World Food Policy Volume/Issue 2-2/3-1 Fall 2015/Spring 2016 | Page 9

World Food Policy

practitioners channel the present
transformation of the rural economy to
better outcomes for those who are most
vulnerable to these changes. Beyond
the scope of the lower Mekong river
basin, this knowledge may also improve
our understanding of transformation
processes in other areas where livelihood
outcomes, especially food security, are
highly dependent on small-scale fresh
water fishery such as in the floodplains
of Bangladesh (Sultana and Thompson
2007), in Myanmar (FAO 2003), or lake
Victoria in Africa (Matsuishi et al. 2006).

The remainder of the article is
structured as follows: The section “Sample
and Methodology” describes the dataset
and how food security and nutrition are
measured. The next section “Results”
introduces and discusses the results. The
last section concludes.

the procedures described in Hardeweg,
Klasen, and Waibel (2013) and United
Nations (2008). Due to attrition 11
households were dropped from the
sample, another 26 observations had to
be dropped as outliers.
We
used
two
different
questionnaires during the household
survey: one long household questionnaire
and a shorter village questionnaire.
The former was administered to the
household head and the person in
charge of decisions about food eaten by
the household. While the household
head can be male or female, the person
making decisions about food is usually
a female household member. The latter
questionnaire was answered by the village
head or deputy village head. The main
household questionnaire usually refers to
the period of the past twelve months and
covers basic data on individual household
members, sections on all possible income
components, such as agriculture and
fishing, information on assets, land
and consumption. Furthermore, a
comprehensive food security section
measuring for example household dietary
diversity, FCS and calorie and protein
intake was applied using the 1-week
recall method. Additionally, to capture
periods of food insecurity across the past
year, other indicators such as months of
adequate food provisioning, CSI and FCS
for different seasons were measured with
a 1-year recall section.

In comparison to data from other
sources, our sample is representative of
the rural population of Stung Treng. As
depicted in Table B1 in Appendix B, the
average household size is 5.22, which is
close to the figure provided by the NCDD
(2009) for the province of Stung Treng.

Sample and methodology
Data

W

e collected the data in
two household surveys in
May 2013 and 2014 in the
Cambodian province of Stung Treng.
The original sample from 2013 contained
600 households which we identified in a
two-stage sampling procedure. In the first
step, 30 villages were sampled from the
list of all 129 rural villages in the province
with probabilities proportional to their
size (PPS) measured as the number
of households. In the second step, 20
households were randomly sampled
from each village’s household list. This
procedure results in equal probability
for each household in the province to
be part of the sample and is based on
9