Can Large Language Models Be Used to Code Text for Thematic Analysis? An Explorative Study
Abstract
In
practice,
thematic
analysis
of
text
involves
six
stages,
among
which
coding
is
particularly
cognitively
demanding,
labor-intensive,
and
time-consuming.
This
study
investigates
compares
the
potential
two
large
language
models
(LLMs),
namely
ChatGPT-4
OpenAI
o1-preview,
to
perform
coding,
with
goal
reducing
time
effort
required
by
human
researchers.
Our
results
indicate
that
both
exhibit
decreased
comprehensiveness
as
document
length
increases,
demonstrate
low
accuracy,
primarily
due
limitations
in
textual
comprehension
reasoning.
These
findings
highlight
significant
challenges
using
LLMs
support
analysis,
emphasizing
need
for
oversight
rigorous
validation
ensure
analytic
accuracy
validity.
Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown
Published: April 30, 2025
Language: Английский