In the previous lecture we looked at the evolution of population geography and various approaches used in studying it.

The lecture outlines the main sources of population data.  Furthermore, it gives the merits and demerits of these sources of population data.

As stated above, population geography became popular because many countries were collecting population data and using almost every ten years.  Therefore, it is important to understand the way this data is being collected and also to be able to detect the main errors and biases that often occur in population data.  Indeed, before analyzing any population data it is essential to check for the weaknesses of this data.




After studying carefully this lecture the student should be able to:

i)     Discuss the main sources of demographic data;

ii)        Explain the importance of population data;

iii)      Suggest ways of improving the population data.


At global and national level very few data are collected primarily for demographic purposes.  Actually most data are produced for, or are by-products of administrative exercises.  The collection of this data is often supervised by the government or international organizations such as the United Nation.

Therefore, sometimes the demographers have relatively little control over the way in which this data can be analyzed and collected.  In general due to this problem the population geographer is forced to analyze population data the way it is presented or used models to analyze them.

The main sources of population data are vital registration systems, censuses, migration controls, health centres, population systems and migration controls are established for legal purposes.  These legal purposes include birth and death certificates, passports, identification cards, work permits, residence qualifications and citizenship.

On the other hand, censuses collect data that is essential for planning purposes like the provision of basic services such as power, water and transport.  In many areas, politically the population of an area determines the level of electoral representation and possibly also the pattern of resource allocation from the government.  In several areas before a new project is introduced demographic data is needed to justify the existence of this project.

For example, in order to establish a dispensary or hospital in Tanzania you need at least to have a catchments area with about 10,000 people.  The same also applies in other social services such as education and clean water supply.

Unlike vital registration system and census, sample surveys are mainly designed for research purposes.  They are mainly used to assess the impact of programmes or existing policies in order to influence the government and other funding agencies.  In addition, pure academic purposes of surveys only exist in universities and other higher institutions of learning.

Normally they look at the impact of economic, social and environmental factors on population or how government policies influence the levels and patterns of health, mortality, fertility and migration.

Usually largely government and other international organizations produce population data.  The quantity and quality of data generated depends on the financial strength of the government or these organizations.

The quality of data also depends on the main objective of collecting this data and the overall attitudes of the population on the importance of data that leads to better response to the questionnaire.  Furthermore, the quality of data depends on the accuracy of recording and processing which in recent years has been very much simplified by the use of computers and other data transfer networks (Newman, & Matzke, 1984).

2.3     CENSUSES

Wilson (1985) defines a census as the total process of collecting, compiling and publishing data on the demographic, social and economic situation of all persons in a specific territory and at a particular time.  Undeniably, a census is the largest single data collecting exercise carried out in a country.

Since it requires a lot of preparation like identifying enumeration areas, questionnaires and advertisement, a considerable amount of human resources are required to implement it from the beginning to the end.  After collecting the data, the tabulation and publication also requires a lot of time and money.

On most occasions preparation alone may take about 5 years and the processing of data may take another five years.  For example, in Tanzania the preparation for the 1988 population census began in 1984 and the volume of analysis for the same census was not out by 1994 (Newell, 1998: URT, 1988).

Therefore, under these circumstances some scholars question the validity of holding censuses that are a bit expensive and more time consuming when compared to other sources of data.  Although it is difficult to justify the importance of censuses on strictly economic terms,

it is important to remember that in most Third World Countries it is the main source of all population data.  Also the disadvantages of census in most countries can be judged on long-term basis and the utilization of this type of data spreads over a wide range of activities which fall under the government,

industries, services, research organizations and others.  Nevertheless, a few countries carry censuses for prestige purposes or simply because they have been induced by funding agencies such as the United Nations (Newell, 1988:15).

2.3.1        Characteristics of Censuses

As observed by Shryock and Siegel (1976: 29) the essential features of a population census recommended by United Nations includes:

(i) Individual Enumeration

When a census is being carried out it requires individuals to be listed separately together with their personal characteristic.  It is only on rare occasions that group enumeration can be applied in order to cover communities, which are less developed.

Yet, in this type of census details of a population are lacking and so tabulations become a mere process of adding and subtraction.  It is also impossible to cross tabulate the results.

(ii) Universality within a Defined Territory

Ideally a census to be valid it must cover the whole country and if it includes those who are absent it is a de jure census.  When this condition is not fulfilled for whatever reason e.g. war, or isolated tribes it is essential to indicate the proportion of population that is not covered.

(iii) Simultaneity

To have any comparative logic, censuses are supposed to be taken at a fixed date and time.  For example, in Tanzania the 1967, 1978 and 1988 censuses were assumed to be taken on the last Sunday of August of the respective years.  More often the enumeration canvas does not need to be completed on the same day but the official time still remains the mid-night of the census day.

Nevertheless, the more time that is used in enumeration of population, the more difficult it becomes to avoid omissions or duplications.  Sometimes during the census it is normal to ask questions which inquire a predetermined date before the census or after the census. For example, in both the 1978 and 1988 censuses in Tanzania respondents were asked a question on the usual place of residence during the previous census for all people who were aged ten years and above.

(iv)  Defined Periodicity

The United Nations recommends that censuses should be taken at regular intervals so that the information collected by censuses can be comparable in a fixed sequence.  This has major advantages in appraising the past and predicting the future.

On this basis, most countries take their censuses at regular interval of between 5 to 10 years.  This helps to analyze the data in a more conventional way.  Moreover, censuses can be divided into groups by using the way the census is carried out.

The de facto census enumerates individuals by asking where he was during the census night.  It uses a canvassing method that requires a trained enumerator to ask respondents on various population issues.  This method of census taking is very common in areas where the majority of the people are illiterate.

The de jure records the census data by using the usual place of residence.  In this type of census respondents usually fill the questionnaire and it is more used in developed countries such as in Western Europe.

It is also important to note that some censuses are very simple indeed containing no more than simple factual questions.  Others are more complex because they can extract a lot of information from the respondents.

With the rise in reluctance of the respondents in filling detailed questionnaires in many developed countries, there is a tendency to make censuses shorter and more specific.  Usually they include simple questions such as name, age, sex, relationship to the head of household, marital status, race/religion/ethnic group, education, occupation, employment status, migration, housing enmities.

Also in countries with poor vital statistics registration systems it includes questions on the number of children ever born and how many are still alive.  They may also include questions on transport and distance to important social amenities (Newell, 1988: 16).

In fact, in Tanzania, both the 1978 and 1988 population censuses included a general questionnaire, which had only five questions and a detailed questionnaire that had about thirty-two questions (URT, 1988).


Although several countries have taken censuses for quite a long period, errors in census taking are very common and this problem in more acute in developing countries.  These are:

2.4.1        Incomplete Coverage

Incomplete coverage may occur simply because an area is missed or because certain sub-groups are hard to cover completely such as the homeless, nomadic people, infants, students, vagrants, seamen and so on.

In de jure censuses errors of coverage can be increased to a large extent if an incomplete or out-of-date frame i.e. lists of addresses or villages is used.  This may happen when it is difficult to produce new frames or it is time consuming and expensive to produce.

2.4.2        Quality of Data

The other major error is more related to quality than quantity of data.  This type of error is caused by either the respondents misunderstanding the questions or mis-recording of enumerators.  In developed countries this error is common in the categorization of industry and employment activities such as housework and formal work.

In developing countries it is very common to mix agro-pastoralists with pure agriculturists.  Sometimes the error is caused by respondents just intending to give the wrong answer especially in activities, which they believe are less prestigous, or does not disclose household’s information that they consider to be secretive.  Typical examples include children born out of wedlock or divorced people many indicate that they are single.

2.4.3        Data Processing

The processing of data is also a very important activity if at all the census is used for economic and social planning.  It is a stage that needs to be organized properly as the stage of enumerating.  Always the processing of data involves checking and coding/entering data into computers and producing tables, which will be easily understood by users of data.

It needs to be carried out properly to avoid errors and biases.  Nevertheless, despite being very careful in data entry, it may not be possible to uncover all errors.  Errors of coverage will still be there and these are only solved by conducting a post-enumeration survey.  Indeed, if the census is well organized it is possible to get the results within two years.

However, in area that adequate processing capacity is lacking such as in developing countries like Tanzania, the results can be delayed for up to five or six years.  By the time such data is ready for use, some population issues will be out-of-date and they will need to use projections (Newell, 1988: 16-18).


The main aims and objectives of demographic surveys are very diversified but the common feature is that they normally involve taking a sample of the population to represent the whole population.  This sample must be taken in such a way that it can explain or describe as accurately

as possible important aspects of demography of a population such as fertility mortality and migration together with the main factors that influence them.  A second important aim of population surveys is to explain some aspects of these demographic events by correlating between the variables measured by the survey (Wilson, 1985: 217).

Most researchers resort to sample surveys because the census covering the whole country is very expensive.  But at the same time resorting to sample surveys exposes the researcher to sampling errors that calls for a carefully prepared sample design.  Using designs of sample surveys it is possible to identify two main types of surveys:

2.5.1    There is the single -round survey, which normally involves only one interview with respondents.  More often this type of design permits the use of large samples and it is more convenient in countries where

addresses are not well defined or where the population is highly mobile.  Actually it includes surveys like those of the World Fertility Survey that covered more than 60 countries or the on-going surveys of Demographic and Health Surveys that so far have been carries out in several countries.

2.5.2    The multi-round surveys, which include a variety of designs such as

(i) Surveys, which re-interview the same respondent several times.  These are commonly known as follow-up or panel surveys.  A few examples of these surveys include the United Kingdom’s, National Survey of Health and Development and United States National Fertility Survey in 1975.

(ii) Some surveys re-interview the same or similar respondents so that they can produce indications of trends overtime.  The United Kingdom’s General Household Survey, Tanzanian Household Budget Survey, the Labour Force Survey and the United State’s current Population Surveys use a rotating sample of households that act as representative sample of the total population.

In fact some multi-round surveys use geographical areas of residents at each round, regardless of whether the people were resident in the unit during the previous round (Wilson, 1985: 217; Newell, 1988: 19-20).  Sometimes multi-round designs are known as prospective surveys especially when they trace the experience of respondents over two or more rounds.

This may lead to a bit of confusion in distinguishing this type of survey design to those of retrospective design.  However, it is important to note that nearly all surveys whether single-round or multi-round, collect information for the immediate pre-survey period and so in one way or another they are retrospective (Wilson, 1985).

Since the survey questions only a small fraction of the total population, the problem of relating the survey findings to the larger population is a complex one.  Therefore, the success of any survey will depend very much on a good sample design.

Currently, designs have turned out to be a very important subject in statistics, though the application of its rules differs from one researcher to another, among these, the most important is the sampling technique.  As a matter of fact, sampling is the strategy of collecting data from a part of the population with respect of drawing inferences about the whole.

Generally, before any sampling is done it is important to draw a sampling frame, which can be divided into strata.  The sampling in each stratum such as rural and urban can be carried out independently in each stratum.

Usually in unstratified sampling, the overall-sampling fraction is specified, but within any particular division of the population the proportion selected may actually differ just by chance from this fraction.  However, in stratified sampling this source of chance variation is controlled and each stratum receives the exact sample planned.

This has the advantage of reducing the sampling error by an amount, which depends on the integral homogeneity of the strata. Such variations may be introduced to increase the efficiency of sampling by taking into account the operational costs and population variability.  Also stratification is used to allow over sampling of small but significant sub-populations.

This is done in order to ensure that there is an adequate representation within the planned total sample.  In fact this reduces the sample error for sub-groups in question, while increasing it for the population as a whole.  Sometimes the sampling error of this type can be reduced by using appropriate weighting estimations.  Nevertheless, despite the division of the survey into strata, the final choice of the sample can be picked at random (Wilson, 1985, 217).

In fact, sample surveys have turned out to be essential sources of data because as stated before censuses and vital registrations are very expensive.  In most developing countries it is impossible to include many questions in a census because these will make the census almost unmanageable.

Above all sample surveys can be organized and executed relatively quickly and they can also collect more detailed information than the census.  It also allows to include attitudinal information or in-depth studies.  Besides this the few interviewers can be trained (Newell, 1988: 20).


It is believed that the tradition of registration began several centuries ago in China.  However, the earliest registration data was obtained from Parish Registers of pre-industrial Europe.  History reveals that as early as 1935 every priest in England was required to make weekly records of baptism, marriages and burials even though it was some decades before the registration system was established throughout the country. Even though, the data was not arranged systematically to produce statistics some demographic analysis could be done.

It is possible the first compulsory civil registration system started in Scandinavia in the seventeenth century and spread to Europe and North America only in the early nineteenth century.  Indeed, nowadays the registration system in

developed countries is more established and gives more elaborate data on many aspects of social change, even though the relatively expensive and complex infrastructure militates against its effective use in some poor countries (Shryock & Siegel, 1976: 20; Newell 1988: 18).

The history above shows that a vital event can be defined as a major change in individual status that leads to a change in the composition of the population.  Typical examples of vital events include births and deaths.The term now also embraces other events such as marriage, adoption, annulments, legitimization, separation, divorce and migration,

which in one sense they are not strictly vital.  Therefore, vital registration is a system, which is devised to collect data on some or all of these events.  Since this system of collecting data requires a well-established recording and reporting network, it is found in very few centres such as in urban areas.  In Tanzania some experiments were made to develop this system in Kilimanjaro and Morogoro Regions and in other regions villages were encouraged to collect some of the vital events.

Otherwise for many areas only some vital registration takes place. Nonetheless, the data collected in vital registration system have been a crucial source for demographic analysis especially in providing information on the dynamics of population.  Since they are collected on a daily basis and totaled on monthly or annual basis, they are very important in analyzing short-term population changes.

As stated before, the origins of vital registrations hinge wholly on the legal basis for they  are required, for example, to validate births, marriages and deaths of individuals.  Although these basic requirements still stand, it is also required for quantitative information for planning purposes.


Please enter your comment!
Please enter your name here