Wilcoxon pair sampling tests were used to detect possible systematic management trends for different groups of advisors. None of the comparisons between the subgroups reached (all ≥ 0.05). Thus, we found no evidence of a systematic orientation of the discrepancy in rating, either for bilingual children or for monolingual children. In statistics, reliability between advisors (also cited under different similar names, such as the inter-rater agreement. B, inter-rated matching, reliability between observers, etc.) is the degree of agreement between the advisors. This is an assessment of the amount of homogeneity or consensus given in the evaluations of different judges. Second, it must be decided whether subjects evaluated by multiple coders should be evaluated according to the same coder sentence (completely cross-design) or whether different subjects are evaluated from different subsets of coders. The contrast between these two options is shown at the top and bottom of Table 1. Although completely cross-concepts may require a higher total number of ratings, they allow for a systematic distortion between coders, which can be evaluated and controlled in an ERROR estimate, which can improve overall IRR estimates. For example, CCIs may underestimate the true reliability of some designs that are not fully cross-referenced, and researchers may need to use alternative statistics that are not well distributed in statistical packages to evaluate ERREURS in some studies that are not completely cross-referenced (Putka, Le, McCloy, – Diaz, 2008). The syntax for calculating ICCs with SPSS and the R-package is shown in Table 6. Both procedures offer point estimates, confidence intervals, degrees of freedom and significance tests for the zero hypothesis as ICC 0. In practice, only point estimates are generally reported, although confidence intervals can provide additional useful information, especially when CCIs are low or when the confidence interval is large due to a small sample size.
The results of the significance tests are generally not reported in the IRR studies, since irr estimates for trained coders will generally be greater than 0 (Davies – Fleiss, 1982). Many research projects require an evaluation of IRRs to show the extent of the interim agreement between coders.