228
G. M. Oliveira et al.
(KMO) sample adequacy measurement and Bartlett Sphericity Test were applied to
evaluate the observed correlation coefficients. A KMO value of 0.935 was found,
indicating an almost perfect sample adequacy (varying between 0 and 1, the closer
KMO is to 1, the better the sample). Bartlett’s Test of Sphericity returns a null significance value thus, rejecting the null hypothesis (p < 0.05) and confirming no relation
between the set of variables. Additionally, the Anti-Image Correlation Matrix test
was applied to determine variables suitability in the WeGIx model. Twelve indicators
with values under 0.5 of Measures of Sampling Adequacy (MSA) were identified: 41
IPBL, 39 NATURA, 37 IWMM, 36 PMOT, 35 LF, 34 I, 31 BA, 20 WQH, 18 PSS, 17
PWSS, 15 GGW and 13 DR. In accordance, all these indicators were removed from
the initial WeGIx set of variables. Table 3 presents the three components extracted
by the factor analysis using eigenvalues higher than 1.0 as the extraction criterion.
The three components of Table 3 explain 83.5% of the variance.
Component 1—“Urban Issues”—aggregates twenty-two variables from different
dimensions that strongly contribute to WeGIx objectives, explaining 53.7% of data
variability. This component includes important issues in urban areas, spaces where
social and environmental problems are aggravated due to higher population density
and consequently enhanced consumer behaviour and road traffic.
The second component extracted—“Emissions and Income”—explains 19.1% of
data variance, suggesting that income and power purchasing growth relates to the
increase in CO 2 and pollutants’ emissions, revealing the direct relationship between
income, general consumption and gaseous emissions.
Last component—“Accidents and Fires”—explains 10.1% of data variability,
associating two apparently unlikely variables: 5 DCA and 40 RFF. One possible
explanation may relate to the lack of infrastructures and geographical isolation of
some rural or insular areas, characterized by limited transport accessibility due to
large distances and to poor quality of roads’ paving. These factors may contribute
to the increasing number of deaths by car accidents because, in such geographic
isolated regions, the extensive time needed to reach victims by emergency health
assistance aggravates these situations and their outcomes (Vidal et al. 2018). On the
other hand, rural forest fires also tend to occur in geographic isolated areas.
A simple linear additive method of aggregation was used to combine the final
set of twenty-nine WeGIx indicators by addition of two arithmetic means of NI i,j
indicators: twelve positive NI
+
p|i and seventeen negative NI
−
n|i indicators, for each
municipality i, according to Eq. 4.
WeGIx |i =
12
p=1 NI
+
p|i
12
−
17
n=1 NI
−
n|i
17
; i ∈ [1, 308]
(4)
For each year of analysis, Portugal always has a reference value of zero
W eGIx |Portugal = 0 and the W eGIx |i value for each municipality i floats above or
below. For a specific municipality i, if W eGIx |i > 0 then, life conditions in that territory are better than the national average. On the other hand, if W eGIx |i < 0 it means
that the municipality life conditions are drifting from the SDGs when compared with
the national mean value for a certain year.
G. M. Oliveira et al.
(KMO) sample adequacy measurement and Bartlett Sphericity Test were applied to
evaluate the observed correlation coefficients. A KMO value of 0.935 was found,
indicating an almost perfect sample adequacy (varying between 0 and 1, the closer
KMO is to 1, the better the sample). Bartlett’s Test of Sphericity returns a null significance value thus, rejecting the null hypothesis (p < 0.05) and confirming no relation
between the set of variables. Additionally, the Anti-Image Correlation Matrix test
was applied to determine variables suitability in the WeGIx model. Twelve indicators
with values under 0.5 of Measures of Sampling Adequacy (MSA) were identified: 41
IPBL, 39 NATURA, 37 IWMM, 36 PMOT, 35 LF, 34 I, 31 BA, 20 WQH, 18 PSS, 17
PWSS, 15 GGW and 13 DR. In accordance, all these indicators were removed from
the initial WeGIx set of variables. Table 3 presents the three components extracted
by the factor analysis using eigenvalues higher than 1.0 as the extraction criterion.
The three components of Table 3 explain 83.5% of the variance.
Component 1—“Urban Issues”—aggregates twenty-two variables from different
dimensions that strongly contribute to WeGIx objectives, explaining 53.7% of data
variability. This component includes important issues in urban areas, spaces where
social and environmental problems are aggravated due to higher population density
and consequently enhanced consumer behaviour and road traffic.
The second component extracted—“Emissions and Income”—explains 19.1% of
data variance, suggesting that income and power purchasing growth relates to the
increase in CO 2 and pollutants’ emissions, revealing the direct relationship between
income, general consumption and gaseous emissions.
Last component—“Accidents and Fires”—explains 10.1% of data variability,
associating two apparently unlikely variables: 5 DCA and 40 RFF. One possible
explanation may relate to the lack of infrastructures and geographical isolation of
some rural or insular areas, characterized by limited transport accessibility due to
large distances and to poor quality of roads’ paving. These factors may contribute
to the increasing number of deaths by car accidents because, in such geographic
isolated regions, the extensive time needed to reach victims by emergency health
assistance aggravates these situations and their outcomes (Vidal et al. 2018). On the
other hand, rural forest fires also tend to occur in geographic isolated areas.
A simple linear additive method of aggregation was used to combine the final
set of twenty-nine WeGIx indicators by addition of two arithmetic means of NI i,j
indicators: twelve positive NI
+
p|i and seventeen negative NI
−
n|i indicators, for each
municipality i, according to Eq. 4.
WeGIx |i =
12
p=1 NI
+
p|i
12
−
17
n=1 NI
−
n|i
17
; i ∈ [1, 308]
(4)
For each year of analysis, Portugal always has a reference value of zero
W eGIx |Portugal = 0 and the W eGIx |i value for each municipality i floats above or
below. For a specific municipality i, if W eGIx |i > 0 then, life conditions in that territory are better than the national average. On the other hand, if W eGIx |i < 0 it means
that the municipality life conditions are drifting from the SDGs when compared with
the national mean value for a certain year.