Generating Random Samples

1
Generating Random Samples

• SAS, EXCEL, JMP, SPSS

2
Population of Data ? Sample

• Data should be in a dataset where each row
  represents an individual unit, and column fields
  represent variables measured on the individual
• A new column field will be created that contains
  a pseudo-random number generated by statistal
  software package (usually Uniform(0,1) r.v.)
• Data are sorted by the random number, or simply
  selected for analysis (depending on software)

3
All ATL/HNL 10/2004 Delta Flights (N31)
4
SAS Code to Generate Random Sample (n10)
Data Pop Input Day FlightNum TailNum DestCity
TaxiTime Sched Actual Distance RanxRanuni(seed
) / Put in a number where seed is
/ Cards 1 317 N834MH HNL 29 580 567 4496 31 31
7 N840MH HNL 16 580 576 4496 Proc Sort By
Ranx Data Sample Set Pop If _N_ lt
10 Run / You can run any procedures now on
the Sample /
5
Generating Random Samples in EXCEL

• In a new column past the data, type in the first
  row containing data on a unit RAND()
• COPY and PASTE that cell until all N units have a
  random number
• While that range of N cells is still highlighted,
  click on COPY ? EDIT ? PASTE SPECIAL ? Click on
  VALUES (This changes them from Dynamic to
  Static)
• Place Cursor in Data Field, Click on DATA ? SORT
  ? Choose Column with Random Values and Choose
  appropriate Header Row Status
• After data are sorted, COPY and PASTE (SPECIAL
  VALUES) the first n units to a new worksheet

6
Random Numbers (Not Sorted Yet)
7
Data Sorted By Random Number
8
EXCEL Random Sample (n10)
9
Random Samples in JMP

• Import Data into JMP
• Click on TABLES ? SUBSET ? RANDOM SAMPLE ? Enter
  Your Sample Size (e.g. 10) in Box
• Subset Dataset Appears in New Table
• Analyze the Data in that Table

10
Random Samples in SPSS

• Import Data into SPSS
• Older Versions Click Transform ? Random Seed ?
  Enter a numeric seed
• Newer Versions Click Transform ? Random
  Generators ? Active Generator Initialization ?
  Fixed Value ? Enter a numeric seed
• Click Data ? Select Cases
• Under Select, choose Random Sample of Cases.
  Click on Sample box.
• Choose Exactly n (10) of the first N (31), click
  Continue
• Click OK. Slashes will appear in the case numbers
  of the cases not in the sample, and an indicator
  variable representing Filter will be attached to
  end of dataset.
• Analyze the Data