KSKevin G. Stephenson

Kevin G. Stephenson

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Kevin G. Stephenson has a total of 147 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Parsing the Pediatric Mood Spectrum in U.S. Youth: Transdiagnostic Severity Versus Phenotypic Specificity

    by Kevin G. Stephenson, Phoebe A. Rodda, Carla Mazefsky, Eric A. Youngstrom

    Objective: Emotion dysregulation and irritability occur across pediatric disorders, complicating distinctions between general symptom burden and specific clinical presentations. Person-centered analyses can look at profiles of scores to identify groups with similar presentations. We examined whether modeling shared and domain-specific symptom variance identified differentiated profiles in youth. Method: Caregivers of 2,187 US youth ages 6–18 completed 87 items assessing emotion dysregulation, depression, mania-related symptoms, sleep-related impairment, and attention and behavioral problems. An ordinal bifactor model estimated one general and 13 specific factors. Latent profile analyses used standardized scores on all 14 factors, with an expanded raw-score analysis as a comparator. We examined caregiver-reported diagnoses and functioning and approximately two-week classification agreement among 166 recontacted caregivers. Results: The general factor accounted for 62% of common variance (ωH=.94), although model fit was mixed (CFI=.884; RMSEA=.058; SRMR=.092). A provisional four-class solution distinguished population-modal (67.4%), internalizing (11.8%), attention/hyperactivity (11.0%), and unusual-thought/elevated-mood (9.8%) profiles. Entropy was .83, but model enumeration and initialization checks did not establish a unique solution. Profiles showed overlapping diagnostic and functional correlates. Recontact agreement was 78.9% (κ=.60). Irritability-specific reliability was limited, constraining interpretation of its weak profile differentiation. Conclusions: General symptom burden coexisted with differentiated residual symptom patterns. These findings support evaluating severity alongside symptom configuration, without equating profiles with diagnoses or weak residual irritability signals with limited clinical relevance. Measurement refinement, model-selection verification, and confirmation of classification-error-adjusted associations are needed before clinical application, but profiles could simplify clinical interpretation of sets of scales.

  • Open Access

    In Search of Subtypes of Children With Developmental Disabilities Using the WISC-V: A Latent Profile Analysis

    by Jordyn L. Esprit, Eric A. Youngstrom, Soo Youn Kim, Megan Norris, Ann Levine, Eric M. Butter, Kevin G. Stephenson

    Purpose: Identifying subtypes within autism spectrum disorder (ASD) has often been attempted to better understand the significant heterogeneity within the spectrum. Profiles of cognitive functioning have frequently been investigated as possible subtype definitions, but most studies have relied on relatively small sample sizes and higher-functioning samples, and few have used advanced statistical techniques to obtain estimates of underlying profiles of IQ in ASD. The purpose of this study was to use latent profile analysis to identify possible subgroups of children with ASD and other neurodevelopmental disabilities using the Wechsler Intelligence Scale for Children-5th Edition (WISC-V) across a range of ability levels. Methods: We used latent profile analysis with a clinical sample of 727 youth (ages 6–16) referred for neurodevelopmental evaluations. We also completed a multigroup bifactor confirmatory factor analysis of the clinical sample compared to the WISC-V standardization sample. Results: We did not find distinct cognitive profiles based on the WISC-V index scores; results indicated a continuum of the g factor of intelligence instead of distinct patterns of strengths and weaknesses across IQ subscales. A multigroup bifactor confirmatory factor analysis revealed partial strong invariance when comparing the two clinical groups to the WISC-V standardization sample, but full strong invariance within the neurodevelopmental groups and substantive explained variance of the processing speed index above and beyond FSIQ. Conclusion: Alternative methods and measurements are needed to identify potential cognitive subtypes of children with autism and other neurodevelopmental disabilities. Processing speed (but not other WISC-V indexes) may be of unique clinical interpretability above and beyond FSIQ.

  • Open Access

    Evaluating the Autism Spectrum Rating Scale Parent and Teacher Reports (6 to 18 years) to Predict an Autism Diagnosis in a Clinical Sample

    by Reese Costenbader, Megan Norris, Courtney E. Rice, Kelsey B. Shively, Kevin G. Stephenson

    Rating scales are widely used in autism spectrum disorder (ASD) evaluations to provide information about developmental history and behavior across settings. Despite frequent use, limited research has examined the diagnostic utility of the Autism Spectrum Rating Scale (ASRS) in clinical settings. This study evaluated the ability of the ASRS to predict ASD diagnoses in a clinical sample. We conducted a chart review of 2,200 youth ages 8–18 years (M = 10, SD = 3) who completed comprehensive autism evaluations and either received an ASD diagnosis ( n = 1101) or did not ( n = 1119). We analyzed parent and teacher ASRS forms using t -tests and receiver operating characteristic analyses. The DSM subscale scores were significantly higher in the ASD group, though effect sizes were small. Teacher ratings performed slightly better than parent ratings, but overall diagnostic accuracy was poor (AUC = 0.62). Findings suggest the ASRS may aid screening but has limited diagnostic utility in high base-rate clinical settings.

  • Open Access

    Latent Profile Analysis of Parent-Reported Mood Symptoms in Youth: A Person-Centered Approach to Classification

    by Eric A. Youngstrom, Yinuo Liu, Amanda J. Thompson, Kevin G. Stephenson, Jennifer Kogos Youngstrom, Robert L. Findling

  • Open Access

    Long-Term Stability of IQ Scores for Children With Neurodevelopmental Disabilities: Stable Global IQ But Unstable Index, Subtest, and Profile IQ Scores

    by Kevin G. Stephenson, Megan Norris, Soo Youn Kim, Nicholas C. C. Russell, Ann Levine, Caroline Murphy, Eric A. Youngstrom, Eric M. Butter

    Purpose: Intelligence testing is an important tool for clinicians to help in diagnosis and treatment planning for children with neurodevelopmental disabilities. Although interpreting IQ profiles is a common clinical practice, there are concerns regarding the validity and reliability of such scores. Previous research has shown relatively stable global measures of IQ, but less stable scores within index or subtest score. More research is needed to better understand the long-term stability of IQ scores within autistic and other neurodevelopmental populations who may have more instability in their scores over time. Methods: We tested the long-term stability of IQ scores using the same test over time (Stanford-Binet, Fifth Edition; SB-5) in a large clinical sample of 650 youth with autism ( n = 236) or other neurodevelopmental disabilities ( n = 414), over a period of 4 months to 11 years. In addition to IQ scores, we also tested consistency of scatter scores, overall profiles, and strengths and weaknesses using linear mixed effects models. Results: Results indicated overall consistency of FSIQ scores (ICC = 0.86), slightly less stability for VIQ and NVIQ, and low stability for abbreviated IQ as well as index and subtest scores. The consistency of cognitive profiles, scatter scores, and strengths and weaknesses was poor. Younger age was the best predictor of higher FSIQ instability. Conclusion: Long-term stability of IQ in neurodevelopmental disabilities appears similar to results from other studies with different clinical and nonclinical groups. Implications and recommendations are discussed.


  • Construct Validity of IQ in the Presence of Scatter: A Measurement Invariance Approach

    by Kevin G. Stephenson, Soo Youn Kim, Moira J. Wendel, Eric M. Butter, Eric A. Youngstrom

    Practitioners of IQ tests are often taught that patterns of scatter among IQ subtests can lead to noncohesive full-scale IQ (FSIQ) scores. This notion exists despite a lack of empirical support and evidence for the robustness of FSIQ in the presence of scatter. However, to date, no study has directly tested the measurement invariance of groups varying in the amount of subtest scatter. We aimed to use multigroup confirmatory factor analysis to test the measurement invariance of the Stanford-Binet Intelligence Scales, fifth Edition (SB-5) for individuals with high vs. low scatter in a clinical sample of 5,352 individuals (ages 2–22) referred for comprehensive evaluations due to concerns for neurodevelopmental conditions. There was overall evidence for non-invariance in our sample, suggesting a limited impact of subtest scatter on the construct validity of IQ as measured by the SB-5. In addition, the explained variance of specific factors was low, even for individuals with high scatter among subtests. This study provides additional support for the robustness of FSIQ, even in cases of high scatter.


  • Open Access

    Does WISC-V scatter matter when it comes to academic achievement in neurodevelopmental disorders? It depends

    by Soo Youn Kim, Ann Levine, Nicholas C. C. Russell, Eric M. Butter, Kevin G. Stephenson

    The interpretability of FSIQ is often believed to rely on variability across index scores, although studies show that FSIQ is the strongest predictor of academic achievement, regardless of scatter. This study investigated whether IQ scatter affects the predictive validity of FSIQ on academic achievement in youth with neurodevelopmental disorders. A retrospective chart review was conducted with patients aged 6-16 evaluated at a developmental assessment clinic between 3/2018 and 12/2022 (N = 76, 74% male, 76% White). Moderation analyses were conducted using the FSIQ as a predictor, index scatter as a moderator, and academic achievement factor as a dependent variable. Index scatter was a significant moderator for the predictive effect of FSIQ on broad achievement and broad written language. The Johnson-Neyman technique revealed that the effect of FSIQ on broad achievement was significant when the index scatter was ≤ 54 in our sample. The results are consistent with previous studies, as the robust predictive validity of FSIQ was overall demonstrated. However, this study also provides novel information suggesting that the predictive validity of the FSIQ for academic achievement in neurodevelopmental disorders may weaken as index score scatter increases, a relationship that became statistically nonsignificant within our sample at highly elevated levels.

  • Identifying Latent Cognitive Profiles in Autism Spectrum Disorder and Attention-Deficit/Hyperactivity Disorder Using the Stanford-Binet Intelligence Scales—5th Edition

    by Soo Youn Kim, Eric A. Youngstrom, Megan Norris, Ann Levine, Eric M. Butter, Kevin G. Stephenson

    There is limited information concerning the presence of empirically derived, person-centered latent cognitive profiles in youth with autism spectrum disorder (ASD) or attention-deficit/hyperactivity disorder (ADHD) and whether these profiles are diagnostically useful. The aim of this study was to identify empirically driven cognitive subgroups in youth with ASD or ADHD and examine predictors of those profiles. A retrospective chart review was conducted with patients seen at a developmental assessment clinic who were identified with ASD or ADHD aged 2 to 16 years ( n = 1,679, M age = 8.4, SD age = 3.1). A Latent Profile Analysis with Stanford-Binet–Fifth Edition composites resulted in 14 profiles, which were roughly parallel to each other across various levels of cognitive functioning. Several profiles were characterized by a relatively large discrepancy between the Nonverbal IQ and Verbal IQ. Younger age and higher IQ were significant predictors of those with scattered profiles, whereas diagnoses (i.e., ASD or ADHD), sex, and emotional–behavioral functioning were not.