網路教育問卷調查新技術與線上模糊統計分析建構1/2

Abstract

內容： 本研究的目的乃建構網路模糊問卷調查與模糊統計軟體的設計。另外對於模糊數的無母數檢定我們提供了一個較為一般的方法，可以針對梯型、三角型，區間型的模糊樣本同時進行處理。而利用新設計的模糊問卷，可以較清楚地記錄受訪者的感覺，此外我們所提供之對於模糊數的無母數檢定方法比其他傳統檢定方法限制更少，對田野調查應用範圍更廣。 我們解決有關：1.當回答樣本非實數值時（例如基於模糊邏輯之區間或多值）模糊統計量處理。2.當所施測的樣本數很大時，如何有效率的在網路上記錄受測者所建構的隸屬度函數？3.建構網上即時模糊統計分析。本研究預期可完成實用之網路教育問卷調查新技術設計工具，線上模糊統計分析操作手冊與模糊統計分析決策說明各一份。使教育研究學者能迅速且精確獲取問卷結果，未來在教育調查研究過程中，更能有效且精確發揮功能。 本年度已完成的內容有兩個方向，第一個方向為發展模糊理論，而相關係數為本年度發展之重點。相關係數在教育領域上應用的很多，但是目前並沒有一套簡單、可行的理論可以用於模糊數據，因此，研究團隊在本年度發展了模糊相關的計算方式。第二個方向為發展模糊網路問卷系統，目前已經建構一個平台，研究者可以自行上去設計模糊問卷，並實施線上施測與蒐集數據。 對教育政策之建議與貢獻： 在教育政策研究課題中，常應用問卷抽樣調查或田野調查進行分析研究。而在資料收集與分析時，敘述統計能夠簡單且快速地描述資料的基本結構。例如在很多公共議題上，特別是當我們迫切地想了解大眾的看法及教育課題上的主流意識時，使用眾數會比平均數要來得恰當。但是傳統的統計量夠用嗎？從認知心理學的觀點，人的認知與訊息的處理包含明確和不明確的兩類知識。而事件的邏輯常是多值邏輯而非二元邏輯，人們的主觀感覺很難用單一數值表達的，因此適宜用模糊數來進行心智性的評估。且傳統的數字易犯過度應用和過度解釋之誤，如能應用模糊的數值資料，較能避免這樣的疑慮。 以模糊理論的觀點來分析教育統計資料，應比傳統統計學方法更為可行，而這也是教育政策執行者，所需加以著重的新統計方法。因為模糊統計，對傳統數值模型無法解釋複雜人文現象的改進更具有重大啟示。傳統調查與統計分析，常限於僅是限於單一值的統計量，無法表現出人類思維及主觀意思的複雜性，故為了要將人類思維做更合理且精確的描述，發展與使用模糊軟計算方法是極需要的。 調查研究是目前各領域常用的方法，可用於描述性、解釋性或探索性的研究。依據特定研究目的，從人類現實社會中蒐集資料以驗證事實的一套合乎科學的研究方法。它探討人類行為模式與思想理念，提供決策參考或作為學術理論基礎。但是在科學研究中所提假設，定義範圍及其發展成可測量的經驗指標時，由於問卷的設計上遣辭用句，語意順序以及問題型式關係，結果就產生相當差異。 傳統問卷設計與分析，常用勾選的方式讓受訪者在數個選項中選唯一一個答案。如調查人類的心情時，人的心情是否可以用好或不好來完全分隔清楚呢？其實絕大多數人的內心是介於中間，當受訪者的回答不只一個時，單一選答結果，可能造成資料難以精準地反映事實真相。模糊問卷調查利用模糊邏輯概念來改良傳統問卷，允許受訪者在數個選項中填寫隸屬度，改進傳統問卷。 此外，隨著網路的發展，除了傳統問卷外，新興的網路調查亦成為新的選擇。網路調查雖然只有一個較為短暫的歷史。但因調查研究的受訪者，正可借助以個人行為為主體的互聯網進行網路調查研究。所以網路調查顯現出來的獨特優點，開始勃興，甚至在一定範圍內呈現取代傳統調查方式而成為一種主流調查方式的趨勢。因此，本研究的研究成果，除了可以使用在各種教育問卷的資料蒐集上 (包括傳統問卷、模糊問卷)，也可以利用第二年所研發的統計軟體，進行數據分析。如此一來，教育研究者無須怯步於艱深的統計理論與計算模式，而可以輕鬆的使用本研究所研發的統計軟體進行分析。 中長期而言，本研究計畫所擬完成的研究成果，預期可以達成的具體貢獻和效益為透過本研究所完成實用之網路教育問卷調查新技術設計工具，教育研究學者能迅速且精確獲取問卷結果，未來在教育調查研究過程中，更能有效且精確發揮功能。此外模糊統計與模糊問卷系統為國內首創，若能將此研究構想加以推廣，必能提升國內教育研究能量與提升國內教育測驗研究效能。

Content: The purpose of this project is to develop a methodology for getting a fuzzy data on the educational sampling survey research by using the software The Geometer`s Sketchpad (GSP). We construct an online fuzzy testing/decision process (including nonparametric methods) to deal with fuzzy data such as trapezoid, triangular, and interval-valued type. On the other hand, using fuzzy questionnaire designed can help respondents to record their thoughts more precisely. Additionally our method for nonparametric tests with fuzzy data is more powerful than the traditional ones. We will solve: 1. How to process the fuzzy data when the reply is not a real value. 2. How to record the membership function on line, especially when the sample size is large? 3. Construing an on line fuzzy statistical testing and decision system. Finally we are expected to present: a guide of well designed educational fuzzy questionnaires techniques, a guide with collecting fuzzy data and a manual how to process the nonparametric testing and decision online system. We have already completed two goals this year. First is to develop the fuzzy correlation. Correlation is the most important statistical approach in education, but for now, there is no simple way to calculate the correlation index for fuzzy data set. We developed some approaches to calculate fuzzy correlation this year. Second, we constructed a fuzzy survey system. The researchers can log in to the system and design their own fuzzy questionnaires, collect data and analyze the data online. Recommendation for Educational Policy: In the research of educational policy, the sampling survey is always used to evaluate and understand public opinion on certain issues. The traditional survey forces people to choose fixed answer from the survey, but it ignores the uncertainty of human thinking. For instance, when people need to choose the answer from the survey which lists five choices including \"Very satisfactory,\" \"Satisfactory,\" \"Normal,\" \"Unsatisfactory,\" \"Very unsatisfactory,\" despite of the fact that the answer of the question is continual type, we may be only allowed to choose one answer. It limits the flexibility of the answer and forces people to choose fixed answers. When the survey proposes to have the answer for sleeping hours of a person, it will be difficult to describe the feeling or understanding reasonably unless the fuzzy statistics are adopted. Traditional statistics deals single answer or certain range of the answer through sample survey, and unable to sufficiently reflect the thought of an individual. If people can use the membership function to express the degree of their feelings based on their own choices, the answer presented will be closer to real human thinking. Therefore, to collect the information based on the fuzzy mode should be more reasonable. In the consideration for the question related with fuzzy property, the information itself had the uncertainty and fuzzy property. The development of internet makes the web survey becomes the new options for educational policy maker. Although web survey is relatively new compare to the traditional paper-pencil survey, this becomes quite popular nowadays. Not only because its easiness to collect data, but also it can save a lot of time to code data. In this project, we will design a web survey system, which can be used to design both traditional and fuzzy questionnaire. In the second year, we will also develop a statistical software which can be used to analyze the fuzzy data. Thus, the education researcher will not be impeded by the difficult statistical theory or calculation model, and can easily implement their analysis by using the software developed from this study. In the long run, the results of this project can make contribution for research. Fuzzy statistics is relatively new but it is quite useful to capture the real thought of human. The completion of this project will shed some lights for the educational field in Taiwan.