(ADHD)/Oppositional Defiant Disorder (ODD) Symptoms and First-Order Latent Structure of Sluggish Cognitive Tempo Symptoms

Bifactor Latent Structure of Attention-Deficit/Hyperactivity Disorder (ADHD)/Oppositional Defiant Disorder (ODD) Symptoms and First-Order Latent Structure of Sluggish Cognitive Tempo Symptoms

SoYean Lee G. Leonard Burns Sookmyung Women’s University Washington State University

Theodore P. Beauchaine Stephen P. Becker The Ohio State University Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio

The objective was to determine if the latent structure of attention-deficit/hyperactivity disorder (ADHD) and oppositional defiant disorder (ODD) symptoms is best explained by a general disruptive behavior factor along with specific inattention (IN), hyperactivity/impulsivity (HI), and ODD factors (a bifactor model) whereas the latent structure of sluggish cognitive tempo (SCT) symptoms is best explained by a first-order factor independent of the bifactor model of ADHD/ODD. Parents’ (n ϭ 703) and teachers’ (n ϭ 366) ratings of SCT, ADHD-IN, ADHD-HI, and ODD symptoms on the Child and Adolescent Disruptive Behavior Inventory (CADBI) in a community sample of children (ages 5–13; 55% girls) were used to evaluate 4 models of symptom organization. Results indicated that a bifactor model of ADHD/ODD symptoms, in conjunction with a separate first-order SCT factor, was the best model for both parent and teacher ratings. The first-order SCT factor showed discriminant validity with the general disruptive behavior and specific IN factors in the bifactor model. In addition, higher scores on the SCT factor predicted greater academic and social impairment, even after controlling for the general disruptive behavior and 3 specific factors. Consistent with predictions from the trait-impulsivity etiological model of externalizing liability, a single, general disruptive behavior factor accounted for nearly all common variance in ADHD/ODD symptoms, whereas SCT symptoms represented a factor different from the general disruptive behavior and specific IN factor. These results provide additional support for distinguishing between SCT and ADHD-IN. The study also demonstrates how etiological models can be used to predict specific latent structures of symptom organization.

Keywords: attention-deficit/hyperactivity disorder, sluggish cognitive tempo, oppositional defiant disorder, bifactor models, trait-impulsivity etiological model of ADHD

Recent studies have applied a bifactor latent model to attention- factor accounts for common variance among ADHD symptoms, deficit/hyperactivity disorder (ADHD) symptoms. This model in- and is independent of specific IN and HI factors, whereas specific cludes a general ADHD factor, along with specific inattention (IN) IN and HI factors account for common variance in their respective and hyperactivity/impulsivity (HI) factors. The general ADHD symptoms independent of the general ADHD factor. The bifactor model consistently yields better fit than the two correlated factors (IN and HI) model (e.g., Dumenci, McConaughy, & Achenbach, 2004; Gibbins, Toplak, Flora, Weiss, & Tannock, 2012; Gomez, This document is copyrighted by the American Psychological Association or one of its allied publishers. This article was published Online First October 26, 2015. 2014; Martel, von Eye, & Nigg, 2012; Normand, Flora, Toplak, &

This article is intended solely for the personal use of the individual userSoYean and is not to be disseminated broadly. Lee, Department of Child Welfare and Studies, Sookmyung Tannock, 2012; Toplak et al., 2012; Wagner et al., 2015; Wil- Women’s University; G. Leonard Burns, Department of Psychology, Washington State University; Theodore P. Beauchaine, Department of loughby, Blanton, & the Family Life Project Investigators, 2015). Psychology, The Ohio State University; Stephen P. Becker, Division of Although the bifactor model always fit better than the correlated Behavioral Medicine and Clinical Psychology, Cincinnati Children’s Hos- two-factor model, the specific IN and HI dimensions from the pital Medical Center, Cincinnati, Ohio. bifactor model accounted for little true score variance, over-and- We would like to acknowledge the support of the parents, teachers, and above the general ADHD factor. For example, omega-hierarchical principals, Heather Lang, Don Lee, Samantha Ogden, and Eric Price, of the ␻ ( h) values for the specific IN and HI factors in Wagner et al. Clarkston Washington elementary schools whose assistance made the (2015) were .33 and .08, respectively, with the values from Wil- research possible. G. Leonard Burns would like to acknowledge the sup- loughby et al. (2015) being .25 and .10, respectively. In contrast, port of Jerry Snell. SoYean Lee received support from the Sookmyung ␻ Women’s University Research Grant (1-1309-0012). h values for the general ADHD factor were .86 and .83 in these Correspondence concerning this article should be addressed to G. Leon- two studies, respectively. These results suggest that the specific IN ard Burns, Department of Psychology, Washington State University, John- and HI factors contain too little true score variance to be useful as son Tower 233, Pullman, WA 99164-4820. E-mail: [email protected] specific measures of the IN and HI dimensions independent of the

917 918 LEE, BURNS, BEAUCHAINE, AND BECKER

general ADHD dimension (Burns, Moura, Beauchaine, & McBur- Although one study has applied the trait-impulsivity model to nett, 2014; Wagner et al., 2015). predict bifactor latent structure of ADHD/ODD symptoms, no Findings of better fit for a bifactor structural model are consis- study has applied this etiological model of externalizing liability to tent with the trait-impulsivity etiological theory of ADHD (Beau- model the latent structure of ADHD, ODD, and SCT symptoms. chaine, 2015; Beauchaine, Hinshaw, & Pang, 2010; Beauchaine & The trait-impulsivity model posits that SCT arises from different McNulty, 2013). This etiological model predicts that a general neural substrates from ADHD and ODD, and therefore should not impulsivity factor should account for most of the variance in load on the general disruptive behavior factor with ADHD/ODD ADHD symptoms but has much broader implications for bifactor symptoms. In contrast to the fronto-striatal dysfunction and reward models of externalizing symptom structure than simple application seeking associated with these disorders, SCT may be characterized to ADHD. The trait-impulsivity etiological model hypothesizes by fronto-parietal dysfunction and problems with motivation that common behavioral genetic and neural vulnerabilities underlie (Beauchaine, 2015). all externalizing syndromes, including ADHD, oppositional defi- Below we provide further details on the trait-impulsivity etio- ant disorder (ODD), and conduct disorder (CD) symptoms. At the logical model of ADHD-IN, ADHD-HI, ODD, CD, and SCT neurobiological level, trait impulsivity is conferred by mesolimbic symptoms, followed by predictions from this model for the latent dopamine (DA) dysfunction, as manifested in blunted striatal structure of these symptoms. We then introduce a study to test responding to incentives among those with externalizing behavior these predictions. Our purposes are to provide a more sophisticated disorders (Beauchaine & McNulty, 2013; Gatzke-Kopp et al., evaluation of the independence of ADHD-IN and SCT symptom 2009; Plichta & Scheres, 2014). Affected individuals likely engage dimensions, and test a series of specific predictions from the in impulsive, reward-seeking behaviors to upregulate their under- trait-impulsivity etiological model for the latent structure of responsive mesolimbic DA systems (Beauchaine & Gatzke-Kopp, ADHD, ODD, and SCT symptoms. 2012), which give rise to negative affectivity, reduced hedonic capacity, and trait irritability (Laakso et al., 2003)—core charac- Trait-Impulsivity Etiological Model of ADHD/ODD teristics of externalizing behavior disorders. However, although several studies have examined the bifactor structure of ADHD (see The trait-impulsivity etiological model of ADHD (Beauchaine, above), only one study to date has applied a bifactor latent model 2015; Beauchaine et al., 2010; Beauchaine & McNulty, 2013) to ADHD and ODD symptoms simultaneously (Burns, Moura et assumes that low tonic and low phasic responding in primarily one al., 2014, described in more detail below). neural system—the mesolimbic reward system, or “bottom-up” As investigators have examined whether ADHD best fits within pathway—results in development of HI symptoms early in life a bifactor structure, a largely separate line of research has exam- (Gatzke-Kopp et al., 2009), with IN symptoms arising secondarily. ined whether sluggish cognitive tempo (SCT) symptoms are dis- This neurodevelopmental vulnerability—expressed behaviorally tinct from ADHD-IN symptoms. SCT is characterized by incon- as ADHD-HI or ADHD combined presentations—increases the sistent alertness (e.g., absented minded, alertness fluctuates, likelihood of developing of ODD and CD in high-risk environ- daydreams, loses train of thought, easily confused) and slowness ments characterized by coercive family processes, deviant peer (e.g., thinking is slow, behavior is slow, and drowsiness). Recent group affiliations, and neighborhood violence and criminality psychometric studies have identified a core group of SCT symp- (Beauchaine & McNulty, 2013). This model extends early dopa- toms (i.e., substantial loadings on the SCT factor) that also have minergic theories of impulsivity, including those articulated by discriminant validity with ADHD-IN, anxiety, and depression fac- Gray (e.g., Gray, 1987), Quay (1993), and others (e.g., Beau- tors (Barkley, 2013; Lee, Burns, Snell, & McBurnett, 2014; Will- chaine, 2001). A second “top-down” pathway involves dysfunc- cutt et al., 2014). Research also shows that the SCT dimension is tion in parietal and frontal structures (i.e., the prefrontal cortex, associated with other symptom and impairment dimensions even anterior cingulate). This dysfunction results in the direct develop- after controlling for ADHD-IN. For example, higher levels of SCT ment of IN symptoms with a much lower likelihood of developing predict higher levels of depression, academic impairment, and HI, ODD, and CD symptoms, and low levels of trait-impulsivity. social impairment over-and-above ADHD-IN. In addition, SCT Burns, Moura et al. (2014) tested predictions derived from shows a negative (or a nonsignificant) relationship with external- trait-impulsivity theory by applying a bifactor model to ADHD/ izing problem after controlling for ADHD-IN, whereas ADHD-IN ODD symptoms. Predictions were that the general disruptive be- This document is copyrighted by the American Psychological Association or one of its allied publishers. shows a positive relationship with externalizing problems after havior factor would explain most of the common variance in This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. controlling for SCT (see Barkley, 2013; Becker, Luebbe, Fite, ADHD/ODD symptoms, with the specific HI factor accounting for Stoppelbein, & Greening, 2014; Bernad, Servera, Grases, Collado, no common variance in HI symptoms above-and-beyond the gen- & Burns, 2014; Burns, Servera, Bernad, Carrillo, & Cardo, 2013; eral disruptive behavior factor. It was further predicted that spe- Lee et al., 2014; Penny, Waschbusch, Klein, Corkum, & Eskes, cific IN and ODD factors would account for only a small amount 2009; Servera, Bernad, Carrillo, Collado, & Burns, 2015; Willcutt of common variance above-and-beyond the general disruptive et al., 2014). Finally, although mixed findings have been reported behavior factor. Results were consistent with these predictions, in terms of SCT’s relation to IQ (e.g., Willcutt et al., 2014), some with the general disruptive behavior factor accounting for 69% to studies have found SCT to be unassociated with intelligence 71% of common variance, the specific HI factor accounting for (Becker & Langberg, 2014) or even associated with higher levels 4–6%, the specific IN factor accounting for 14%, and the specific of intelligence in children with ADHD (Marshall, Evans, Eiraldi, ODD factor accounting for 11%. The same results were found for Becker, & Power, 2014). SCT symptoms appear not attributable to mothers’ and fathers’ ratings, for the total sample, for boys and low intelligence. These various findings point to the importance of girls, and for younger and older children (Burns, Moura et al., evaluating the SCT construct in its own right. 2014). STRUCTURE OF ADHD, ODD, AND SCT SYMPTOMS 919

Because vulnerability to development of SCT symptoms is This model assumes that a general disruptive behavior factor believed to be associated with different neural mechanisms than underlies all symptoms, with IN/SCT symptoms best represented vulnerability to ADHD-HI and ADHD-IN symptoms (IN symp- by a single specific attention factor (i.e., SCT and IN do not toms secondary to HI symptoms), a model that includes SCT as represent separate specific attention factors in this model). Model a separate factor that does not load on a general disruptive 3 again represents the application of a bifactor model to all behavior factor should provide a better fit than a model that symptoms. Although this model assumes that a general disruptive assumes the general disruptive behavior factor underlies SCT behavior factor underlies all symptoms, here SCT and IN represent symptoms. Thus, predictions from trait-impulsivity theory have different specific attention factors. implications for the latent structure of the SCT, ADHD, and Model 4 includes a general disruptive behavior factor underly- ODD symptoms. In the next section we present four different ing ADHD/ODD symptoms, along with specific IN, HI, and ODD measurement models for the structure of these symptoms, along factors, and a first-order SCT factor. In this model, the general with a rationale for why one of these models best matches disruptive behavior factor does not underlie SCT symptoms. The predictions afforded by trait-impulsivity theory. model includes a correlation between the first-order SCT factor and the general disruptive behavior and the specific IN factors to Alternative Measurement Models of ADHD, ODD, and evaluate the discriminant validity of the SCT factor from the SCT Symptoms general disruptive behavior and specific IN factors. If the correla- Figure 1 shows four models for the latent structure of ADHD- tion of the SCT factor with either of these two factors is too large IN, ADHD-HI, ODD, and SCT symptoms. Model 1 represents a (Ͼ.85; Brown, 2006, p. 131), it would argue against Model 4 even SCT, IN, HI, and ODD correlated first-order model with separate if Model 4 had better fit than the other three models. If Model 4 has factors for each dimension. With the exception of Garner et al. the best fit and these correlations are moderate, SCT would appear (2014) described in more detail below, this correlated first-order to represent a different factor from the general disruptive behavior model is the model used in all previous studies evaluating the and specific IN factors. independence of the SCT and ADHD-IN factors (e.g., Barkley, 2013; Becker, Langberg, Luebbe, Dvorsky, & Flannery, 2014; Bernad et al., 2014; Burns et al., 2013; Lee et al., 2014; Penny et Predictions From Trait-Impulsivity Etiological Model al., 2009; Severa et al., 2015; Willcutt et al., 2014). These studies for Symptom Structure find support for distinct SCT and ADHD-IN first-order correlated factors with different external correlates. The trait-impulsivity etiological model of externalizing behavior Model 2 represents a bifactor model with a general disruptive (ADHD/ODD) predicts that Model 4 will provide the best fit, and behavior factor along with specific IN/SCT, HI, and ODD factors. also hypothesizes seven more specific outcomes:

Model 1 Model 2

SCT 1-8 IN 1-9 S-SCT/IN

SCT IN HI ODD S-HI G-DB HI 1-9

SCT IN HI ODD ODD 1-9 S-ODD 1-8 1-9 1-9 1-8

Model 3 Model 4

SCT S-SCT SCT SCT This document is copyrighted by the American Psychological Association or one of its allied publishers. 1-8 1-8 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. IN S-IN IN S-IN 1-9 1-9 G-DB G-DB HI S-HI HI S-HI 1-9 1-9

ODD S-ODD ODD S-ODD 1-8 1-8

Figure 1. Models of SCT, ADHD-IN, ADHD-HI, and ODD symptom organization. Model 1: First order correlated factor model of SCT, ADHD-IN, ADHD-HI, and ODD symptoms; Model 2: Bifactor model of the SCT/ADHD-IN, ADHD-HI, and ODD symptoms; Model 3: Bifactor model of the SCT, ADHD-IN, ADHD-HI, and ODD symptoms; and Model 4: First-order SCT and bifactor model of the ADHD-IN, ADHD-HI, and ODD symptoms. G ϭ general; S ϭ specific; SCT ϭ sluggish cognitive tempo; ADHD ϭ attention-deficit/ hyperactivity disorder; IN ϭ inattention; HI ϭ hyperactivity/impulsivity; ODD ϭ oppositional defiant disorder; DB ϭ disruptive behavior. 920 LEE, BURNS, BEAUCHAINE, AND BECKER

Hypothesis 1 conjunction with smaller AIC, BIC, and RMSEA values), a comparison of the loadings of SCT and IN symptoms on the general The first prediction concerns common variance associated with disruptive behavior in Models 2 and 3 allows an additional test of the general disruptive behavior and specific HI, IN, and ODD predictions from trait-impulsivity theory. Trait-impulsivity theory factors. Most of the common variance in ADHD/ODD symptoms predicts that loadings of IN symptoms on the general disruptive should be accounted for by a general disruptive behavior factor behavior factor will be substantially larger than loadings of SCT Ͼ ( 70%) with the specific HI factor accounting for no common symptoms on the general disruptive behavior factor in Models 2 variance above that accounted for by the general disruptive behav- and 3. ior factor (Ͻ5%), and the specific IN and specific ODD factors accounting for only a small amount of common variance above the general disruptive behavior factor (Ͻ15%) (Burns, Moura et al., Hypothesis 6 2014). Such results would indicate that nearly all-common vari- A sixth prediction involves SCT’s ability to predict academic ance in ADHD/ODD symptoms is accounted for by the general and social impairment, after controlling for the general disruptive disruptive behavior factor. behavior and specific IN, HI, and ODD factors. If SCT represents an important clinical dimension, then higher scores on SCT should Hypothesis 2 predict higher levels of academic and social impairment, even after The second prediction concerns latent factor reliability coeffi- controlling for the general and specific factors. cients for factors in the bifactor part of Model 4, more specifically ␻ the omega-hierarchical ( h) values for the total ADHD/ODD Hypothesis 7 scores, along with scores on the IN, HI, and ODD scales (McDon- ␻ The seventh prediction is that Model 4 will be the best model for ald, 1999; Zinbarg et al., 2006). The h value for the total disruptive behavior scores is expected to remain high (Ն.85), even after boys and girls in separate analyses and that this model will show accounting for true score variance associated with HI, IN, and measurement invariance (i.e., like-item loadings and like-item ODD scores. In contrast, HI scores are expected to contain nearly thresholds) across boys and girls. no true score variance, over-and-above the general disruptive behavior factor, with IN and ODD scores containing only a modest Summary amount of true score variance over-and-above the general disruptive behavior factor (ϳ30%) (Burns, Moura et al., 2014). Such At this time only one study has applied a bifactor model to results would indicate that all the content of HI scores, along with ADHD and SCT symptoms (Garner et al., 2014). Although Garner most of the content of IN and ODD scores, reflects the general et al. provided preliminary support for an SCT factor that was not disruptive behavior factor and not specific content of the three part of the bifactor structure of ADHD symptoms; there were individual scales. SCT scale scores are expected to contain ap- several limitations that the current study addresses. First, Garner et proximately 90% true score variance. al. used a small set of items (three items for parents ratings, four items for teacher ratings) drawn from the Child Behavior Check- Hypothesis 3 list/Teacher’s Report Form. In the current study, we used a recently validated parent- and teacher-report rating scale developed ADHD/ODD symptoms are predicted to have substantial load- to assess SCT specifically (Lee et al., 2014). Second, Garner et al. ings (Ͼ.70) on the general disruptive behavior factor, with HI evaluated only ADHD and SCT symptoms. In the current study, symptoms predicted to have low loadings (Ͻ.30) on the specific we examine SCT and ADHD as well as ODD symptoms to more HI factor. ADHD-IN and ODD symptoms are expected to have thoroughly evaluate ADHD and ODD as part of a more general moderate loading on their respective specific factors (ϳ.50). SCT externalizing liability. Third, Garner et al. used an ADHD-referred symptoms are expected to have substantial loadings on the first- sample (168 children, with 92% meeting diagnostic criteria for order SCT factor (Ͼ.70). ADHD). Because examining SCT within ADHD-defined samples may obscure findings in studies that examine how SCT and ADHD Hypothesis 4 are differentiated (Barkley, 2014), we tested our hypotheses in a This document is copyrighted by the American Psychological Association or one of its allied publishers. community sample. If Model 4 results in the best fit, and support This article is intended solely for the personal use of the individual userThe and is not to be disseminatedfourth broadly. prediction concerns the correlation of the SCT factor is obtained for the seven hypotheses, it would indicate that al- with the general disruptive behavior and specific IN factors. The though a general disruptive behavior factor underlies ADHD/ODD magnitude of these correlations will indicate that SCT is disso- symptoms, this factor does not underlie SCT symptoms, providing ciable from the general disruptive behavior and specific IN factors evidence that the SCT dimension is different from a general (correlations Ͻ .85). As noted earlier, if either of these correlations disruptive behavior dimension and a specific IN dimension. Such is greater than .85, this would indicate that the SCT factor failed to findings would have important implications for the Research Do- show discriminant validity with the other factor, suggesting it is main Criteria (RDoC) initiative, which seeks to identify both not a separate construct from the general disruptive behavior or transdiagnostic and unique contributions to different forms of specific IN factor. psychopathology across all relevant levels of analysis, from genes Hypothesis 5 to behavior. Studies such as ours, which dovetail neuroscience- informed theoretical models of psychopathology with factor ana- Even if bifactor Model 4 provides a better fit than bifactor lytic accounts of symptoms, will likely advance our understanding Models 2 and 3 at the global level (larger CFI and TLI values in of etiology, which often facilitates advances in treatment (see STRUCTURE OF ADHD, ODD, AND SCT SYMPTOMS 921

Beauchaine, 2015; Beauchaine, Neuhaus, Brenner, & Gatzke- Table 1 Kopp, 2008), and it has been posited that SCT may itself be Correlations Among the SCT, ADHD-IN, ADHD-HI, ODD, conceptualized as transdiagnostic in nature (Becker, Ciesielski et Academic Impairment, and Social Impairment CADBI Subscales al., 2014; Becker, Marshall, & McBurnett, 2014). for Parent (Above the Diagonal) and Teacher (Below the Diagonal) Ratings

Method 123456

1. SCT — .75 .53 .44 .41 a Participants and Procedures 2. ADHD-IN .74 — .74 .61 .43 a 3. ADHD-HI .35 .71 — .69 .32 a Parent ratings. With approval from the school district and 4. ODD .25 .57 .75 — .24 a a Washington State University’s IRB, all 1,356 children from the 5. AI .65 .67 .35 .30 — 6. SI .45 .53 .44 .53 .52 — four elementary schools in the district were given the cover letter and rating scale to give to their parents. Participation was volun- Note. SCT ϭ sluggish cognitive tempo; ADHD ϭ attention-deficit/ hyperactivity disorder; IN ϭ inattention; HI ϭ hyperactivity/impulsivity; tary and anonymous. Scales were returned from 703 children ϭ ϭ ϭ ϭ ϭ ODD oppositional defiant disorder; AI academic impairment; SI (55.2% girls), ages 5–13 years (M 8.29, SD 2.04). Mothers social impairment. All correlations significant at p Ͻ .001. completed 81.6% of the scales, fathers 11.7%, and other family a The social impairment subscale was not administered to parents. members 6.7%. The race of the children in the four elementary schools was 2% American Indian or Alaskan Native, 1% Asian or Pacific Islander, 2% Black, 6% Hispanic, and 89% White. Sixty- The eight SCT items were rated on a 6-point duration of occur- three percent of children in the four schools received free/reduced rence scale for the past month ranging from 1 (nearly none of the priced meals, thus indicating many low-income families in the time)to6(nearly all of the time) on parent and teacher scales. For district. Ethnicity and free/reduced meal information were not the teacher scale, the ADHD and ODD items were rated on an available for individual children. 8-point frequency of occurrence scale for the past month ranging Teacher ratings. The teacher ratings were collected in the from 1 (never in the past month)to8(10 or more times per day). same four elementary schools the year prior to the parent ratings. For the parent scale, ADHD and ODD items were rated on a With school district and IRB approval, 58 teachers were invited to 6-point frequency of occurrence scale for the past month ranging participate, with 46 participating teachers each rating eight stu- from 1 (nearly occurs none of the time; e.g., 2 or fewer times in the dents randomly selected from their class list by the researchers past month)to6(nearly occurs all the time; e.g., many times per (one teacher rated six students). Teachers provided ratings for a day). For the parent and teacher scales, academic and social ϭ ϭ total of 366 children (56% girls; Mage 8.23, SD 2.08). Given impairment items were rated on a 7-point scale ranging from 1 that teachers and parents completed the measures in different (severe difficulty)to7(excellent performance). All of the teachers years, and the ratings were anonymous, it was not possible to link had been interacting with the students for a minimum of 6 months. ratings. Parents were asked to rate their child’s behaviors in the home/ community and to not consider behavior at school since parents Measure could not directly observe such behavior in the classroom. Teach- ers were asked to rate each child’s behaviors in the classroom. Parents and teachers completed the Child and Adolescent Dis- Earlier studies provide support for the reliability (true score vari- ruptive Behavior Inventory (CADBI; Burns & Lee, 2011). The ance, test-retest, and interrater) and validity of scores from the SCT (eight items), ADHD-IN (nine items), ADHD-HI (nine symptom and impairment scales of the CADBI (e.g., Bernad et al., items), ODD toward adults (eight items; e.g., argues with adults), 2014; Burns et al., 2008, 2013, 2014; Lee et al., 2014; Servera et ODD toward peers (eight items; e.g., argues with peers), academic al., 2015). For example, for mothers and fathers, interrater factor impairment (four items; homework, reading, arithmetic, and writ- correlations were .71, .83, .79, .70, and .86 for SCT, ADHD-IN, ing skills), and social impairment (teacher scale only, two items: ADHD-HI, ODD, respectively, and academic impairment with quality of interactions with adults and peers at school) scales were similar interrater factor correlations for teachers. The parent– This document is copyrighted by the American Psychological Association or one of its allied publishers. used in this study. The eight SCT symptoms were daydreams, teacher factor correlation for academic impairment varied from .55 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. alertness fluctuates, absent-minded, loses train of thought, easily to .67 in earlier studies. Twelve-month stability coefficients (test- confused, drowsy, slow thinking, and slow behavior (Burns et al., retest) varied from .52 (social impairment) to .78 (ADHD-IN). 2013, Table 1, shows the complete wording of the eight SCT For the current study, Cronbach’s alphas for teachers (parents) symptoms). These eight SCT symptoms showed convergent valid- for SCT, ADHD-IN, ADHD-HI, ODD, academic impairment and ity (substantial loadings on the SCT factor) along with discrimi- social impairment scales were .93 (.90), .98 (.96), .97 (.95), .97 nant validity (significantly higher loadings on the SCT factor than (.95), .94 (.90), and .93 (no parent social impairment scale), the ADHD-IN factor, Lee et al., 2014). Earlier studies provide a respectively. One-month test–retest reliabilities for a subsample of more detailed description of the CADBI SCT scale (Burns et al., parent ratings (n ϭ 105) for SCT, ADHD-IN, ADHD-HI, ODD, 2013; Lee et al., 2014). The ADHD symptoms included examples and academic impairment were .80, .83, .85, .71, and .83, respec- specific to the home and school settings. The two ODD scales (i.e., tively. Table 1 shows Pearson correlations among the SCT, oppositionality toward adults and peers) were combined into a ADHD-IN, ADHD-HI, ODD, academic impairment, and social single ODD scale for this study. Wording of ADHD and ODD impairment subscales for parent (above the diagonal) and teacher symptoms was consistent with DSM–IV. (below the diagonal) ratings. 922 LEE, BURNS, BEAUCHAINE, AND BECKER

Table 2 Fit Statistics for Alternative Models of SCT, ADHD, and ODD Symptoms

Models df ␹2 CFI TLI RMSEA (90% CI) AIC BIC

Parents’ ratings (n ϭ 703) Model 1 521 1,859 .963 .961 .060 [.057, .063] 50,617 51,465 Model 2 493 2,467 .946 .938 .075 [.073, .078] 50,628 51,603 Model 3 493 2,400 .948 .941 .074 [.071, .077] 50,491 51,701 Model 4 499 1,512 .972 .969 .054 [.051, .057] 50,246 51,194

Teachers’ ratings (n ϭ 366) Model 1 521 1,216 .979 .978 .060 [.056, .065] 26,978 27,844 Model 2 493 964 .986 .984 .051 [.046, .056] 26,596 27,571 Model 3 493 1,933 .957 .951 .089 [.085, .094] 26,351 27,327 Model 4 499 928 .987 .986 .048 [.044, .053] 26,134 27,086 Note. Figure 1 shows the path diagrams for the four models. Bold entries indicate the best-fit value for each column. SCT ϭ sluggish cognitive tempo; ADHD ϭ attention-deficit/hyperactivity disorder; ODD ϭ oppositional defiant disorder; CFI ϭ comparative fit index; TLI ϭ Tucker Lewis Index; RMSEA ϭ root mean square error of approximation; CI ϭ confidene interval; AIC ϭ Akaike Information Criterion; BIC ϭ Bayesian Information Criterion.

Analytic Strategy and the three specific factors (i.e., “for each factor the ECV is the sum of the squared loadings for that factor divided by the sum of Items were treated as categorical indicators and robust weighted all squared factor loadings (the common variance) for the model;” least squares estimation was used to evaluate the four models Brouwer, Meijer, & Zevalkink, 2013, p. 139). Consistent with (WLSMV estimator in Mplus v.7.3, Muthén & Muthén, 1998– predictions from the trait-impulsivity etiological model of ADHD/ 2012). Global model fit was evaluated with the comparative fix ODD, for parents (teachers), common variance accounted for by Ն index (CFI; ideal study criterion .95), Tucker-Lewis index (TLI; the general disruptive behavior factor was 70% (69%). The spe- Ն ideal study criterion .95), and root mean square error of approx- cific HI factor accounted for 2% (3%), the specific IN factor for Յ imation (RMSEA; ideal study criterion .05). Covariance cover- 12% (13%), and the specific ODD factor for 16% (15%). Table 3 age was approximately 99% for parent and teachers. In addition, shows these values. These results are almost identical to the first because the WLSMV estimator does not provide information study that examined ADHD and ODD symptoms with bifactor criteria, all analyses were repeated with the MLR estimator (items modeling (Burns, Moura et al., 2014). still treated as categorical, thus a logit link) to obtain Akaike information criterion (AIC) and Bayesian information criterion (BIC). The model with the smallest AIC and BIC values was Hypothesis 2: Latent Factor Reliability Coefficients considered to have the best global fit. for Model 4 The four models (see Figure 1) do not represent nested models so it is not appropriate to use the chi-square difference test to Table 3 shows latent factor reliability coefficients (amount of ϩ determine if one model provides a better fit than another (Little, true score variance) for the total disruptive behavior (nine IN ϩ 2013, pp. 339–341). If Model 4 has the best global fit (i.e., best nine HI eight ODD symptoms), IN (nine IN symptoms), HI CFI, TLI, RMSEA, AIC, and BIC values) and support occurs for (nine HI symptoms), ODD (eight ODD symptoms), and SCT ␻ the seven hypotheses, then such outcomes would suggest that (eight SCT symptoms) scores. The values for IN, HI, and ODD Model 4 was the best of the four models. scores represent the amount of variance the IN, HI, and ODD scores accounted for by a blend of the general disruptive behavior Results factor and the relevant specific factor (e.g., because the ␻ values conflate the variance of the general and specific factors, the This document is copyrighted by the American Psychological Association or one of its allied publishers. ␻ Model Fit hierarchical values are more relevant for evaluating the subscale This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Table 2 shows fit measures for the four models. Model 1 estimated 186 parameters, Model 2 estimated 214, Model 3 esti- 1 Although a second-order model (i.e., the SCT, ADHD-IN, ADHD-HI, mated 213, and Model 4 estimated 208. As predicted, for both and ODD first-order factors and a disruptive behavior second-order factor) was not central to our evaluation of the trait-impulsivity model of exter- parent and teacher ratings, Model 4 was the best model (i.e., nalizing liability, we also evaluated such a model—that is, parents: smaller chi-square values, larger CFI and TLI values, and smaller ␹2(523) ϭ 2351, CFI ϭ .950, TLI ϭ .946, RMSEA ϭ .071 (.068-.073), RMSEA, AIC, and BIC values than the other models). We now AIC ϭ 50702, and BIC ϭ 51540; teachers: ␹2(523) ϭ 1869, CFI ϭ .960, evaluate the seven predictions for Model 4.1 TLI ϭ .958, RMSEA ϭ .084 (.080-.088), AIC ϭ 27292, and BIC ϭ 28138). The second-order model fit significantly worse (ps Ͻ .001) than the first-order model, thus indicating that the second-order disruptive Hypothesis 1: Common Variance for the General and behavior factor did not account for the covariation among the first-order Specific Factors in Model 4 factors. Research on a general (higher order) factor underling all psychopathology represents a different research question than the current study’s Expected common variance (ECV) is the percent of explained use of a bifactor model to test predictions from the trait-impulsivity variance accounted for by the general disruptive behavior factor etiological model of externalizing liability (see Caspi et al., 2014). STRUCTURE OF ADHD, ODD, AND SCT SYMPTOMS 923

Table 3 Standardized Loadings for Model 4—The SCT First-Order and The Bifactor Disruptive Behavior Model

Bifactor model of ADHD/ODD symptoms First-order SCT General DB Specific IN Specific HI Specific ODD Symptoms Parents Teachers Parents Teachers Parents Teachers Parents Teachers Parents Teachers

SCT symptoms Daydreams .77 .86 Alertness Fluctuates .81 .92 Absent-minded .83 .93 Loses train of thought .85 .91 Easily confused .83 .88 Drowsy .64 .70 Slow thinking .80 .80 Slow behavior .71 .75 ADHD-IN symptoms Close attention .64 .66 .55 .65 Sustaining attention .71 .78 .58 .57 Listen .76 .74 .42 .53 Follow through .74 .72 .51 .64 Organizational skills .69 .70 .57 .64 Concentration .67 .70 .57 .62 Loses things .68 .72 .49 .53 Easily distracted .77 .79 .48 .52 Forgetful .71 .79 .43 .48 ADHD-HI symptoms Fidgets/squirms .79 .93 .31 Ϫ.15 Leaves seat .89 .97 .36 Ϫ.14 Runs/climbs .89 .97 .19 Ϫ.06ns Playing quietly .88 .90 Ϫ.04ns .27 Talks excessively .91 .91 .14 .06 Driven/on the go .84 .86 Ϫ.20 .31 Blurts .84 .82 Ϫ.25 .47 Awaiting turn .88 .89 Ϫ.16 .35 Interrupts/intrudes .84 .88 Ϫ.26 .42 ODD symptoms Argues .64 .76 .54 .50 Loses temper .65 .67 .62 .66 Refuses .66 .72 .65 .58 Annoys .62 .65 .61 .63 Blames .64 .66 .60 .66 Gets annoyed .65 .68 .63 .63 Resentful .58 .59 .70 .74 Vindictive .52 .53 .71 .74 ECV (%) .70 .69 .12 .13 .02 .03 .16 .15 ␻ .90 .93 .98 .98 .96 .98 .96 .98 .97 .98 ␻ h total scale .87 .87 ␻ h subscale .38 .46 .02 .00 .46 .39 Note. These are the loadings from Model 4 in Figure 1. The standard errors for the loading varied from .01 to .04. SCT ϭ sluggish cognitive tempo; IN ϭ inattentive; ADHD ϭ attention-deficit/hyperactivity disorder; HI ϭ hyperactive/impulsive; ODD ϭ oppositional defiant disorder; DB ϭ disruptive ϭ ␻ϭ ␻ ϭ Ͻ behavior; ECV expected common variance; omega; h omega hierarchical. All loadings were significant (p .05) unless noted as nonsignificant

This document is copyrighted by the American Psychological Association or one of its allied( publishers. ns). This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

␻ ␻ scores, see next paragraph). For parents (teachers), values for specific factors. For parents (teachers) hierarchical values for IN, IN, HI, and ODD scores were .96 (.98), .96 (.98), and .97 (.98), HI, and ODD scores (after controlling for the general disruptive respectively. The ␻ values for the total disruptive behavior scale behavior factor) were .38 (.46), .02 (.00), and .46 (.39), respec- scores (blend of the general and three specific factors) were .98 for tively. These results are consistent with predictions from trait parents and .99 for teachers. The ␻ coefficient for the SCT scores impulsivity theory and are almost identical to the first application was .90 for parents and .93 for teachers. of the bifactor model to ADHD/ODD symptoms (Burns, Moura et ␻ The hierarchical values indicate the unique amount of variance in al., 2014). scores that was due to the general or specific factor. These values are more relevant for evaluation of factors in the bifactor part of Hypothesis 3: Symptom Factor Loadings for Model 4 Model 4 than the ␻ values. For total disruptive behavior scores for parents (teachers), the general disruptive behavior factor accounted Table 3 shows loadings from Model 4. All eight SCT symptoms for 87% (87%) of the variance after controlling for the three had substantial loadings on the first-order SCT factor (Ͼ.70). 924 LEE, BURNS, BEAUCHAINE, AND BECKER

Loadings of IN, HI, and ODD symptoms on the specific and Model 4 Factors general factors were consistent with predictions from the trait- impulsivity model. For example, loadings of these 52 symptoms SCT on the general disruptive behavior factor were substantial (all but four Ͼ.60 for parents and teachers) with loadings of the HI symptoms on the specific HI factor being low (all less than .47 G-DB with most close to zero). IN symptoms had moderate loading on the specific IN factor (.42–.58 for parents and .48–.65 for teachers) Academic with the ODD symptoms also having moderate loading on the S-IN Impairment specific ODD factor (.54–.71 for parents and .50–.74 for teachers). S-HI Hypothesis 4: Correlation of SCT Factor With General Disruptive Behavior and Specific ADHD-IN S-ODD Factors From Model 4 ϭ For parents, the first-order SCT factor correlated .59 (SE .03) Figure 2. Structural regression model. Each latent variable was defined with the general disruptive behavior factor and .52 (SE ϭ .03) with by manifest variables (not shown in the path diagram). The same structural the specific IN factor. For teachers, the first-order SCT factor regression model was used to regress social impairment on the four factors correlated .44 (SE ϭ .05) with the general disruptive behavior and from Model 4. The SCT factor is a first-order factor with the G-DB, S-IN, .72 (SE ϭ .04) with the specific IN factor. These correlations S-HI, and S-ODD being part of the bifactor model. G ϭ general; S ϭ indicate that first-order SCT showed discriminant validity with the specific; SCT ϭ sluggish cognitive tempo; IN ϭ inattention; HI ϭ hyper- ϭ ϭ general disruptive behavior and specific ADHD-IN factors. activity/impulsivity; ODD oppositional defiant disorder; DB disruptive behavior. Hypothesis 5: Loading of the SCT and ADHD-IN Symptoms on the General Disruptive Behavior Factor in Models 2 and 3 A structural regression analysis was used to regress the academic impairment factor on the first-order SCT factor, the general The average loading of SCT symptoms on the general disruptive disruptive behavior factor and specific IN, HI, and ODD factors behavior factor in Model 2 was much lower than the loadings of from Model 4 to determine each factor’s unique effect on aca- the IN symptoms on the general disruptive behavior factor (Par- demic impairment (see Figure 2). For parents and teachers, higher ents: SCT—M ϭ .46, SD ϭ .06, range ϭ .37–.53; IN—M ϭ .81, scores on SCT predicted higher levels of academic impairment, SD ϭ .04, range ϭ .74–.86; Teachers: SCT—M ϭ .32, SD ϭ .12, even after controlling for the other factors (ps Ͻ .05). The general range ϭ .13–.44; IN—M ϭ .76, SD ϭ .03, range ϭ .73–81). disruptive behavior factor, and the specific IN and specific HI Similar results occurred for Model 3. These results indicate the factors also uniquely predicted academic impairment (ps Ͻ .01). general disruptive behavior factor underlies IN symptoms more Table 4 shows these partial standardized regression coefficients. than SCT symptoms. For teachers’ ratings, the social impairment factor was regressed on the first-order SCT factor, the general disruptive behavior Hypothesis 6: First-Order and Unique Relationships of factor, and specific IN, HI, and ODD factors to determine each Factors From Model 4 With Academic and Social factor’s unique effect on social impairment. Higher levels of SCT Impairment predicted higher levels of social impairment, even after controlling for all the other factors (p ϭ .001). Higher scores on the general Higher scores on the SCT factor were associated with higher disruptive behavior factor and the specific ODD factor also pre- scores on the academic impairment factor (parents: r ϭ .46, SE ϭ dicted higher levels of social impairment, controlling for the other .03, p Ͻ .001; teachers: r ϭ .73, SE ϭ .03, p Ͻ .001) with Ͻ

This document is copyrighted by the American Psychological Association or one of its allied publishers. factors (ps .001). significant results occurring for general disruptive behavior (par- This article is intended solely for the personal use of the individual user and is not to be disseminatedϭ broadly. ϭ Ͻ ϭ ϭ Ͻ ents: r .35, SE .04, p .001; teachers: r .45, SE .05, p Hypothesis 7: Sex Effects .001), specific IN (parents: r ϭ .35, SE ϭ .04, p Ͻ .001; teachers: r ϭ .66, SE ϭ .04, p Ͻ .001), and specific HI (parents: r ϭ .12, The analyses for Models 1, 2, 3, and 4 were repeated for parents’ SE ϭ .05, p Ͻ .01; teachers: r ϭ .19, SE ϭ .05, p Ͻ .001). The ratings of boys and girls. The sample size for teacher ratings was specific ODD factor was not related to academic impairment too small for a separate evaluation of the models for boys and girls. (parents: r ϭ .02, SE ϭ .04, ns; teachers: r ϭ .01, SE ϭ .04, ns). For parent ratings, Model 4 was again the best model for boys and Higher scores on the social impairment factor (only available for girls. ECV and reliability values for boys and girls for were almost teacher ratings) were associated with higher scores on the SCT, identical to values reported for the total sample for Model 4, and r ϭ .51, SE ϭ .04, p Ͻ .001, general disruptive behavior, r ϭ .52, regression results yielded the same results as the total sample with SE ϭ .04, p Ͻ .001, specific IN, r ϭ .28, SE ϭ .05, p Ͻ .001, and one exception: For parent ratings of girls, the general disruptive specific ODD factor, r ϭ .31, SE ϭ .04, p Ͻ .001. The social behavior factor did not predict academic impairment after control- impairment factor was not associated with the specific HI factor ling for the first-order SCT factor and the three specific factors, (r ϭ .08, SE ϭ .05, ns). ␤ϭ.06, SE ϭ .09, ns. STRUCTURE OF ADHD, ODD, AND SCT SYMPTOMS 925

Table 4 factor. In addition, higher levels of SCT predicted higher levels of Regression of Academic Impairment and Social Impairment academic and social impairment, even after controlling for the Factors on the Model 4 Predictors general disruptive behavior along with three specific factors, align- ing with other recent studies linking SCT to impairment (Becker, Social 2013; Bernad et al., 2014; Severa et al., 2015). These results Academic impairment impairment provide additional support that SCT and ADHD-IN represent Predictors Parents ␤(SE) Teachers ␤(SE) Teachers ␤(SE) different dimensions, suggesting that SCT symptoms do not align First-Order SCT .19 (.06) .19 (.04) .26 (.08) with the externalizing spectrum of psychopathology (Becker et al., General DB .24 (.05) .37 (.05) .41 (.04) 2013), and are consistent with the trait-impulsivity model that SCT Specific ADHD-IN .25 (.05) .52 (.09) .10 (.09)ns and ADHD/ODD symptoms develop from different neural sub- ns Specific ADHD-HI .12 (.05) Ϫ.19 (.05) .08 (.05) strates. As outlined above, externalizing spectrum disorders, in- Specific ODD .02 (.04)ns .01 (.04)ns .31 (.04) cluding ADHD, ODD, and CD, share a common neural sub- Note. ␤ϭpartial standardized regression coefficient; SE ϭ standard strate—dysfunction on mesolimbic DA responding, whereas SCT error; SCT ϭ sluggish cognitive tempo; ADHD ϭ attention-deficit/ may be linked to fronto-parietal dysfunction (see, e.g., Beauchaine hyperactivity disorder; IN ϭ inattention; HI ϭ hyperactivity/impulsivity; ODD ϭ oppositional defiant disorder; DB ϭ disruptive behavior; DB ϭ & McNulty, 2013; Zisner & Beauchaine, in press). disruptive behavior. All coefficients were significant (p Ͻ .01) unless noted as nonsignificant (ns). Clinical Implications Our study demonstrates that it is possible to develop predic- Model 4 was also invariant across boys and girls for like-item tions about the latent structure of the SCT, ADHD, and ODD loadings and like-item thresholds (i.e., the CFI, TLI, and RMSEA symptoms from a specific etiological model, the trait- values all indicated improvement in fit going from the model with impulsivity theory of externalizing liability. To date, almost no no constraints to the model with the constraints on the loadings and confirmatory factor analytic studies of externalizing behavior thresholds). Fit of the model with constraints was good, have used etiological models to predict latent structure of ␹2(1,159) ϭ 1,844, CFI ϭ .981, TLI ϭ .982, and RMSEA ϭ .030 symptom sets in this manner. Continuation of such efforts (.025–.034). With girls coded 0 and boys coded 1, the latent might yield more useful diagnostic models that are grounded in Cohen’s d values for the SCT, general disruptive behavior, specific both neurobiology and sound psychometrics (e.g., the connec- IN, specific HI, and specific ODD factors were 0.21 (p Ͻ .05), tion of the trait-anxiety etiological model with the anxiety and 0.21 (p Ͻ .01), 0.00 (ns), 0.30 (p Ͻ .05), and Ϫ0.12 (p ϭ .26), depression dimension, Beauchaine, 2015). respectively. As noted above, this approach is consistent with objectives of the National Institute of Mental Health RDoC initiative, which Discussion seeks to characterize how genetic and neural vulnerabilities map onto transdiagnostic symptom dimensions in efforts to A bifactor model applied to ADHD-HI, ADHD-IN, and ODD better capture etiological substrates of psychopathology. It has symptoms, in conjunction with a separate first-order SCT factor, been suggested that SCT might ultimately fit within the RDoC provided the best model for the latent structure of these symptoms. framework (Becker, Ciesielski et al., 2014), and while specu- Of note, results were remarkably similar across parent and teacher lative in the absence of data, it is possible that SCT and ratings. The general disruptive behavior factor accounted for disruptive behavior both relate to the cognitive systems domain nearly all of the common variance in ADHD-HI symptoms, and with different cognitive constructs being most relevant (e.g., most of the common variance in ADHD-IN and ODD symptoms. working memory, cognitive control), and perhaps the arousal Our results also suggest that the ADHD-HI, ADHD-IN, and ODD and regulatory system domain (although, again, different reg- subscale scores contain too little true score variance to be viewed ulatory processes may be implicated). SCT and disruptive be- as specific measures of these constructs, independent of the gen- havior may further part ways in that SCT may be more related eral disruptive behavior factor (e.g., Reise et al., 2013, recommend to negative valence systems whereas disruptive behavior may ␻ a minimum h of .50 for specific factors in the bifactor model to be more related to positive valence systems. Research is needed This document is copyrighted by the American Psychological Association or one of its allied publishers. be useful with our values for specific HI, IN and ODD being much to evaluate these hypotheses in a more formal manner, as well This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. lower than .50). In other words, clinical interpretations of ADHD- as by evaluating multiple constructs across the various units of HI, ADHD-IN, and ODD subscale scores must keep in mind that analysis within the RDoC framework (e.g., genes, circuits, most of the true score variance in these three subscales is ac- physiology, behavior). Such research would not only advance counted for by the general disruptive behavior factor. These out- our understanding of the structure and etiology of psychopa- comes replicate findings of Burns, Moura et al. (2014) with par- thology but would also be highly useful for informing our ents, extend findings to teachers, and are consistent with understanding of the heterogeneity within psychopathologies. predictions from the trait-impulsivity etiological model of ADHD Our findings also have implications for the organization ADHD and ODD (Beauchaine, 2015; Beauchaine et al., 2010; Beauchaine and ODD in the DSM–5 (American Psychiatric Association, & McNulty, 2013). 2013). In DSM–5, ADHD is in the general neurodevelopmental Another important contribution of this study is demonstration disorders category, whereas ODD is in the general disruptive, that a general disruptive behavior factor does not underlie SCT impulse-control, and conduct disorders category. Placement of symptoms. The SCT factor showed discriminant validity with the ADHD and ODD in such different categories seems to imply general disruptive behavior factor and, importantly, the specific IN different causal processes. However, as suggested by theoretical 926 LEE, BURNS, BEAUCHAINE, AND BECKER

models (Beauchaine, 2015), behavioral genetics (e.g., Tuvblad et Barkley, R. A. (2014). Sluggish cognitive tempo (concentration deficit al., 2009), and the outcome of this study and the earlier one (Burns, disorder?): Current status, future directions, and a plea to change the Moura et al., 2014), creation of categories without careful consid- name. Journal of Abnormal Child Psychology, 42, 117–125. http://dx eration of etiological models may be problematic. Linkages of .doi.org/10.1007/s10802-013-9824-y specific etiological models with specific symptom dimensions will Beauchaine, T. (2001). Vagal tone, development, and Gray’s motivational likely prove more useful in furthering our understanding of psy- theory: Toward an integrated model of autonomic nervous system func- chopathology than the categorical approach. tioning in psychopathology. Development and Psychopathology, 13, 183–214. http://dx.doi.org/10.1017/S0954579401002012 Beauchaine, T. P. (2015). Future directions in emotion dysregulation and Limitations youth psychopathology. Journal of Clinical Child and Adolescent Psy- chology, 44, 875–896. http://dx.doi.org/10.1080/15374416.2015 The use of a single rating scale and only two measures of .1038827 impairment represent limitations. A multiple traits (general, spe- Beauchaine, T. P., & Gatzke-Kopp, L. M. (2012). Instantiating the multiple cific, and first order factors) by multiple sources (mothers, fathers, levels of analysis perspective in a program of study on externalizing teachers) by multiple methods (rating scales and interviews) by behavior. Development and Psychopathology, 24, 1003–1018. http://dx multiple occasions (repeated assessments) design would provide a .doi.org/10.1017/S0954579412000508 stronger test of Model 4 in addition to a larger community sample Beauchaine, T. P., Hinshaw, S. P., & Pang, K. L. (2010). Comorbidity of with a wider age range. Such a design would allow for separation attention-deficit/hyperactivity disorder and early-onset conduct disorder: of trait effects from method and source effects, and would provide Biological, environmental, and developmental mechanisms. Clinical for a better identification of common and unique external corre- Psychology: Science and Practice, 17, 327–336. http://dx.doi.org/10 lates of factors (e.g., negative emotionality, executive functions, .1111/j.1468-2850.2010.01224.x peer relationship problems) and how these correlates change Beauchaine, T. P., & McNulty, T. (2013). Comorbidities and continuities across time. It would also be more ideal to have each teacher only as ontogenic processes: Toward a developmental spectrum model of rate a single child’s behavior. externalizing psychopathology. Development and Psychopathology, 25, Other studies are needed to further evaluate other domains of 1505–1528. http://dx.doi.org/10.1017/S0954579413000746 Beauchaine, T. P., Neuhaus, E., Brenner, S. L., & Gatzke-Kopp, L. (2008). adjustment recently linked to SCT symptoms such as peer Ten good reasons to consider biological processes in prevention and functioning (Becker, 2014), emotion dysregulation (Flannery et intervention research. Development and Psychopathology, 20, 745–774. al., 2014), and internalizing symptoms (Becker, Luebbe et al., http://dx.doi.org/10.1017/S0954579408000369 2014; Bernad et al., 2014; Servera et al., 2015) within Model 4. Becker, S. P. (2013). Sluggish cognitive tempo: Research findings and Longitudinal studies should also evaluate the structure of dis- relevance for pediatric psychology. Journal of Pediatric Psychology, 38, ruptive behavior and SCT across time. Finally, inclusion of 1051–1057. http://dx.doi.org/10.1093/jpepsy/jst058 measures of anxiety and depression in addition to measures of Becker, S. P. (2014). Sluggish cognitive tempo and peer functioning in SCT, ADHD, and ODD symptoms within the same study would school-aged children: A six-month longitudinal study. Psychiatry Re- allow simultaneously evaluation of the trait-impulsivity and search, 217, 72–78. http://dx.doi.org/10.1016/j.psychres.2014.02.007 trait-anxiety etiological models of youth psychopathology Becker, S. P., Ciesielski, H. A., Rood, J. E., Froehlich, T. E., Garner, A. A., (Beauchaine, 2015). Such a study would provide an understand- Tamm, L., & Epstein, J. N. (2014). Uncovering a clinical portrait of sluggish ing of SCT’s relationship with the trait-anxiety etiological cognitive tempo within an evaluation for attention-deficit/hyperactivity dis- model in addition to the trait-impulsivity model, thus allowing order: A case study. Clinical Child Psychology and Psychiatry. Advance a more sophisticated evaluation of the etiology of SCT than the online publication. http://dx.doi.org/10.1177/1359104514554312 current study. Becker, S. P., Fite, P. J., Garner, A. A., Stoppelbein, L., Greening, L., & Luebbe, A. M. (2013). Reward and punishment sensitivity are differen- tially associated with ADHD and sluggish cognitive tempo symptoms in Summary children. Journal of Research in Personality, 47, 719–727. http://dx.doi .org/10.1016/j.jrp.2013.07.001 Consistent with predictions from the trait-impulsivity etio- Becker, S. P., & Langberg, J. M. (2014). Attention-deficit/hyperactivity logical model of externalizing liability, a single, general dis- disorder and sluggish cognitive tempo dimensions in relation to execu- ruptive behavior factor accounted for nearly all of the common tive functioning in adolescents with ADHD. Child Psychiatry and Hu- This document is copyrighted by the American Psychological Association or one of its alliedvariance publishers. in ADHD-HI symptoms and most of the common man Development, 45, 1–11. http://dx.doi.org/10.1007/s10578-013- This article is intended solely for the personal use ofvariance the individual user and is not to be disseminated broadly. in ADHD-IN and ODD symptoms, whereas SCT 0372-z symptoms represented a factor different from the general dis- Becker, S. P., Langberg, J. M., Luebbe, A. M., Dvorsky, M. R., & ruptive behavior and specific IN factors. These results provide Flannery, A. J. (2014). Sluggish cognitive tempo is associated with additional support for SCT and ADHD-IN representing differ- academic functioning and internalizing symptoms in college students ent symptom dimensions. with and without attention-deficit/hyperactivity disorder. Journal of Clinical Psychology, 70, 388–403. http://dx.doi.org/10.1002/jclp.22046 References Becker, S. P., Luebbe, A. M., Fite, P. J., Stoppelbein, L., & Greening, L. (2014). Sluggish cognitive tempo in psychiatrically hospitalized chil- American Psychiatric Association. (2013). Diagnostic and statistical man- dren: Factor structure and relations to internalizing symptoms, social ual of mental disorders (5th ed.). Washington, DC: Author. problems, and observed behavioral dysregulation. Journal of Abnormal Barkley, R. A. (2013). Distinguishing sluggish cognitive tempo from Child Psychology, 42, 49–62. http://dx.doi.org/10.1007/s10802-013- ADHD in children and adolescents: Executive functioning, impairment, 9719-y and comorbidity. Journal of Clinical Child and Adolescent Psychology, Becker, S. P., Marshall, S. A., & McBurnett, K. (2014). Sluggish cognitive 42, 161–173. http://dx.doi.org/10.1080/15374416.2012.734259 tempo in abnormal child psychology: An historical overview and intro- STRUCTURE OF ADHD, ODD, AND SCT SYMPTOMS 927

duction to the special section. Journal of Abnormal Child Psychology, Gomez, R. (2014). ADHD bifactor model based on parent and teacher 42, 1–6. http://dx.doi.org/10.1007/s10802-013-9825-x ratings of Malaysian children. Asian Journal of Psychiatry, 8, 47–51. Bernad, M. M., Servera, M., Grases, G., Collado, S., & Burns, G. L. http://dx.doi.org/10.1016/j.ajp.2013.10.009 (2014). A cross-sectional and longitudinal investigation of the external Gray, J. A. (1987). Perspectives on anxiety and impulsivity: A commen- correlates of sluggish cognitive tempo and ADHD-inattention symptoms tary. Journal of Research in Personality, 21, 493–509. http://dx.doi.org/ dimensions. Journal of Abnormal Child Psychology, 42, 1225–1236. 10.1016/0092-6566(87)90036-5 http://dx.doi.org/10.1007/s10802-014-9866-9 Laakso, A., Wallius, E., Kajander, J., Bergman, J., Eskola, O., Solin, O., Brouwer, D., Meijer, R. R., & Zevalkink, J. (2013). On the factor structure . . . Hietala, J. (2003). Personality traits and striatal dopamine synthesis of the Beck Depression Inventory-II: G is the key. Psychological As- capacity in healthy subjects. The American Journal of Psychiatry, 160, sessment, 25, 136–145. http://dx.doi.org/10.1037/a0029228 904–910. http://dx.doi.org/10.1176/appi.ajp.160.5.904 Brown, T. A. (2006). Confirmatory factor analysis for applied research. Lee, S., Burns, G. L., Snell, J., & McBurnett, K. (2014). Validity of the New York, NY: Guilford Press. sluggish cognitive tempo symptom dimension in children: Sluggish Burns, G. L., Moura, M. A., Beauchaine, T. P., & McBurnett, K. (2014). cognitive tempo and ADHD-inattention as distinct symptom dimen- Bifactor latent structure of ADHD/ODD symptoms: Predictions of dual- sions. Journal of Abnormal Child Psychology, 42, 7–19. http://dx.doi pathway/trait-impulsivity etiological models of ADHD. Journal of Child .org/10.1007/s10802-013-9714-3 Psychology and Psychiatry, 55, 393–401. http://dx.doi.org/10.1111/jcpp Little, T. D. (2013). Longitudinal structural equation modeling. New York, .12165 NY: Guilford Press. Burns, G. L., Moura, M. A., Walsh, J. A., Desmul, C., Silpakit, C., & Marshall, S. A., Evans, S. W., Eiraldi, R. B., Becker, S. P., & Power, T. J. Sommers-Flanagan, J. (2008). Invariance and convergent and discrimi- (2014). Social and academic impairment in youth with ADHD, predom- nant validity between mothers’ and fathers’ ratings of oppositional inately inattentive type and sluggish cognitive tempo. Journal of Abnor- defiant disorder toward adults, ADHD-HI, ADHD-IN, and academic mal Child Psychology, 42, 77–90. http://dx.doi.org/10.1007/s10802- competence factors within Brazilian, Thai, and American children. Psy- 013-9758-4 chological Assessment, 20, 121–130. http://dx.doi.org/10.1037/1040- Martel, M. M., von Eye, A., & Nigg, J. (2012). Developmental differences 3590.20.2.121 in structure of attention-deficit/hyperactivity disorder (ADHD) between International Journal of Behavioral Develop- Burns, G. L., & Lee, S. (2011). Child and Adolescent Disruptive Behavior childhood and adulthood. ment, 36, 279–292. http://dx.doi.org/10.1177/0165025412444077 Inventory–Parent and Teacher Versions 5.0. Pullman, WA: Author. McDonald, R. P. (1999). Test theory: A unified treatment. Mahwah, NJ: Burns, G. L., Servera, M., Bernad, M. M., Carrillo, J. M., & Cardo, E. Erlbaum. (2013). Distinctions between sluggish cognitive tempo, ADHD-IN, and Muthén, L. K., & Muthén, B. O. (1998–2012). Mplus user’s guide (7.3 depression symptom dimensions in Spanish first-grade children. Journal edition). Los Angeles, CA: Muthén and Muthén. of Clinical Child and Adolescent Psychology, 42, 796–808. http://dx Normand, S., Flora, D. B., Toplak, M. E., & Tannock, R. (2012). Evidence .doi.org/10.1080/15374416.2013.838771 for a general ADHD factor from a longitudinal general school popula- Burns, G. L., Servera, M., Bernad, M. M., Carrillo, J. M., & Geiser, C. tion study. Journal of Abnormal Child Psychology, 40, 555–567. http:// (2014). Ratings of ADHD symptoms and academic impairment by dx.doi.org/10.1007/s10802-011-9584-5 mothers, fathers, teachers, and aides: Construct validity within and Penny, A. M., Waschbusch, D. A., Klein, R. M., Corkum, P., & Eskes, G. across settings as well as occasions. Psychological Assessment, 26, (2009). Developing a measure of sluggish cognitive tempo for children: 1247–1258. http://dx.doi.org/10.1037/pas0000008 Content validity, factor structure, and reliability. Psychological Assess- Caspi, A., Houts, R. M., Belsky, D. W., Goldman-Mellor, S. J., Harrington, ment, 21, 380–389. http://dx.doi.org/10.1037/a0016600 H., Israel, S.,...Moffitt, T. E. (2014). The p factor: One general Plichta, M. M., & Scheres, A. (2014). Ventral–striatal responsiveness Clinical psychopathology factor in the structure of psychiatric disorders? during reward anticipation in ADHD and its relation to trait impulsivity Psychological Science, 2, 119–137. http://dx.doi.org/10.1177/ in the healthy population: A meta-analytic review of the fMRI literature. 2167702613497473 Neuroscience and Biobehavioral Reviews, 38, 125–134. http://dx.doi Dumenci, L., McConaughy, S. H., & Achenbach, T. M. (2004). A hierar- .org/10.1016/j.neubiorev.2013.07.012 chical three-factor model of inattention-hyperactivity-impulsivity de- Quay, H. C. (1993). The psychobiology of undersocialized aggressive rived from the attention problems syndrome of the Teachers’ Report conduct disorder: A theoretical perspective. Development and Psycho- Form. School Psychology Review, 33, 287–301. pathology, 5, 165–180. http://dx.doi.org/10.1017/S0954579400004326 Flannery, A. J., Becker, S. P., & Luebbe, A. M. (2014). Does emotion Reise, S. P., Bonifay, W. E., & Haviland, M. G. (2013). Scoring and dysregulation mediate the association between sluggish cognitive tempo and modeling psychological measures in the presence of multidimensional- This document is copyrighted by the American Psychological Association or one of its allied publishers. college students’ social adjustment? Journal of Attention Disorders. Ad- ity. Journal of Personality Assessment, 95, 129–140. http://dx.doi.org/ This article is intended solely for the personal use of the individual uservance and is not to be disseminated broadly. online publication. http://dx.doi.org/10.1177/1087054714527794 10.1080/00223891.2012.725437 Garner, A. A., Peugh, J., Becker, S. P., Kingery, K. M., Tamm, L., Vaughn, Servera, M., Bernad, M., Carrillo, J. M., Collado, S., & Burns, G. L. A.J.,...Epstein, J. N. (2014). Does sluggish cognitive tempo fit within (2015). Longitudinal correlates of sluggish cognitive tempo and a bi-factor model of ADHD? Journal of Attention Disorders. Advance ADHD-inattention symptom dimensions with Spanish children. Jour- online publication. http://dx.doi.org/10.1177/1087054714539995 nal of Clinical Child and Adolescent Psychology. Advance online Gatzke-Kopp, L. M., Beauchaine, T. P., Shannon, K. E., Chipman, J., publication. http://www.ncbi.nlm.nih.gov/pubmed/25751712 Fleming, A. P., Crowell, S. E.,...Aylward, E. (2009). Neurological Toplak, M. E., Sorge, G. B., Flora, D. B., Chen, W., Banaschewski, T., correlates of reward responding in adolescents with and without exter- Buitelaar, J.,...Faraone, S. V. (2012). The hierarchical factor model of nalizing behavior disorders. Journal of Abnormal Psychology, 118, ADHD: Invariant across age and national groupings? Journal of Child 203–213. http://dx.doi.org/10.1037/a0014378 Psychology and Psychiatry, and Allied Disciplines, 53, 292–303. http:// Gibbins, C., Toplak, M. E., Flora, D. B., Weiss, M. D., & Tannock, R. dx.doi.org/10.1111/j.1469-7610.2011.02500.x (2012). Evidence for a general factor model of ADHD in adults. Journal Tuvblad, C., Raine, A., Zheng, M., & Baker, L. A. (2009). Genetic and of Attention Disorders, 16, 635–644. http://dx.doi.org/10.1177/ environmental stability differs in reactive and proactive aggression. 1087054711416310 Aggressive Behavior, 35, 437–452. http://dx.doi.org/10.1002/ab.20319 928 LEE, BURNS, BEAUCHAINE, AND BECKER

Wagner, F., Martel, M. M., Cogo-Moreira, H., Maia, C. R. M., Pan, P. M., Zinbarg, R. E., Yovel, I., Revelle, W., & McDonald, R. P. (2006). Estimating Rohde, L. A., & Salum, G. A. (2015). Attention-deficit/hyperactivity generalizability to a latent variable common to all of a scale’s indicators: A ␻ disorder dimensionality: The reliable ‘g’ and the elusive ‘s’ dimensions. comparison of estimators for h. Applied Psychological Measurement, 30, European Child & Adolescent Psychiatry. Advance online publication. 121–144. http://dx.doi.org/10.1177/0146621605278814 http://dx.doi.org/10.1007/s00787-015-0709-1 Zisner, A., & Beauchaine, T. P. (in press). Midbrain neural mechanisms of Willcutt, E. G., Chhabildas, N., Kinnear, M., DeFries, J. C., Olson, R. K., trait impulsivity. In T. P. Beauchaine & S. P. Hinshaw (Eds.), Oxford Leopold, D. R.,...Pennington, B. F. (2014). The internal and external handbook of externalizing spectrum disorders. New York: Oxford Uni- validity of sluggish cognitive tempo and its relation with DSM–IV versity Press. http://dx.doi.org/10.1093/oxfordhb/9780199324675.013 Journal of Abnormal Child Psychology, 42, ADHD. 21–35. http://dx.doi .003 .org/10.1007/s10802-013-9800-6 Willoughby, M. T., Blanton, Z. E., & the Family Life Project Investigators. (2015). Replication and external validation of a bi-factor parameteriza- tion of attention deficit/hyperactivity symptomatology. Journal of Clin- Received March 4, 2015 ical Child and Adolescent Psychology, 44, 68–79. http://dx.doi.org/10 Revision received July 25, 2015 .1080/15374416.2013.850702 Accepted August 14, 2015 Ⅲ This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.