Residual Plots: Single Factor Experiment

title 'Residual Plots: Wound Healing';

options linesize=79;

data wounds;
   infile 'gpi.dat';

input Material Gpi;

PROC GLM;
     CLASS Material;
     MODEL Gpi = Material;
     OUTPUT out=resstuff predicted=gpipred residual=resid;

PROC PLOT data = resstuff;
   plot resid*gpipred;

PROC UNIVARIATE plot data = resstuff;
    var resid;


RUN;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

 Explanation of some commands:

   Under PROC GLM: the OUTPUT statement tells SAS we're going to save
some of our results in a SAS data file called resstuff. In that file
will be 2 variables: The fitted ( y ) values in "gpipred", and the
residuals ( e ) in "resid".

  PROC PLOT is a plotting procedure that produces plots that are not
flashy. It will be used to plot the residuals vs. fitted values, so we
have to tell this procedure these values are stored in "resstuff". Then
we give the command to plot these values.

 PROC UNIVARIATE computes a lot of summary statistics for univariate
data. It will give a lot of information that we don't need about the
residuals, so should be deleted if you use this for any assignment
questions. However, it does give us a QQ-plot of the residuals. To use
this procedure, you need to specify the SAS data file where your values
are stored, then specify the name of the variable you want to examine in
the VAR statement.

 The output starts on the next page.

 NOTE: I've deleted the first part of the output, since that is on your
 previous handout.

                         Residual Plots: Wound Healing                        3

          Plot of resid*gpipred.  Legend: A = 1 obs, B = 2 obs, etc.

  resid |
        |
   5.00 +                                                      B
        |
        |
        |
        |
   3.33 +       A   A
        |
        |
        |
        |
   1.67 +                     B                               B       B
        |
        |
        |
        |
   0.00 +
        |
        |
        |
        |
  -1.67 +       B   B
        |
        |
        |
        |
  -3.33 +                     A                               A       A
        |
        |
        |
        |
  -5.00 +
        |
        |
        |
        |
  -6.67 +
        |
        |
        |
        |
  -8.33 +
        |
        |
        |
        |
 -10.00 +                                                      A
        |
        ---+-------+-------+-------+-------+-------+-------+-------+-------+--
           0      10      20      30      40      50      60      70      80

                                        gpipred

                         Residual Plots: Wound Healing                        4

                           The UNIVARIATE Procedure
                               Variable:  resid

                                    Moments

        N                          18    Sum Weights                 18
        Mean                        0    Sum Observations             0
        Std Deviation      3.70479287    Variance            13.7254902
        Skewness           -1.0122023    Kurtosis            1.70216837
        Uncorrected SS     233.333333    Corrected SS        233.333333
        Coeff Variation             .    Std Error Mean      0.87322805


                          Basic Statistical Measures
 
                Location                    Variability

            Mean     0.000000     Std Deviation            3.70479
            Median   1.666667     Variance                13.72549
            Mode     1.666667     Range                   15.00000
                                  Interquartile Range      3.33333


                          Tests for Location: Mu0=0
 
               Test           -Statistic-    -----p Value------

               Student's t    t         0    Pr > |t|    1.0000
               Sign           M         1    Pr >= |M|   0.8145
               Signed Rank    S         2    Pr >= |S|   0.9373


                            Quantiles (Definition 5)
 
                            Quantile       Estimate

                            100% Max        5.00000
                            99%             5.00000
                            95%             5.00000
                            90%             5.00000
                            75% Q3          1.66667
                            50% Median      1.66667
                            25% Q1         -1.66667
                            10%            -3.33333
                            5%            -10.00000
                            1%            -10.00000
                            0% Min        -10.00000


                              Extreme Observations
 
                  ------Lowest------        -----Highest-----
 
                      Value      Obs           Value      Obs

                  -10.00000        1         1.66667       12
                   -3.33333       10         3.33333       17
                         Residual Plots: Wound Healing                        5

                           The UNIVARIATE Procedure
                               Variable:  resid

                              Extreme Observations
 
                  ------Lowest------        -----Highest-----
 
                      Value      Obs           Value      Obs

                   -3.33333        8         3.33333       14
                   -3.33333        4         5.00000        2
                   -1.66667       18         5.00000        3

# Stem and Leaf plot of residuals
# Appears to have bell-curve shape

               Stem Leaf                     #             Boxplot
                  4 00                       2                |
                  2 33                       2                |
                  0 777777                   6             +--+--+
                 -0 7777                     4             +-----+
                 -2 333                      3                |
                 -4
                 -6
                 -8
                -10 0                        1                0
                    ----+----+----+----+


# HERE IS THE QQ-PLOT

                                Normal Probability Plot
                5+                                    ++*+++ *
                 |                               ++*++*
                 |                        *+**+** *
                 |                  * **+++
               -3+            * *++*++
                 |         ++++++
                 |    +++++
                 |++++   *
              -11+
                  +----+----+----+----+----+----+----+----+----+----+
                      -2        -1         0        +1        +2