Lessons learnt from the 2001 Italian census and innovations leading towards a continuous census

Fabio Crescenzi, Giuseppe Sindoni, Donatella ZindatoIstat (Italian National Institute of Statistics) - Department for Censuses, Administrative and Statistical Registers

Group of Experts on Population and Housing CensusesGeneva, 23-26 September 2014

Population census before 2011

Up to and including 2001, the Italian census was taken with the conventional methodology of complete field enumeration

census forms were delivered and collected by enumerators and self-filled in by respondents

information was collected and processed without making use of any sampling techniques (the same economic, human and organisational resources were allocated to every household).

The need of a new strategy

As 2011 census was being planned, a number of factors raised questions about the possibility of continuing to rely on the conventional methodology:

the huge organizational effort imposed on municipalities, exposed to a sudden and time-concentrated increase of workload

the need of improving dissemination timeliness

the increasing difficulty of questionnaires’ delivery by enumerators, due to changes in population life-style and structure

an increasing feeling of dislike towards the census

the increasing public concern for confidentiality.

Main features of 2011 census strategy

Completeness and simultaneity of the count

Methodological and technical innovations

fieldwork guided by registers

use of new data collection techniques and new territorial instruments designed to improve coverage and quality of the enumeration

Modular strategy (set of modules to be applied flexibly according to the size of the municipality)

Crucial role of the census web management and monitoring system – SGR (backbone of the enumeration process)

From a door-to-door to a register-supported census

The 8092 Municipal Population Registers (MPRs) were used as lists of households to which census questionnaires were mailed out.

A mixed mode design was put in place: Internet return at any post office in Italy return to Municipal Collection Centres (where

specialist assistance was also available).

In a second phase, the paper questionnaire could also be returned directly to enumerators in charge of the recovery of non-response (the other return options were still available till the end of enumeration).

Lesser use of enumerators for a greater flexibility

Dramatic reduction (about 40%) of front-office staff, thanks to the use of questionnaires mail out

of a multimode data collection system (where enumerators were just one of the possible return modes, and hierarchically the last one).

Back-office operations and under-coverage recovery

Independently of the return mode chosen by respondents, all paper questionnaires converged to Municipal Census Offices (MCOs)

questionnaires had to be reviewed and, if necessary (incomplete/inconsistent questionnaires), the household was re-contacted.

for each revised and completed questionnaire, summary data had to be entered in the SGR.

Besides the recovery of non-response, enumerators were also in charge of under-coverage recovery.

Targeted recovery of unregistered individuals

In order to manage the potential under-coverage, data provided by different sources (such as the revenues agency or foreigners permits to stay archives) were used to set up a list of persons not included in registers but potentially residing in each municipality.

In largest municipalities, an additional list has been based on the pre-census Address Numbers’ Survey, containing information on potentially inhabited housing units for which there was no corresponding entry in the municipality records.

A modular strategy

Different fieldwork schedules have been established according to the size of the municipality (3 different size categories have been considered).

Major change concerning only the largest municipalities (i.e. those with at least 20.000 inhabitants and all province capitals) short form/long form strategy: information on socio-economic variables collected

on a sample basisdata so produced are significant at a census area

(grouping of contiguous and homogeneous enumeration areas) level.

Need of a flexible IT management system The management of a modular and flexible strategy implied

a higher level of complexity and a multiplication of risk factors.

The system was designed to guarantee flexibility to fieldwork organization within each MCO.

The coexistence of different return modes required an IT system (SGR) constantly updated on the status of each questionnaire, enabling enumerators to be directed to households who had not yet returned the questionnaire.

Auxiliary lists used fore the targeted recovery of unregistered individuals were integrated and loaded into the system.

Lessons learnt: flexibility the winning card

The 2011 population census introduced methodological innovations towards the planning of a register-based census.

Significant changes in census methodology and techniques large reduction of municipalities front office workload.

Respondents’ burden reduced thanks to the use of new data collection techniques and to the short form/long form strategy quite satisfying spontaneous return rate.

Quite surprising proactive attitude of respondents “problem” (need to bring forward various operations originally scheduled for a later phase).

Lessons learnt: no more paperThe logistics of the mail out/mail back process, entrusted to a contractor, proved to be very complex and with many points of failure

not all addresses in the municipal registers were successfully processed by the contractor

about 2,000,000 questionnaires had to be delivered directly by enumerators

information about questionnaires returned to post offices was not always reported promptly, making it impossible for municipal offices to keep track of which households had to be contacted to prompt return of the questionnaire.

Still need for a change

Innovations designed for 2011 census not enough to achieve a stable and enduring balance between census costs and benefits decennial census data quickly outdated costs remain high and too concentrated in time

(huge one shot activities)use of administrative data not suitable to the

potential offered by the Italian context

need of a totally different approach, combining a most effective use of administrative sources with

the spread over years of the census fieldwork

Italian approach to rolling census

Almost every day people leave traces in administrative information systems enormous amount of spatial data potentially available for statistical analyses.

Administrative data frequently not updated, affected by coverage errors, obtained by non harmonized classifications and definitions.

Italian approach to a continuous census use of administrative data sources + sample surveys rotating through a multi-year period of time, in order to produce spatially detailed data every year.

Integration of registers and surveys

The main administrative source will be the Municipal Population Registers (MPRs), which will be used for counting usual residents and producing key data on the demographic structure of population and households.

The C-sample survey will be specifically designed to measure under- and over-coverage of MPRs (respectively, people usually resident in the municipality who are not registered and people registered but no longer usually resident in the municipality).

The D-sample survey will be designed for producing estimates of socio-economic variables.

The C-sample survey

Part of a sophisticated surveillance system aimed at keeping under control the count from MPRs based on the integration of the C-sample results with various

integrated systems of administrative data currently in use at Istat.

C-sample estimates will be produced through the dual system (capture-recapture) method. The first capture will be represented by the population register

while the second will be carried out by a complete enumeration within a sample of enumeration areas (or addresses).

Over-coverage will be estimated by field follow-up after linkage between the 1st and the 2nd capture.

The C-sample survey

Short form-only survey questionnaires will be completed by enumerators using hand held devices.

About 500,000 households will be surveyed every year, according to a sampling scheme by municipality size. first year (municipalities > 35,000 inhabitants): 320,000

households years 2nd-5th : 510,000 households per year

Intended as a statistical test on the coverage errors of MPRs MPRs whose under/over-coverage errors will result over a given threshold will be revised based on the use of administrative data derived from the System of Integrated Microdata (SIM).

The C-sample survey

The C-Sample survey is intended as a statistical test on the coverage errors of MPRs

MPRs whose under/overcoverage errors will result over a given threshold will be revised based on the use of administrative data derived from the System of Integrated Microdata (SIM)

The de jure population in the census year (2021) will be derived from the MPRs (revised in the course of the C-sample survey 5 years-cycle)

The D-sample survey Main aim of the D-sample survey is to replace the 2011

census long form, in order to collect data (not included in population registers nor in the SIM) on the core topics included in E.U. Regulations.

Besides, the D-sample survey will meet user needsby providing more frequent updates, thus removing the

decline in accuracy over the decadepossibly by satisfying new information needs, as the

field collection of core data will be replaced by the integration of administrative data sources.

The D-sample survey

Designed to give accurate yearly estimates for output areas > 100.000 inhabitants, while a larger sample pooled across different years will be necessary to obtain more detailed territorial estimates: estimates for output areas 35,000 - 100,000 will obtained by pooling

the sample of 3 consecutive years; estimates for output areas < 35,000 will be obtained by pooling the 5

year samples.

Paperless data collection, with a mixed mode system including CAWI and CAPI at the MCCs/with enumerators.

Which benefits from a continuous census?

Disadvantages deriving from higher sampling errors at the finest territorial levels (municipal and sub-municipal) will be balanced by a more frequent statistical information supply.

Continuous operations should bring a significant growth of fieldwork efficiency and thus benefits in terms of increased quality.

The demand of public financial resources will be diluted over time and continuous operations should make service contracts more attractive and possibly cheaper than in “one shot” operations.

Challenges towards a continuous census Several technological issues to be solved

to reduce as far as possible the use of paper questionnaires different organizational solutions, based on local patterns in returning census questionnaires and differences in ICT usage

many areas in Italy not covered by data transmission networks for mobile devices (need of a bi-modal approach, allowing to store data locally and postpone upload to the server).

Need to carefully evaluate the impact of multimode data collection on data quality quality and treatment of administrative sources.