THE PSYCHOMETRIC PROPERTIES OF THE ADHD BELIEFS SCALE-REVISED

Attention-deficit/hyperactivity disorder (ADHD) is one of the most common disorders of childhood, affecting approximately 1-2 students in every classroom across the United States. Teachers play a vital role in the assessment of student behavior and their academic performance; therefore, they need to possess an adequate level of knowledge and understanding of the various disorders that may occur during childhood and adolescence, including ADHD. Reliable and valid measurement instruments are essential for an accurate assessment of teacher knowledge of ADHD. A dearth of studies, however, has addressed the psychometric properties of questionnaires assessing teacher knowledge. The current study investigated the internal consistency, dimensionality, test-retest reliability, and construct validity of one of these measures, the ADHD Beliefs Scale-Revised, in a sample of in-service teachers (N = 226). A principal components analysis revealed two components, Beliefs about the Neurobiology of ADHD, and Beliefs about the Role of Parents in ADHD, with poor and acceptable internal consistency, respectively. Additionally, the test-retest reliability of the ADHD Beliefs Scale-Revised was found to be acceptable, and preliminary evidence of construct validity was found, despite limitations of the study. Implications for educators are discussed and suggestions for future studies are advanced.


Statement of the Problem
Attention Deficit/Hyperactivity Disorder (ADHD) is among the most commonly diagnosed disorders of childhood (American Psychiatric Association, 2000; Barkley, 2006). Core symptoms include inattention, hyperactivity, and impulsivity that often lead to serious behavioral and academic problems for children, especially in the classroom (Barkley, 2006;Faraone et al., 1993;Raggi & Chronis, 2006). Research has found, for example, that children with ADHD are at greater risk for poorer academic performance, grade retention, and school drop-out (Barkley, 2006); hence, teachers are often the first to notice difficulties associated with ADHD.
Given that children spend the majority of their day at school, teachers play a vital role in the assessment of student behavior and their academic performance. For teachers to work effectively with students, it is important that teachers have an adequate level of knowledge and understanding of the various disorders that may occur during childhood and adolescence, including ADHD. Research has revealed that teachers often receive limited training concerning ADHD; however, they typically report that they would be interested in receiving more training (Pisecco, Huzinec, & Curtis, 2001;Vance & Weyandt, 2008). Studies have also found that teachers' knowledge about ADHD is minimal and that they commonly hold misperceptions about the disorder (Weyandt, Fulton, Schepman, Verdi, & Wilson, 2009).
Collectively, this body of literature suggests that teachers, and ultimately students, could benefit from additional teacher training concerning ADHD.
To be able to accurately assess the knowledge level of teachers regarding ADHD and other disorders, reliable and valid measures are essential to the process. To date, only one study has assessed the psychometric characteristics of a teacher knowledge questionnaire; therefore, information is virtually nonexistent concerning the reliability and validity of such instruments. Due to the dearth of studies regarding the psychometric properties of questionnaires assessing teacher knowledge about ADHD and the importance of psychometrically sound instruments, the current study attempted to address this issue by assessing the internal consistency, factor structure, test-retest reliability, and construct validity of one of these measures, the ADHD Beliefs Scale-Revised (Vance & Weyandt, 2008;Weyandt et al., 2009).

What is Attention Deficit Hyperactivity Disorder?
Attention-Deficit/Hyperactivity Disorder (ADHD) is a neurologically based developmental disorder, characterized by symptoms of inattention, impulsivity, and hyperactivity that are developmentally inappropriate and cause impairments in major life activities (American Psychiatric Association, 2000). The prevalence of ADHD is estimated to range from 3% to 7% in the United States school-aged population and has been found across various cultures (American Psychiatric Association, 2000).

ADHD and Academic Difficulties
Children with ADHD commonly experience cognitive and academic problems (Faraone et al., 1993;Raggi & Chronis, 2006), such as difficulty following directions, focusing on tasks, and remaining attentive and seated. In addition, they often demonstrate a number of behavioral problems, such as noncompliance and aggressive behavior (Barkley, 2006). Moreover, these students are more likely than their peers to receive lower grades, fall behind academically, score lower on standardized assessments, receive special education services and other student services, repeat grades, drop out of high school (Faraone et al., 1993) and to not attend college (DuPaul, Weyandt, O'Dell, & Varejao, 2009).
Despite the clear evidence that students with ADHD commonly experience various academic problems, relatively little research exists concerning academic interventions, compared to research regarding behavioral and pharmacological interventions (DuPaul, Weyandt, & Janusis, 2011). Stimulant medication and behavior-modification strategies are the most common interventions for children with ADHD as they have been shown to significantly reduce ADHD symptoms (Barkley, 2006;Spencer, Biederman, & Wilens, 2000). These interventions, although often effective for remediating behavior problems, especially when implemented both in the home and in school settings, have not been equally successful at increasing academic achievement (DuPaul et al., 2011). Given the lack of evidence-based methods for improving the academic performance of students with ADHD, meeting the academic needs of these children can be challenging for educators, especially their teachers, who are often among the primary agents of intervention for these students.
Teachers play an important role in identifying children with ADHD. While previous estimates indicated that on average, one in every twenty school-aged children is diagnosed with ADHD, (American Psychiatric Association, 2000), more recent numbers suggest that up to 10% of students are diagnosed with ADHD (Wolraich et al., 2012). Teachers often witness difficult and disruptive student behavior, as well as problems associated with inattention. Teachers therefore possess valuable clinical information and are often the first to initiate referrals for psychological assessment (Sax & Kautz, 2003;Weyandt et al., 2009). Teacher referrals, however, are not always warranted, as the information they are based on is not always accurate. For example, in a study by Glass & Wegar (2000), teachers were found to overestimate the prevalence of ADHD in their classrooms. Other research has reported similar results with teachers identifying a higher proportion of students as having ADHD than prevalence rates would indicate (Havey, Olson, McCormick, & Cates, 2005;Weiler, Bellinger, Marmor, Rancier, & Waber, 1999). Based on research that suggests that teachers have a tendency to over-identify, and some may actually under-identify ADHD in their classrooms (Glass & Wegar 2000;Fabiano et al., 2013;Havey et al., 2005;Sciutto & Eisenberg, 2007;Weiler et al., 1999), it is plausible that many teachers do not possess accurate and adequate knowledge about the disorder.

Teaching Experience and Knowledge of ADHD
Theoretically, increased teaching experience should lead to increased exposure to a variety of student characteristics; therefore, it is likely that increased teaching experience is associated with greater knowledge about various childhood disorders, including ADHD. Research by Weyandt et al. (2009), however, questions the accuracy of this hypothesis, as findings revealed that teaching experience was negatively correlated with knowledge of ADHD; specifically, increased teaching experience was associated with less knowledge about ADHD. The researchers noted, however, that extensive psychometric information for the scale they used, a revised version of The ADHD Beliefs Scale, was not available, although previous analyses using the original version of the scale among parents of children with ADHD indicated adequate internal consistency. Given the lack of psychometric information regarding the ADHD Beliefs Scale-Revised, the results of Weyandt et al. (2009) and others should be interpreted cautiously. In an earlier study, Vance and Weyandt (2008) explored professor perceptions of college students with ADHD, using the ADHD Beliefs Scale-Revised and findings revealed that college professors' perceptions of college students with ADHD did not differ by educational level, years of teaching experience, previous experience with students with ADHD or ADHD training. In a related study, Vereb and DiPerna (2004) examined teacher knowledge of ADHD, years of experience teaching students with ADHD and teacher ratings of ADHD treatment acceptability. Results did not provide evidence for an association between teaching experience and knowledge about ADHD, or between teaching experience and the acceptability of behavior management interventions for ADHD. Vereb and DiPerna (2004) created their own instrument for their study and examined its content validity qualitatively by having a panel of experts rate the importance of each item of the questionnaire, where items that received a low importance rating were eliminated. In addition, the internal consistency of the four subscales of the instrument was assessed, for three of which Cronbach's alpha was acceptable while for the fourth it was poor (alpha coefficients ranged from .58 to .81). No other quantitative methods were used to further assess the reliability and validity of the instrument, therefore the findings should be interpreted cautiously. In a study conducted by Kos, Richdale, and Jackson (2004), teachers with more years of teaching experience tended to overestimate their knowledge about ADHD compared to teachers with less experience, although no significant association between years of teaching experience and degree of actual ADHD knowledge was found. In-service teachers were also compared to pre-service teachers on measures of ADHD knowledge. Although, among in-service teachers, amount of teaching experience was not associated with greater ADHD knowledge, in-service teachers scored higher than pre-service teachers on a measure of ADHD knowledge.
Experience teaching students with ADHD was, however, related to greater knowledge about ADHD, as well as amount of ADHD training. Kos and colleagues (2004) developed their own survey for the purposes of their study, but no psychometric data on its reliability or validity were reported, which once again underscores the need for careful interpretation of the findings. Collectively, these studies suggest that increased teaching experience may not result in greater knowledge about ADHD, although the scarcity of studies on the psychometric properties of the various instruments used to assess teacher knowledge about ADHD calls to question the validity and reliability of these findings. Similar to findings reported by Kos et al. (2004), where previous experience teaching students with ADHD was associated with greater knowledge about ADHD, Sciutto, Terjesen and Bender-Frank (2000) reported that the extent to which teachers had taught children with ADHD in the past and teachers' confidence in their ability to effectively teach a child with ADHD, was positively related to their knowledge about the disorder. They also found that many teachers, however, held common misperceptions about the disorder, especially regarding the effects of sugar intake on ADHD symptoms and the long-term prognosis of the disorder. Sciutto et al. (2000) developed their own ADHD knowledge measure and reported "good internal consistency", but no further reliability or validity information was provided. Similarly, Anderson, Watt, Noble, & Shanley (2012) found that in-service teachers possessed both more actual and perceived knowledge about ADHD than pre-service teachers, which is an indication that increased teaching experience was associated with more knowledge about ADHD in this sample. Anderson et al. (2012) administered a revised version of an instrument developed by West, Taylor, Houghton, & Hudyma (2005), which was based on the instrument originally created by Sciutto et al. (2000). Anderson et al. (2012) reported acceptable to good internal consistency for the subscales of their version of this measure, but no further psychometric information about this scale has been published. In addition to examining whether teaching experience is associated with teacher knowledge about ADHD, the extent to which teacher knowledge of the disorder relates to other variables, such as teacher perceptions of students with ADHD, has also been studied.

Teacher Perceptions of Students with ADHD
Labeling, that is, assigning a diagnostic label to a student such as a learning disability, autism or ADHD, can influence the way teachers interact with and evaluate students. Perhaps the most famous study on the impact of labels was conducted by Rosenthal and Jacobson in 1966, where students who had been randomly selected to receive the label "likely to demonstrate unusual intellectual achievement" showed significantly greater gains in cognitive ability than students who were not assigned that label. Results suggested that the expectations teachers held for students based on the labels, had an impact on student performance. Although in some cases, labels may have positive effects, they can also lead to more negative outcomes, such as decreased teacher expectations and negative stereotypes of students. In another landmark study, Foster and Ysseldyke (1976) found that teachers held negative expectations of students with a diagnostic label, such as emotional disturbance, learning disability, and mental retardation, compared with students without a label, even for students engaging in normal behavior that was inconsistent with the labels. In a more recent study, Batzle, Weyandt, Janusis, and DeVietti (2010), explored K-12 grade teachers' ratings of children, both with and without an ADHD label. Results revealed that the teachers rated children with an ADHD label less favorably than children without an ADHD label on measures of behavior, cognitive functioning, and personality. Similarly, in a study by Ohan, Visser, Strain, and Allen (2011), in-service and pre-service teachers responded differently to questions about children who had a diagnosis of ADHD than to questions about children who did not have an ADHD diagnosis. Participants' negative expectations and negative emotions increased when a child was labeled "ADHD", and their confidence in their ability to instruct the child decreased. Findings reported by Liljequist and Renk (2007) corroborate the results reported by Ohan et al. (2011), wherein externalizing behaviors, such as aggression, impulsivity, or hyperactivity, which are among the core symptoms of ADHD, were found to "trouble" teachers more than internalizing behaviors, such as withdrawal and depression.
Teacher perceptions of students can affect their interactions with students, which can influence students' academic outcomes (Feldman & Theiss, 1982).
Negative teacher expectations of students can thus serve to exacerbate students' problems and thereby create self-fulfilling prophecies, where students are perceived negatively, which adversely affects their academic outcomes, which, in turn, confirms teachers' original negative perceptions of these students (Eisenberg & Schneider, 2007;Harris, 1994). Research has also demonstrated that teacher perceptions of students with ADHD can affect other students' perceptions of those students (Atkinson, Robinson, & Shute, 1997).
It is plausible that teachers' level of ADHD knowledge contributes to their interactions with and perceptions of students who have the disorder. For example, Sherman, Rasmussen, and Baydala (2008) conducted a systematic review of the literature and concluded that a variety of teacher factors, such as tolerance of classroom behaviors, acceptability of various treatments for ADHD, as well as their level of knowledge and training regarding ADHD, can have an impact on the academic and behavioral outcomes of students with ADHD. In fact, Ohan, Cormier, Hepp, Visser, and Strain (2008) found that teachers who retained greater levels of ADHD knowledge were more likely to believe that for students with ADHD, professional assessment services are beneficial, that academic support is helpful, as well as making various adjustments in the home and school environment. Teachers who were more knowledgeable, however, also reported less confidence in their ability to manage these students than those with less knowledge. Ohan and colleagues (2008) used a questionnaire designed by Jerome, Gordon, and Hustler (1994). No psychometric information for this scale was reported, which renders analyses difficult regarding how different aspects of ADHD knowledge contribute to teacher behavior and attitudes toward students with ADHD, and decreases the confidence with which the results can be interpreted.
Although the results of the aforementioned studies suggest that teachers often lack adequate training and knowledge about ADHD and that their knowledge level has an impact on their perceptions of students with ADHD, the lack of psychometric data concerning the measures likely impacts the validity and reliability of these findings.
Psychometric studies are sorely needed to determine the underlying properties of questionnaires that are used to determine teacher knowledge are about this disorder.

Psychometric Studies on ADHD Knowledge Questionnaires for Teachers
To date, the only study that has examined the psychometric qualities of an instrument measuring teacher knowledge and attitudes about ADHD was conducted by Hepperlen, Clay, Henly, and Barké in 2002. Hepperlen and colleagues (2002) created the Test of Knowledge about ADHD (KADD) as an indirect attitude measure using the "error-choice technique". The error-choice technique involves a series of multiple choice questions presented in the format of a test or exam, including questions about general knowledge topics that are unrelated to ADHD. The authors' rationale for using the error-choice method was to reduce response bias and social desirability responding. Hepperlen et al. (2002) surveyed approximately 100 teachers and found that the scale comprised one global factor with acceptable internal consistency. The researchers noted, however, that evidence regarding the validity of the KADD was lacking. Due to the unconventional approach to knowledge and attitude measurement and the limited evidence for its validity, the KADD (Hepperlen et al., 2002) was not chosen for use in the present study. Additional instruments of ADHD knowledge, however, have been validated psychometrically, albeit in different populations, and were therefore considered more appropriate for the purposes of current study.
The ADHD Beliefs Scale was originally designed by Johnston and Freeman (2002) to measure beliefs of parents of children with ADHD about the disorder, but has also been modified for use with teachers and college professors as the ADHD Beliefs Scale-Revised (Vance & Weyandt, 2008;Weyandt et al., 2009). The scale reflects a variety of beliefs concerning ADHD, such as the causes of ADHD (e.g., "ADHD is related to neurological functioning in the brain" or "Some children develop ADHD because they want attention") and various treatment options (e.g., "A combination of medication and behavior management is best for treating ADHD" or "Limiting a child's sugar intake can be an effective treatment for ADHD"). The most recent version of the scale (Johnston, Seipp, Hommersen, Hoza, & Fine, 2005) originally contained 27 items or statements, which participants respond to on a 7 point Likert-scale, ranging from disagree to neutral, and from neutral to agree. A principal components analysis (PCA), conducted in a combined sample of 253 mothers and fathers of children with ADHD, yielded a four-factor solution that accounted for more than 50% of the variance in scores, and indicated that four items should be omitted due to inconsistent factor loadings (Johnston et al., 2005). The first factor was labeled Belief in Behavior Management (eight items, α = .73), the second factor Belief in Medication (six items, α = .77), and the third and fourth factors were named Belief in Psychological Causes/Treatments (five items, α = .74) and Belief in Diet/Vitamin Treatments (four items, α = .71), respectively. Results reported by Johnston, Hommersen, & Seipp (2008) indicate that the original parent-version of the ADHD Beliefs Scale has good construct validity, as parents' beliefs were related to their experience with ADHD treatment and their attributions for the causes of their children's behavior.

Purpose of the Present Study
A review of the literature revealed that a substantial number of studies have examined teacher perceptions and knowledge about ADHD and have explored the effect of teacher knowledge on interactions with students. None of the studies, however, properly addressed the psychometric properties of the measures used to assess teacher knowledge about ADHD. Because reliability and validity are fundamental characteristics of any measurement instrument, a rigorous examination of the psychometric properties of such instruments is of great importance. The present study explored the factor structure of one of these instruments, the ADHD Beliefs Scale-Revised, as well as its test-retest reliability, internal consistency, and construct validity. Although other instruments have been used in the literature (e.g., the 20-item scale prepared by Jerome and colleagues (1994), the Knowledge of Attention Deficit Disorders Scale by Sciutto et al. (2000), and the Test of Knowledge about ADHD (KADD) by Hepperlen et al., (2002)), the ADHD Beliefs Scale was chosen for use in this study due to the a) number of published studies using the ADHD Beliefs Scale or the ADHD Beliefs Scale-Revised, b) psychometric information available for the parent version of the scale (Johnston & Freeman, 2002;Johnston et al., 2005;Johnston et al., 2008: Vance & Weyandt, 2008Weyandt et al., 2009), and c) seven point Likert-scale response format of the ADHD Beliefs Scale, which is more appropriate for the proposed analyses (Reise, Waller, & Comrey, 2000) as opposed to the two to three response options format offered by other instruments (e.g., Jerome et al., 1994;Sciutto et al., 2000).
Based on previous research, the first hypothesis of the present study was that four factors would emerge, including Belief in Behavior Management, Belief in Medication, Belief in Psychological Causes/Treatments, and Belief in Diet/Vitamin Treatments (Johnston et al., 2005). Although Johnston et al. (2005) did not provide extensive psychometric information about the ADHD Beliefs Scale, such as the degree of correlation between factors, it was predicted that the four factors would be correlated due to their conceptual nature (e.g., beliefs in psychological treatments were expected to be associated with beliefs about behavior management). Reliability coefficients were expected to be adequate, internal reliability coefficients equal to or higher than 0.70, and test-retest reliability equal to or higher than 0.60.

Chapter II: Method
Procedure Participants for the present study included in-service teachers, employed at the primary and secondary educational level. School administrators in various school districts in Rhode Island, Connecticut, and Massachusetts were contacted and asked for permission for the researcher to contact teachers working in the district. Emails were sent to a contact person (e.g., principal, assistant principal or school psychologist) at each school, who was asked to send an email to all teachers at the school. Information in the email directed participants to a secure website hosted by SurveyMonkey, where an online survey was accessible. Potential participants were instructed to read a consent form once they entered the website and confirm they understood the content by clicking on a statement of endorsement. Participants who provided consent were then directed to the ADHD Beliefs Scale-Revised and a demographic questionnaire designed by the researcher. Participants were also provided with information regarding how to contact the researcher if desired. Before beginning the survey, participants were asked to choose a six digit number that was easy to remember but difficult to trace to them, such as a parent's date of birth. They were then asked to provide that number on the questionnaire. Approximately two to three weeks later, this procedure was repeated. To match the answers from the first administration to those of the second administration, participants were asked to provide the six digit number they chose during the first administration. To encourage participation, participants were offered to register for a drawing, by providing their email address, where they had a chance of winning one of two $50 gift cards.

Participants.
A convenience sample of 260 in-service teachers in Rhode Island, Connecticut, and Massachusetts was recruited, 233 of whom were eligible for participation and completed all study questionnaires. While exact response rates could not be calculated due to a lack of information concerning the number of teachers working in each school district as well as the number of teachers who received the participation email, eight school districts out of 30 agreed to participate in the study, that equals a participation rate of 26.7%. The final sample was smaller than the desired sample of 300, which was determined by Comrey and Lee's recommendation (1992) of an N of 200-300 for factor analysis, and Nunnally's (1978) recommendation for a minimum of 300 participants when assessing internal consistency. According to Guadagnoli & Velicer (1988), however, a sample of 100-200 is sufficient for factor or principal components analysis, provided that factor loadings are high. Participants were expected to be representative of the sex and race/ethnicity demographics of teachers in the United States; the majority of participants were expected to be White/Caucasian and female (Feistritzer, 2011 Participants were informed that to be eligible for participation they needed to be at least 18 years of age, working as teachers, i.e., not as school support/guidance staff, and to be able to read and write in English. Those who did not fulfill these criteria were excluded from participating in the study. Three participants identified as school support staff members (i.e., school psychologist, school nurse, and guidance counselor); therefore, their answers were not included in the analyses.
As for future studies. The demographic questionnaire is presented in Appendix D.

Design
The current study investigated the: 1) dimensionality of the ADHD Beliefs Scale-Revised; 2) internal consistency of the ADHD Beliefs Scale-Revised; and 3) test-retest reliability of the ADHD Beliefs Scale-Revised, and 4) construct validity of the ADHD Beliefs Scale-Revised; exploratory analyses included examining group differences on the ADHD Beliefs Scale-Revised as well as the correlation between perceived level of preparation to teach students with ADHD and actual ADHD knowledge as measured by the ADHD Beliefs Scale-Revised.
Hypothesis 1 stated that an exploratory principal components analysis (PCA) would reveal a factor structure of four underlying factors. To address hypothesis 1, an item analysis, followed by an exploratory PCA of the ADHD Beliefs Scale-Revised, was conducted. Items that poorly discriminated among participants as measured by extremely high or low means and little variance were eliminated, as recommended by Redding, Maddock, & Rossi (2006). The item-total correlation was analyzed, wherein items with an item-total correlation of less than 0.25 were removed. Although guidelines for item elimination based on item-total correlation coefficients suggest using 0.30 (Ferketich, 1991;Kline, 1993) or 0.40 (Nunnally, 1978)  however, that had a relatively complex loading but made sense conceptually and fit well with its respective component was retained.
Hypothesis 2 posited that internal reliability coefficients would be adequate, that is, equal to or higher than 0.70. Internal consistency was assessed by calculating Cronbach's alpha for each factor, using Nunnally's criteria (1978) of 0.70 or higher for a satisfactory internal consistency coefficient. Given that the correlation between factors was low (r = 0.151), a global ADHD Beliefs Scale-Revised score was not calculated, nor a global internal reliability coefficient.
Hypothesis 3 held that test-retest reliability would be adequate, that is, equal to or higher than 0.60. To assess test-retest reliability, the ADHD Beliefs Scale-Revised was administered twice, with approximately a two to three week interval, and the correlation between scores on the first and second administration was calculated.
Given that any ADHD training participants may have received in the interim could have affected the test-retest reliability, participants were asked about the amount of ADHD training they had received, and their answers from the first and second administration of the questionnaire were compared.
Additionally, the construct validity of the ADHD Beliefs Scale-Revised was explored. Multiple regression analyses were used to determine whether years of experience teaching students with ADHD, as well as amount of ADHD training, were predictive of scores on the ADHD Beliefs Scale-Revised.
Finally, multiple regression and correlational analyses were conducted posthoc to examine group differences on the ADHD Beliefs Scale-Revised based on educational setting, teaching experience, and age, as well as the association between perceived level of preparation to teach students with ADHD and actual ADHD knowledge as measured by the ADHD Beliefs Scale-Revised. Further, univariate ANOVAs were conducted to examine sex differences in ADHD knowledge as measured by the scale.

Chapter III: Results
To test the hypotheses, two sets of analyses were conducted. Specifically, the first set of analyses involved: a) analyses of the factor structure and internal consistency of the ADHD Beliefs Scale-Revised; b) an analysis of the test-retest reliability of the scale, and c) an exploration of the construct validity of the questionnaire. The second set of analyses were conducted post hoc to explore group differences in ADHD knowledge and the correlation between self-perceived preparedness to teach students with ADHD and teacher knowledge about ADHD, as measured by the scale.

Item Analysis, Dimensionality, and Internal Consistency
SPSS version 22 was used to conduct all analyses. An item analysis involving a comparison of item means, skewness, kurtosis, and item-total correlations was conducted, where items that had an item-total correlation lower than 0.25 were removed. To determine the appropriate number of components, Horn's parallel analysis and Velicer's MAP procedure were employed, as recommended by O'Connor (2000). After the initial PCA (N = 226 with listwise elimination of cases with missing data), complex items (i.e., loading on more than one component with coefficients greater than 0.40, not loading onto any components with coefficients greater than 0.40, or loading on components that did not make sense conceptually) were removed. One item (item 27) that made sense conceptually and fit well with its respective component, despite its complex loadings, was retained, however, to form a component that contained two items, instead of only one item.
The remaining items were entered into a second and third PCA with an orthogonal (Varimax) rotation, given the minimal correlation between factors, yielding the final version of the ADHD-Beliefs Scale-Revised. Internal consistency was assessed using Cronbach's alpha and Pearson's bivariate correlation was then calculated to assess the test-retest reliability of the ADHD Beliefs Scale-Revised.
Multiple regression analyses were conducted to assess the construct validity of the questionnaire, as well as potential predictors of teacher knowledge. ANOVAs were conducted to analyze potential group differences.  skewness and/or kurtosis greater than |1.0| and were therefore discarded. Table 3 depicts the corrected item-total correlation for each of the remaining 20 items as well as the overall Cronbach's alpha for the scale if each of these items were deleted. As stated previously, a more lenient criterion for item-total correlation was adopted than has been suggested by some (e.g., Ferketich, 1991;Nunnally, 1978;Kline, 1993) due to the resulting low number of items, wherein items with an itemtotal correlation below 0.25 were discarded. As shown in Table 3 The results of the MAP analysis suggested retaining 2 components whereas the parallel analysis indicated that 3 components should be retained. Tables 4 and 5 contain information regarding loadings for 2 and 3 components, respectively. 27. ADHD is related to parents' use of poor discipline strategies.
.449 .691 .008 a suggested total of six items. Given that the two-component structure was more parsimonious, had fewer complex loadings, had more than one item loading on each factor, and appeared more readily interpretable, two components were retained. A second PCA using the seven remaining items was conducted; results can be found in Table 6. 9. It is likely that medications used to treat ADHD are effective because they alter the neurotransmitters in the child's brain.
.670 .069 13. ADHD results from parents being inconsistent with rules and consequences.
.013 .892 27. ADHD is related to parents' use of poor discipline strategies. .065 .855 The second PCA on the seven remaining items revealed that item 3 had a complex loading, suggesting it should be discarded. A third PCA was therefore conducted for the six remaining items; results can be seen in Table 7. 9. It is likely that medications used to treat ADHD are effective because they alter the neurotransmitters in the child's brain.
.658 .039 13. ADHD results from parents being inconsistent with rules and consequences.
.034 .902 27. ADHD is related to parents' use of poor discipline strategies. .112 .890 Table 8 provides information regarding the eigenvalues for each of the two components, labeled Beliefs about the Neurobiology of ADHD (component 1) and Beliefs about the Role of Parents in ADHD (component 2).  Nunnally (1978) is acceptable during the initial stages of scale development. The factors were only minimally correlated: r = 0.151, p = 0.022, providing support for the orthogonal (varimax) rotation.

Test-Retest Reliability
The test-retest reliability of the ADHD Beliefs Scale

Construct Validity
Although the internal consistency of one of two subscales, Beliefs about the Neurobiology of ADHD, was below acceptable limits, the construct validity of the scale was explored via a series of multiple regression analyses. The validity analyses, however, should be interpreted with caution, given the less than optimal internal consistency of one of the two subscales.
To investigate the construct validity of the scale, the association between the self-reported level of ADHD training completed and scores on the two components,

Beliefs about the Neurobiology of ADHD and Beliefs about the Role of Parents in
ADHD was examined. The following variables, all measured on a 7 point Likert scale ranging from "never" to "frequently" or "substantial", were entered as predictors into two multiple regression models: a) ADHD coursework taken at the university/college level, b) professional development training regarding ADHD, c) books read about ADHD, d) magazines read about ADHD, and e) research journals read about ADHD, for each of the two subscales. Results for each of the two models can be found in tables 9 and 10, respectively. Potential violations of the assumptions of multiple regression were identified for some of the models (see Appendix E for an evaluation of the assumptions for multiple regression). To control the overall Type I error rate the Benjamini-Hochberg linear step-up procedure (1995), a modified version of the Bonferroni approach, was utilized. As shown in Table 9  To further explore the construct validity of the scale, overall teaching experience (measured in years), experience teaching students with ADHD (measured on a 7 point Likert scale ranging from "never" to "frequently"), and number of students with ADHD taught were entered as predictors into a multiple regression model, with scores on the Beliefs about the Neurobiology of ADHD and Beliefs about the Role of Parents in ADHD components as the dependent variables. Table 11 and 12 include the results of these analyses, respectively.

Post Hoc Analyses: Teacher Knowledge Calibration
The correlation or agreement between perceived preparedness to teach students with ADHD (measured on a 7 point Likert scale ranging from "disagree" to "agree") and teacher knowledge about ADHD as measured by the two components of the Of those who participated in the study, 82% indicated that they would be interested in receiving ADHD training. No significant correlation was found between interest in ADHD training and perceived preparedness to teach students with ADHD: r = -0.095, p = 0.156.

Chapter IV: Discussion
Teacher knowledge and attitudes concerning ADHD, one of the most commonly diagnosed disorders of childhood, have been found to predict the academic performance of students with the disorder (Sherman et al., 2008). It is therefore critical to investigate whether teacher knowledge and beliefs regarding ADHD are associated with outcomes of students with ADHD; however, in order to accurately interpret this information, reliable and valid measures are necessary to measure teacher knowledge.
The purpose of the current study was to examine the dimensionality, internal consistency, test-retest reliability, and construct validity of a questionnaire intended to measure teacher knowledge and beliefs about ADHD, given significant gaps in the literature concerning the underlying psychometric properties of such measures.
Moreover, the present study sought to identify group differences in ADHD knowledge and to assess the correlation between teacher perceptions of their preparation to work with students with ADHD and their actual knowledge as measured by the ADHD Beliefs Scale-Revised, the measure of interest in the current study.

Psychometric Findings of the ADHD Beliefs Scale-Revised
Results revealed that the factor structure of the ADHD Beliefs Scale-Revised in this sample was rather different from what was hypothesized based on the original version of the scale. In the present study, a two-factor structure emerged, as opposed to the four factor structure of the original version developed by Johnston and Freeman (2005). The two components were labeled Beliefs in the Neurobiology of ADHD and Beliefs about the Role of Parents in ADHD. The former component includes beliefs about the physiological aspects of ADHD (e.g., "It is likely that medications used to treat ADHD are effective because they alter the neurotransmitters in the child's brain"), while the latter reflects beliefs about the role of parents in ADHD as causal agents (e.g., "ADHD results from parents being inconsistent with rules and consequences"). Based on the psychometric findings, the number of items on the ADHD Beliefs Scale-Revised decreased substantially, from 27 items to only six. As noted previously, the ADHD Beliefs Scale was originally designed to assess the knowledge and beliefs concerning ADHD among parents of children with the disorder. Given these divergent findings, it appears that the psychometric characteristics of the ADHD Beliefs Scale may not be equivalent across teacher and parent populations. While this was somewhat surprising, the divergent findings, however, make sense conceptually. Specifically, being the parent of a child with ADHD is clearly different from being the teacher of a student with ADHD, especially given the distinct responsibilities and experiences inherent in each of these roles.
Furthermore, the present findings suggest that the ADHD Beliefs scale may not measure teacher knowledge and beliefs as well as it measures parent knowledge and beliefs of ADHD.
The elimination of items resulted in the loss of several statements reflective of common misconceptions of ADHD (e.g., limiting a child's sugar intake can be an effective treatment for ADHD) as well as beliefs about various behavior management strategies (e.g., behavior management is an effective treatment for ADHD) which may be problematic due to previous findings suggesting widespread misconceptions about the disorder among teachers (Sciutto et al., 2000;Weyandt et al., 2009) as well as the relevance of behavior management strategies for classroom management. Evaluating these beliefs is important as they may possibly predict various teacher behaviors and their acceptance of interventions for students with ADHD.
In addition to the scant number of items on the final version of the scale, the internal consistency of one of the two subscales, Beliefs about the Neurobiology of ADHD, was below acceptable limits. One of the issues contributing to this finding may be the fact that the subscale only included four items. In contrast, the internal consistency coefficient of the other subscale, Beliefs about the Role of Parents in ADHD, however, was above acceptable standards although it only comprised two items. Together, the limited number of items and the low internal reliability of one of the two subscales, suggest that in its current form, the scale may be a less than optimal measure of teacher knowledge of ADHD.
Although the internal consistency of the subscale of the ADHD Beliefs Scale-Revised in the present study was lower than expected (i.e., Cronbach's alpha = 0.635), the 2-3 week test-retest reliability exceeded acceptable limits (i.e., r = 0.795), despite the small number of participants completing the retest. These findings are only preliminary, but indicate that scores on the ADHD Beliefs Scale-Revised are relatively stable over time. Such findings have important implications for intervention studies seeking to experimentally examine the impact of teacher training concerning ADHD, which requires measures to be temporally stable so that any differences in scores can be attributed to the intervention and not measurement instability.
Another focus of the present study was to evaluate the construct validity of the Despite inconsistent findings in the literature concerning the relationship between teaching experience and knowledge of ADHD, the association between overall teaching experience, experience in teaching students with ADHD, and scores on the two subscales was assessed to further explore the construct validity of the scale.
Results revealed that while controlling for overall teaching experience, self-reported experience teaching students with ADHD was positively associated with scores on the Beliefs about the Neurobiology of ADHD component, whereas the absolute selfreported number of students with ADHD taught was negatively associated with Beliefs about the Neurobiology of ADHD. Similarly, while accounting for overall teaching experience and self-reported experience teaching students with ADHD, the absolute self-reported number of students with ADHD taught was negatively associated with scores on the Beliefs about the Role of Parents in ADHD subscale. Together, these findings suggest that while holding both overall teaching experience and experience teaching students with ADHD constant, a higher self-reported number of students with ADHD taught is associated with poorer knowledge about ADHD. It is important to note that these findings do not allow for causal inferences due to the lack of experimental manipulation and the self-report nature of information concerning teaching experience and exposure to students with ADHD. Further, although findings indicate a negative relationship between number of students with ADHD taught and knowledge about the disorder, this is not necessarily an indication that the scale lacks validity. It is possible that those with less knowledge about the disorder may overestimate the prevalence of ADHD in their classrooms (Glass & Wegar 2000;Havey et al., 2005;Sciutto & Eisenberg, 2007;Weiler et al., 1999), and thus report having taught a higher number of students with ADHD than those who are more knowledgeable. Conversely, however, experience in teaching students with ADHD was associated with a higher score on the Beliefs about the Neurobiology of ADHD component while holding overall teaching experience and number of students with ADHD taught constant. This finding is in accordance with those reported by Anderson et al., (2012), Kos et al. (2004), and Sciutto et al., (2004), who found that experience teaching students with ADHD was positively associated with knowledge about the disorder. Although making sense of these conflicting findings is challenging, is it clear from the results that the manner in which teaching experience is measured can alter its relationship with knowledge and beliefs about ADHD.

Post-Hoc Analyses: Group Differences and ADHD Knowledge Calibration
In terms of group differences, analyses revealed that while holding overall teaching experience constant, special education teachers as a group had a higher score on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised, compared with general education teachers in this sample. No group differences were found, however, for the other subscale, Beliefs about the Role of Parents in ADHD. Hence, these findings indicate that special education teachers may be slightly better aware of the neurological aspects of the disorder, while the groups seem to hold similar beliefs concerning parental behaviors as a causal factor in ADHD. Additionally, the results suggested that in the present study female teachers had somewhat higher scores on both subscales of the ADHD Beliefs Scale-Revised than males. Although findings pertaining to group differences should be interpreted cautiously due to unequal group sizes, an examination of variance homogeneity did not suggest any major violations of assumptions.
Also of interest was the agreement or calibration between teacher perceptions of their preparedness to teach students with ADHD and their actual level of knowledge as measured by the ADHD Beliefs Scale-Revised. Findings revealed a weak association between perceived preparedness to teach students with ADHD and scores on the Beliefs about the Neurobiology of ADHD component and no significant correlation between perceived preparedness and scores on the Beliefs about the Role of Parents in ADHD component. Therefore, it appears that teacher knowledge calibration regarding ADHD in this study was relatively poor, and teachers may thus not have been aware of potential gaps in their knowledge and training concerning ADHD. These findings are corroborated by a number of findings from other studies supporting the notion that teachers would benefit from additional training regarding ADHD (e.g., Jones & Chronis-Tuscano, 2008).
Although the present findings did not support an association between teacher interest in receiving ADHD training and perceived preparedness to teach students with ADHD, 82% of participants indicated that they would be interested in receiving training regarding ADHD. This suggests that in this sample, teachers who were less knowledgeable about ADHD were no more or less likely than teachers who were more knowledgeable about ADHD to report being interested in receiving ADHD training.
On a more positive note, however, the majority of participants did endorse being interested in additional training, a finding also reported in other studies (e.g., Pisecco et al., 2001;Vance & Weyandt, 2008). This finding has important implications for practice in the schools and teacher preparation, particularly the training of both preservice and in-service teachers.

Limitations and Future Directions
A major limitation of the present study was the small, largely homogeneous convenience sample. The goal was to recruit a minimum of 300 in-service teachers, ideally of diverse backgrounds. Due to substantial difficulty in the recruitment of participants, however, data were collected from 260 participants who were mostly White/Caucasian and female. The relatively small sample size may partially explain the less than optimal psychometric findings. In addition, the homogeneous nature of the sample and the fact that it was a convenience sample may restrict the generalizability of the findings.
Due to the reported psychometric properties of the ADHD Beliefs Scale-Revised as measured in the current study, particularly the internal consistency of one of the two components and the low number of items, all subsequent analyses using the scale should be interpreted with caution. Moreover, the current results highlight the need to carefully investigate the psychometric properties of all measures prior to collecting data rather than assuming they are reliable or valid. Previous findings regarding teacher knowledge of ADHD obtained using instruments that have not been validated psychometrically should therefore also be cautiously interpreted.
In light of the current findings regarding the ADHD Beliefs Scale-Revised and its poor psychometric properties in this sample, future studies are needed to further develop and validate measures of teacher knowledge and beliefs concerning ADHD.
Such measures should include a variety of questions or items pertaining directly to teacher experiences in the classroom and with students with ADHD, and ideally, focus on a larger and more diverse sample of teachers. Furthermore, studies that explore the link between teacher knowledge and beliefs about ADHD, teacher behavior, and student outcomes are sorely needed. Given the important role teachers play in identifying students with ADHD and providing these students with appropriate instruction and interventions, teacher knowledge is likely a major contributor to the academic success and overall well-being of students with ADHD.
assistants at the University of Rhode Island. Participation in this project is completely anonymous. Your information will not be shared with any organization.
Decision to quit at any time: You may choose not to participate at any time. [ ] I choose not to participate.

. Assumption of residual independence for multiple regression assessing the association between self-reported level of training completed and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
As depicted in Figure 1, the residual does not appear to be severely affected by participant number, indicating that the time at which participants completed the questionnaire did not have a large impact on the results. The range of residual values, however, appears to widen slightly over time, indicating that the time at which participants responded to the questionnaire is associated with somewhat greater response variability.

. Assumption of residual homoscedasticity for multiple regression assessing the association between self-reported level of training completed and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
While no major violations of the assumption of residual homoscedasticity can be identified in Figure 3, the residual appears to follow somewhat of a downward trend.

Figure 4. Assumption of residual independence for multiple regression assessing the association between self-reported level of training completed and scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
As depicted in Figure 4, the residual does not appear to be affected by participant number, indicating that the time at which participants completed the questionnaire did not impact results.

Figure 5. Assumption of residual normality and linearity for multiple regression assessing the association between self-reported level of training completed and scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
No major deviations from the assumption of linearity can be identified in Figure 5. The distribution of the residual, however, suggests minor deviations from the assumption of normality, but not necessarily severe enough to constitute an assumption violation.

Figure 6. Assumption of residual homoscedasticity for multiple regression assessing the association between self-reported level of training completed and scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
According to Figure 6, the distribution of the residual appears to follow a distinct downward pattern that suggests a violation of the assumption of residual homoscedasticity.

Figure 7. Assumption of residual independence for multiple regression assessing the association between self-reported teaching experience and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
As depicted in Figure 7, the residual does not appear to be affected by participant number, indicating that the time at which participants completed the questionnaire did not impact results.

Figure 8. Assumption of residual normality and linearity for multiple regression assessing the association between self-reported teaching experience and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
No major deviations from the assumption of linearity can be identified in Figure 8. The distribution of the residual, however, suggests some deviation from the assumption of normality.

Figure 9. Assumption of residual homoscedasticity for multiple regression assessing the association between self-reported teaching experience and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
According to Figure 9, the distribution of the residual appears to follow a distinct pattern suggesting a violation of the assumption of residual homoscedasticity.
Further, a ceiling effect appears to be present.

Figure 10. Assumption of residual independence for multiple regression assessing the association between self-reported teaching experience and scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
As depicted in Figure 10, the residual does not appear to be affected by participant number, indicating that the time at which participants completed the questionnaire did not impact results.

Figure 11. Assumption of residual normality and linearity for multiple regression assessing the association between self-reported teaching experience and scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
No major deviations from the assumption of linearity can be identified in Figure 11. The distribution of the residual, however, suggests some deviation from the assumption of normality.

Figure 12. Assumption of residual homoscedasticity for multiple regression assessing the association between self-reported teaching experience and scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
According to Figure 12, the distribution of the residual appears to follow a distinct pattern suggesting a violation of the assumption of residual homoscedasticity.

Figure 13. Assumption of residual independence for multiple regression assessing the association between educational setting, teaching experience and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
As depicted in Figure 13, the residual does not appear to be affected by participant number, indicating that the time at which participants completed the questionnaire did not impact the results.

Figure 14. Assumption of residual normality and linearity for multiple regression assessing the association between educational setting, teaching experience and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
No major deviations from the assumption of linearity or normality can be identified in Figure 14.

Figure 15. Assumption of residual homoscedasticity for multiple regression assessing the association between educational setting, teaching experience and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
According to Figure 15, the distribution of the residual does not suggest a violation of the assumption of residual homoscedasticity.

Figure 16. Assumption of residual independence for multiple regression assessing the association between educational setting, teaching experience and scores on the Beliefs about the Role of Parents of ADHD component of the ADHD Beliefs Scale-Revised
As depicted in Figure 16, the residual does not appear to be affected by participant number, indicating that the time at which participants completed the questionnaire did not impact the results.

Figure 17. Assumption of residual normality and linearity for multiple regression assessing the association between educational setting, teaching experience and scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
No major deviations from the assumption of linearity can be identified in Figure 17. The distribution of the residual, however, suggests a deviation from normality which may be an indication of an assumption violation.

Figure 18. Assumption of residual homoscedasticity for multiple regression assessing the association between educational setting, teaching experience and scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
According to Figure 18, the distribution of the residual appears to be heteroscedastic, indicating a violation of the assumption of homoscedasticity.

Figure 19. Assumption of variance homogeneity for ANOVA assessing sex differences in scores on the Beliefs about the Neurobiology of ADHD component of the ADHD Beliefs Scale-Revised
Judging by Figure 19, the distribution of scores appears to be relatively homogeneous across the two groups indicating that the assumption of variance homogeneity is met.

Figure 20. Assumption of variance homogeneity for ANOVA assessing sex differences in scores on the Beliefs about the Role of Parents in ADHD component of the ADHD Beliefs Scale-Revised
While the variance across groups as depicted in Figure 20 is not fully homogeneous, this does not constitute a violation of the assumption of variance homogeneity.