Return to Chapter 1 Page 3
Data labeling might seem insignificant or just time consuming, but it is
one of the most fundamental steps toward accurate data analysis. If there are not proper
labels on the variables and the meaning of the values is not properly specified, there
will likely be numerous mistakes made during the analysis. The labeling is actually very
simple. Even though it may sometimes seem monotonous, the process will be worthwhile.
Follow these steps for each variable:
- To access the labeling system in the Data Editor where the data base is stored, double
click on the variable name at the top of the column and a Define Variable box will
- In the Variable Name box, enter the new variable name that you would like to use (each
variable must keep a distinct name under 8 characters).
- In the Change Settings area, click on the Labels button.
- In the Define labels box, type in a Variable label. The label allows a
longer description of the variable, whereas the variable name is limited to under 8
characters. If you have a systematic naming system for your variables (s1, s2, s3, s4,etc)
then you can give a better description with the variable label (i.e. Breastfeeding up to
four months, age at interview, etc.).
- In the same Define labels box, find the area to enter Value labels. The
values are the possible values that the variable can take. If the variable is lived in
an urban area, then in the value box you might enter 1 with a value
label as urban and click on Add to place it in the lower box.
Then you would enter value 0 and value label rural and Add,
and so on for any values that might apply and press OK.
- Finally, if there are any values that are defined as missing, such
as 9, 999, etc. then click on the Missing Values box. Enter the discrete values, or
range of values that apply, and press Continue and then OK. With the
nutritional z-scores, unfortunately the missing values function does not allow missing as
a value (it only allows for an inclusive range to be set to missing). To set the values
lowest to <-4 SD and>4 SD to highest to missing for weight for age z, height for age z, and weight
for height z, the recode option must be used.
THE CODE BOOK is an essential part of navigating through a data
set. Usually, a code book is drawn up from the data set to give a list of the variables
and the variable labels as well as all of the values and meanings of the values that each
variable may take. Additionally, a code book may list all of the missing values, or the
range of acceptable values for each variable. In SPSS, it is possible no only to write all
of the labeling directly into the computer, but it is also possible to print up this code
book on hardcopy after the labeling is complete.
Here is how to create a code book in SPSS:
- Open SPSS, but do not open any particular data set.
- In the Data editor, click on File, Display Data Info...
- Type in the appropriate file path and name in the box and press Open.
- The file information will be displayed in the Output screen, starting with System
Info and followed by Variable information.
- The font can be changed to condense the output to fewer pages by double clicking on the
output to make it editable. Once the output is in edit mode (output is surrounded by hash
marks and an edit menu appears at the top of the screen), click on Edit, Select All,
and when all of the data is blocked out, select the font and type size (Times New Roman, 8
point works well) and then Print.
to top of page