Symposium
2001/32 12 July 2001 English only
|
Symposium on Global Review of 2000
Round of
Population and Housing Censuses:
Mid-Decade Assessment
and Future Prospects
Statistics Division
Department of Economic and Social Affairs
United Nations Secretariat
New York, 7-10 August 2001
Population
and housing censuses in Italy *
Aldo
Orasi and Angela Ferruzza **
CONTENTS
B. Census development
programme and preparation phase. 1
C. New units of enumeration
and new units of analysis
D. Monitoring system for the
fieldwork
E. Optical reading for data
capturing
F. Data processing and data
quality analysis
1.
The census can be more relevant to current needs for monitoring
progress towards national goals improving the geographical and territorial
matters. In fact, the census is the most important opportunity to collect data
in small geographic areas. In Italy there is also the unique opportunity to
collect detailed information for educational attainment, status in employment
and branch of economic activities.
2.
The international community plays a very
important role in ensuring the continuity and maintaining the quality of
censuses. In fact, given the very high level of expenses of census operations,
it is important that the international community emphasizes the necessity to do
the census to collect this kind of information.
3.
The Italian experience will be described
briefly in this paper. But because the Italian population and housing censuses
will be held on 21 October 2001, it is important to say that it is too early
for Italy to give an evaluation of the situation. Perhaps in the near future
ISTAT will have more information to describe the problems encountered and
actions taken by census programmes, to speak about advantages and disadvantages
of the newly adopted technologies, to explain what it is possible to learn from
the current round of censuses, what are the specific areas that need to be
targeted as priorities and so on.
4.
New units of enumeration and new units of
analysis will be used in the next Italian census. The census is based on the
traditional method of enumeration, with questionnaires delivered to homes by
about 100,000 designated enumerators, relying on the local authorities for the
fieldwork. A computerized system is monitoring the fieldwork processes. Optical
reading will be used for data capturing and the most innovative equipment for
data processing. The Internet will be used to disseminate information during
the census operations and to disseminate census results at the end of the operations.
5.
The population general census and the general
census of dwellings represent the main enumeration about the state of the
population and dwellings. In Italy the census is a complete enumeration, and it
is conducted every 10 years. It is a direct enumeration, and it is based on the
traditional completion of questionnaires by respondents. Therefore, it is a
very expensive and irreproducible enumeration.
6.
With regard to the organization of the census
out of ISTAT, besides the ISTAT regional departments, there are some other
institutions that will be actively involved in the census execution. Among
these, particularly the municipalities will perform an essential role for the
success of the census operation.
7.
Figure 1 shows the external organization of
census operations in Italy. With regard to the legal aspects, the population
and housing census is based on a Census Law for financing and on a Census
Regulation for all the other technical and organizational aspects. Censuses are
based also on articles 56 and 57 of the Italian constitution.
8.
Traditionally the main aims of the population
and dwellings censuses are:
·
To cover the total population on a uniform
geographic basis and to enumerate the population’s characteristics;
·
To cover the total housing on a uniform
geographic basis and to enumerate the housing’s characteristics;
·
To update and to review the population register
using population census data; and
·
To determine the legally resident population.
9.
Taking into consideration past experiences, it
would be useful if the municipalities could make comparisons between the census
and the population register directly. It is important that the population
register of resident persons represent the most important demographic
characteristics. Furthermore, the general trend is that of a transition from a
questionnaire-based enumeration to a register-based one through progressive
mixed solutions.
10.
To test the innovations and modifications for
the collection of data in the 2001 Italian census, the following steps have
been carried out:
·
Pre-test (April 1998);
·
First pilot survey (25th October 1998);
·
Second pilot survey (April 2000).
11.
An important innovation in the next census will be the introduction of
new units of enumeration and analysis: “buildings”, on which a specific census
will be carried out and, regarding the population census, the “persons
temporarily resident” in a dwelling (but usually resident elsewhere).
12.
The availability of a good informative database is important,
especially for local public administrators who need to know about the
consistency and characteristics of the Italian stock of buildings, their
seismic risks and so on. Past censuses took into account only residential
buildings (having at least one dwelling), and data were collected from the
households living inside. The collected data on buildings were therefore
“dwelling attributes”. They were repeated for each household (dwelling) in the
building and they were often incoherent. In short, the data did not allow an
unequivocal identification and enumeration of buildings. In order to solve
these quality problems and to achieve a better knowledge at this unit of
analysis, a specific census on buildings will be carried out, using a single
form for each building, to be filled in by the enumerators some days before the
population census.
13.
One of the most relevant aims of the population census is the
“counting” of usually resident persons, so that public administrative registers
of the population (called “Anagrafi”) could be revised and updated for each of
the 8,100 Italian municipalities. These registers are sometimes old or
incorrect, especially because of delays in enrolling and removing people, and
they are biased. These biases are to be corrected according to the law on the
basis of census counts, and municipalities will have to officially verify
whether the census or the population register is correct.
14.
Even if correctly updated, the public register of a municipality may
not exactly measure the true number of people living there. In fact, the living
habits of certain groups of people make the traditional definition of
“resident” quite unsuitable and improper. People like students or employed
persons, who spend part of the year or of the week far from their home (living
in another place), or persons who spend a long time at their relatives’ house,
cannot identify a single place of usual residence, as theoretically required.
15.
A new unit of analysis, the “person temporarily living in a dwelling”,
will be used in the Italian census. This is a person who is registered as a
usual resident in a certain municipality, where he will be enumerated as a
resident, but who lives, lodges or works in a another municipality, where he
will be enumerated as “temporarily resident”.
16.
Taking into consideration this kind of residential pattern should also
improve data on housing. In the 1991 census a dwelling was classified as “not
occupied” if no resident (from the public register) lived there. This criterion
overestimated the number of “not occupied” dwellings, which included those
occupied by temporary residents. The new definition will better identify houses
that are actually empty.
17.
The census involves great organizational
complexity, since the fieldwork of enumeration is delegated to 8,100
municipalities, which are characterized by different demographic sizes,
different efficiency and different levels of automation. ISTAT will be
supported by its own regional offices to control the performance of census
fieldwork. For monitoring census operations a web application was developed to
monitor, in real time, census activities and to quickly take measures against
anomalies and delays.
18.
Moreover, an auxiliary model computerized
system will help the municipalities to carefully control the number of
distributed and gathered questionnaires, to compute daily partial aggregation
of enumeration units, to summarize data for each enumeration area and to
compute the final aggregation of enumeration units.
19.
ISTAT is going to adopt optical reading for
data capturing of census questionnaires and automatic processing for coding
some alphabetic variables. In order to realize this plan, the three main stages
are entrusted to one single “provider of service”, which will undertake the
graphic design of the questionnaires, their printing, transportation (to and
from the municipalities) and their scanning.
20.
ISTAT determined that the use of a single
provider would avoid the possibility that one supplier could shift responsibilities
onto another, in case that the required standards, for any of the different
activities (printing, transportation of questionnaires and, mostly, optical
reading of the forms) would not be achieved.
21.
Immediately after the scanning process, the
main alphabetical variables (name of the municipalities, citizenship, level of
education) will be automatically coded, relieving the municipal offices of the
burden of a lengthy and difficult clerical coding task.
22.
To assure the best quality of this expensive
and important phase, a certified and reliable company will be delegated to
perform both the monitoring of the optical reading process and the controls on
the quality of acquired data and images.
23.
In the census, there are three main stages that
modify the data:
·
Identifying the statistical units;
·
Quantitative adjustment; and
·
Editing and imputation operation on the value.
24.
In the first stage we have to transform some sequences
of characters that are recorded (we can call it “ physical raw data”) into
objects that have some meaning in our context ("logical raw data”).
25.
For the second stage, we need a method to
reconstruct total missing values and to identify duplication. Furthermore, we
need controls of the territory codes. Also, we aim to have a method to do this
automatically as much as possible.
26.
For the third stage, we have to develop methods
to correct systematic errors and stochastic errors. In order to do this, we
shall use “rules“ to identify and analyse the erroneous data. In imputing the
values, we will utilize mainly the probabilistic approach whenever we cannot
identify the right value in a deterministic way with a very high degree of
confidence. In doing this, we intend to use all the information to determine
the right values.
27.
In order to evaluate census data quality, a
survey has been planned to evaluate coverage errors affecting the census
enumeration. In particular, both undercoverage and overcoverage errors will be
estimated by means of a post-enumeration survey (PES) carried out on a sample
of census enumeration areas (CEAs).
28.
This survey is intended to be an external
source for controlling the quality of the population census. The results of the
control surveys will not be used to correct census data, but simply to provide
estimates of the accuracy of the census data. Instead census results will be
checked with administrative records from the population register.
1. Issue 1: Strategies for involving stakeholders in
census activities
29.
Who are the stakeholders in Italy? The
municipalities will perform an essential role for the success of the census
operation. The municipalities are the stakeholders in the collection,
dissemination, planning and use of population data. Municipalities have the
most important role in the collection of data. But they are also the most
important users of data. This is a unique opportunity to collect some of the
information about the population and housing for each municipality and for the
municipalities divided in enumeration areas. For example, in addition to the
number of people, the census makes it possible to know the relationship of
persons, educational attainment, current activities status, status in
employment, branch of economic activities, information related to journey to
place of work or place of study, the number of buildings, the number of
dwellings, people living in institutional households and so forth. So municipalities
have the unique opportunity to collect and then to use these data. They know
this, but ISTAT is trying to involve them in using statistics through national
reports, the Internet, and appropriate training classes.
30.
Moreover, Italy has a special commission for
the 12 biggest municipalities. In this commission census projects and problems
have been introduced and argued about. This commission is continuing its job
now and will help during the census operations.
31.
A special census commission comprising many
professors and official delegates from institutions collaborated with the
census department to project the enumeration structures and contents, to define
the dissemination plan and to discuss some methodological aspects. In this
commission official delegates of some institutions and of some municipalities
have been involved in the selection and development of topics included in the
census questionnaire. Data about the 1991 census dissemination were also used
in the selection of topics.
32.
In addition, in 1999 a national meeting of the
Italian Statistical Society (SIS) was organized on the topic of censuses to
suggest and discuss ideas and hypotheses.
2. Issue 2: Strategies for
choosing among data-collection methods as sources of demographic and social
statistics: censuses, sample surveys and administrative records
33.
Taking into account costs, periodicity,
geographic coverage, respondent burden, timeliness, content and statutory considerations,
Italy has used traditional census data collection. Sample surveys can give some
information but not at the geographic level of enumeration areas.
Administrative records have only some of the collected information and at the
moment their quality is heterogeneous. Moreover, not all of them are
computerized. Perhaps the next census will be the one in which Italy will use a
mixed system of traditional methods plus administrative records.
3. Issue 5: Identifying and
resolving problems of census mapping
34.
In Italy, as in other countries, the census
data-collection process provides a unique opportunity to build complete
small-area mapping. Census enumeration areas are called “Sezioni di censimento”
(SDC) and SDC features are defined to allow enumerators to easily identify
borders of their assigned region, to correctely count statistical units and to
avoid any possible double counting. Natural boundaries for SDCs are streets,
railway lines and hydrological features such as rivers and lakes.
35.
In 1991 for the first time, an advanced methodology of census mapping
was used. “CENSUS”, a complete digital database in ArcInfo format and at a
scale of 1:25,000, was developed. It integrated remote sensing images, IGMI
maps (Italian National Mapping Agency), technical maps at the regional level
and municipality information. “CENSUS” contains administrative boundaries (20
Regions, 95 Provinces and 8,100 communes, corresponding respectively to level
NUTS2, NUTS3 and NUTS5 of the European nomenclature); populated localities
boundaries (“centri abitati” and “nuclei abitati”) and SDCs (about 330,000)
boundaries and coordinates of SDC centroids. Out of 330,000 SDCs in the 1991
census, about 70 per cent were in “centri abitati”, about 10 per cent in
“nuclei abitati” and the remaining 20 per cent in “case sparse”. “Centri
abitati” and “nuclei abitati” constitute about 10 per cent of the Italian area,
and “case sparse” constitute about 90 per cent of the Italian area. With regard
to population, 91 per cent is in
“centri abitati”, 3 per cent in “nuclei abitati”and 6 per cent in “case
sparse”.
36.
CENSUS 2000 is the mapping project for the
2000-2001 round of censuses. It starts with maps from the 1991 census and
combines them with available cartographic data such as digital orthophotomaps,
technical maps and vector data from other
government agencies, local authorities or private companies. These sources
contain a large amount of information on populated areas and such features as streets, rivers and railways.
The main objectives of CENSUS 2000 are: i) to redesign and reduce in size
census sections in extra-urban areas to get the integration of all census
mapping, including for the first time data from the agriculture census. “Case
sparse” in 1991 were too big to be used in agriculture and environmental
analysis; ii) for the first time, to design economic areas; iii) to improve
integration with other territorial databases of public interest; and iv) to use
data extraction from digital maps of private companies and local authorities.
Four types of populated localites are mapped in the first stage: 1) centri
abitati (populated areas), 2) nuclei abitati (small populated areas), 3)
località produttive (economic areas) and 4) case sparse (remaining areas). In
the second stage, using techniques differentiated for each type of locality,
SDCs are mapped.
37.
To fully understand the potential benefit of the census collection
process it should be considered not only the output of census data but also all
products of the considerable fieldwork, like collection of data on streets,
address numbers, and other attributes that can be used in address matching and
in geocoding of administrative records. All census outputs can be used in
developing strategies of small-area data dissemination. The digital census
mapping database can be used for area mapping, and data access via the intranet
or the Internet and travel-to-work flow data can be used in functional zones
mapping.
38.
Most of the administrative databases contain addresses which allow,
when coordinates of address numbers are known, a complete matching of data to
points on a digital map. In any case, an address is a respondent identifier and
due to confidentiality restrictions, statistical data have to be geographically
aggregated.
39.
ISTAT uses a procedure of address matching to link addresses to SDC
codes, so that data of each SDC can be aggregated. For each SDC, as an
attribute of the SDC polygon, the list of streets is available, and for each
street the ranges of address numbers in the SDC, called “Itinerario di sezione”
(SDC path). After orthographic correction and normalization of the addresses,
the procedure links the SDC code to each address. An important application is
the geocoding of the ASIA register (the Italian businesses register). Applications
will be extended in the next few years also to geocoding of records from
population and social registers, including school and hospital registers.
Chieppa, A., and G. Massimini (1999). Edit and imputation: issues
involved in building a system. Proceedings
of European Workshop on the Preparation of the Census Fieldwork. Joint
ISTAT-ECE-Eurostat Meeting, Rome, 12-14 April.
Chieppa, A., and F. Panizon
(2001). Data quality control system for the 2001 Italian population census.
Paper presented at International Conference on Quality in Official Statistics,
organized by Statistics Sweden and Eurostat, Stockholm, 14-15 May.
Balestrino, R., F. Baiocchi and A. Reale (1999). Census forms optical reading: taking
the opportunity. Proceedings of European
Workshop on the Preparation of the Census Fieldwork, Joint
Istat-ECE-Eurostat Meeting, Rome, 12-14 April.
Corrado H.C., and D. Zindato (1999): Enumeration structure units of
analysis and contents of Italian 2001 Population Census. Proceedings of European Workshop on the Preparation of the Census
Fieldwork, Joint Istat-ECE-Eurostat Meeting, Rome, 12-14 April.
Crescenzi, F. (1999): New challenges and opportunities in redesign and
updating territorial database. Proceedings
of European Workshop on the Preparation of the Census Fieldwork, Joint
Istat-ECE-Eurostat Meeting, Rome, 12-14 April.
Egidi, V. (1999). “Le
strategie dell’ Istat per i Censimenti del 2000 (Istat strategies for 2000 Censuses). Paper presented at the Conference: Verso i Censimenti del 2000,
Udine, 7-9 June.
Ferruzza, A., and S. Mastroluca (1999). Italy 2001 Census development
program and the first pilot surveys, Proceedings
of European Workshop on the Preparation of the Census Fieldwork, Joint
Istat-ECE-Eurostat Meeting, Rome, 12-14 April.
Massimini, G., and P. Valente (1998). Processing the Italian Population
and Housing Census data. Paper presented at International Seminar on Census
Methodology, Portsmouth (UK), 29 April–1 May.
Vivio, R., (1999). Towards 2001 Italian Housing Census: enumeration
structure, contents and proceedings. Proceedings
of European Workshop on the Preparation of the Census Fieldwork, Joint
Istat-ECE-Eurostat Meeting, Rome, 12-14 April.