FILENAME website HTTP 'http://www.public.asu.edu/~rgcox2/ECN410/data/poverty2000.xlsx';

data file;
   n=-1;
   infile website recfm=s nbyte=n length=len;
   input;
   file  "poverty2000.xlsx" recfm=n;

   put _infile_ $varying32767. len;
  run;

  /*
Variables in the data set are,

  p_poverty : proportion of the state's population that lives below the poverty line
  unemrate : unemployment rate expressed as a propotion
  p_rural : proportion of the population living in rural areas
  lnpcinc_adj : natural log of percapita income adjusted for inflation
  p_hs_grads_19plus : proportion of the state's population age 19 and over that are high school graduates
  state : the state
  */
PROC IMPORT OUT= WORK.poverty
            DATAFILE= "poverty2000.xlsx"
            DBMS=xlsx REPLACE;
       *RANGE="A1:D64";
       /*sheet="NAME"*/
       getnames=yes;

RUN;

proc means data=poverty;
run;

/* fixed effects, use dummy variables for the states*/

proc genmod data = poverty;
  class state /param=glm;
  model p_poverty= unemrate p_rural lnpcinc_adj p_hs_grads_19plus state;
  repeated subject=state; /*use this statement to obtain robust standard errors*/
  *store p1;
  run;

proc glmselect data=poverty;
  class state;
 model p_poverty= unemrate p_rural lnpcinc_adj p_hs_grads_19plus state / select = rsquare ;
run;
quit;


proc glmselect data=poverty;
  class state;
 model p_poverty= unemrate p_rural lnpcinc_adj p_hs_grads_19plus state / select = AIC ;
run;
quit;


/* random effects */
proc reg data=poverty;
model p_poverty= unemrate p_rural lnpcinc_adj p_hs_grads_19plus;
run;
quit;