ch11up.gif (1608 bytes)

              Random Sampling with SPSS

Random sampling need not always be done by hand.  Sometimes random sampling may be completed as part of a computer analysis.  SPSS permits us to select a sample and then to determine if it possesses qualities that indicate that it captures characteristics of the entire sample.   Xcell does not include routines
to complete similar sampling.  This exercise is designed to give you a chance to examine this use of computers to help you secure a random sample.

As a starting point, we need a data set for analysis. 

If you have not already done so, go to the Data Sets location and download the file nca.sav  
under the category, "NCA Study of  'How Americans Communicate.'"   If this source no longer 
is available from the NCA website, then, download the file
nca.sav from the "SPSS Files" 
category on the "Data Sets" web page.
Start your SPSS program.  From the "File" menu load the data file "nca.sav" from the location 
where the you saved it after downloading.

Purpose of the Exercise:

The purpose of this activity to to complete a random sample of 200 individuals (20% of the total 
sample) from the original survey of 1001 Americans.   Then, there will be an effort to see if the sample
is representative of the population on a key survey and population characteristic.

In the original sample, 52.2% (523) were women and 47.8% (478) were men.  In the original survey, 
the variable of respondent sex was identified by the variable called "sex."  This item coded a person as
male (coded as 1) or female (coded as 2).

Steps in Conducting the Analysis

The process involves drawing a sample and then running a simple descriptive output with SPSS.  We
will guide you through these steps. 

Draw the random sample. To draw a random sample of 200 (20%) at random from
the original survey, you need to follow several steps. From the "Data" menu click "Select Cases."  From the dialog box that emerges, select "Random sample of cases."  For the options 
"Unselected Cases Are" select "filtered" if it is not already selected.  The dialog box should appear
as follows:

        Click on "Sample..." to bring up the Sample Size dialog box.  Select the "Approximately"
        option and insert 20% in the adjacent box to indicate a random sample of "20% of all
        cases."  The dialog box should resemble this on
e.       

Click on "Continue" and "OK" on the subsequent menu box.  Now a random sample has
been completed.  The unselected cases have not disappeared from the file, but they are
not used in subsequent analyses.

Check for a representative sample.  To determine if this sample is 
representative, a descriptive output should be run for the SEX variable. To do so, click 
on the"Analyze" menu, and on other menus select "Descriptive Statistics" and then 
"Frequencies."  In the dialog box for "Frequencies," click on "sex" variable that appears in 
the left window of the dialog box. Transfer this variable to the window marked 
"Variable(s):" by clicking on the arrow key in the middle of the dialog box.  When you
have completed this process the dialog box should look as follows.

Click on the "Statistics" button, to open the following dialog box:

Place check marks in the boxes to select "Mean," "Std. deviation," and "S.E. mean."

Click on "Continue," and then click "OK" in the "Frequencies" dialog box.

Interpret Your Results

Though researchers may use statistical tools to compute confidence intervals, in this case you may wish to rely on your own judgments of the facts. 

Print out your results.

How well did this sample of  200 capture the proportions of the larger sample?

What are the reasons you might wish to use a reduced sample size in research?