/* module 8: problems distributed on handout 8-1 */

/* 8-1.1 */
data one;
  do i = 1 to 1000;
    x1 = ranbin(36,5,.5);
    x2 = ranbin(37,50,.5);
    x3 = ranbin(37,100,.5);
    x4 = ranbin(37,500,.5);
    x5 = ranbin(37,1000,.5);
    x1_1 = ranbin(37,5,.1);
    x2_1 = ranbin(37,50,.1);
    x3_1 = ranbin(37,100,.1);
    x4_1 = ranbin(37,500,.1);
    x5_1 = ranbin(37,1000,.1);
    output;
  end;
run;
proc chart data=one;
  vbar x1;
  vbar x2;
  vbar x3;
  vbar x4;
  vbar x5;
  vbar x1_1;
  vbar x2_1;
  vbar x3_1;
  vbar x4_1;
  vbar x5_1;
run;
* The rule of thumb for whether a binomial distribution;
* would be well approximated by a normal distribution  ;
* is if the lesser of np and n(p-1) is five or more then;
* the normal distribution becomes a good approximation.;
* By that, the plots for x1 and x1_1 should not look normal,;
* x2_1 should be borderline, and the rest should
* look normal.  This is reflected in the generated;
* histograms.  For x3_1, np is 10, and it doesn't look;
* entirely normal.;