filename tui_data 'N:\share\exam\tuition.csv';
filename loc_data 'N:\share\exam\location.csv';

data tuition;
infile tui_data firstobs=2 dlm=',' missover dsd;
informat unitid 6. col_name $50. public 1.;
input unitid col_name tuition90 tuition91 public;
tuitionrise=(tuition91-tuition90)/tuition90;
run;

proc contents data=tuition; run;
proc means data=tuition; run;

data location;
infile loc_data firstobs=2 dlm=',' missover dsd;
informat unitid 6. city $20. state $2.;
input unitid city state;
if 	state="CT" or 
	state="ME" or 
	state="MA" or 
	state="NH" or 
	state="VT" or
	state="RI" or 
	state="NJ" or 
	state="NY" or 
	state="PA" then region="Northeast";
else if state="IL" or 
	state="IN" or 
	state="MI" or 
	state="OH" or 
	state="WI" or 
	state="IA" or 
	state="KS" or 
	state="MN" or 
	state="MO" or 
	state="NE" or
	state="ND" or 
	state="SD" then region="Midwest";
else if state="DE" or 
	state="DC" or 
	state="FL" or 
	state="GA" or 
	state="MD" or 
	state="NC" or 
	state="SC" or 
	state="VA" or 
	state="WV" or 
	state="AL" or
	state="KY" or 
	state="MS" or 
	state="TN" or 
	state="AR" or 
	state="LA" or
	state="OK" or 
	state="TX" then region="South";
else if state="AZ" or 
	state="CO" or 
	state="ID" or 
	state="MO" or 
	state="NV" or 
	state="NM" or 
	state="UT" or 
	state="WY" or 
	state="AK" or 
	state="CA" or
	state="HI" or 
	state="OR" or 
	state="WA" then region="West";
else region="Other";
run;

proc contents data=location; run;
proc means data=location; run;

proc sort data=tuition;
by unitid;
run;

proc sort data=location;
by unitid;
run;

data merged;
merge tuition (in=one) location (in=two);
by unitid;
if one=1 or two=1;
in_one=one;
in_two=two;
run;

proc contents data=merged; run;
proc means data=merged; run;

proc freq data=merged;
tables in_one*in_two;
run;

data final;
set merged;
if in_one=1 and in_two=1;
run;

proc means data=final; run;
proc means data=final; run;

proc freq data=final;
tables state*public;
run;

proc means data=final;
class public;
var tuition90 tuition91 tuitionrise;
run;

proc freq data=final;
tables region*public;
run;

proc glm data=final; 
class region; 
model tuitionrise=region; 
means region / waller;
run;

proc reg data=final;
model tuitionrise=public;
run;

proc glm data=final;
class region;
model tuitionrise=public region/solution;
run;