搜索
1AnalysisofQuantitativeDataSocialResearchMethods2113&6501Spring,2007March26~April2,20072QuantitativeAnalysis:convertdatatoanumericalformandstatisticalanalyses•quantification(量化):theprocessofconvertingdatatoanumericalformat(將資料轉換成數字形式)3PartI.DealingWithData•(過錄)編碼簿製作、原始資料編碼(acodebook&codingdata)•資料輸入(enteringdata)•清理資料(cleaningdata)4CodingData(資料編碼)•DataCoding:systematicallyreorganizingrawdataintoaformatthatismachinereadable(將原始資料有系統地轉化成電腦程式可閱讀的形式)•Needacodingprocedure&acodebook–Codingprocedure(編碼程序):asetofrulesstatingthatcertainnumbersareassignedtovariableattributes(一些設定的規則，以確定變數屬性的編號為何)5CodingData(資料編碼)•Acodebook(編碼簿):adocumentdescribingthecodingprocedureandthelocationofdataforvariablesinaformatthatcomputerscanuse(一份描述編碼程序及變數位置的文件)–Shouldbewell-organized,detailed(有條理、詳細)–Makemultiplecopies!(多影印幾份，妥善保存)–Precoding(事前編碼)done–Codeeachquestionnaire(每份問卷都需編碼)6Acodebookexample71.您認為系上提供了您哪些協助？(可複選)help_d1□選課2□生活3□獎學金4□就業資訊5□其它，請說明：_________help_do______8:Skip9:MissingCodebookExample8•請問影響您繼續升學的主要原因：(可複選)1_a1□目前所就讀的科系有很大的發展前途1_a2□為了滿足父母的期待1_a3□家庭經濟足以支持您繼續升學1_a4□個人成績優異1_a5□對從事學術研究有興趣1=yes,0=no1_a6□目前就業率不高8=skip,9=missing1_a7□目前就業市場不需所就讀科系之人才1_a8□為了延後兵役1_a9□繼續升學可獲得較佳的升遷機會1_a10□繼續升學可獲得較高的薪水1_a11□個人認為取得碩士文憑以上是重要的1_a12□社會普遍升學風氣興盛1_a13□因為同學都升學1_a14□老師的建議•勾選完後請將您認為最重要的三個原因圈選出來1mi_1,1mi_2,1mi_388=skip,99=missing9編碼注意事項:•製作編碼簿(acodebook)–記得加上兩個變數:編碼者號碼及受訪者ID–Ex:變數名稱—coder&ID•製作SPSS資料檔–輸入變數名、變數註解、數值及註解、遺漏值等•請用紅筆編碼，綠筆用來訂正•編碼者請在問卷左上方或右上方簽名•資料輸入者也請在問卷封面上簽名10Enteringdata(資料輸入)•Datarecords:–eachrow(列)representsarespondent(case)–eachcolumn(行):aspecificvariable•Fourwaystoenterdata:–Codesheet–Direct-entrymethod–Opticalscan–Barcode•Recommend:doubleentry(重複輸入)11CleaningData(清理資料)•Whycleanyourdata?Needaccuracy(準確度)!•Doubledataentrytoensuredataaccuracy.•Proceduresto“clean”data(清理資料的步驟):–Usefrequencytablestodocodecleaning(利用次數分配表檢查變數屬性數值、遺漏值等)–Contingencycleaning(consistencychecking)(一致性檢查)•EX:cross-classifyingtwovariables•Checkcontingencyquestions(檢查條件式問項答案)•Willdiscussedinmoredetailslater12PartII.DataAnalysis13Review:QuantitativeDataAnalyses•Univariateanalyses(單變量分析):asinglevariable–Distributions,centraltendency,variation•Bivariateanalyses(雙變量分析):theanalysisoftwovariables–Thescattergram,percentagingatable,contingencytables•Multivariateanalyses(多變量分析):analyzingmorethantwovariablessimultaneously14Pleasenote:somecalculationsarenotsuitabletoallvariables•Continuousvariables(ratiovariables)(連續變數):avariablewhoseattributesformasteadyprogression–Ex:age•Discretevariables(間斷變數):avariablewhoseattributesareseparatefromoneanother,ordiscontinuous–Ex:gender–Nominalorordinalvariables–Canuserawnumbers,percentages,ormodes(butnotveryinteresting)15Needaresearchplanforquantitativedataanalysis:•Thinkaboutyourresearchquestion(s)&researchpurposes(先想想你的研究問題及目的)[記得:分析資料不能偏離主題!]•Thinkaboutyourhypotheses(想想你的假設)•Mainvariables:dependentvariables&independentvariables(依變數及自變數是什麼?)•Proposeaplanfordataanalysis,timeline,hoursofwork,etc.(擬定計劃，如何進行及分工…)16Adoableprocessfordataanalysis•DataCleaning(willdiscussedindetaillater)•Understandyourdata–Lookatfrequencytablesofkeyvariablesagain(再看一次重要變數的次數分配表)–Findanypotentialproblemsforfurtherdataanalysis(ex:missingdatapercentage?)(尋找問題:遺漏值太多?其他?–Producesometablesfordescriptiveanalysis(製作表格，注意表格格式)–Havesomesensefordatarecording(想想變數如何重新編碼)17DataAnalysis:datarecodingandcreatenewvariables•Whydatarecodingandcratenewvariables?–Formoremeaningfulanalyses–Ex:dummyvariables,collapseresponsecategories•Datarecodingandediting:Thinkaboutyourhypotheses,havepurposes•Remembertocheckconsistencyafterrecording(usefrequencytables)(新舊變數次數分配要一致)•Addvariableandvaluelabels(新變數要有變數及數值註解)18Bivariateanalyses(雙變量分析)•確定你的研究性質•建議:想想你的依變數，如果是名義或次序量數，只能用cross-tabulation與卡方來分析相關性。•依變數如果是等距或等比量數，可採用複迴歸。自變數要先分析變數之間的共線性。19BivariateTables•Contingencytables:valuesofthedependentvariablearecontingenton(dependon)valuesoftheindependentvariables•Howtopercentageatable?–Note:thedirectionofpercentagingintablesisarbitrary–Ingeneral,wefollowthis“tip”:“row”by“column”(dep.varbyindep.var)–UsuallydonebySPSSofothercomputerprograms•Howtoreadatableofpercentages?2021Readingapercentagetable:•Aruleofthumb(thetip):–ifthetableispercentageddown,readacross(表格是直行百分比，依橫列來解讀)–Ifthetableispercentagedacross,readdown22BivariateTableswithoutPercentages•Usethemeaninabivariatetable(i.e.,subgroupcomparison)23Anexampleofcontingencytableanalysis:Q:性別與大學主修之關係24卡方值(Χ2)是17.825，自由度是1，由表得知此卡方值大於10.827(p=0.001)，所以得到此卡方值的機率應低於0.001，觀察到的性別與大學主修之間的關係不可能只是因抽樣誤差而產生。由列聯表中可看出，約半數的男性選擇念人文社會或自然科學，但女性僅有略多於四分之一的比率(26.6%)在大學時選擇主修自然科學，性別與大學主修之差異在統計上是顯著的。majorofcollege-basedonv13bm*1性別Crosstabulation7910518442.9%57.1%100.0%49.7%73.4%60.9%26.2%34.8%60.9%803811867.8%32.2%100.0%50.3%26.6%39.1%26.5%12.6%39.1%15914330252.6%47.4%100.0%100.0%100.0%100.0%52.6%47.4%100.0%Count%withinmajorofcollege-basedonv13bm%within1性別%ofTotalCount%withinmajorofcollege-basedonv13bm%within1性別%ofTotalCount%withinmajorofcollege-basedonv13bm%within1性別%ofTotal人文社會商科自然工程majorofcollege-basedonv13bmTotal男女1性別Total25Anexampleofdataanalysis:教育成就的世代差異Q:什麼因素可