A new Method of Treating Frequency Curves in Fishery Statistics, with some results
A new Method of Treating Frequency Curves in Fishery Statistics, with some results
复制标题
渔业统计中频率曲线处理的新方法，取得了一些成果
DOI：
10.1093/icesjms/4.2.207
复制
发表时间：
1929
影响因子：
3.3
通讯作者：
W. C. Hodgson
中科院分区：
文献类型：
--
作者：
H. J. Buchanan;W. C. Hodgson
本文所讨论的方法是将一定思路贯彻到逻辑结论中的结果。无论如何，这些想法本身并不完全是新的，而且肯定是在许多渔业统计学家的头脑中独立产生的，但似乎没有任何关于它们的应用和发展成为一种确定方法的出版物。它们的适用性取决于分别处理每一个鱼类测量样本，或只一起处理在时间和空间上非常有限的”区域”内的数据。在过去，人们一般认为，最有用的统计图片的长度频率分布和growthrate是通过处理非常大的样本，并通过大量的样本组合范围内不是非常有限的时空区域。数据中的”粗糙度”一般被认为是由于采样不足和测量中的微小误差造成的，因此被平滑掉了，由此产生的平滑曲线被认为比原始的粗糙数据给出了更正确的图像。本方法建立在一组根本不同的假设之上，结果也有根本不同。如果这些都是正确的，统计”常数”由通常的方法是相当不正确的和”平均大小”，”标准偏差从平均值”，和其他常数在一般使用的传统符号只是和意味着没有真正确定的。“增长率”如果从这些常数计算出来，将是毫无意义的，因为它是在假设始终处理同一人口的情况下从不同人口的增长率计算出来的。在接下来的几页中，我们将看到，有强有力的证据表明，鱼类测量原始数据中的”粗糙度”，
The method which is the subject of the present paper is the result of carrying a certain train of ideas to its logical conclusion. The ideas themselves are at any rate not entirely new, and have certainly arisen independently in the minds of many fishery statisticians, but there does not seem to be any published account of their application and development into a definite method. Their applicability depends on treating each sample of fish measurements separately, or on treating together only such data as are limited to a very restricted" area" in time and space. In the past it has generally been assumed that the most useful statistical pictures of length frequency distribution and growthrate are to be obtained by treating very large samples, and by a considerable amount of combination of samples ranging over a not very restricted time-space area. The" roughnesses" in the data, which have generally been considered to be due to insufficient sampling and to slight errors in measurement, are thus smoothed out, and the resulting smooth curves have been considered to give a more generally correct picture than the original rough data. The present method is founded on a fundamentally different set of assumptions, and the results are fundamentally different. If these are correct, the statistical" constants" given by the usual method are quite incorrect and the" mean size"," standard deviation from the mean", and other constants in general use are conventional symbols only and mean nothing really definite." Growth-rate" if calculated from these constants will be meaningless since it is calculated from the growth-rate of different populations on the assumption that the same population is dealt with throughout. It will be shown in the following pages that there is strong evidence that the" roughnesses" in the original data of fish measurements, however