/*import excel file*/
PROC IMPORT OUT= work.seuss DATAFILE= "C:\Users\stevensonja2\Desktop\seuss\seuss.xlsx"
            DBMS=xlsx REPLACE;
RUN;

data seuss1;
set seuss;
if Year >=1930 and year <=1939 then decade=1930;
if Year >=1940 and year <=1949 then decade=1940;
if Year >=1950 and year <=1959 then decade=1950;
if Year >=1960 and year <=1969 then decade=1960;
if Year >=1970 and year <=1979 then decade=1970;
if Year >=1980 and year <=1989 then decade=1980;
if Year >=1990 and year <=1999 then decade=1990;
run;

data seuss1;
set seuss1;
if Year >=1930 and year <1957 then time=1;
if Year >=1957 and year <=1999 then time=2;
run;

DATA Seuss1;
set Seuss1; 
If time = 1 and Type = "literacy" and Genre = "rhyme" then new= "1LR";
If time= 1 and Type= "literacy" and Genre= "prose" then new= "1LP";
If time= 1 and Type= "moral" and Genre= "rhyme" then new= "1MR";
If time= 1 and Type= "moral" and Genre= "prose" then new= "1MP";
If time= 2 and Type= "literacy" and Genre= "rhyme" then new= "2LR";
If time= 2 and Type= "literacy" and Genre= "prose" then new= "2LP";
If time= 2 and Type= "moral" and Genre= "rhyme" then new= "2MR";
If time= 2 and Type= "moral" and Genre= "prose" then new= "2MP";
run;

proc anova data = seuss1;
      class decade;
      model T_N = decade;
	  ods graphics on;
   run;

proc anova data = seuss1;
      class time;
      model T_N = time;
	  ods graphics on;
   run;

Proc GLM data=Seuss1;
Class new;
MODEL T_N = new;
MEANS new / HOVTEST=LEVENE WELCH;
RUN;

/*NORMALITY TEST:*/

PROC GLM DATA= Seuss1;
CLASS Time Genre Type;
MODEL T_N=Time|Genre|Type;
LSMEANS Time Genre Type /STDERR OUT=ADJMEANS;
OUTPUT OUT= norm_test r=resid p=predicted;
RUN;
Proc univariate data=norm_test normal plot;
Var T_N;
Run;

Proc Mixed;
Class Time Type Genre;
Model T_N=Time|Type|Genre;
LSmeans Time|Type|Genre / pdiff adjust=tukey;
Run;