Choudhary PK, Nagaraja HN: conformity assessment tests based on probability criteria. Review of statistical planning and inference. 2007, 138 (4): 1102-1115. 10.1016/j.jspi.2007.03.056. With regard to the comparison of the two levels of compliance, the point estimates for all the variables concerned did not differ significantly between Fleiss`K and Krippendorffs Alpha, regardless of the observed concordance or the number of categories (Table 2). As we suggested in our simulation study, the confidence intervals for Fleiss`K were narrower when the asymptotic approach is used than in the Bootstrap approach. The relative difference between the two approaches was reduced as the observed concordance was low. There was no difference between the bootstrap confidence intervals for Fleiss`K and Krippendorffs Alpha. The concordance between the methods can be assessed by the following distribution: (y ij· – y ij`) ~ N (μ D, ) and, therefore, Lin  defined the TDI as a limit, κ p, which covers a large proportion of even measurement differences of two devices or observers within the boundary, i.e.
the value of κ p, the P(| D| < κ p) = p, where D is the difference in pairs. In the hypothesis of the mixed model in (1), D is the even difference based on one of the replicates, D = (y ijl – y ij`l`) and that is why, on the basis of D, κ p is effectively called global TDI to assess overall conformity . It is shown that the distribution of D is then D ~ N (μ D, σ D) with μ D = β j – β j and , or in case of different error variants between devices. Hoy D, Brooks P, Woolf A, Blyth F, March L, Bain C, Baker P, Smith E, Reliure R. Assessing risk of bias in prevalence studies: modification of an existing tool and evidence of interrater agreement. J Clin Epidemiol. 2012;65 (9):934-9. Finally, we would like to point out that the method proposed here is simple, because the TDI estimate is directly derived from a probability interval of a variable normally distributed in its initial scale, without any further transformation.
Subsequently, the natural possibility of drawing conclusions about this estimate is to deduce the corresponding IT. The expression of our TI proposition corresponds to the exact one-sided IT defined by Hahn in 1970  for at least a predetermined proportion of a population of normal distribution, with the particularity that the indicated proportion is found with a search algorithm to ensure that the confidence limits are symmetrical by 0. . .