In addition to examining the measurement properties for the CAARS 2 Content Scales scores, demographic group differences were also assessed for the CAARS 2 Associated Clinical Concern Items and Impairment & Functional Outcome Items (note that MI and DTF analyses are not item-level analyses and, therefore, are not included in this section).

Item-level analyses were based on the determination of an Associated Clinical Concern Item as “Endorsed” or an Impairment & Functional Outcome Item as “Elevated” (for more information about item elevations, see chapter 4, Interpretation). For the Associated Clinical Concern Items regarding suicidality¹ and self-injury, “Endorsed” refers to a response of at least 1, “Just a little true; Occasionally“. For Depression and Anxiety screener items, “Endorsed” refers to a response of at least 2, “Pretty much true; Often/Quite a bit” (with the exception of individuals aged 70 years or older for the Depression screener item, for whom “Endorsed” refers to a response of at least 1, “Just a little true; Occasionally“).

To understand the potential for bias in these items, differences are presented as Cliff’s delta effect size values (Cliff’s d), which compare the frequency of each response option between the two groups. Values closer to -1 indicate that the first group tended to endorse lower values than the second group, values closer to 1 indicate that the first group tended to endorse higher values than the second group, and values close to 0 represent little difference between the groups (Romano et al., 2006). The size of the group differences in terms of item-level endorsement for gender, race/ethnicity, country of residence, and EL² was explored with the same matched samples described in the corresponding analyses for the Content Scales earlier in this chapter.

Results of these item-level analyses (presented in Tables 10.21 to 10.26) are based on subsets of the Normative Samples and are described in earlier sections of this chapter that correspond to the target variable. Effect sizes were negligible or very small for nearly all comparisons. The median effect size across all comparisons was Cliff’s d = |0.07| for Self-Report (ranging from 0.00 to |0.27|) and |0.04| for Observer (ranging from 0.00 to |0.17|). Overall, these observed effect sizes demonstrate how minimal the differences in the distribution of response frequency are between the various demographic groups, such that responses to these items are not unduly influenced by the individual’s gender, race/ethnicity, country of residence, or EL. Note that overall, endorsement was relatively low (as compared to individuals with clinical diagnoses; see chapter 9, Validity, for more information) for these severe items, as individuals analyzed here comprise the general population. The very small differences observed between demographic groups on the Associated Clinical Concern Items and the Impairment & Functional Outcome Items further support that the CAARS 2 item-level indicators are generalizable and can be fairly applied across diverse populations.