抽樣調查不完全回收的問題研究

機構典藏 > College of Business and Management > Department of Management Sciences > Thesis > Item 987654321/33108

Please use this identifier to cite or link to this item: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/33108

Title:	抽樣調查不完全回收的問題研究
Other Titles:	A study of incomplete response problem in sampling survey
Authors:	賴正隆;Lai, Cheng-lung
Contributors:	淡江大學管理科學研究所碩士班張紘炬;Chang, Horng-jinh
Keywords:	抽樣調查;回收率;未回收樣本;Sampling Survey;Truncated Binomial Distribution;Nonresponse
Date:	2004
Issue Date:	2010-01-11 03:24:25 (UTC+8)
Abstract:	調查是取得統計資料的一種方法，一般可以分為普查與抽樣調查兩種。考慮人力、財力及時間等因素時，若不使用抽樣調查，則會喪失或減低其原有的價值；而且可以節省大量的人力、物力、縮短時間，也可能比大規模的普查更有效率。但是過低的回收率所產生的無反應偏差(nonresponse bias)卻是一個嚴重的問題。通常抽樣調查只看到以所回收的部分樣本資料推估母體特性，事實上，所回收的部分樣本資料只能作為可回收子母體的樣本，不能代表不可回收子母體，所推估的結果也只是推估可回收子母體的特性，若以其作為推估原母體的特性，準確度當然令人懷疑。因此，本研究以韓森-霍維茲(Hansen- Hurwitz)所提出有關對未回收樣本之處理的觀念，以簡單隨機抽樣法，探討抽樣調查中，只以回收的樣本資料推估原整個母體的影響；以及探討如何處理以所設計的資料蒐集方式無法回收樣本資料的問題，本研究將採固定數額追蹤方式追蹤原未回收資料的樣本，並結合原已回收的樣本資料，提出新的估計量，以為調查實務應用之參考。 Survey is one of the ways to collect statistical data which include census and sampling approaches. To consider the factors of manpower、financial and time, sampling survey is more effective than census. Besides, it could gain important information somehow. However, non-response always got the serious mistakes to influence result. So we need to discuss the response rate to check if it has big biased problem. In practical, people often use the response date to substitute to the whole date of the number of surveys. People will doubt if it is accurate enough to only use the response data. The method of my paper is from the paper of “ The Problem of Non-response in Sampling Survey “ of Hansen and Hurwitz which thought of the approach to deal with non-response problems. Not only I try to discuss the mistakes which we just use response sample to estimate population, but also consider non-response sample to correct the former biased problem. I make the constant number from non-response sample to be the new non-response sample instead of the original non-response sample to find a new sample by using the simple random sampling method. I provide the new approach to provide the decision persons to make the sample what they want when they do the sampling survey.
Appears in Collections:	[Department of Management Sciences] Thesis

Files in This Item:

File	Size	Format
	0Kb	Unknown	353	View/Open

Loading...