ST 555 -- Summer 2013

  Homework #5 -- Due WEDNESDAY, 19 June 2013

Among the controversial votes during the past NC legislative session
were some taken early in the morning.  To assess whether the times 
were unusual, the Raleigh News and Observer (story, 13 Jan 2012) did 
some analysis of the times when votes occurred.  Our goal is to 
replicate some of their study to assess the story's accuracy.

In the 'ncleg' directory are files of the result of votes for the last
several sessions.  The files are named '20xxVotes.txt' where xx goes
from 01 to 09 (five files).  Your task is to read them all in, put them
together, and do some analysis.

The format of the data is
  roll call sequence number (numeric)
  chamber (S or H, character)
  date    (yymmdd10.)
  time    (hhmmss.000, so use hhmmss8. and read another numeric
            variable and drop it, that is 
     input seq chamber $ date yymmdd10. time hhmmss8. junk prefix ....;
            and then drop junk)
  prefix  (Chamber, bill, resolution etc. -- character)
  billid  (numeric)
  yes votes
  no votes
  excused absences
  excused votes
  didn't vote
  total yes/no votes
  ... and more stuff that we don't care about
*** the data are semicolon delimited, but there's a space between the
date and the time, so use two delimiters, space and semicolon, that is,
dlm=' ;' as well as dsd ***

a) Read in the data from each of the five files.  Create a variable
HOD (hour of the day) using the SAS function hour(x) which gives the 
hour from a time variable.  Also create a variable WHEN (character or 
numeric) that roughly corresponds to morning, afternoon, evening, and
night (4 values).

b) Check to see if the total yes/no votes matches the sum of the two.
(Since these are integers, no need to worry about rounding.)

c) Put the datasets from these five files together for analysis.

d) Create a frequency table of HOD and chamber.  (how many votes at
each hour for each chamber)

e) Find the mean number of votes for each value of WHEN.  For each 
chamber (separately), do an analysis of variance of number of votes 
to determine whether these means are significantly different.


FOR ALL EXERCISES: Hand in (paper) (1) The answers to any questions 
asked. (2) Your program * With at least as many comments as you have 
toes! * (3) Your output