places, estimated with the adjustment of the 
network and obtained from cartography at 
small scale, shows a relevant difference 
between them as it can be seen in the above 
table. 
The places: 
• GAMBARIE 
• PENTIMELE 
are not in the city and they have been 
identified from a cartography at small scale. 
So, in this case, even if the difference are too 
big, such difference may be accepted. 
On the contrary, the places: 
• RAVAGNESE 
• RC CENTRO 
Are situated inside the city (here it’s available 
the cartography at large and medium scale) so 
the identification of errors take us not to accept 
such differences. 
COORDINATE DIFFERENCE 
PLACES 
AN (m) 
AE (m) 
Gambarie 
75.59 
-26.23 
RAVAGNESE 
(Saracinello) 
-350.87 
245.94 
PENTIMELE 
-24.70 
-5.42 
RC Centro 
242.00 
411.39 
In conclusion it’s to mark that that the results 
are good but the sample was not big enough to 
allow us to formulate reliable considerations. 
On the contrary, a bigger sample would have 
allowed us to do a proper statistical analysis 
and to obtain a more consistent validation of 
the survey data. 
In the appendix, it’s explained the developing 
of a methodology for data analysis with 
multivariate statistical inference that is very 
useful to solve problem like that. In fact the 
distribution free inference allow to judge the 
fixing between histograms and distributions, 
the dependence or the independence of the 
samples and their dispersion in both cases. 
Appendix 
A) DISTRIBUTION-FREE INFERENCE 
Distribution-free significance tests (sometimes 
called non-parametric significance tests, 
although the two terms are not synonyms) 
collect a broad category of significance tests, 
that can be used instead of classical 
significance tests. 
Distribution-free significance tests present 
some advantages over classical ones: 
♦ ###they ask soft, a priori, hypotheses on 
the behaviour of statistical populations; 
♦ ###their understanding is usually easy and 
generally quite immediate; 
♦ ###they don't ask heavy calculation, 
neither for the storage of information nor 
for the processing time. 
However some disadvantages over classical 
ones belong, by definition, to distribution-free 
significance tests: 
♦ ###they waste information; 
♦ ###they tend to be too conservative, i.e. 
the null hypothesis is accepted too often.