Page 387 - Special Topic Session (STS) - Volume 4

STS2319 Lakshman N. R. et al.
(PAPI) techniques which are prone to numerous data issues, carrying huge
implications for the quality of data analysis.
PAPI records data onto paper forms, which are then manually compiled
and entered to come up with a data set for analysis. While a lot of surveys
have benefitted from PAPI over time, the limitations of the method are
potentially compromising for data quality. With surveys becoming more
comprehensive and complex, PAPI might suffer from issues in data accuracy
because enumerators would need to exert additional effort to accurately
navigate through the more complicated logic and skips built into the surveys.
At some point, more complicated surveys could cause enumerator burden to
set in and potentially compromise the data gathered. Further, PAPI involves
dealing with heaps of paper and reprints, entailing tedious manual data
encoding, which is prone to human error. The encoding process also involves
time costs, which could cause delays in data availability and analysis.
The advent of information technology has brought forth an alternative that
could potentially address the limitations of PAPI in the form of computer-
assisted personal interviewing (CAPI). With CAPI, interviewers use a handheld
device instead of a paper questionnaire to record interview responses. The
main advantages of CAPI include the elimination of paper forms and
associated print outs, increased data accuracy due to automated skipping
mechanisms and logical checks, and faster data availability virtually
eliminating manual data entry allowing for almost immediate data analysis.
Further, CAPI arguably may lower costs for larger sample sizes, which is
beneficial for national statistical systems conducting multi-topic nationally
representative surveys. CAPI also possesses additional features such as the
ability to integrate images, video and audio recordings, timestamps, and
global positioning system (GPS) information into the questionnaire.
In theory, CAPI is expected to address the limitations of PAPI when it
comes to improved data quality, timeliness, and costs. Yet there is very limited
literature that rigorously and empirically looks at the cost and benefits of
transitioning from CAPI to PAPI in in developing economies, particularly in
Asia and the Pacific. Much of the earlier research on the implications of
transitioning to CAPI has usually focused on developed economies (Couper
and Burt 1994; Nichols and de Leeuw 1996; Banks and Laurie 2000). Only two
studies have made systematic and rigorous attempts at studying the impact
of transitioning to CAPI. Caeyers et al. (2012) empirically assessed CAPI’s
benefit in terms of data quality, cost, and timeliness from a randomized control
trial done on Tanzanian households. Meanwhile, Fafchamps et al. (2012),
attempted to quantify CAPI’s advantage for data quality when it comes to
collected data on sales and profits for microenterprises in Ghana. Both studies
only provide an African context and have contrasting findings, thereby calling
for more research in this area, especially with external validity across different
contexts and types of surveys.

376 | I S I W S C 2 0 1 9