• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 2
  • 2
  • Tagged with
  • 2
  • 2
  • 2
  • 2
  • 2
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

在抽樣調查中運用追蹤樣本對無反應所產生偏誤的修正探討

許石山 Unknown Date (has links)
抽樣調查是統計學很重要的一部份,抽樣方法恰當與否關係著調查結果的成敗,然而無反應卻是一個很棘手的問題。基於成本考量,一般都會假設其對估計結果不會造成太大的影響,然而事實證明,在很多時候無反應的確會影響到估計結果的準確度,準確度的高低必需與真實母體參數值做比較,我們較難取得真實母體的資訊,所以我們使用模擬的方式在不同情況下來證實使用二次訪問(也就是追蹤調查)其估計結果的準確度較一般訪問(不進行二次訪問)的準確度高出許多。這個結果在某些情況下是特別明顯的,當有反應者與無反應其所包含的資訊差異很大且無反應的比重較高時,一般訪問的估計結果偏誤情況會比二次訪問的估計結果偏誤嚴重。 針對無反應問題,許多學者提出一些補救的方法,例如加權法、插補法等等,然而至今沒有一個較明確且通用的方法可以有效改善無反應的偏誤問題,造成無反應的原因很多,我們只能針對這些可知的原因加以修正調查方式或內容;我們也不能假設有反應者與無反應者所代表的資訊會很相近,進而宣稱“可以只用有反應的樣本來估計母體參數並不會造成太大的偏誤”,我們只能從反應率方面著手來降低偏誤情況,提高反應率是我們可以努力的方向,從開始設計問卷到最後實施調查,中間很多環節都會直接或間接影響到反應率,我們只要針對這些環節多加考慮、規劃,盡可能免除會造成無反應的因素,相信對最後的反應率提高是很有幫助的。另一方面,追蹤訪問也是一個提高反應率的有效方法,針對第一次無反應的有效樣本再次進行重新訪問,如此也可以使偏誤降低,本研究著重在不同情況下進行二次訪問的效率探討及比較。
2

敏感性問題之無反應因素探究-以收入與投票意向為例 / Deeper Look into Factors Causing Nonresponse Phenomenon in Reaction to Sensitive Questions:Using Income Level and Voting Inclination as Examples

林彩玉, Lin, Tsai-Yu Unknown Date (has links)
調查研究被廣泛的使用於現今社會,而研究中出現頻繁的「無反應」問題卻常為研究者帶來統計分析過程中的困擾,基於預防重於治療的前提下,「無反應」因素的探究有其必要性。本研究的目的即在現有的理論基礎上,針對「敏感性」問題於「無反應」(收入及投票意向問題)中的情況,應用卡方檢定與logit模型驗証其於本研究的適切性,並進一步建立新的「無反應」問題解釋模型,以幫助我們瞭解此種「無反應」問題在現有台灣調查研究中可能存在的因素,進而減低對研究結果的影響。 影響「無反應」因素的驗證部份,大致上與過去相關的研究結果相同,只有少數的變項有所差異。另外,在模型配適的解釋部份,收入的研究結果中會影響收入「無反應」的因素有:受訪者年齡、受訪者職業、訪員性別、訪員經驗以及調查問題型式。投票意向的研究結果顯示,影響投票意向「無反應」的因素有:受訪者的性別、省籍、政治參與度、對統獨立場的看法、對族群認同的看法、受訪者的性別與對統獨立場看法的交互作用以及受訪者的省籍對族群認同看法的交互作用。本研究的研究結果主要適用於敏感性問題中影響收入與投票意向問題之「無反應」因素情況。 / Investigative process is widely used in our society today. However, statistical analysts are often bothered by the frequent occurrence of the nonresponse problem. Hence, it is important to investigate the cause of the nonresponse situation. The purpose of this research is to create a new model to explain the nonresponse problem in certain sensitive situations. There are two types of questions of sensitive nature; namely, income level and voting inclination, that lead to the nonresponse. We have explored two methodologies, Chi-square test and Logit model, for their applicability to the study. The study will help us to understand what are the factors influencing the nonresponse for investigative process conducted in Taiwan. These findings may be used to reduce the impact of the nonresponse in regards to how we can interpret the results. In terms of verifying influential factors of the nonresponse, there are only a few different outcomes in comparison with previous research results. To find factors in regard to income level, the following five elements are identified: the age of the interviewee, the occupation of the interviewee, the gender of the interviewer, the experience of the interviewer, and styles of the questionnaire. As for the voting inclination, we found the following factors: interviewee's gender, birth origin, degree of political participation, viewpoint toward the issue of unification/independence, and opinion of ethnic identity; interaction between interviewee's gender and his/her viewpoint toward the issue of unification/independence, and interaction between interviewee's birth origin and opinion of ethnic identity. The main application of this study is to identify the various factors causing nonresponse phenomenon in reaction to sensitive questions of income level and voting inclination during an investigative process.

Page generated in 0.1928 seconds