Agriculture Reference
In-Depth Information
table 6.10
dimensions accounting for over 75% of the Variation among 12 study sites
(Villages) and based on a multiple correspondence analysis of study-site-level,
Researcher-Proposed Indicators of agroecosystem Health and sustainability
dim
Inertia
a
(%
b
)
Indicator
c
(
r
2
)
d
categories (coordinates
e
)
1
0.188 (18.03)
Indicator-resource ownership/LUU
(0.72)
H (1.00)
L (−0.72)
2
0.175 (34.76)
Distance to Nairobi (0.72)
H (−1.41)
L (0.14)
M (0.73)
3
0.146 (48.58)
Coffee production (0.58)
A (−0.44)
P (1.32)
4
0.115 (59.32)
Sheep and goats per LUU (0.48)
H (0.82)
L (−0.59)
5
0.095 (68.13)
Coliform counts/LUU (0.48)
H (−0.82)
L (0.59)
6
0.084 (75.85)
Kale yield/acre (0.51)
H (−0.85)
L (0.60)
A, absent; C, close; Dim, dimension; E, missing; F, far; H, high; L, low; LUU, land-use unit; N, none; P,
present
a
Principal inertias (also the average squared correlations; Greenacre, 1993)
b
Cumulative percentage of total inertia accounted for by the dimensions based on adjusted principal
inertias (Greenacre, 1993)
c
The indicator most highly correlated with the scores derived from optimal scale values
d
Correlation coefficient
e
Also the optimal scale values for the categories
participatory tools. Since the process is adaptive, the lists can be further refined as
more understanding of the system is gained. In addition, the larger data sets serve
as reference points to help interpret results from the more refined suites.
The use of correspondence analysis had three main advantages. First, the rela-
tionships between various indicators were complex and did not often fit the multivar-
iate normal distributional requirements of most analytical approaches for continuous
data. More crucial though was the presence of nonresponses and zero values that
carried special meaning but would be discarded in most analytical methods for
continuous data. Finally, interpretation of the indicators would only be meaningful
relative to a set of cutoff or threshold values, necessitating a transformation of con-
tinuous variables to an ordinal scale. Correspondence analysis provided a means for
summarizing and graphically presenting the data in a way that enabled identification
of important trends and for reducing the dimensions of the indicator suite.
6.4.3 p
r A C t i C A l i t y
A n D
A
p p l i C At i o n
The process used in this research project is predicated on two main assumptions. The
first is that there exists a community with a set of common goals and values within
the targeted study sites. The second is that there exists sufficient capacity for collective
action to enable negotiation and compromise if the goals and values are competing or
conflicting and for the development of a community-based monitoring system.
The combination of participatory methods and soft system methodology pro-
vides a means through which goals and values can be stated and negotiated. The
Search WWH ::
Custom Search