Statistics and Quantitative Analysis U4320 Segment 1: IntroductionProf. Sharyn O’HalloranPreliminariesn Professor Sharyn O’Hallorann Office Hours: Wednesday 9:00 to 11:00n Contact information:n Office: 727 IABn E-MAIL: [email protected] PHONE: X4-3242Overviewn How to describe numerical data.n How to make inferences about populations from samples.n How to evaluate the relation between variables, factors or events.Example 1: Educationn Main Finding: n Black and Hispanic students are far less likely to attend college than are white students.Example 1: Education (cont.)n Dependent Variable: n What is the phenomenon to be explained?n Here, it is the percent of minorities enrolled in collegen Independent Variable:n What factors might help explain this phenomenon?n We can think of many…n Evidence:n 34% of Whites 18-24 were enrolled in college in 1991.n 24% of Blacks 18-24 were enrolled in college in 1991.n 18% of Hispanic 18-24 were enrolled in college in 1991.Example 1: Education (cont.)n How would you represent this data graphically?Example 1: Education (cont.)Percent Attending College by Category 010203040CategoryWhiteBlackHispanic% Attendingn Measurement Problemsn What does 34% represent?n What is the reference group? n What is the sample population? Example 1: Education (cont.)n What is the causal relation between race & education?n Education is the dependent variable or the thing to be explained.n Race is the independent variable or the causing factor.n This is a causal model or a path diagram.n Simplest model:Race Education(Independent) (Dependent)Example 1: Education (cont.)n What does the article suggests? Race Income EducationIndependent Intervening DependentExample 1: Education (cont.)n Income is an intervening variable. n Because minorities tend to have lower incomes they are less able to afford education. n Implicationn Race affects education via incomen Policy Prescription:n Article argues that to improve educational attainment, need to ensure funding for minorities.n But what if income is not the problem?n What if the relation between race and higher education is due to discrimination or cultural factors?n How should we redirect government policy?Example 1: Education (cont.)n A second hypothesis postulated is that:n Race Income Dropouts(Independent) (Intervening) (Dependent)Course type Income OpportunitiesExample 1: Education (cont.)n Policy Implication: n Raising the minimum wage willincrease dropout rates. n Moral: n Different models of the world lead to different policy predictions.Example 1: Education (cont.)Example 2: EnvironmentIncinerator Health ProblemsIndependent Dependentn What other factors might intervene here?n How would these interpretations change the implied policy prescriptions?Harper’s Indexn http://www.harpers.org/harpers-index/listing.php3n Numbers are present as facts, as if they speak for themselvesn But numbers rarely speak for themselvesn As we have seen, they can have many different interpretations and causesn This course will teach you how to speak for the numbers (or else someone will do it for you)Goals of the Coursen The purpose of the course is introduce professional students to basic data analysis skills.n Develop techniques to test and evaluate competing models of how the world works.n Approachn Hands on / learning by doingMaterialsn Textn Wonnacott and Wonnacott (4th Edition)n Course Packetn Software n SPSS for Windows (Also available at the CU Bookstore)n Excel n SIPA Skills Coursen Datan 1998 GSS Data set available on the SIPA servern See “Why Take Statistics” in Course PacketTeaching Assistantsn One TAn Head Sections and Office Hoursn Weekly Labs Mandatoryn Meet in classroom, then move to labn One PRAn Gradingn Office HoursSupportn Website n http://www.columbia.edu/itc/sipa/U4320y-003n Newsgroupn Class bulletin boardn Check regularlyn Be-Nice Policyn Class Notesn PowerPoint slides available on website after classn Not a substitute for attending lectureGradingn Weekly Assignments (40%) n No late workn Presentation countsn Midterm (30%) n In classn 1 3x5 index cardn Final Paper (30%). n Can work in