抽樣樣本數的推估

The Estimate of Sample Size

鄧天德(Tien-The Teng)

《北市教大社教學報》 10期 (2011/12) Pp. 191-189

https://doi.org/10.7035/PSCTSCHP.201112.0189

摘要

不論自然科學或人文科學，量化研究仍然占大多數。而量化研究常需透過抽樣的方法，再依據樣本的統計量來推估母群體的參數。本文旨在討論在不同的假設條件下，如何抽取適量的樣本數來推估母體的參數。抽樣前應先確認母體的兩大特性，包括母體的分配模式與數量。母體分配模式分為常態分配及非常態分配，數量方面分為無限母體及有限母體。其次討論推估抽樣數的基本條件，主要有三：1.顯著水準α2.樣本的標準差s3.預期的抽樣誤差值e本文依上述兩大分類及三個基本條件，分別提供五個抽樣公式。

關鍵字

抽樣 ； 統計量 ； 參數 ； 常態分配 ； 無限母體

並列摘要

No matter what kind of research subject, natural sciences or humanities, the quantification research still occupied the majority. The sampling method is often essential in the quantification research. This article is for the purpose of discussing how to sampling an adequate sample size to estimate the population parameters within the given conditions in order to reduce the subjective element in selection to a minimum. In the sampling procedures, a researcher should first confirm the population distribution model and the population quantity, which are the sample size concerned. Generally, the population distribution models are divided into both the normal distribution and the non-normal distribution. The quantitative aspect is divided into both the infinite population and the limited population. Next, a researcher should determine the sample size within the following given conditions, which the sample size depends on: 1. Level of significance, 2.Sample standard deviation, 3. Expected sampling error. According to the above classification and the conditions, five sampling equations are recommended separately.

並列關鍵字

sampling ； statistic ； parameter ； normal distribution ； infinite population ； limited population