Page 324 -

11.4 Analyzing text content 313

11.4.2.3 Making comparisons of data
Both during the coding process and the stage afterwards to interpret the results,
you are encouraged to make comparisons at multiple levels. First, you can com-
pare instances under different coding categories. For example, if you are inves-
tigating the difficulties that older people experience when using computers, you
can compare the frequency with which each capability (physical, cognitive, per-
ceptual) is reported. You can also compare the degree of impact between different
capabilities.
Second, you can compare the results between different participant groups. You
may find that the capabilities and computer usage behaviors vary substantially
among the users. You can further investigate this diversity via different dimensions:
Is the diversity related to age, educational background, or community and family
support? To answer these questions, you need to subdivide the data set and compare
the results among subsets.
Third, you can compare the findings in your data to previously reported literature.
Do your findings align with the existing literature or is it contradictory? If your find-
ings differ from existing literature, can you explain why? Is the existing literature
incorrect? Or is the difference caused by a different context? Sometimes the need
to compare your findings with a related population or tasks may facilitate you to
conduct additional studies to collect more data. For example, if you observe some in-
teresting computer usage behavior in children with autism, you may want to conduct
the same study for neurotypical children to investigate whether there is a difference
between the neurotypical children and children with Autism.
11.4.2.4 Recording the codes
When you find an item in the content that you wish to describe with a code,
you should note exactly what you are coding and which codes you are assign-
ing. The “what” should provide enough detail to unambiguously identify the
relevant content—you might quote specific paragraphs, sentences, or phrases
from a textual document or identify start and end points for time intervals in an
audio recording. The code(s) that you are assigning will come from your list of
codes. Note that coders often find that they want to describe a given item with
multiple codes.
A variety of strategies can be used to record codes. One low-tech approach might
involve marking up text with comment and highlighting tools from a word processor.
Another possibility might be to use a spreadsheet with columns for the identification
of the item being coded and for the codes being assigned.
A number of commercial and academic software packages provide dedicated
support for text analysis, including tools for creating codebooks, coding documents,
and searching and querying for material associated with given codes. Other tools
support automated text content analysis via features for searching, counting, sorting,
and conducting basic statistics. Examples of commonly used text analysis software
include SAS (https://www.sas.com), GATE (the General Architecture for Text
Engineering) (https://gate.ac.uk), and Carrot2 (http://project.carrot2.org).