The primary intended use of the CAARS 2–ADHD Index is to accurately distinguish between ratings from individuals with and without ADHD. To validate that the CAARS 2–ADHD Index is effective at this classification activity, scores on the CAARS 2–ADHD Index were compared between individuals from the general population and individuals with all presentations of ADHD (i.e., ADHD Inattentive, ADHD Hyperactive/Impulsive, and ADHD Combined). A description of the Total Sample, which contains the General Population group used here, can be found in Table 6.4 in the Standardization Phase section of chapter 6, Development, and details about the ADHD Reference Samples can be found in Tables 7.11 to 7.15 of chapter 7, Standardization.

CAARS 2–ADHD Index probability scores were computed based on raw scores and classified into one of five score categories, ranging from Very Low to Very High (see Table 4.6 in chapter 4, Interpretation). The percentage of each sample with CAARS 2–ADHD Index probability scores in each score category was compared, with results presented in Table 12.15. Individuals from the General Population group (without a clinical diagnosis) overwhelmingly received CAARS 2–ADHD Index probability scores between 1% and 39%, falling in the Very Low and Low ranges (89.2% for Self-Report; 85.6% for Observer). Conversely, and as expected, individuals diagnosed with ADHD were much more likely to have CAARS 2–ADHD Index probability scores greater than 59%, falling in the High and Very High ranges (93.7% for Self-Report, and 81.7% for Observer). Although there is a gradient of possible scores for both groups, it is clear from the skewed distributions of scores shown in Table 12.15 that the probability score captures the increased likelihood of high scores reflecting ratings from individuals diagnosed with ADHD. These results provide support for the use of the CAARS 2–ADHD Index for accurate classification of individuals with and without ADHD.

Another way to examine the ability of the CAARS 2–ADHD Index to predict group membership is to dichotomize scores into two groups: below 60% (i.e., Very Low, Low, or Borderline ranges, where individuals from the general population would be expected to score) and 60% or greater (i.e., High or Very High ranges, where individuals diagnosed with ADHD would be expected to score). Once the data were dichotomized, a second set of analyses investigated the dichotomized classification category of an obtained probability score in contrast with an individual’s actual group membership via confusion matrices. Using the approach outlined by Kessel and Zimmerman (1993), the confusion matrices were then used to derive classification accuracy statistics (see Inconsistency Index in chapter 6, Development). Predicting diagnostic status depends on the prevalence of ADHD in the population. The prevalence (or base rate of ADHD in the relevant referral population) can vary widely depending on the purpose of the evaluation and the setting. For example, whereas an ADHD prevalence rate of around 10% or less might be expected in a general population screening context, expected rates might be around 50% in a clinically referred sample and as high as 80% among referrals to an ADHD specialty clinic. Accordingly, the classification accuracy statistics, assuming a 50% base rate, are summarized in Table 12.16, and the Positive Predictive and Negative Predictive Values based on varying base rates are provided in Table 12.17. The overall correct classification rate was 93.1% for Self-Report and 85.5% for Observer.