/* describemath.sas */
title 'Gender, Ethnicity and Math performance';
title2 'Basic descriptive statistics on the exploratory sample';

proc format; 
     value ynfmt  0 = 'No'  1 = 'Yes';
     value crsfmt   4 = 'No Resp';
     value nfmt 
                1 = 'Asian'
                2 = 'Eastern European'
                3 = 'European not Eastern' 
                4 = 'Middle-Eastern  and Pakistani'
                5 = 'East Indian'
                6 = 'Other   and DK' ;

data math;
     infile '/home/brunner0/441s20/exploremath.data.txt';
     input id course precalc calc gpa calculus english mark lang $ sex $ 
           nation1 nation2 sample;

/* Computed Variables: totscore, passed, grade, hsgpa, hscalc, hsengl, 
                       tongue, ethnic */

     totscore = precalc+calc;
     if (50<=mark<=100) then passed=1; else passed=0;
     /* Some missing final marks were zero, and 998=SDF and 999=WDR */
     if mark=0 then grade=.;
        else if mark > 100 then grade=.;
        else grade=mark;
     /* Missing HS marks were zeros */
     if 65 le gpa le 100 then hsgpa = gpa; /* Else missing is automatic */
     if 0 < calculus < 101 then hscalc = calculus;
     if 0 < english < 101 then hsengl = english;
     /* There were just a few French speakers */
     if lang='French' then tongue='Other  '; else tongue=lang; 
     label tongue = 'Mother Tongue (Eng or Other)';
     /* Rater 1 knows Middle Eastern names -- otherwise believe Rater 2 */
     if nation1=4 then ethnic=nation1; else ethnic=nation2;

     /********************************************************************/

     label 
           precalc  = 'Number precalculus correct'
           calc  = 'Number calculus correct'
           totscore = 'Total # right on diagnostic test'
           passed = 'Passed the course'
           grade = 'Final mark'
           hsgpa = 'High School GPA'
           hscalc = 'HS Calculus'
           hsengl = 'HS English'
           lang = 'Mother Tongue'
           nation1 = 'Nationality of name acc to rater1'
           nation2 = 'Nationality of name acc to rater2'
           tongue = 'Mother Tongue (Eng or Other)'
           ethnic = 'Judged Nationality of name'        ;

     format course crsfmt.;
     format passed ynfmt.;
     format nation1 nation2 ethnic nfmt.;

/*********************************************************************/

proc means;
     title3 'Quantitative Variables';
     var precalc calc totscore hsgpa hscalc hsengl grade;

proc freq;
     title3 'Categorical variables';
     tables course sex ethnic tongue passed;

proc sgplot;
     title3 'Bar chart of ethnic background';
     vbar ethnic;

proc univariate normal plot; 
     /* The normal option gives tests of H0: Data are normal.  
     Plot produces graphics, some of which are informative. */
     title3 'Detailed look at grade';
     var grade;

proc freq;
     title3 'Frequency Distribution of Final Grade';
     tables grade;

/* The following suggests that profs in Courses 1 and 3 bumped  
   the marks discontinuously, while the prof in Course 2 did  
   something more sophisticated. */

proc freq;
     title3 'Grade separately by course';
     where course < 4; /* Exclude no response */
     tables grade*course / norow nopercent nopercent;

quit;