ST 445 -- SPRING 2010 
     HOMEWORK 6 
     DUE THURSDAY, 29 APRIL 2010 
     The files are in 'apr10' directory.
     ** updates 21 April **                ***                ***
     file 'ncecon2010.variables' has been corrected
     file 'NCpres2008.txt' has been modified to make it (hopefully)
      easier to read -- still formats with tabs is tricky -- suggestion:
      read votes as character variable, say 'mccainvotec' and then
      convert to numeric with 
        mccainvote = input(mccainvotec,comma7.) ;
      ***                ***                ***                ***

      In the file 'NCpres2008.txt' are county-level summaries of the
      voting in the 2008 presidential race.  This file is tab-delimited
      includes both the percentages of votes for the two candidates
      as well as the raw vote totals.  (These data were taken from 
      the New York Times site:
        http://elections.nytimes.com/2008/
               results/states/president/north-carolina.html )
  
      a) Read in the data in this file and create a dataset with the
      relevant variables.

      In the file 'nc2010econ.dat' are selected economic, educational,
      and demographic data for North Carolina at the county level.  The 
      list of variables is given in the file NC2010econ.variables.  
      (Data courtesy of Tammy J Lester, NC Dept of Commerce.)
 
      b) Read in these related variables and create a dataset

      c) Merge these two datasets by county name.

      d) The main task of the assignment is to describe the voting data 
      and investigate some relationships with one or two of the economic 
      variables.  Particularly appropriate would be correlations (proc 
      corr) or crosstabulations (proc freq), as well as scatterplots.  
      Summarize your conclusions in a paragraph or two, using charts, 
      graphs, etc. as a supporting appendix.  Make effective use of
      labels and titles; use of user-written formats is encouraged.

 Note:
 1) Be aware that many counties in North Carolina are quite small,
    both in area and population, and so that analysis of count data
    from these counties requires care.  My suggestion is to limit
    your analysis to the largest counties, and avoid the smallest.
 2) You should be aware that the largest counties will have the 
    largest counts, the smallest counties the smallest -- of anything
    -- regardless of any cause/effect.  For example, Mecklenberg and
    Wake counties have the highest numbers of doctors and the highest 
    incidence of cancer while Camden and Clay counties have small numbers
    of both.  But does the apparent positive correlation here mean that
    we should try to reduce the number of doctors to reduce cancer
    incidence?  Base any relationship on RATES: number/population.