The classical Bland Altman agreements (BA LoA) define an area in which about 95% of the differences in normal distribution between measures by type should be[1][2][3]. In cases of non-normal differences, the use of empirical quantiles has been proposed as a robust alternative[2][4][5]; However, significant research efforts in the past have proposed the application of nonparametric quantile estimation to the evaluation of 2.5% and 97.5% perzentiles as nonparametric loA. We conducted a simulation study on 15 nonparametric quantile estimates to determine nonparametric prediction intervals and evaluate their performance based on the coverage probability of a newly generated observation. The objective of this study was to propose a non-parametric and robust alternative to classical BA LoA if the normality hypothesis is not maintained and/or if the sample size is small to moderate. For n=30, none of the evaluators achieved the nominal coverage probability of 0.95. The coverage probability of SQp1 was close to 0.95 and ranged from 0.934 to 0.938. Note that for sample sizes up to n=40 observations, the smallest and largest difference is used as nonparametric quantitative estimates for the percentiles 2.5% and 97.5% respectively. For SQIp, HDp, HDplc, SVp1, SVp2 and SVp3, the coverage probabilities were at least 0.921, 0.911, 0.910, 0.920, 0.914 and 0.923 respectively. Neither SQp2 nor KLp are defined for n<40. Barnhart HX, Haber MJ, Lin LI. An overview of the assessment of conformity to continuous measures. .

. .