table 1 Demographic data for the participants
Full size table

Clinical diagnoses were established using best-practice evaluations for developmental concerns and included (at minimum) a clinical diagnostic interview and comprehensive developmental history, cognitive or developmental assessment, and adaptive behavior assessments. For those patients for whom ASD was being evaluated, an Autism Diagnostic Observation Schedule- Second Edition (ADOS-2; Lord et al., 2012) was also administered. In addition, clients presenting with other concerns (e.g., executive functioning, anxiety, depression) were further assessed in those areas. Clinical diagnoses were made by licensed psychologists or pediatricians practicing at the autism specialty center. Participants with only an ASD diagnosis had, on average, parent adaptive behavior composites that fell in the low average range (M = 73.88, n = 91) and teacher adaptive behavior composites that fell in the below average range (M = 83.23, n = 91). Overall, the average adaptive behavior composites for youth with ID, GDD, and comorbid diagnoses fell in the extremely low range for both parent (M = 62.25, n = 24) and teacher ratings (M = 65.54, n = 24) on the ABAS-3.

Measures

Clinical reports for clients diagnosed with ASD, ID, and/or GDD seen in the past five years (2015–2020) were reviewed for demographic data, autism screening results, and measures of intellectual/developmental, adaptive behavior ability. Parents received and completed a background history form upon their child’s (i.e., participants’) diagnostic evaluation, and relevant demographic data extracted for the purposes of the current study included age of the child, child’s gender, child’s race/ethnicity, and parent education level. For the purposes of this study, the following measures were examined: the ABAS-3, the Social Communication Questionnaire-Lifetime (SCQ-Lifetime; completed by the parent; Rutter et al., 2003a, 2003b), and intellectual/developmental measures administered in the participants’ evaluation. Complete General Adaptive Behavior Composite (GAC) and domain ratings and scores (i.e., Conceptual, Social, and Practical domains) for both the parent and teacher ABAS-3 forms were required to be included in the current study.

Adaptive Behavior Assessment System, Third Edition (ABAS-3)

The primary measure of interest for the current study—the ABAS-3—is a norm-referenced instrument used to assess adaptive skills that are necessary for independent functioning in daily living (Harrison & Oakland, 2015). The ABAS-3 is used to inform diagnostic and eligibility decisions during comprehensive evaluations, as well as assist in formulating intervention plans based on identified strengths and challenges. This measure spans the age range of birth to 89 years of age and consists of five different rating forms that can be used to obtain input from critical informants as relevant for the individual’s age and environments. For this study, rating forms were obtained from parents and teachers. The ABAS-3, Parent/Primary Caregiver Form was administered for participants ages 0–5 years and the ABAS-3, Parent Form was administered for participants between 5 and 21 years. In addition, the ABAS-3 Teacher/Daycare Provider Form was administered for participants between 2 and 5 years of age, and the ABAS-3 Teacher Form was administered to participants between the ages of 5 and21 years. Respondents are asked to rate the individual’s abilities on a 4-point scale (0–3), ranging from “is not able to” perform the skill up to “always or almost always” performs the skill in question.

Ratings on the ABAS-3 result in raw scores that are converted to scaled scores for 9 specific adaptive skill areas (M = 10, SD = 3), as well as standard scores for adaptive skill domains of Conceptual, Social, and Practical skills (M = 100, SD = 15). Scores are compiled to form the GAC, an indicator of overall adaptive functioning (M = 100, SD = 15). Scores obtained on the skill areas, domains, and the GAC compare the individual’s reported skills to those in the normative sample of the same age. The present study compares Parent and Teacher standard scores for the Conceptual domain, Social domain, Practical domain, and GAC. The Conceptual domain assesses communication, academic, and self-management skills. The Social domain assesses interpersonal, leisure, and relationship skills. The Practical domain assesses community functioning, safety, and self-care skills.

The ABAS-3 was standardized using a standardization sample that was representative of the U.S. population except for White individuals and people from higher educational backgrounds being overrepresented (Harrison & Oakland, 2015). In terms of reliability, across all the standardizations samples, the ABAS-3 GAC has an internal consistency between 0.96 and 0.99. The reliability coefficients for adaptive domains vary by rating form and range between 0.72 and 0.99. Test–retest reliability also shows acceptable ranges of reliability (0.62–0.86). Interrater reliability estimates of GAC and adaptive domains fall in the moderate to strong level of consistency (0.67–0.92), varying by rating form. Cross-form consistency is of relevance to the present study as it measures the consistency between different informants using different forms. Cross-form consistency between parent and teacher ratings fell in the low to moderate range (0.41–0.57), which Harrison and Oakland (2015) note is to be expected and provides evidence for the need to obtain ratings from both parents and teachers during a comprehensive evaluation.

Validity of the ABAS-3 was established, in part, through analyzing intercorrelations for each skill area, all domains, and the GAC (Harrison & Oakland, 2015). Harrison and Oakland (2015) note that the intercorrelation findings support the expected among skills and domains, while also indicating that each skill area measures a separate construct. Factor analysis revealed that a one-factor model best fit the observed data from the standardization sample, which is consistent with prior research and conceptualization of most adaptive functioning assessments measuring a global factor. However, a three-factor model was also a close fit to the standardization sample data and supports the interpretation of three separate adaptive domains. In addition, assessment of the extent to which scores on the ABAS-3 were associated with scores on other measures of similar abilities (ABAS-II and Vineland-II) showed evidence of moderate to strong convergent validity. Finally, multivariate analysis of variance (MANOVA) analyses of the ABAS-3 also showed that adaptive composite scores were able to differentiate children with delays when compared to typically developing children. Specifically, youth with neurodevelopmental disorders were rated lower on the ABAS-3 compared to neurotypical youth with large effects for subsamples of youth diagnosed with ASD (ES = 1.48–2.51) and ID (ES = 2.26–2.91). Youth with ADHD scored lower than the neurotypical group, but significant effects were only evident for the self-direction subscale on the parent and teacher reports (ES = 1.53, 1.71 respectively).

Autism Symptom Severity Screener

Parents ratings of current ASD symptoms and behaviors were assessed using the SCQ-Lifetime (Rutter et al., 2003a, 2003b). The SCQ-Lifetime is based on the Autism Diagnostic Interview-Revised (ADI-R; Rutter et al., 2003a, 2003b) algorithm items that gather parent report of social interaction, communication, and restricted, repetitive behaviors. More specifically, the SCQ-lifetime is designed to capture an individual’s developmental history and symptom presentation between the ages of 4 and 5. The authors of the SCQ-Lifetime recommend that children be referred for a comprehensive evaluation if they receive a cutoff score of at least 15.

Measures of Intellectual and Developmental Assessment

Due to the retrospective nature of the study and variability in assessment data obtained, several different measures of intellectual and developmental functioning were used across participants. These included verbal and nonverbal index scores from the Differential Ability Scales, Second Edition (DAS-II; Elliot, 2007) early years (n = 16, M Verbal = 81.88, SD Verbal = 24.56, M Nonverbal = SD Nonverbal =) and school-age versions (n = 3, M Verbal = 106, SD Verbal = 13.53, M Nonverbal = 106, SD Nonverbal = 7), Leiter International Performance Scale, Third Edition (Leiter-3; Roid et al., 2013; n = 28, M = 74.5, SD = 18.40), Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI-II; Wechsler, 2011; n = 35, M Verbal = 93.20, SD Verbal = 14.29, M Perceptual Reasoning = 94.94, SD Perceptual Reasoning = 14.59), Wechsler Intelligence Scale for Children, Fifth Edition (WISC-V; Wechsler, 2014; n = 22, M Verbal = 93.73, SD Verbal = 19.16, M Visual Spatial = 90.95, SD Visual Spatial = 20.20), Wechsler Preschool and Primary Scale of Intelligence, Fourth Edition (WPPSI-IV; Wechsler, 2012; n = 3, M Verbal = 87, SD Verbal = 10.82, M Visual Spatial = 92.33, SD Visual Spatial = 15.28), and Wechsler Adult Intelligence Scale, Fourth Edition (WAIS-IV; Wechsler, 2008; n = 5, Verbal = 87.60, SD Verbal = 20.20, M Visual Spatial = 85.20, SD Visual Spatial = 23.35).

The Mullen Scales of Early Learning; (Mullen; Mullen, 2001) is a developmental measure that assesses young children’s Expressive and Receptive language, Fine and Gross Motor, and Visual Reception abilities. The Mullen is unlike the intelligence tests listed above, as participants’ performance on subtests are reported as t-scores. To account for these differences, ratio verbal and nonverbal IQs were calculated from the Mullen Expressive language and Visual receptive subtests respectively. Ratio verbal and nonverbal IQs were calculated by taking the age equivalent in number of months, dividing is by the clients’ age at the time of their evaluation and multiplying my 100. This was completed for three participants (M Expressive = 25.61, SD Expressive = 17.48; M Visual Receptive = 40.46, SD Visual Receptive = 3.44).

Procedures

Approval was granted from the university’s Institutional Review Board to analyze secondary data from a large database housed within the autism specialty clinic. Data were collected within the context of clinical evaluations where the primary referral concerns included behaviors related to ASD, developmental concerns, and comprehensive neuropsychological assessments. Prior to or during these evaluations, parents were given the ABAS-3 protocol to complete. Similarly, teacher forms were sent to teachers prior to the clients’ evaluation date and returned to the primary clinician. The completed ABAS-3 forms were checked for errors and scored using scoring software. Participant data was de-identified and imported to SPSS 26 for statistical analysis.

Data Analysis Plan

Descriptive statistics were computed for participant demographic data as well as the measures used in the current study. Each participant received an individualized evaluation, and the primary referral concern was variable, which resulted in missing variables across the sample. The evaluations were not identical because of clinical decision-making by psychologists at the time of the evaluation, and organization changes across the 5 year period. It is not uncommon for autism clinicians to adjust their assessment plan to match the individual need and presentation of their client. For example, a nonverbal IQ test (e.g., Leiter International Performance Scales, Third Edition) may be more appropriate for clients with limited verbal abilities, and a developmental test (e.g., Mullen) for younger children. As a result, verbal and nonverbal IQ standard scores were not available for every child.

The current study replicated a comprehensive analysis plan completed by Jordan et al. (2019) and McDonald et al. (2016). A power analysis was conducted to determine the statistical power required for mean score comparisons, informant correlations, and for the multiple regression. The statistical power for the mean score comparisons was estimated at 0.98 across the total sample [d = 0.4, α = 0.0125, (two-tailed), and N = 115 pairs; (G*Power 3.1.9.6, Faul et al., 2007)]. For the correlations between pairs, the estimated statistical power was 0.92 to detect a relationship as small as 0.3. Post hoc power analyses were conducted for the multiple regressions conducted in this study. The sample size was 115 and 2 to 4 predictors were used with an alpha level of p < 0.5. The post hoc analyses revealed medium to large effects for the GAC, Conceptual, and Practical score differences (\({f}^{2}\)= 0.27, 0.23, and 0.36 respectively), indicating and adequate power (0.99–1.0). However, post hoc analysis revealed a small effect for the Social domain (\({f}^{2}\)= 0.10; Cohen, 1988) and power detection of 0.87.

Agreement and consistency between parents and teachers (i.e., research question one) were evaluated with intra-class correlations coefficients (ICC). A one-way random effects model was used for the ICC because there were different informants across the entire sample. An acceptable reliability between informants on the average measures is at least 0.70 (Koo & Li, 2016). To test whether discrepancies between informants were associated with child diagnosis (i.e., research question two) we assessed parent and teacher differences were assessed with a series of analyses including a paired samples t-test of the GAC, Conceptual, Social, and Practical domains for the entire sample and a split sample based on diagnostic category (i.e., ASD or other). Correlations alone are not sufficient to measure agreement (Altman & Bland, 1983; Ranganathan et al., 2017) as correlations only refer to the direction and magnitude of the relationship between variables. ICCs and Bland–Altman plots and correlations were generated to examine the data for trends in the data by regressing the mean ABAS-3 scores on the parent-teacher difference scores. (i.e., research question three). Similar to previous research (Jordan et al., 2019; McDonald et al., 2016), the difference scores between parent and teacher pairs were plotted on the y-axis and the mean score of each pair was plotted on the x-axis. In other words, Bland–Altman plots are a statistical method to estimate agreement between two measures where the x-axis represents how well the measures agree on average, and the y-axis represents the exact discrepancy value for each parent-teacher pair (Bland & Altman, 1986). Bland–Altman plots include upper and lower limits of agreement (LOA), which shows how much difference is displayed across the entire sample based on the mean observed difference. Four Bland–Altman plots were generated to represent informant pairs’ ratings on the GAC and three ABAS-3 domains. As in Jordan et al (2019), the scatterplots and accompanying correlations were used to further evaluate whether predictor variables moderated the relationship between parent and teacher discrepancies on the ABAS-3. Finally, we tested if any client factors predicted parent/teacher discrepancies on the ABAS-3 (i.e., research question four) by building a linear regression model. Predictors were included in the model if there was a significant correlation between the predictor and magnitude of discrepancy.

Results

Parent and Teacher Reliability Estimates

Parent and teacher reliability was analyzed with ICCs between the informants’ ratings on the ABAS-3 GAC and three domains. For all scores on the ABAS-3, there were significant correlations between the parent and teacher ratings. However, poor reliability (i.e., below the acceptable reliability of 0.70) was found between parent and teacher informants on the GAC, Conceptual, Social, and Practical domains with ICCs (Koo & Li, 2016). The average ICC between parent and teacher GAC was 0.60 with a 95% confidence interval from 0.42 and 0.72 (F [114] = 2.47, p < 0.0001). The ICC between the average measure on the three ABAS-3 domains was 0.67 (95% CI 0.50–0.76; F [114] = 2.90, p < 0.0001) for parent/teacher Conceptual scores, 0.43 (95% CI 0.18–0.61; F [114] = 1.77, p < 0.01) for parent/teacher Social scores, and 0.61 (95% CI 0.44–0.73; F [114] = 2.58, p < 0.0001) for parent/teacher Practical scores.

Cross-Informant Group Comparisons

Paired sample t-tests were conducted to further evaluate the discrepancy between parent and teacher ratings on the ABAS-3. Cross-informant comparisons were conducted on the GAC and three domains for the entire sample, as well as subsamples of participants with only an ASD diagnosis and comorbid presentations (refer to Table 2). Teachers rated youth with neurodevelopmental disorders as having higher Conceptual, Social, Practical, and general adaptive skills (i.e., GAC) compared to their parents’ ratings, and all comparisons were significant for the total sample with small to medium effect sizes. Similarly, teachers rated the subsample of youth with only ASD as having higher levels of adaptive skills on the GAC (d = 0.70) as well as on the Conceptual, Social, and Practical domains (d = 0.47, 0.68, 0.13). Finally, teachers rated autistic youth with ID or GDD as having more skills overall, but parent-teacher ratings for the Social domain was the only significant difference (p < 0.05) with a small effect size of d = 0.47.

Table 2 Parent-teacher discrepancies across ABAS-3 scores
Full size table

Trends in Parent-Teacher Ratings

Bland–Altman plots were created to determine whether there were systematic trends across the informants’ difference and mean scores on the GAC and three ABAS-3 domains for the total sample. The difference scores were calculated by subtracting the teacher scores from the parent. Thus, negative difference scores represent an informant pair with higher teacher ratings. Refer to Online Resources 1–4 for sample data that illustrate the parent, teacher, difference, and mean scores used for the Bland–Altman plots. The vertical (y) axis represents the informants’ difference scores; the horizontal (x) axis represents the mean score for each pair of ratings (Figs. 1, 2, 3, 4). The solid line represents the mean difference scores, the dotted lines represent the upper and lower limits of agreement (LOA) for the difference scores, and the diagonal dotted lines represent the regression line.

Fig. 1
figure 1

Bland–Altman plot of parent/caregiver-teacher scores on the General Adaptive Composite. Scatter plot with a light gray background and dark gray data points representing parent-teacher GAC difference scores regressed onto the mean scores. The plot includes one solid black line representing the mean difference score, two dotted lines representing the limits of agreement (or confidence intervals), and one diagonal, dotted line representing the regression equation

Full size image
Fig. 2
figure 2

Bland–Altman plot of parent/caregiver-teacher scores on the ABAS-3 conceptual domain. Scatter plot with a light gray background and dark gray data points representing parent-teacher conceptual difference scores regressed onto the mean scores. The plot includes one solid black line representing the mean difference score, two dotted lines representing the limits of agreement (or confidence intervals), and one diagonal, dotted line representing the regression equation

Full size image
Fig. 3
figure 3

Bland–Altman plot of parent/caregiver-teacher scores on the ABAS-3 social domain. Scatter plot with a light gray background and dark gray data points representing parent-teacher Social difference scores regressed onto the mean scores. The plot includes one solid black line representing the mean difference score, two dotted lines representing the limits of agreement (or confidence intervals), and one diagonal, dotted line representing the regression equation

Full size image
Fig. 4
figure 4

Bland–Altman plot of parent/caregiver-teacher scores on the ABAS-3 practical domain. Scatter plot with a light gray background and dark gray data points representing parent-teacher practical difference scores regressed onto the mean scores. The plot includes one solid black line representing the mean difference score, two dotted lines representing the limits of agreement (or confidence intervals), and one diagonal, dotted line representing the regression equation

Full size image

Visual analysis indicated large difference ranges for the GAC, Conceptual, Social, and Practical parent and teacher ratings. For example, on the GAC (Fig. 1), there are 55 points separating the upper LOA (19.66) and lower LOA (-35.85), which suggests high variability in the difference scores and less agreement between informants. This is also suggested by the magnitude of the difference scores, which supports the findings from the paired sample t-tests that informant scores are significantly different. The mean difference scores for all four plots are negative (i.e., teacher ratings tend to be higher). In addition to calculating the mean difference value, upper and lower LOA were calculated to create a visual representation of the difference between the parent and teacher difference scores and 95% of future measurement pairs. Anything outside the LOA would be considered unacceptable agreement (Bland & Altman, 1986). Figure 1 displays the GAC difference and mean scores between parent and teacher pairs. All the paired GAC scores fell within the LOA, which suggested acceptable agreement across all pairs. There were a few outliers on the Conceptual, Social, and Practical plots, but many of the paired scores fell within the upper and lower LOA. In the Bland–Altman regression analyses, the mean scores were regressed onto the difference scores. There were significant relationships between the difference and means for the GAC composite (B = − 0.377, t = − 3.846, p < 0.001), Conceptual (B = − 0.331, t = − 3.360, p < 0.001), Social (B = − 0.479, t = − 4.131, p < 0.001). and Practical domain (B = − 0.341, t = − 3.455, p = 0.001).

Predictors of Parent Teacher Differences

In addition to the analysis of the Bland–Altman plots, potential predictors of the difference scores across the total sample for the GAC, Conceptual, Social, and Practical domains using a linear regression model. The variables included were participant age at the time of their evaluation, verbal and nonverbal IQ, autism severity (represented by parent report on the SCQ-Lifetime), parent education, and final diagnosis or diagnoses (Table 3). There were several significant correlations between the parent-teacher difference scores and the included variables (refer to Online Resource 5 for all correlations). Nonsignificant correlations were found between parent-teacher difference and nonverbal IQ and parent education; thus, those two variables were not included in the final regression model. The parent-teacher difference scores on the GAC were predicted by the SCQ-Lifetime scores, which explained 30% of the variance in GAC difference scores (B = − 0.70, p < 0.05). Specifically, as SCQ-Lifetime scores increased, parent-teacher difference scores decreased by 0.70 on the GAC. SCQ-Lifetime scores also predicted differences on informants’ ratings on the Conceptual (B = − 0.54, p < 0.05), Social (B = − 0.51, p < 0.05), and Practical domains (B = − 0.86, p < 0.01). Conceptual, Social, and Practical scores explained 46, 49, and 13% of the variance in parent-teacher difference scores respectively.

Table 3 Regression analyses: variable predicting parent-teacher GAC, conceptual, social, and practical difference scores
Full size table

Discussion

Given the increasing prevalence rates for ASD and other neurodevelopmental disorders, clinicians require access to measures that allow them to identify client’s current functioning as well as measures that provide useful information for intervention planning. Adaptive behavior measures are necessary in evaluations, not only because they are needed for diagnostic purposes, but because of the need to identify skills for intervention. Gathering data from multiple informants provides information about the client’s performance across settings (e.g., school, home), which is consistent with best practice in evaluation (Floyd et al., 2015). While discrepancies on behavior measures are common for youth with ASD (e.g., Stratis et al., 2015), they are potentially problematic as they complicate clinical decision-making including clinical interpretation, differential diagnosis, treatment planning, and evaluating treatment efficacy (Hawley & Weisz, 2003). Research has identified discrepancies between parent and teacher informants when rating behavior in youth with ASD (Dickson, 2018; Jordan et al., 2019; McDonald et al., 2016), and less often, with children with ID (Ellison, 2016). It is important for clinicians and other stakeholders to understand what patterns of discrepancies might be encountered in clinical settings, and to recognize what individual factors might be more likely to result in such discrepancies. Thus, we aimed to determine the extent to which parents and teachers agreed on a measure of adaptive behavior for youth with ASD, ID, and/or GDD. Specifically, we asked the following research questions: (1) What is the level agreement between parent and teacher ratings on the ABAS-3? (2) Are discrepancies between informants associated with the child’s diagnosis? (3) What are the trends in the magnitude of discrepancies across the entire sample? and (4) Do client factors predict parent/teacher discrepancies on the ABAS-3? The current study adds to this line of research by including a sample of participants with diverse cognitive abilities across a wide age range and aimed to better understand factors that might predict rating discrepancies across multi-informants.

Parent-Teacher Agreement

With regard to the level of agreement between parent and teacher ratings (i.e., research question one), while the paired ratings were significantly and positively correlated, there was poor reliability. This was observed for the GAC and all three domains on the ABAS-3. This finding is similar to previous research by Jordan et al. (2019) where positive associations between parent-teacher informants were identified on the GAC and Practical domain. However, in the current study, the magnitude of the relationship between parent-teacher GAC, Conceptual, and Social scores was slightly larger than those in Jordan et al. (2019). Unlike previous research, reliability was interpreted in the current study using Koo and Li’s (2016) guidelines for intraclass correlation coefficients (ICC) where 0.70 is considered acceptable reliability between two measures. Relying solely on correlations between informants may be insufficient for measuring agreement, because correlational data only provides insight to the magnitude of the relationship between variables. The poor reliability found between parent-teacher informants in the current study suggested that parent and teacher informants were not reliable with each other in reporting youth adaptive behavior, which warranted further investigation.

Parent-Teacher Discrepancies Across Subsamples

Discrepancies between informants were further examined and compared to investigate if there was an association between diagnoses and patterns of discrepancies (i.e., research question two). Teachers rated all youth in our sample as having more adaptive behavior skills compared to parent report on all ABAS-3 composite and subdomain scores. These results are different from previous research where parent-teacher discrepancies were only found for the GAC and Conceptual rating (Jordan et al., 2019). In the current study, subsamples were examined to determine how parents and teachers differ in rating adaptive behavior for youth with ASD only and youth with co-occurring ID or GDD. Teachers rated participants with ASD (n = 91) with more General Adaptive, Conceptual, Social, and Practical skills than their parents. There were no significant differences between informants rating the subsample of participants with ID and co-occurring diagnoses, except on the Social domain. However, all difference scores on the ABAS-3 were negative, which indicated higher teacher ratings and scores. This may be an artifact of the small sample size (n = 24), but this finding is consistent with previous research demonstrating more agreement between informants’ ratings with decreasing IQ (Stratis & Lecavalier, 2015) and suggest that parents and teachers are more likely to agree on skill development in autistic youth with ID as they demonstrate clearer deficits in adaptive behavior. Parents and teachers rating participants with ASD and ID/GDD disagreed on level of social skills. The assessment of social skills may be more difficult due to amount of subjective judgement used (Lecavalier & Butter, 2010). However, this finding is somewhat surprising given research that suggests impairments in social skills may be more easily observed in youth with ASD and ID (Wilkins & Matson, 2009). Nonetheless, teachers consistently rated youth with ASD, ID, and/or GDD with more adaptive behavior skills. This may be related to the differences across settings (i.e., home versus school). Specifically, teachers may have more opportunities to observe social skills as they can readily compare individual students to their same-aged peers. Teachers are highly effective when identifying friendships, peer rejection, and social concerns in their classroom (e.g., Lane & Menzies, 2005), which is likely explained by their exposure and direct interaction with multiple children in one school year as well as across their teaching career.

Trends Across the Total Sample

Following the data analysis plan used in previous research (Jordan et al., 2019; McDonald et al., 2016), the current study evaluated the magnitude of parent-teacher discrepancies across the total sample’s range of scores (i.e., research question three). There were significant, negative relationships between the parent-teacher difference scores and mean difference score for the GAC and three ABAS-3 domains. Thus, differences in parent-teacher ratings differed significantly across the sample’s range of scores, specifically, larger parent-teacher score differences were observed with higher teacher scores. All Bland–Altman plots had negative means (i.e., higher teacher scores) and downward trends (i.e., increasing negative difference scores). This is unlike previous research where there were no significant differences were found (Jordan et al., 2019), but this finding is somewhat similar to McDonald et al (2016) who found positive, significant relationships between informant difference scores and mean scores. However, results from the current study are different than McDonald et al. (2016) in that—in the current study—teachers rated youth higher on the ABAS-3 whereas parents in previous research rated autistic youth without ID as having more adaptive skills on the BASC-2. This may represent the differences in sample characteristics and measure selection. This discrepancy across individual informant pairs (i.e., t-test) as well as all pairs included in the current study (i.e., Bland–Altman) may represent differences in informants’ understanding of adaptive behavior as well as adaptive behavior expectations.

Predicting Parent-Teacher Discrepancies

The current study extended on previous research investigating child and family factors that explain discrepancies between multiple informants on behavior rating scales of youth with ASD, ID, and GDD (Dickson, et al., 2018; Jordan et al., 2019; McDonald et al., 2016; Stratis & Lecavalier, 2015). Several child factors were significantly associated with parent-teacher difference scores (i.e., research question four). The strength of the relationships between age, diagnosis, verbal IQ, and family income were small (r = 0.21–0.28), so it is unclear if this finding has clinical significance. However, autism severity scores on the SCQ-Lifetime had moderate, significant relationships with parent-teacher rating discrepancies for every ABAS-3 domain with correlations ranging from 0.27 to 0.39. This finding is in contrast to studies using more comprehensive measures of ASD. For example, Jordan et al. (2019) did not find any significant relationships in a study including participants’ scores on the ABAS-3 and Autism Diagnostic Interview-Revised (ADI-R; Rutter et al., 2003a, 2003b), which was used to characterize autism symptomology. Similarly, comparison scores from the ADOS-2 were not related to parent-teacher discrepancies on the Vineland-II (Dickson et al., 2018; Kanne et al., 2011). Participant ratings on the SCQ-Lifetime may be more related to informant discrepancies than other autism-specific measures because it is a screening instrument and is often completed by an informant who also completes other behavior measures including the ABAS-3.

Given the significant relationships found between ABAS-3 difference scores and selected child variables, regression analyses were used to determine predictors of parent-teacher discrepancies. SCQ-Lifetime was the only significant predictor of all ABAS difference scores, and specifically, as SCQ-Lifetime scores increased, the GAC, Conceptual, Social, and Practical parent-teacher difference scores decreased. A high score on the SCQ-Lifetime indicates an increased risk of having ASD and having higher scores on an autism screener predicted increased informant agreement in the current study. Informants may find it easier, and thus are more consistent, when rating youth with more obvious or apparent skill deficits. This pattern of clinical symptomology impacting informant agreement has been found in other clinical samples where multiple informants were likely to endorse symptoms when clients have demonstrated elevated ASD symptom severity (e.g., Carlson & Youngstorm, 2003).

Clinical Implications

Discrepancies between informant ratings may be the result of several factors including the informants, the settings in which the informant and client interact, as well as the client themselves. First, it is well understood and agreed that parents and teachers have different relationships and roles in a child’s life. Parents and teachers interact with children in different settings, which reflects contextual variations that naturally disrupts agreement on behavior measures. Therefore, parents and teachers may have varied opportunities to observe adaptive behaviors. Teachers can observe and compare one student to several, same-aged peers, which may enhance their ability to identify strengths and weaknesses in individual children. Research suggests that executive functioning skills predict differences in adaptive behavior (Tomaszewski et al., 2020), and, in some cases, more than IQ (Puliese et al., 2015). These related skills and their impact on adaptive behavior may be more readily observed in schools due to the demands in that setting. However, parents might be better positioned to speak to a client’s current daily living skills (e.g., feeding, hygiene) as well as their development of these skills overtime. One potential explanation for the higher teacher ratings of adaptive skills found in the current study could be that the structures, routines, and supports of school settings contribute to the performance of more adaptive behavior skills at school than at home (which may be less structured environments). Moreover, aspects of the settings like time of day (e.g., parents are more likely to see their child when they are tired) and other conditions may result in discrepant ratings.

The ABAS-3 manual offers other possible explanations behind informant discrepancies including misunderstanding of the instrument, and errors when completing the rating form. Differences in informant experience and training may further influence ratings. Notably, one informant may have enhanced observations skills or preparation for completing the rating form. Relatedly, it is also likely that parents and teachers have a different understanding of adaptive behavior. Informant characteristics like reading skill, language use, and cultural differences may impact understanding and expectations for adaptive behavior as well as accurate completion of a measure.

Individual client characteristics also contribute to discrepant ratings which has implications for differential diagnosis. In the current study, increased autism symptom severity (as measured by the SCQ-lifetime) predicted informant agreement on the ABAS-3. In addition to autism symptomology, Kanne et al. (2011) found significant relationships between adaptive behavior and age, and IQ in youth with autism. Specifically, youth with ASD showed more deficits in adaptive behavior when they were older, and youth with average cognitive abilities demonstrated larger differences in adaptive behavior compared to youth with ID. Clients with lower cognitive abilities are more likely to show commensurate adaptive behavior, whereas clients with higher IQ demonstrate discrepancies between their cognitive and adaptive behavior abilities. Moreover, the current study suggests that parent-teacher data obtained through adaptive behavior measures is more discrepant when youth show less autism symptomology. Clients with subtle autism presentations and solid cognitive abilities may experience adaptive behavior deficits, but in these instances informants may strongly disagree on adaptive behavior rating scales further complicating identification of client strengths, weaknesses, and treatment planning.

Researchers have recommended that clinicians use what they know about informant discrepancies to anticipate informant differences and then to use this information while constructing their assessment plan (De Los Reyes et al., 2015). Clinicians can maximize their time spent conducting family interviews and/or use other assessments to gather more information on adaptive behavior to better understand these differences and the child’s overall adaptive behavior functioning. The ABAS-3 manual also provides guidance to clinicians on how to navigate such discrepancies by calculating the difference score and its statistical significance when informant differences are present (Harrison & Oakland, 2015). However, this solution may not be feasible due to time constraints during clinical evaluations. Thus, seeking out additional information with other assessment tools may be more useful for diagnoses and intervention planning. If the clinician assumes that results on the ABAS-3 will look different between parents and teachers, the clinician could plan to include questions related to settings differences (e.g., opportunities to engage in daily living skills, behavior management practices, consequences). Moreover, direct assessment of adaptive behavior is recommended as a necessary supplement to measures and interviews (Floyd et al., 2015). Behavior observations are advantageous as they allow the clinician to observe clients’ behaviors in structured and unstructured settings and evaluate their response to different demands.

Limitations and Future Directions

Although the current study has potential implications for research and practice, results should be considered within the context of their limitations. First, the current study found differences in parent and teacher reports for the small subsample of youth with ID, GDD, and co-morbid ASD (n = 24) only for the Social domain on the ABAS-3. However, there may be differences in other domains across informant reports that were not detected as a result of the small subsample size and low power. Second, the sample was limited to one geographical area in the United States (i.e., Midwest) and the sample largely consisted of White (83%), male (83%) clients diagnosed with ASD only (79%). Girls and youth of color are more likely to be undiagnosed with ASD compared to their male, White peers (Wiggins et al., 2020), which is evident in the current study as well. Disparities in ASD diagnoses have been attributed to many factors, such as levels of stigma and providers dismissing family’s concerns (Stahmer et al., 2019). When it comes to measurement of children’s behavior, previous research indicates that informant agreement may vary based on the race and ethnicity of the child. For example, Black parents were less likely to report concerns regarding their child’s ASD symptoms on screening measures (Donohue et al., 2017). However, in a separate study, Black children were more likely than White children to fall in the at-risk range on an autism screening tool again, after their parents participated in the follow-up interview that included more questions about autism symptomology (Dai et al., 2021).

Additionally, some research suggests that youth of color with ASD (specifically Black and Hispanic/Latinx youth) demonstrated more severe behaviors and language delays when compared to their White counterparts (Angell et al., 2018). These findings suggest increased severity of ASD symptomology in Black and Latinx youth. As noted earlier, informants tend to be more consistent in their ratings when youth demonstrate more obvious deficits, and the current study showed reduced ASD symptomology predicted greater disagreement across reporters. Given this, informants may demonstrate more agreement on behavior measures when rating youth of color who demonstrate increased severity in symptoms. As such, if the sample in the current study was more racially and ethnically diverse, the reports between parent and teacher informants may have been less discrepant.

In the area of informant discrepancies on adaptive behavior measures, research suggests differences across several individual factors such as age (Hill et al., 2015), diagnoses (Matson et al., 2009), and IQ (Liss et al., 2001). For example, youth with intellectual disabilities have shown more adaptive behavior skills than youth with ASD (Kanne et al., 2011), which also has been found in samples of youth living outside of the United States (e.g., Alvares et al., 2020; Kilincaslan et al., 2019). However, research examining differences in informant ratings of adaptive behavior across youth from diverse racial and ethnic backgrounds is lacking. In a study with a larger, diverse sample, parents from different racial and ethnic background report autism symptomology and concern differently (Azad et al., 2021), so it is likely that there are differences on adaptive behavior measures too. Future research should replicate the current study using a more diverse sample, and in addition to the comprehensive data analysis plan used in the current study, other methods of understanding individual, familial, and cultural factors is encouraged (i.e., culturally relevant assessment practices).

Other limitations surround issues of assessment, measurement, and methodology. More specifically, the current study did not control for type of test used to measure participants’ verbal and nonverbal IQ. In addition, the current study is cross-sectional, and data included in the current study were gathered retrospectively. Studies using alternative methodology could provide further insight to the nature of the relationships among patient characteristics and informant agreement of adaptive behavior skills over time. Fourth, the current study only examined informant agreement on one measure of adaptive behavior and the degree to which these findings generalize to other measures is unknown. Future researchers might compare these findings to informant agreement on other scales (e.g., Adaptive Behavior Evaluation Scale, Third Edition; McCarney &House, 2017) and possibly evaluate the clinical utility of other sources of clients’ adaptive behavior (e.g., individualized evaluation plan goals and progress, occupational therapy case notes). Finally, a review of the original research on the development of the ABAS-3 suggests that its psychometric rigor could have been more robust. The original psychometric research on the ABAS-3 showed low to moderate correlations between parent and teacher forms. As such, studies using the ABAS-3 (including the current study), are limited by the psychometric properties of the instrument.

Conclusion

In the current study, large discrepancies between parent-teacher informants were found on the ABAS-3 for youth with ASD, ID, and GDD. This may be a particularly important consideration for clinicians when conducting evaluations of patients with less severe autism symptomology and cognitive impairments. It is essential that clinicians have access to reliable and valid assessment tools that provide a snapshot of clients’ present functioning as well as provide data that can be used to guide treatment and interventions; however, clinicians should expect there to be discrepancies between parents and teachers on adaptive behavior measures. Obtaining data from multiple informants is best practice and necessary to make diagnostic decisions, but based on the results of the present study, practitioners are encouraged to seek further information from parent-teacher informants (e.g., interviewing, behavior observations, etc.) when there are discrepancies to support clinical impressions and treatment planning.