Precision, time, and cost : a comparison of three sampling designs in an emergency setting
Authors & affiliation
Megan Deitchler, Hedwig Deconinck, Gilles Bergeron
Abstract
The conventional method to collect data on the health, nutrition, and food security status of a population affected by an emergency is a 30 × 30 cluster survey. This sampling method can be time and resource intensive and, accordingly, may not be the most appropriate one when data are needed rapidly for decision making. In this study, we compare the precision, time and cost of the 30 × 30 cluster survey with two alternative sampling designs: a 33 × 6 cluster design (33 clusters, 6 observations per cluster) and a 67 × 3 cluster design (67 clusters, 3 observations per cluster). Data for each sampling design were collected concurrently in West Darfur, Sudan in SeptemberOctober 2005 in an emergency setting. Results of the study show the 30 × 30 design to provide more precise results (i.e. narrower 95% confidence intervals) than the 33 × 6 and 67 × 3 design for most child-level indicators. Exceptions are indicators of immunization and vitamin A capsule supplementation coverage which show a high intra-cluster correlation. Although the 33 × 6 and 67 × 3 designs provide wider confidence intervals than the 30 × 30 design for child anthropometric indicators, the 33 × 6 and 67 × 3 designs provide the opportunity to conduct a LQAS hypothesis test to detect whether or not a critical threshold of global acute malnutrition prevalence has been exceeded, whereas the 30 × 30 design does not. For the household-level indicators tested in this study, the 67 × 3 design provides the most precise results. However, our results show that neither the 33 × 6 nor the 67 × 3 design are appropriate for assessing indicators of mortality. In this field application, data collection for the 33 × 6 and 67 × 3 designs required substantially less time and cost than that required for the 30 × 30 design. The findings of this study suggest the 33 × 6 and 67 × 3 designs can provide useful time- and resource-saving alternatives to the 30 × 30 method of data collection in emergency settings.
Publication date:
2008
Staff members:
Link to publication
Attachments
Deitchler2008_Article_PrecisionTimeAndCostACompariso.pdf (open)Related publications
Anne Marieke Simmelink, Caroline M Gichuki, Frances H Ampt, Griffins Manguro, Megan S C Lim, Paul Agius, Margaret Hellard, Walter Jaoko, Mark A Stoové, Kelly L'Engle, Marleen Temmerman, Peter Gichangi, Stanley Lüchters
2016 Enrolment of young women attending cervical cancer screening to survey effectiveness of HPV vaccinationMarc Arbyn, Davy Vanden Broeck, Johannes Bogers, Pierre Van Damme, Marleen Temmerman, Steven Weyers
2016 Epidemiology of HPV genotypes among HIV positive women in Kenya : a systematic review and meta-analysisSonia Menon, Aibibula Wusiman, Marie Claude Boily, Mbabazi Kariisa, Hillary Mabeya, Stanley Lüchters, Frode Forland, Rodolfo Rossi, Steven Callens, Davy Vanden Broeck