Y
UNITED NATIONS
ECONOMIC AND SOCIAL COUNCIL
Distr.
LIMITED
E/CN.14/K)P/72
17 October 1972 Original: ENGLISH
ECONOMIC COMMISSION FOR AFRICA Working Group on Fertility Levels and Differentials in Africa,and the Prospects for the Future
Addis Ababa, Ethiopia, 18-22 December 1972
INDICATIONS OF CUREENT AND FUTURE METHOJOLOGICAL TRENDS- IN THE COLLECTION OF DATA CN FERTILITY LEVELS AND
DIFFERENTIALS IN AFRICA
H72-1971
E/Cir.14/K)P/T2
INDICATIONS OF CURRENT AND FUTURE METHODOLOGICAL TRENDS IF THE COLLECTTON"
OF DATA ON FERTILITY LEVELS AND DIFFERENTIALS IN AFRICA
Introduction
1. The comparatively unsatisfactory progress in regard to the
acquisition of adequate and reliable vital statistics in the African region in the past is now very well known and hardly requires further documentation. All the same, the situation, as at present, still calls for a critical examination of the potentials for improvement now and in the future. In doing so in this paper, the focus will be on the prospects for the study of fertility and fertility differentials in the continent as shown "by an analysis of recently completed or current or proposed data collection programmes for some known dates between 1965 and 1975.
2. The analysis contemplated has both scientific as well as practical
significance. Firstly, with a view to aiding better comprehension of fertility levels and differentials, there is and always will be the need to identify the available sources of data together with their
potentials and limitations for programming and indicating future
developments. Secondly, and on a oragmatic level,, there is the need to highlight the relevance of a clear understanding of fertility levels and differentials, in relation to population dynamics, for economic, and social development on national, regional and sub-regional levels.
Even granting the limitations of available statistics, the fact that significant variations in fertility axist not only between but within countries and in an inprecise direction calls for further delineation of current and future trends. There is little doubt that these
scientific and practical considerations, at least in part, reflect the working ideas of the programmers cf the World Fertility Survey,.
and of the African Census Programme which has the overall objective
of assisting in collecting better nation-wide demographic data, including those on fertility, in the African countries which have
indioated interest in the Programme.
Description of Identified Progra_mmes_of ..Fertility, Data Collection 3. In vie* of the above objectives, an inventory of 73 different programmes spanning the period 1965-75 has been prepared.
The programmes include sample surveys and censuses which have either been completed or are being currently executed or are planned for execution in the future. The basic source material for the tables
presented derives from data compiled by the ECA Population ProgrammeCentre and published in one of its Working Paper seriesj/ The.
inventories prepared by the Population Council and.the U.S. Bureau l/ ECA/PPC,. "Fertility Data in Surveys and Censuses, Africa,
1965-75,"' ECA/PPC Working Paper No. 1972/3.
E/CH.14/POP/72
Page 2
of the Census for the World Fertility Survey were alsn employed for
this purpose. 1/
4. The programmes under study were broadly distributed among regions of the continent, with West, Central and East Africa having the lion's
share in this order of magnitude: West 34 programmes (47 per cent), Central 10 (16 per cent) and East 17 (23 per cent). Only 8 i.e. 11
per cent of the programmes were based in North Africa a£ against 2
(3 per cent) in other African countries of the South. It should be
appreciated that the above distribution may not necessarily embrace all current, completed or proposed programmes in all the countries covered in this study. Actually, it will be presumptuous not to
accept that access may not have ~been had to all the available sources.
In particular,' the study could have missed-out on a number of import ant surveys executed by various agencies of the Government and by different educational institutions. Nevertheless, after observing a reasonable level of comparability between the present collection and other recent inventories referred to above, it has been assumed that the margin of error may be slight,
5* Of the identified programmes, the great majority were sponsored by the various arms of governments in the African countries concerned.
Exclusively, 24 out of 73 programmes were organized or planned by the government agencies, who also shared or plan to share the admini
stration of 25 other programmes, mainly national censuses (l9)» with
UN aud non-UH agencies. As can be seen from table 1, institutions of higher learning have or will contribute significantly to the execution of the programmes, coming next in order of importance to government agencies. Also evident from table 1 is the significant
contribution of international (Non-UN) voluntary agencies which, in
retrospect, have for so long provided African governments and academic institutions with technical assistance and fund for executing many projects, and in some cases executed the programmes alone. As would be expected, all the census programmes have been or will be, either
separately or jointly, operated by the governments and the UN and its agencies.
6. It has been possible to classify the 73 projects into three very broad types of data collection programmes. From table 1, almost
one-third of the "programmes constitute a one-shot census with fertility data included. Sample survey accompanying or following a census made up slightly over one-fourteenth, while exclusively and separately designed sample surveys account for three-fifths of the entire programmes. The majority of the surveys in the last-named
"~\J World Fertility Survey, "An Inventory of Fertility (and Related) Surveys Conducted or Planned since 1968." Part II, First Ad Hoc Technical Advisory Meeting, The Hague, July 11-14, 1972.
The Population Council, Selected Questionnaires on Knowledge, Attitudes and Practice of Family Planning, Vol. I, Demographic
Div., The Population Council, Inc. New York, August 1967.
Table1.Sponsorsof73CompletedorCurrentorProposedpataCollectionProgrammeson
Fertility (Surveys and Censuses) in Africa for some Known Dates (1965-75)
TypeofDataCollectionProgrammes SponsoringOrganizationsOne-shot censuswith Fertilitydata included SampleSurvey accomp&cyingor followingcensusExclusively SampleSurvey AfricanGovernmentOffices, UNalidiHisAgencies(liainly
;\frecall iGensus Programme) 19
;oemographic,K.A.f. Fertilityetg.mainly.
Source: ECA/PPC Working Paper No. 1972/3.
TOTAL AfricanGovernmentOffices only;v' UNafiditsAgenciesonly
-"on-UN Agencies (population Council,; etc.) only
Universi^ies,Institutes, etc.;only ■.I"4 1 0 0
4 0 0 0
16 1 2 2
0 0 3
5
24(3$) 5(1%) 7(1($)
AfricanGovernmentOffices andNpn-UNAgenciesL Universities,InstitutesandAll Agencies (HIT and Non-UN)
Universities,Institutesand African-Governments TOTAL0 0 0
■ 24(33$)
; *
0 05(7$)
1
5 ■
132(4436)
0
4.
012(1<#)
«: ;9(l2$) 73(100^;)
h3 p <5o
E/CN.14/POP/72
Page 4
category, 32 out: of 44, either embrace those which collect demographic,
social and economic data with a marked coverage of fertility.items
or thpse purposely designed to assemble data on fertility and fertility behaviour, also along with Mother socio-economic variables. In spite
of the fact that: most surveys simultaneously collect information onpast fertility performance as well as on family planning, some of the surveys under review, \l in ally have.been classed as KAP studies, ; mainly intended to collect information on knowledge of, attitudes to ward, and practice of family planning, These KAP surveys, like . most other surveys, are multi-purpose oriented and, therefore, include
the collection of fertility and other-■ items, :
7. The available data on the programmes also permit, the study of the various approaches followed or to be followed in carrying out the ! projects. Three major approaches, shown in table 2, have been singled- out. The commonest one is the "single-round (visit) retrospective": ; which may sometimes collect information on current events- All the census enquiries as well as a large number of sample surveys.employ , or will employ this approach. The preponderance of the use of singlef round retrospective approach tallies with the views expressed at the . First Ad Hoc Technical .Advisory .Meeting of the World ;Feri;ility Survey held at the Hague during 11-14 July 1972. The Committee not only noted the very common use made of the approach, but also observed ■ that the World Fertility Survey has decided in favour of using the
approach in its work, -i/ The Committee did, however, express anxiety
that single-round surveys may be subject to significant memory and implementation biases which may affect estimates of Fertility levels and differentials-2/ qt^e multi-round (several visits) has and will ■, be less frequently used. More often than not, academic institutions have tended or plan to employ the multi-round method either al^ne or in collaboration with other agencies.; The third and the least common is the dual-system approach or the PGE technique which involves the use of retrospective or current survey recording as well as registration
of vital events for obtaining more reliable^estimates of rates of popu lation growth by matching of events and use of the Chandrasekaran - ,
Deming technique. .3/ ■ :
l/ World Fertility Survey, Ibid, p:. 5- . : Also ISI, Report of the First Ad Hoc Technical Advisory Meeting on the World Fertility Survey-, The Hague, 11-14 July 1972, :
p. 16, para, 59.: ;
2/ ISI, Report of the Second-Ad Hoc Technical Advisory Meeting on the World Fertility Survey, the Hague,; 18-21 July 1972,
s p. 9) para. 46. ■ , :
3/ C...Chandrasekaran and W.E. Deming, "A Method of Estimating Birth and Death rates and the Extent of Registration," Journal
of American Statistical Association, Vol.; 44* March 1949? PP- IOI-II5. E.S. Marks, W. Seltzer and K.J.rKrotfci, Population
; Growth Estimate; A Handbook of Vital Statistics Measurements
Population Council, New York, March 1972.; ...\ = - j
Table2.TypeofApproachbySponsoringOrganizationsof73CompletedorCurrentor
Proposed Data Collection Programmes on Fertility (Surveys and Censuses) in Africa for some known Dates. (1965-75)
TypeofApproach SponsoringOrganizationsSingle-roundKulti-round RetrospectiveRetrospective/Current/ /Current/
DualSystemApproach UnknownTotal AfricanGovernmentOffices etc,only UNanditsAgenciesonly
Non-UN Agencies {Population Council, etc.) only African Government Offices/ UN & its Agencies (llainly African Census Programme) African Government Offices/
Non-UNAgenciesUniversity/Institute etc. University/lnstitiite and all Agencies (UN and Non-UN) - . University/institute and : :
AfricanGovernment^Offices'■TOTAL ';; . ■" V ;* ."-■ '.,
12 1 4 22 1 0 6 0".-
46(63#)
6 1 0 0 0
5
2 115(20%;)
3 0 1 2 0 1
o"
07(10%);
3, 0 0 •0 0 1 1 0
5(7£)
24. ■2--
■5 ' ■ i-\ 1 , ■9'i"
173(100%)
13 0 roSource:Idem*
E/CN.14/POF/-72
Page 6
Table 3. Sample Size and Areal Coverage of 73 Completed or Current or
' Proposed Data Collection Programmes on Fertility (Surveys and
": Censuses) in Africa for some known dates (• 1965 - 75_A
i Sample Size ;
(<p ersons)
; ; i Under 2,000 ;
' / 2,000- 5,000
| ' ' 5,000-20,000
; ■ 2to, ooo v 50,000
50,000-100,000 ■| 100,000-300,000 i
300,000-= or more ; - Census Enumeration.
:- No Information
< ' ; TOTAL
Nation wide Surveys
> 0 0
4
4 17 : 5
—
4
Areal
Regional/
Localized ■ Surveys
3 ■ '
; 5 . ■ 5 . ■
l ■ 0 l 0 o 6
25(34/0 ] 21(2995)
Coverage Census Not Enumer- Ascer- ation tained 0 . 0 X>
1 0 0 0
24 o
2
24(33$) 3(456)
Total .
< .
3 1
9
6 1 85 .
24 12 ;73(100/,)
<
Note: " *• " means :not applicable Source: Idem ■ '. ':
8, , The -discussion of the sample size and areal coverage of the ■ . prdgrammss will of necessity be confined to sample surveys. Since goyernmepts and ti^eir agencies are connected with the programmes, it
■will be expected that a national out-look will colour the planning of
the programmes. Thus 67 per cent of the entire programmes, including
all census enumerations-, have or will' attempt a nation-wide coverage.
Considering the ascertained 46 sample surveys (Table 3)j over half (25)
also have or will attempt a nation-wide coverage. .The number of regional
and localized surveys, even though less (21) is also appreciabily high.
Generally, the size of the samples indicated for the respective survey programmes falls on the large side, mainly in keeping with the
E/CK.I4/POP/72
Page 7
recommendations of the Technical Advisory Committee of the World Fertility Survey.i/ As can be sesn from table 3, the nation-wide survey programmes, by their very nature, have larger samples than regional and localized ones*
In additions table 4 indicates that government sponsorship of surveys, either alone or in conjunction with other agencies, is associated with tb.e use of comparatively larger samples. Academic institutions, especially if operating aL.ne, use or plan to use relatively small-^ samples. The
international voluntary organizations (Non-Uff) fall in the same category as the academic institutions with interest in small-scale experimentally-
oriented projects.:
9« Some projeot organizers have considered the relative merits of ^j enumerating from a sample of areas over that, from a sample of households.
Not all sample designs were available for study, but of those accessible, the majority favour area sampling or some fdrm of clustering on various technical and administrative grounds..3/
\] ISI, Second Report, Ibid, p.11s para. 57 recommends that 5000 women in the reproductive ages may be considered as the approximate lower bound for a sample size3 if the main objective of the individual aui-veya is to prepare estimates of fertility differentials at the national level.
This, the committee added,implies a national sample size of about 20,000
to 30,000 persons.,
2/ For a discussion of -the merits, see EGA.Manual, on Demographic Sample.
Surveys in Africa, E/CN.14/CASB7/i7, 17 August 1971, PP.95-96.
- Also "Use of Sampling in Population and Housing Censuses" (E/CN0I4/
SM/4; E/CN.14/CPH/8), para 51
- Also "The collection of Data on Fertility and Mortality in African Censuses of Population", E/CN.h/CPH/14, para. 6
- Also ISI Second Report, Ibid, p.10, paras. 51 and 55 recommends that in a majority oF^ases, probability sampling could only be accomplished by ■ using area samples and that sub-sampling of households within clusters
should be avoided because it is a generally complicated field procedure
than enumeration of complete clusters,,
3/ On clustering or area sampling see, for example, Direction de la Statlstiqu et de la Comptabilite Rationale, La Population de Yaounde en 19<>9v PP»
9-10, June 1970; EGA, Enqueue Den >graphique au Sen gal_ 1970-71« Rapport de mission de Rajakovio Zdenko, Gonseiiler regional (22 Ncvembre - 10 Deoembre 1969), pp, 7-9, paras. 30-35? Republic of Liberia, Bulletin of the Population Growth Survey, Selected Demographic Indices 1^711
Publication Number LPGS - q" 1972, p.vr Republique Algerienne
Deraooratique et Populaire, Etude Statistique Nationale da Population
(Premiers Resultats Provisoxresj,Septembre 1971> P» 9; Imperial
Ethiopian ISovernment, Population of Addis Ababa.) Statistical Bulletin o,
Addis Ababa, July 1972~, p-4*On the us© of a 3aople of households, see Sidiki Philippe Coulibaly^et D. Ian Pool, "Queique Differentiollas de la Fecondite en Haut-Volta", Conference Afoicaine sur la Population, Accra. Ghana^ 9-18 December 1971>
POP. Conf* 3/b7~PP= V-T*
. T-K. Eumekpox-- Rural VJomen and Attitudes to Family Planning, Contraceptive Praotice and Abo1FETon~nr Southern 'l'o«o» A Report or a fij-Q.t Purvey in tne Ihritime "SiHonToT" Republic of Togo", Mimeograph copy, University of Ghana.
Aoora. August iyib, pp, z-}*
PtO* Ohadike, "Urbanization- Migration and Migrants in. Zambia: A. Survey or Patterns. Variations and'Change in Lusaka," Proceedings of the EL even tL International African Seminar on Town and Country in East and- Central Africa, Lusaka, September 1972, pp« 34
Table 4. Sponsoring Organizations and Sample Size of 73 Completed or Current or Proposed Data Collection Programmes on Fertility (Surveys and Censuses) in Africa for some known dates (1965-75)
Sponsoring Organizations AfricanGovernment Officesetc.only UNanditsAgenciesNon-UN Agencies (Pop. Council etc.) only
AfricanGovernmentOffices/UN Agencies
AfricanGovernmentOffices/Non-UN Agencies University/lnst itut e
etc.onlyUniversity/lnstitut e and all Agencies (UN and Non-UN) University/institute
andAfricanGovernment Offices TOTALUnder 2,000 0 0 1 0 0 0 2 0 3
2.000-- 5,000 0 1 1 0 0 2 1 0
5
5,000- 20,000 3 0 0 0 1. 2 2 1
9
Sample 20,000- 50,000 4 0 2 0 0 0 0 0 6
50,000- 100,000 0 0 0 1 0 0 0 0 1
Size 100,000- 300,000 2 0 0. 3 0 1 2 0 8
300,000- or more
5
0 0 0 0 0 0 05
Census Enumer ation 4 1 0
19
0 0 0 0 24No inform ation 6 0 1 1 0 2 2 0 12
Total -24 2
5
24 "1 79
1 73o CO
3
3 ro Source:Idem. is*-f , E/CN.14/POP/72
Page 9
The Potentials of Programmes for the Study of Fertility Differentials 10. Judging from "the content of questionnaires designed for the pro grammes, ther* appfterB to exist immense possibilities for the study of fertility differentials. Such possibilities have always existed in the past without being fully explored at the tabulation and analysis stages of survey programmes. In particular, organizers of national surveys have tended to undertake less detailed cross-tabulations of fertility indices with socio-economic variables. They have shown greater interest, judging from published data, in differentials related to geographic units, 'such as administrative districts or urban and^ rural areas or industrial and ' agricultural zones- This has been done much to the neglect of variations by socio-economic and cultural variables.
11. For purposes of discerning and'interpreting objectively the different
types of prevailing levels and patterns of fertility in the continent, greater interest has to be shown in the cross-tabulation of fertilitydata not only with geographic but also socio-economic and cultural factors.
12. Ab already indicated, the projects (current, completed and planned)
under review here contain basic information needed to make this possible.
In the first place, almost all the programmes covered, collected or plan to colleot social and economic charateristics pertaining either to each member of the household or to females and their spouses respond ing to the fertility questions asked. The oross-tabulation of these variables with the various fertility indices would promote the study of fertility differentials.
13« Under the African Census Programme which covers the census projsots included in this analysis, a list of recommended items has been
established by the Consultative Group on the African Census Programme at
its first meeting in Addis Ababa in January 1971 -J-/ The minimum list*
as well as asking current and retrospective questions on fertility also, includes questions on age, sex, and migration status. The list of
additional items divided into two priority groups also totally concern the basic socio-economic variables of interest in the study of fertility and other demographic differentials. For our purposes, a lot depends on which recommended variables the countries concerred choose to colleot. The countries may even decide to include other items of special interest and delete some from the recommended list, although it would, for purposes of enhancing comparability, be desirable to have a standard questionnaire.
14* ..The types of fertility questions asked or proposed, and which" could be examined against the variables deal with the parity performance and or current fertility of the respondents. Questions on parity seek
\] ECA/PPC, African Census Programme Newsletter, No, 6, August 1972,
pp. 4-5» See appendix A for a complete list of items recommended.e/CN.14/K)P/72
Page 10
to record the total number of live births, while those on current
fertility attempt to reoord. the number of births .during a perio/d . ■ . ■ pripr to the survey, say, six or twelve months. Where the programmes relate to surveys, some of the questions also delve in some depth into the past fertility performance of respondents through collecting pregnancy or birth histories. In addition, some glimpses into future - fertility developments are sometimes given by questions on prospective fertility desires covering ideal and expected family size and also attitudes toward family planning*
15* In conclusion, it seems apparent that the projects examined here could, yield meaningful results for the study of fertility
differentials on a national as well as on an international basis.
To this end, increased and concerted effort towards obtaining more detailed and meaningful tabulations of all geographic, social,
eoonomic and cultural factors with current and retrospective fertility data, where possible, should be made by all organizations which
sponsor and promote the execution of surveys and censuses.' An ideal goal for all such organizations is one geared to producing nationally reliable and internationally comparable information through the
evolution and adoption of relatively standard definitions, terminologies and tabulation procedures where national circumstances and conditions permit. Recapitulating some recommendations of the First Ad Hoc Technical Advisory Meeting on the World Fertility Survey, "The focus should be on a common oore of fertility questions and the sample' should be large enough to provide reliable data. In addition to the
"core" questions, optional additional areas of investigation should
be suggested"!/
16. In the interest of enriching knowledge on several methodological matters, use should be made of several "ancillary" questions for improving acouraoy of information and the reliability of data obtained
.*." To this end, "a review of methodological research undertaken and of outstanding methodological problems"2/ should be undertaken.
17* Actually, whether programmes are current or contemplated, there is the need for undertaking additional methodological research in collecting and processing survey data. It is »ot possible to indioate here all the areas needing attention. However, some of the
fundamental and often neglected dimensions of the problem generally
JJISI, Report of the First Technical Advisory Meeting, Ibid, p. 16,
para. 59
2/ Ibid, p.16 para. 60.
11
underline the need for greater depth analysis and interpretation of survey results. Thus- as a result of ths high degree of statistical variability of the indices used in studying fertility differentials, especially where regional comparisons are required, there is the
need to encourage the calculation of standardized or adjusted fertility rates. Pew studies have done so in the past, in spite of the rich analytical dividends involved in doing soc
18*, There are also other missing gaps3such as failure to indicate levels of sampling and especially non-sampling errors, and also of variance estimates. In particular, non-sampling error3 arising mainly from faulty response and recording of fertility and related, data have often been rated to be more important than sampling errors^ and tc / have vitiated survey estimates of fertility and ether vital rates.^
As suggested "by the Second Technical Commii.tce of the World Fertility Survey, especially in connection with the use of single-round surveys, particular attention should be paid to the assessment of the impact of specific biases on the validity of results? 2/ vhile specific
provisions should be made for subsequent calculation of variance estimates, possibly through the use of simple replication methods,!/
II H«K« Som» Recall Lapse in Demographic Enquiries (Bombay, Asia Publishing House), (in press); Robert Nadot, Afriqu.e Hoir^, Madagascar, Comores, Demographie Compe-jge* 3. Feoondites Niveau
(Institut National de la Statistique et des Etudes Eoonomiques, Paris, 1966), Annexe V, "Application de la methode Som an Cas de
Haute Volta et du Tchad"o
2/ ISI, Report of the Second Ad Hoc Technical Advisory Meeting qh
World Fertility Survey Ibid, p.9p para, 46. "
3/ Ibid, po 12, para. 62
e/cn.u/pop/72
Appendix
Appendix A
ITEMS RECOMMENDED FOR INCLUSION IN THE
AFRICAN CENSUS PROGRAMMEThe minimum list of items recommended for inclusion in the
census programme is as follows:
1• Name
2. Place where found at the time of enumeration 3» Place of birth
4. Sex 5. Age
6. Children "born alive 7» Children living
8, Live "births in the past 12 months ("by sex) 9» Deaths in the past 12 months (by sex and age)
The additional items were listed in two priority groups as
follows:
First Priority:
1. Type of (economic) activity
2. Occupation
3- Educational attainment
4* Marital statusSecond Priority:
1. Relationship to head of household 2* Ethnic group (or citizenship)
3- Literacy
4» School attendance 5- Industry
6. Status (employer, employee, etc.)
7- Usual place of residence8, Duration of residence
9- Previous place of residenoe 10. Religion
11. Number of wives