 Working
through the Data
6.5 Coding and data entry
In evaluation studies, information is collected about
individuals, then analyzed by group to find patterns
and trends.
Making sense out of data entails counting responses
and tabulating how frequently each occurs, which requires
data to be coded and organized.
Coding
-
code individual responses in accordance with a master
coding sheet to allow easy computer identification,
e.g., male = 1, female = 2; no = 0, yes =1.
- responses already in the form of numbers, as for
age or grade or numeric scales (1-5 from least… to
most…), are entered as is
- blank responses are coded as well, to show where
the informant failed or refused to answer
the question. The number used is one that will not
appear in other
coding, like 99, if this is well outside
the range of other possible responses
Scales
The response scale for a question on a form or in
an interview is usually written from least to most,
with the numbers increasing accordingly. Sometimes,
however, the ‘least’ option is the most
positive response (least dissatisfied, least bullied.)
It is important that all the coded scales are in the
same direction (all positive or all negative) to enable
data from different questions to be compared or cross-tabulated.
This reordering has to be done at the coding stage
rather than on the data collection form to avoid confusing
informants.
Data entry
Data can be entered onto coding forms by hand or
directly into a computer.
- To find and reduce coding errors:
- have two people enter data separately then compare
results (double-entry)
- or have someone verify data entry by spot-checking
for errors against the original coding form
or questionnaires.
Common errors to look for are:
- numbers outside the range of codes (6 for a 1-5 response
option)
- transposed numbers (1.52 instead of 1.25.)
Automatic range checks are provided by most commercial
data entry programs.
|
