![](bg3.png)
We conclude remarking the importance of recog-
nizing the relationships existing among different in-
terventions. In practice, if each intervention was in-
dependently planned, there would be no difference
between to plan q interventions and to plan q times
an intervention. What will make the quantum leap
is identifying the dependencies existing among dif-
ferent types of interventions, setting up a hierarchical
system that will allow to start well-coordinated and
highly correlated tasks, according to a bottom-up ap-
proach aiming at privileging the construction of basic
common infrastructures.
3 IMPACT ANALYSIS
We here introduce a methodology for carrying out
the analysis of the impact of a planned intervention.
It is based on the concept of indicator. Indicators
have been introduced in statistics and are currently
used in a variety of areas, among which the manage-
ment control (Smith, 2009); here we use indicators
for carrying out the analysis of the impact of interven-
tions. An indicator is a mathematical function defined
over a finite or infinite domain commonly defined as
D = D
1
× D
2
× ··· × D
n
, where each D
i
is a finite set
of numbers (real, integer or natural) and n ∈ N de-
scribes the quantity of homogeneous data which we
want to get concise information from. In the man-
agement control, statistical indicators are used to get
concise information about some specific aspect of re-
ality; depending on the type of analysis we are carry-
ing on — pre-analysis, post-analysis, feasibility anal-
ysis, benchmarking etc. — many different categories
of indicators can be used. In the recent literature there
are several proposals providing sets of indicators, or-
ganized by category, level of aggregation, homogene-
ity, correlation etc. (see, e.g., (European Commission,
2010; eGEP, 2012; Ojo et al., 2005; Understand, 2006
)).
From what we discussed before, it is clear that the
Indicators Set (IS) plays a critical role in the whole
process of planning, designing, and evaluating inter-
ventions; the following points are therefore crucial:
1. The definition of a correct and complete Indica-
tors Set able to model the scenario.
2. The indicators in the IS must be easily measured
and constantly monitored before, during, and af-
ter the intervention. Information sources must be
reliable for the whole duration of the process.
3. In order to improve the reliability, the IS should be
chosen to be partially redundant, i.e. there should
be some correlation between different indicators
and, if possible, information sources should be
chosen to obtain independently values of corre-
lated indicators.
With distinct information sources providing the
values of the indicators, it is possible on one side to
havea precise picture of the real evolutionof the inter-
vention/project, on the other a variation in the correla-
tion between related indicators might point out some
errors in the measure or in the update of an indicator
and, in the long run, can help in the assessment of the
information sources themselves.
Given an indicators set I = {i
1
,i
2
,..., i
n
}, we
define an aggregation (of the indicators) A =
{A
1
,A
2
,..., A
k
}, where A
i
⊆ I for any i and A
i
∩ A
j
=
/
0 for i 6= j; in other words, an aggregation is a par-
tition of I, conceptually based on a high level of ho-
mogeneity. From the decisionmaker point of view,
both indicators and aggregations belong to concep-
tual categories whose level is not sufficiently high.
The decisionmaker prefers to reason about concrete
objectives, directly related to benefits for citizens, en-
terprises, concerns, public administration etc. When
defining a main topic for an intervention (e.g., the area
of ICT) it is easy to define a set of (concrete) possi-
bly interesting objectives O = {o
1
,o
2
,..., o
m
}. Once
O has been defined, we expect it very slowly changes
as time passes, so that we can assume without loss of
generality O is fixed. For each item o
i
∈ O it is possi-
ble to identify its correlations to some indicators in I
or, more simply, to elements in A.
In this way, when interested in an objective o
i
, the
decisionmaker can be easily informed about the in-
volved indicators, related to o
i
. It will be sufficient
to make explicit all the correlations and store them
into some suitable supporting system. Notice that we
can consistently extend our assumption of static sets,
what leads us to static correlations. Identifying ele-
ments of sets and their correlations can be done once;
later, only limited maintenance will be required.
The decisionmaker is also interested in con-
textualizing information (according territory, socio-
economics, politics etc.). We assume for simplicity
one semantic coordinate of contextualization. Hence,
we introduce a set of contexts R = {r
1
,r
2
,..., r
ℓ
}
(e.g., the main politic units, or regions, of a given
country). It is possible to introduce more sets of
contexts, all of them to be considered as orthogonal.
On the base of the context analysis, and of laws and
rules, high priority objectives can defined, immedi-
ately identifying the involved indicators.
In order to describe all this knowledge we exploit
the mathematical concept of graph; for basic defini-
tions on graphs (simple graph, tree, forest, walk etc.)
see for instance (Diestel, 2006). In particular we are
Planning,DesigningandEvaluatingMultipleeGovernmentInterventions
87