Introduction To Correlation with SAS

							Introduction To Correlation
         with SAS

           Sam Gordji
       ccsam@olemiss.edu
            Weir 107
    List of Statistical Packages
       Available through OIT

•    SPSS (Windows)
•    SAS (Windows & Unix)
•    Mathematica (Windows)
•    Matlab (willow, Unix)
•    Free PGI Fortran compiler
        General Information
•   IT staff will assist users
    – To access these packages
    – To find the proper procedure to analyze
      their data


•   For assistance please email
    – assist@mcsr.olemiss.edu
        Correlation with SAS
•   This seminar covers correlation with SAS
•   SAS is available in
     – Weir Student Lab (4 PCs)
     – Weir 107
     – Several other labs around campus
•   SAS is also installed on willow (Unix server)
•   SAS performs statistical analysis including
    procedure correlation
        Statistical Package:
          SAS/PC (cont.)
•   50 copies are available for faculty use
•   For each copy purchased, a faculty
    member can get a free copy (to be
    installed on a student PC)
•   The cost is $200 per copy per year
    (starting Aug. 2009)
•   To obtain a copy of SAS please email
    – assist@mcsr.olemiss.edu
         Installation of SAS
• To obtain a DVD for SAS send an email
  to:
  – assist@olemiss.edu



• The instruction for installation will be
  emailed to you
                  Links for SAS
•    The main webpage
http://www.sas.com/


•    For information on documentation for
     SAS visit
http://support.sas.com/onlinedoc/913/docMainpage.jsp


•    SAS Resources for Faculty and
     Students:
http://www.mcsr.olemiss.edu/mathematica/fall_sem/SASResources.ppt
      SAS’ Three Main Files
•   *.sas (a SAS program file created by the
    user, an input file)
•   *.log (a file created by SAS containing
    the “log” after the user’s program is
    finished running)
•   *.lst (a file containing the output)
•   Examples of the three SAS files
    – my.sas
    – my.log
    – my.lst
       Running SAS on Windows
Click on the “SAS” icon to activate SAS
Create your program (below) in a window named “Editor- Untiled1”
Blue Text is your program
/* comments go between slashes and stars */
   data test;   /* every SAS program starts with “data” , test is a
                          given name for this program */
   input a b c; /* variables are a, b, and c */
   cards;       /* input data will follow “cards” */
   123          /* variable a has 2 observations, 1 and 2 */
   2 3 -1       /*variable b has 2 observations, 2 and 3, and so on */
   ;            /* semicolon signals the end of the data */
   proc print; proc means;
                /* proc print prints the data, proc means
                          obtains several statistics including the mean */
   run;         /* run, runs the program */
     Running SAS on Windows
            User’s SAS Program




Click “run” and then “submit” the above example
Running SAS on Windows
     listing file (output of SAS)




Input data:



Results:
Running SAS on Windows
            SAS log file




   first 3 lines of the user’s input




      Lines 6 & 7 end data & proc print



        Lines 7 & 8 proc means & run
              Correlation
• Correlation coefficient measures the
  relationship between two variables
• The correlation coefficient is always
  between -1 and 1, -1.0<=C<=1.0
• Values near 0 are indication of no
  relationship and values near 1 indicate a
  strong relationship between the variables
• Negative values indicate negative
  relationship
               An Example of Correlation
                              corr748re.sas
• Below is an example of correlation analysis
• Data is entered into “Editor”
• Download corr748re.sas program below from:
  http://www.mcsr.olemiss.edu/educationsubpage.php?pagename=jancamp09_inc
Output from Correlation
       Analysis
   Non-parametric Correlation
• Proc corr also performs non-
  parametric correlation

• Proc corr below performs non-
  parametric correlation and obtains
  Spearman, Kendall Tau, and
  Hoeffding correlation
Non-parametric Correlation
        Non-parametric Correlation
Assumes That the Input Data Is Non-Parametric




             Pearson




             Spearman




            Kendall Tau
           Some SAS Proc
•   Proc (Procedure)
    – proc means (obtains mean, standard
      deviation)
    – proc anova (performs simple AOV)
    – proc plot (plots)
    – proc lp (performs Linear Programming)
    – proc reg ( performs regression)
    – proc corr (performs correlation)
                Questions
•   Email contact
    – Email your questions to assist@olemiss.edu

•   To look at this presentation and other
    materials visit, Computing Camp at
    www.mcsr.olemiss.edu

•   Please fill out the feedback form and
    leave your email address so we may
    contact you for follow up questions
            For Further Practice


•   Go to

–   http://www.mcsr.olemiss.edu/educationsubpage.php?p
    agename=jancamp09_sam.inc

    download and run the following examples

    – mort1.sas
    – mem_corr.sas