29
legal restricions on isheries. Two supervising ield oicers coordinated the administraion of the survey
and conducted this rapid assessment. Summaries of each village, based on the rapid assessment and key
informant interviews, are provided in Annex I. The informaion collected during the rapid assessment
considered the following quesions:
1. What: What types of ishery resources are
exploited by people in the community?
2. Who: Who in the community exploits the
resources? To what extent do ishermen only ish as their sole source of income and
livelihood, or do they have other sources of income and livelihoods?
3. When: Time of year, month, moon, ide, day,
etc.?
4. Where: Where are the resources gathered?
What are the use rights?
5. Why: What are the resources gathered
for? Household consumpion? Selling in the market? Approximately how much do each
use?
6. How
: How are the resources gathered ishing equipment, methods? Source of equipment
if any? Source of spare parts, maintenance, and fuel if needed?
7. How: How are the resources distributed? If
sold, how is it sold? When, where, to whom? If traded, how? For what, when, where,
with whom? If given to kinsmen or other families, is it reciprocal? Does local ish get
processed in any way? Fishermen use ice, do drying, smoking, pickling, etc.?
5.4 Survey Questionnaire for Individual Fishermen
The sample survey form is included in Annex 2. It was translated and pretested in Somalia and includes
variables meant to provide a descripion of the background to ishing legal and illegal in Somalia,
as laid out in the conceptual framework. The sample includes ten towns represening all areas of Somalia,
from ciies to small villages along the Indian Ocean and Gulf of Aden coast, including the longstanding
autonomous states of Somaliland, Puntland and other federal states. Names of towns sampled and sample
size are in Table 1, and Figure 2 shows the map of sampled locaions.
sizes
Somali land
Punt land
Other Total
Maydh 36
36 Berberra
36 36
Eyl 36
36 Garacad
36 36
Bosaso 36
36 Lasqoray
36 36
Baargaal 36
36 Kismayo
42 42
Mogadishu 42
42 Hobyo
36 36
Total 72
180 120
372 We designed the survey to select a representaive
sample of ish landing sites from several of the main coastal administraive regions of the country,
including Somaliland, Puntland, Galmudug, Jubbaland and Benadir. This was not a random sample but a
purposeful sample that ensured all coastal regions
were covered. The inal sites sampled were selected from a list of 20 Adeso-recommended sites and
factored in budget, iming and security consideraions. Due to the extensive coastline, distances covered,
and taking into account cost factors and diferences in local dialects, we used several diferent survey
teams in the diferent surveyed areas of Somaliland, Puntland and three other southern coastal states.
Surveying ishermen at a landing site At each landing site, we used a systemaic sampling
design in which every third or ith house was selected depending on community size, and interviewed the
head of the household. In some cases, ishermen were selected at random at community landing
30
30-40 individuals per community to ensure adequate representaion. Time and budget did not allow for a
more rigorous sampling frame such as developing a list of all ishing households and randomly selecing
respondents or for including a larger number of surveyed communiies or sampled individuals at
each landing site. While it could be argued that the inal sample of individuals was not truly random, we
conducted some staisical analysis of diferences between three regions as shown in the below results
that we consider to generally represent the country’s ishermen given the sampling caveats noted above.
Ideally, the Somali government should undertake more rigorous surveys such as the “FRAME” survey,
using the FAO methodologies
34
. The FRAME survey is a much more extensive census-based approach of
all landing sites and provides beter informaion for determining ishing efort. When combined with other
surveys on landings data, it can help to determine whether the various isheries are experiencing
overishing or not. These surveys provide basics of catch and landings data only; however, they are not
designed to collect any informaion on IUU ishing as this one did.
While the ield supervisor conducted key informant interviews, several locally hired ield enumerators
conducted individual oral interviews with ishermen and in some cases with women working in isheries
markeingprocessing, and recorded their verbal answers on a hard copy quesionnaire. The ield
supervisors also conducted village meeings in each surveyed landing site. The results of these meeings
and other key informant interviews and observaions are found in Annex I. Data from the survey forms
was coded into Excel iles in the ield and sent to supervising researchers for review and quality control
checks. Ater compleion of the ieldwork, the original survey forms were sent to supervising researchers
to double check ield coding entries and correct them where needed. Data iles for each village were
then merged and analyzed using SYSTAT staisical sotware.
5.4 a Statistical Method Used
Throughout the paper, commonly used descripive staisics mean, standard deviaion, frequency
distribuion displayed as percent distribuion are
34
htp:www.fao.orgdocrep004Y2790Ey2790e00.htmContents Sample-Based Fishery Surveys - A Technical Handbook. FAO Fisheries Technical Paper 25.
grouped data Blalock 1979. The groupings used are Somaliland, Puntland and other.
Since the values are derived from a sample of cases, we needed to determine whether the diferences
observed in the samples could have occurred by chance alone. To accomplish this, we use inferenial
staisics appropriate to the data type to determine the probability of diference between the groups. Chi
Squared χ2 is used to determine whether frequency distribuions difer between groups, Kruskal-Wallis
H ordinal analysis of variance for diferences of variables measured on an ordinal scale e.g., low,
medium, high; small, medium, large, and Analysis of Variance F-Raio for coninuous variables measured
on a metric scale Tabichnick Fidell 2007, Blalock 1979.
Since a test of staisical signiicance tells us nothing about the strength of the relaionship Ziliak
McCloskey 2011, we also present efect sizes to provide this important informaion Cohen 1994.
Efect sizes in the analyses presented in this report vary between 0.0 and 1.0, with 0.0 indicaing no
efect and 1.0 indicaing that all of the diferences between the groups can be explained by the grouping
the strongest efect. For chi-square tests we use Cramer’s-V, a coeicient appropriate when there are
more than two rows and columns in the table being tested. The efect size we used for Analysis of Variance
is R2, which is the squared muliple correlaions between the groups and the variable described.
Like most researchers, we consider an efect size of less than 0.3 a small efect, 0.3 to 0.5 medium, and
greater than 0.5 as large.
5.5 Survey Results
5.5.1 Background Informaion This secion of the report provides informaion on
