Planning a Census By the End of
the Nineties

Eliahu Ben-Moshe

Census and Demography Department

Tel. 972-2-655 3474

FAX 972-2-655 3531

e-mail: msebm@pluto.mscc.huji.ac.il

When planning a Census there are several basic principles underlying the general
planning process. Not all of them are absolutely clear at the beginning of the
process; some of them develop during the process and became clear only by the
end of it. However, they provide useful insights into the general planning process.
I would like to begin this introduction on the main features of the Israeli Census with
a review of the main planning principles. This will help to provide an understanding
of the presentations of the workshop.

1. Main Planning Principles Of Israel's Current
Census (1995)

In the planning of Israel's current Census six basic underlying principles may be
identified:


1.	Pervasive but intelligent use of state of the art technological developments
2.	Extensive use of existing external information sources
3.	Cooperation with other public agencies and the private sector
4.	Struggle for integration at every step
5.	Simplification of processes
6.	Increase the use of statistical methodology

1.1 Intelligent use of state of the art technological
developments

On this issue the emphasis should be put on the intelligent use of technological tools.
As you will see during the next days, in the planning and operation of the Israeli
Census we used state of the art technology extensively. However, we did this always
bearing in mind that our main objective was to satisfy real Census needs, and not
simply for the sake of using new technologies. For this reason we worked hard to
adapt these technologies to Census needs, and not the other way around.

A very important issue that needs to be addressed regarding technology is the
problem of the continuous and rapid development of new technologies. I should say
that we do not have a general solution for this problem, but we acquired some
experience that may be useful for the future.

The first step should be to identify the basic technology that may be useful in
Census planning. This implies ruling out other developments that are not relevant.
For example, since we intended to conduct an enumerator based door-to-door Census
that included dropping off questionnaires that were to be completed by the
respondent, we ruled out at a relatively early stage the technology of notebook or
laptop computers. This decision was made primarily for economic reasons. The basis
was primarily economic: we couldn't afford to buy thousands of this portable
computers for all the enumerators. Furthermore, since the use of portable
computers would force us to change our collection methodology, from self-filled
questionnaires to interviewer-filled questionnaires, this will result in a significant
increase in the number of enumerators (and computers) .

A second step is to identify what are the expected future developments in the
technology we decide to adopt. For example once we decided to use scanning and
OCR in the data capture process (ODE), we tried to anticipate the future
developments in the field. When a similar approach was used in Switzerland, it was
based on 386 and 486 PC's. We identified from the very beginning that the coming
development for scanning and OCR systems will be based on a UNIX platform.
Therefore we decided to base our system on that platform despite the fact that the
new machines were not yet on the market. We also decided not to invest in the
development of OCR capable of recognizing handwritten Hebrew since it was
improbable that this would become marketable in the future. Indeed, as of today we
do not know of any OCR technology capable of "reading" handwritten Hebrew.

1.2 Extensive use of existing external information sources

We searched for and found several external sources of information that were used
extensively in the planning process and during the field work of the Census.

One of the most important sources of external information was the National
Population Register (NPR). This Register suffers from several problems that made
it not suitable as the basis for an administrative Census. Since the NPR also
serves as the Voters Register, all the people who have voting rights are included.
Furthermore, we estimate that more than half a million records (around 10%) in the
NPR are of people who left the country (emigrants), many of them many years ago.
This excess of unrecognizable records is the first problem. The second problem is
that many of the people registered in the NPR and who are actually living in the
country are registered at the wrong address. We estimate that more than 20% of the
people living in Israel are not registered at their correct address in the NPR. In
many instances they are registered in the wrong locality.

Despite all these shortcomings, the NPR was used extensively as a source for
addresses and rough population estimates in order to improve population coverage
and in the process of automated mapping.

We used other external sources in the mapping process. In general, every available
map was used. This reduced to almost zero the need for the expensive field
collection of geographic information. During the mapping process we used aerial
photos, detailed conventional maps, and even tourist maps to capture any useful
piece of geographic information. Naturally, we also used the limited stock of digital
maps that was available before the cartographic digitizing for the Census was
initiated.

We also took advantage of the existing division of the population by Poll-Areas that
is recorded in the NPR. Estimates based on Poll-Areas and their borders served in
the process of dividing the different localities into size-homogenous enumeration
areas (EA).

We also used extensively materials from the 1983 Census regarding population,
addresses, maps and the geographic infrastructure that supported this Census.
Although Israel's population grew by more than 38% (from 4.1 million to 5.7 million
people, half of it due to immigration) between 1983 and 1995, our 1983 Census
provided us with useful information to plan the 1995 Census.

1.3 Cooperation with other public agencies and the private
sector

We tried whenever possible to take advantage of the existing expertise in other
governmental or public agencies. One of the most outstanding and successful was
the collaboration with The Survey of Israel, the national mapping agency that
assisted the CBS in preparing the digital mapping infrastructure (GIS).

A similar approach was used with private firms. The parts of the mapping process
that could be suitably performed in the private sector were identified and tenders
were issued to private firms specializing in photogrametry and digitizing. Another
example of this type of collaboration with the private sector was the development of
the Optical Data Entry system that will be discussed tomorrow. This system that
allowed the data capture of information from the Census questionnaires was
developed in close cooperation with private firms.

1.4 Struggle for integration at every step

The integration of the different systems of an undertaking of this magnitude is a
very complex matter. On one hand the advantages are clear. By combining efforts
through integration we may avoid duplicating operations, thereby reducing the number
of operations and saving money. In this way we may also exploit the advantages of
one system for the benefit of others. But on the other hand, the dependence of the
different systems on each other may complicate and slow down the work in all of
them. This may jeopardize Census time-tables. The strategy we adopted was to
struggle whenever possible for integration but without creating absolute dependence
among the different projects. This means that on occasion we chose not to fully
integrate processes in order to avoid the shortcomings mentioned above.

1.5 Simplification of processes

This principle is not easy to accomplish in such a complicated undertaking as a
Census. However it was one of the principles we stated from the beginning, and we
tried to adhere to it. We simplified some of the processes, especially those related
to the enumerators' field work. This was accomplished primarily by removing from
the enumerators' work load those cases that were very uncommon. These cases
were assigned to the enumerator supervisors to resolve.

We also invested efforts in simplifying the questionnaires that were intended to be
completed by the respondents. Questions that could not be stated in a simple form,
without too many explanatory notes, were removed from the Census questionnaires.

We found that extensive explanations in the questionnaires worked as an "alibi" for
the Census planners, but were not effective, because most people fill out the
questionnaire without reading them. In the same way, field procedures that were too
complicated had a very low probability of being completed successfully. Planners
could always say (as an "alibi") that they provided exact instructions, but actually
nobody was ready or able to follow after them because they were to complicated.
Therefore, we sought ways to simplify them, even if the price was loosing some
accuracy.

1.6 Increase the use of statistical methodology

In the 1995 Census we expanded greatly on the use of statistical methodology. We
used a method of indirect standardization in the process of delineation of
enumeration areas (EA) to enable a reasonably equal work load distribution for each
enumerator. A statistical approach was also used to classify EA's by level of
difficulty. This made it possible to remunerate enumerators who faced special
problems and worked harder during the Census fairly, within the limits of a pre-
established budget. Sampling was used in several operations including the control
enumerators' performance during the field work and during the quality control
process of the Data Capture system (ODE). In addition, a special sampling
procedure was developed to get a representative sample for the Post Enumeration
Survey immediately following the Census enumeration. Today we are using several
statistical models in the last stages of the editing of the Census file and the
imputation of missing variables and records.

The planning of the Census was developed on the basis of these main principles. In
the next days we are going to deal with those parts of the Census where use was
made of new technological developments. In order to help you to understand the
coming presentations I will now provide you with a short description of the Israeli
Census, emphasizing the three main developments on which we will be focusing in the
workshop:


1.	The use of Geographical Information System technology;
2.	The use of an Optical Data Entry for data capture; and
3.	The extensive use of computerization during the Census field work

Planning a Census By the End of the Nineties

1. Main Planning Principles Of Israel's Current Census (1995)

1.1 Intelligent use of state of the art technological developments