ST500                           Session 6
_review assignment 1 (c) - minitab output _Visit other statistical Web
sites _Regression: Y is the dependent variable, X1, X2, X3 ... are the
independent variables .  SAS:        proc reg;
                model Y= X1--X3 /p cli influence vif  selection=stepwise;
                output out=resid p=py r=ry rstudent=rs; Also:
            proc print data=resid;            RSREG
SPSS:       regression variables = (collect)/
                descriptives = default/    statistics = default/
                dependent= Y/  method =  enter X1 to X3  stepwise X4 to
                X6/ residuals=outliers(sresid resid )/ save= dffit pred
                sresid resid cook lever dfbeta /
            list execute
MINITAB:    reg Y 3 X1 - X3 res c10 fits c11;
              hi c12; tresiduals c13;                 Also: cookd c14;
              STEPWISE dfits c15;                      BREG vif .
            print Y X1 - X3    c10 - c15 stop
_ Analysis of Variance: Y is the dependent variable, F1, F2, F3 ... are
the factors, and X is a covariate.
     SPSS:  manova Y by F1 (min,max),F2(min,max) with X/
                 pmeans=tables (F1 by F2)/ residuals plot/ plot=stemleaf
                 boxplots normal/ design=F1, F2, F1  bv F2/
            list execute
     MINITAB:    glm Y = F1 I F2;
               covariates X; fits c10; residuals c11; sresiduals c12;
               hi c13; means F1 I F2.
            print Y F1 F2 X c10 - c13 stop
     SAS:   proc glm; class F1 F2;
               model Y = F1 I F2 X/ p cli solution; Ismeans F1 I F2;
               output out= resid p= PY r= RY h=h cookd=cd dffits=df
                                rstudent=rs ;
               Ismeans F1 lF2;
            proc print;         Assignment 3 - analysis of variance
            email to me before class on Thurs, Jan 11.
Please use a3a for SAS, a3b for SPSS and a3c for Minitab.

 1. Run the following regression with sas using '../data/weight.dat' as
 data file.
     Note: the variable called sex, which is alpha will not work in proc
     reg, so a numeric
variable, nsex = 1 (male), = 2 (female), must be created in the data step.
   a. proc reg; model weight = nsex; b. proc reg; model weight = height;
   c. proc reg; model weight = nsex height;
2. Run the following procedures with sas using '../data/weight.dat'.
   a. proc ttest; class sex; var weight; b. proc glm; model weight =
   height; c. proc glm; class sex; model weight = height sex;
3. How does the output in 1.a.- 1.c. compare to 2.a.-2.c.?
  Explain how height and sex are related to weight. Are height and sex
  'statistically significant'
effects? (We will discuss in class.)  4. Run spss to get the analysis of
variance in 2.c.(as in 1., nsex must be created) 5. Run minitab to get the
analysis of variance in 2.c. (either read weight.dat with a format as in
assign 1 (c) and then create a numeric col for sex, OR read 'assign1.datx'
and then 'code (-1) 77 c3 c3'. Remember, -1 was missing, and was replaced
in the weight.dat file with 77) 6. copy the '../data/weight.dat' file
into your home directory and add the following line of data as the bottom
line of the file: Harry    m      55    .     15
     Use proc reg and proc glm in sas to get a predicted value for
     harry's weight for the models
in 1.c. and 2.c. (Use 'p' and 'cli' after a '/' at the end of the model
statement. )




NOTE about running minitab:(in both batch and interactive modes)

ALWAYS after using
echo outfile 'file';
            noterm.  batch
use         nooutfile which causes minitab to write the file 'file.LlS'.

ALWAYS after using
journal 'file' use         nojournal which causes minitab to write the
file 'file.MTJ'.