Environmental Engineering Reference
In-Depth Information
organisation) and other, and past promise was added to capture the important impact
that iteration and continuity are hypothesised to have (bayne 1999).
as appendix 14-5 shows, the analysis of 35 cases shows that a health commitment
contains up to five such catalysts and as few as none. The most frequently employed
across the initial 30 health commitments analysed were, in order, priority placement
(13), specified agent (11), other international organisation (10), money mobilised (8),
past promise (7), multi-year timetable (6), G8 body (6), and one-year timetable (4).
the wHo was explicitly invoked in only two commitments, once in 2000 on HIv/
aIDS, malaria, and tb (where compliance came in at +100) and again in 2005 on
polio (where the compliance score was +11 percent). From 2003 onward, the catalysts
of priority placement and money mobilised have been the catalysts of choice.
the effect of the catalyst variables on commitment compliance was formally
tested in an initial study of 30 cases in a series of multivariate ordinary least squares
(olS) models. because of the few degrees of freedom, due to the small sample
size and many independent factors, it was not practical to incorporate the agency,
institutionalisation, and structure variables used in earlier finance and development
studies in a single grand model. Instead, each of the three categories of variables
was tested separately. For the agency variables (the compliance catalysts), a full
(unrestricted) OLS model of commitment compliance was tested first, and the
best subset of catalysts (optimal nested sub-model) was selected using a forward-
backward search optimisation of akaike information criteria (aIc) scores. the
results generally agreed.
In the restricted optimised model, two catalysts had highly significant, strong
positive effects on compliance. The presence of a specific timetable of one year
or less tended to increase compliance with that commitment by an average
0.65 compliance points. This variable is significant above the 99 percent confidence
level. Its effect and direction remain robust among all considered models. Similarly,
delegating some responsibility for the implementation of a health commitment to
the wHo tended, on average, to improve compliance with that commitment by
0.55 compliance points. This variable is significant at the 92 percent confidence
threshold. It should be considered as a significant explanatory factor of compliance,
as its slightly deficient significance is an inevitable artefact of the small sample size
(cf. long 1997). the model has satisfactory explanatory power (adjusted r-squared
of 0.147), particularly given the small sample size and the need to estimate the three
groups of independent variables separately.
To identify more specifically what caused compliance to differ from the overall
level of compliance with all commitments in a given year, a further regression was
conducted measuring the dependent variable by mean-adjusted compliance. this
variable was constructed by removing the overall summit compliance for the given
year from each individual commitment compliance score, thereby controlling for
overall year-specific effects. Here the only significant variable identified both in the
original (full) and optimised (nested) model is 'other international organisation'. but
it reduced rather than improved compliance, and did so by 5.9 compliance points.
This variable is significant at the 94 percent confidence level.
