Peil J, Rother P
Gegenbaurs Morphol Jahrb. 1984;130(3):461-72.
A discussion of the possibilities is given of parametric and of nonparametric description of measured aging processes. As examples serve the counted volume portions of different histological structures in the human submandibular gland. As best suited approach the nonparametric evaluation of data by empirical regression is recommended. It is studied which qualitative changes are shown by the empirical regression curves if the sample size will be modified. For this the original sample size of 42 measured values is reduced in 2 steps to 35 and 29 values by removing at random firstly 7 and secondly additional 6 values. The tendencies in the aging courses represented by empirical regression curves will hardly altered by these reductions whereas a nonrandom elimination of outstanding points yields to marked changes in the calculated empirical regression curves. The influence of different local sample variances and densities of the measured points on the shape of the aging curves is stressed.
讨论了测量衰老过程的参数化和非参数化描述的可能性。以人类下颌下腺中不同组织结构的计数体积部分为例。建议采用经验回归对数据进行非参数评估。研究了如果修改样本量，经验回归曲线会呈现哪些定性变化。为此，通过首先随机去除7个值，然后再额外去除6个值，将原始的42个测量值样本量分两步减少到35个和29个值。这些减少几乎不会改变经验回归曲线所代表的衰老过程中的趋势，而有选择地去除异常点则会使计算出的经验回归曲线产生显著变化。强调了不同局部样本方差和测量点密度对衰老曲线形状的影响。
