UNITED NATIONS
ECONOMIC AND SOCIAL COUNCIL
Distr.
LIMITED
E/CN. 14/CAS. 10/18
16 May 1977Original: ENGLISH
ECONOMIC COMMISSION FOR AFRICA"
Tpnth session of thp Conferpnce of African Statisticians Addis Ababa, 17-22 October 1977
AFRICAN HOUSEHOLD SURVEY CAPABILITY PROGRAMME Some aspects of household survey methodology
CONTENTS
Paragraphs Pages
Introduction „<.„.„., 1-10 1- 2
Repeated-visit economic survey 11-12 2 - 3
Background 11-12 2-3
Sample design ... 13-30 3 - 9
Survey pro cedures ... 31 — 47 10 - 15l
Questionnaires 16-37
Working sheets ... D o - „ 38-42
Single-visit economic survey 48 - 106 43 - 52
Survey design and content 48-50 43
Notes on questionnaire 51-53 43
Questionnaire „.»...,. 53 — 61
Labour force survey .... 107 - 113 63 - 65 Survey design and content 107 - 113 63 - 65
Questionnaire • •.. 67 - 70
INTRODUCTION
lo The ninth session of the Conference of African Statisticians indicated that it would welcome the circulation of specimen survey questionnaires, Such documents can often make a useful contribution tc the design of specific national inquiries and they help to avoid duplication of effort in developing methodology appropriate to the African region.
2- Resources available in the "3CA secretariat are at present too limited for any comprehensive review of African survey designs and questionnaires.
It is also doubtful whether such an exercise .would Toe profitable because many of the previous surveys were of an ad hoc nature and not consistent with the current aim of establishing continuing and integrated survey programmes within the context of national statistical systems-
3« The secretariat therefore proposes to respond to the wish of the Conference by distributing information on survey designs in 'which ECA is directly involved, and which can be regarded as components of ongoing pro grammes c In general1, ths material will relate to methodology already applied
or under active consideration by countries of the regions plus tentative iedas on new survey arrangements which may meet special conditions or data requirements. In the initial stages there will be no attempt to produce anything approaching a polished surveys manual<» It seems more important to circulate relevant information as it becomes available and the present paper is a first attempt to present details of economic surveys.
4* To some extent the population field has already been covered by the
ECA/UNESCO Manual on Demographic Sample Surveys for Africa which comprises a
technical and practical body of operational survey detail. However it was written before the need for integrated survey programmes became apparent and some of its proposals may need revision with respect to the programming of survey activitiess use of core questions, etc.
5. The secretariat would like to suggest that ECA should not be the only
agency which distributes information on survey methodology in the African region. One of the essential concepts of the African Household Survey Capability Progr-tmme is that surveys will develop more q/\ickly if countries help cne another in technical matters. This can be achieved cults easily through bilateral exchange of information and the more general circulation of relevant documents» However such arrangements are not working well at the moment and a special effort will be needed on the part of survey statisticiansc It also has to be borne in mind that most surveys can bo organized in a number of different ways and the approaches of individual bxpsrtb arc j.nflusnoed by thei^ training and rracti'sO oxp?7iL-i?r:e!< I?CA is therefore anxious that the views of its own staff should not be the only ones circulated in the region* Many national statisticians hava equally interesting experience and it is important that they should contribute to the overall process of methodological development.
E/CN.14/CAS.10/13
Page 26. In the present paper information is given on two kinds of household economic survey. The first is a repeated-visit arrangement of the type favoured by African countries because it provides not only structural data but also details of expenditure patterns, seasonal changes, etc. The second is an untested survey design drafted "by the secretariat in the hope that it may be of use to countries which need basic household economic data but lack the resources for repeated visits to a sample of reasonable size. It con centrates on the fundamental details which any co-operative household can make available during a single lengthy visit and then goes on to use these
details-in constructing a broad but comprehensive set of household accounts.
It must be emphasised that the second survey arrangement is strictly
experimental and some of its procedures may prove to be over-enthusiastic.
7. The chcioe of the two survey designs in this paper results from the priority which African countries are giving to household economic data. In
this connection it is interesting to note that there are some differences between the African approach and that in other parts of the world.
8. Household surveys in industrialized countries are labour force oriented and also to some extent have to keep track of changes in patterns of con sumption expenditure- They provide essential data for estimating the level of employment and cost of living which are the two main considerations under present conditions. They can also be relatively modest in scale because a great deal of other information on household conditions and activities is available from administrative sources-
9. Information available at ECA on survey activities in other developing regions is at present rather limited. Latin America appears to be con
centrating on labour force and basic income data along the lines proposed in
■the "Atlantida" study prepared ~by "the US Bureau of the Census- In the Far East the Indian National Sample Survey has been developing independently
during the last quarter of a century as a rather comprehensive data collection instrument but notably also excludes the current African emphasis on details of consumption expenditure.
10. In the future there will be a need for Africa to keep its data collection priorities under review, mainly because lengthy budget surveys are expensive.
The single-visit method suggested here probably goes too far in the effort to reduce costs but it should be noted that it is still a more intensive inquiry
than those proposed for household income and expenditure surveys in the Atlantida study-.
REPEATED-VISIT ECONOMIC SURVEY
Background
11. The survey design described in this section is being considered for
application in the Sudan and is based on previous experience in Ghana, Kenya
and Lesotho. It covers both urban and rural areas and aims to obtain a com prehensive record of household transactions during a period of one year.E/GN.14/CAS.10/18
Page 3
12. Although the presentation here has "been generalised as far as possible, it still reflects its origin in the documentation relating to a specific country. This perhaps points to one important conclusion: survey designs are generated from a body of internationally accepted concepts, definitions, classifications and basic probability theory, but each survey must neverthe less be tailor-made to meet particular conditions and requirements.
Sample design Rural sample using census frame
13«■ It is usual to base the sampling arrangements for household surveys on the geographical frame available from a previous population census. Four possible methods for use in rural areas are examined in the following notes.
They are all self-weighting and incorporate an income stratification of the ultimate-stage household sample. A self-weighting sample makes it somewhat easier to process the rather detailed records from an income-expenditure survey and the sample is more efficient if larger proportions of higher- income households can be selected. Figures relating to the Sudan have been retained for purposes of illustration.
14* Basic data
Supervision Areas (SA) In whole country
^numeration Areas (EA) In whole country.
Average number per SA Households
Average number per SA Average number per EA
2 18
2,049
,462 ,800,142
832 108
7.636068
.308 -997
Other abbreviations: PP3 Probability proportional to size.
EP Equal probability.
S/F Sampling fraction.
15« PPS selection at first and second stages. 50 SAs, two EAs in each selected SA and 2,400 households at the third stage.
Selection chances in unsr.ratified sample*
Any SA (PPS): 50h /2,049-142 (where h is number of households
s .-,, \ sin SA)
Any EA in this SA (PPS): 2h /h (where h is number of households e S in EA) e
Any one of 24 households in this EA: 24/h
e ' ■
Overall selection chance for any rural
household: 50 x 2 x 24/2,049*142 = 2,400/2,049*142.
Suppose the PPS sample gives 12,000 households in the 100 EAs selected.
Then the sampling fraction in each EA will be: 2,400/12,000 x 120/h , e.g.
in an EA of 275 households the number to be selected is 0*2 x 0.436 x 275*** 24*
In this form there is a fixed third-stage sampling fraction applicable to the entire rural sample plus a deflator which deals with the size variation of individual EAs* The arrangement is used below in introducing an income stratification of households. * "■- :
Three income strata are needed along the lines of the following illustration:
Total households . Number to be selected Overall sampling
btratum ±n selected eas i/ . at third stage fraction
1 2,000 800 0.400 ■-
2 4,000 800 00200
3 6,000 800 O^ljj
12,000 2,400 0«200
Examples:
Large EA with average income distribution
vj. j. tt v -> ^ Sampling. To be , Stratum Households fraction selectsd
2 200 o'.O4O 8 S/F deflator: 120/600
3 300 0.027 _8 .- <-: " °'2
600 24
Large EA with high proportion of lovr-income households:
„. . T, . .. , Sampling To be S tr atum Househol ds ^ A. -ixj
fraction selected
1 - 0,096
2 100 0.048 5 S/F deflator: 120/500
3 400 0.032 13 - 0*24
' <'■"■■■■■■ 500 -. .. 18 _ . __;. ■
Small zone with large proportion of high-income households:
1 20 0.686 14
2 40 0.343 14 S/F deflator: 120/70
3 10 0,228 _2 - 1.715
70 30
1/ The strata are arranged so that each accounts for an appToximately stare of total
E/OT.14/CAS..10/18
Page 5The number of households to be selected is not affected by variations in EA size "but is affected by variations in the income distribution* The maximum
(48 in this illustration) occurs when the EA has households only in Stratum 1 and the minimum (16) occurs when there are households only in Stratum 3»
In some ways the arrangement may be self-defeating. For example high- income farmers may constitute an EA with few households because their farms are large; their transactions are more complicated to investigate and it is a disadvantage if it is necessary to apply a "deflator" which significantly
increases the workload.
There is a further cause for doubt (probably unjustified) in relation to PPS selection of SAs and EAs. If there is soms unknown variable or
characteristic associated with SA or EA size there is the possibility that a sample concentrating on larger EAs would have a bias which could not be
easily detected-
Use of this arrangement would require revision of the draft Working Sheet 3 (see page 41 )•
16, EP selection in three stages. 50 SAs, 100 EAs and 2,400 households.
Selection chances in unstratified sample:
Any SA, of size ns: 50/2,462
Any BA in a selected SA: 100/£n , the same sampling fraction having been applied to each selected SA,£ n being the total number of EAs in the
sample.
Any household in a selected EA: 2,400/12,000, the same sampling fraction being applied to each selected SA, 12,000 being the total number of households
in the sample•
Overall selection chance for any rural household:
50 x 100 x 2,400/2,462 x<n x 12,000.
Assuming the same overall income distribution as in the previous
illustration the strata sampling fractions would be 0.400, 0,200 and 0.133
in every selected EA. Applying these in the three examples given abovo the
numbers of households to be selected would be 120, 73 and 17 • These are of
course extreme cases but they serve to indicate that enumerators* workloads
could vary widely in the event of large differences in EA sizes. A pre
liminary effort would therefore be needed to group smaller EAs and divide
larger ones as far as possible to achieve roughly constant sizes. Similarly the frame of SAs would need modification to ensure that each SA containsapproximately the same number of EAs. This would be a formidable task
involving the entire rural part of the country and would appear to rule out
the use of two area stages selected on an EP basis.^7« EP selection :in two stages. 100 £*As and 2,400 Loueo
Hers -the chance of selecting any FA would "he 100/7.8* 800 and the chance of selecting any household,in-a -selected EA £f4OO/l23OOQ? giving an overall
selection chance for .any. rural: household of 2^400/2,256^000. Strata sampling
fractions would be. the came: in; <=■!! s«lfcufce& FA ticThe method provides a inore efficient sample at the expanse of an increase in transport problems« As beX'ors it would be high?.?.'" desirable to reduce the si^.e yar?nt?.^r of EAo before making the firci; yi;age selection*.
1&° ??5 selection in two rtagesP . 100 Bus and 2?4OO households*
If it were necessary to accommodate .p. oi^vJfloant FiA size variation in a two-stag3 sample the first stage could .}>» sGl?oted or. a PPS "^aslaa Then
the chanoss of selecting any EA. would bo 100h^/2;049^142 and the chance of choosing any one of 24 households ir. any selected . cue would be 24/h , giving the overall selection ohance of 2,400/2,049,142 for znj individual household.
Strata sampling fractions for individual gor.es would then.be deflated as indicated in the illustration.
19 o Conclusion
If transport problems can be overcome and ii1 the frame of EAc can be modified to avoid a largo size variation i'i.- wcvC.d be desirable to use n
two-stage EF s&mpleo It it is not possible to reduce the variation in EA siaes £ PPS selection of PAs would, be the so3.ut5.ono
In the event cf serious transport and Qommiui^oaticna difficulties it might be necessary to use a thrGe--r?tage sample with PPS selection at the f irs b two std^os. However this would aot be -iocira^ls beo?usc- it would significantly reduce f?amp3 ing efficiencyc A better alternative world Vie to exclude those parts of the country which arc spaisely populated or economically less impor tanV
The average of 24 households to j^ chcsori in selected BA& is perbaps a little high.. The.number ooulo. be reduced to 18 .l:;' ±t Lva'eiQG the t-ask of
r^tocrn and pr^onsfr-ing s c^f f fji^nxricantly ^aaicx-o Rural sample "based on local
£0. It nay transpire that. -Mho frame fron the les-i. population census io
difficult to use for sampling "because of oae or both of the following reasons:
(1) it was initially J .precise or has not l:een kept systematically .up to date;
(2) it has been„■ at least partially invalidated .by a change in local government
structure.
E/CN.14/CAS. 10/18 Page .7-."
21. If only the first of th9£u difficulties is encountered the solution is straightforward^ the frame has to be improved and Drought up to date, after which it oan be used along the'lines' indicated above. Cn the other
hand, if it is seriously inconsistent with new local government arrangements
and consequently r.lso out of date- aner approach is necessary. This wasthe situation in the Sadan and it is likely to be found in a number of other
African countries.22. A comprehensive national cartographic exercise would require considerable
resources and delay the initiation jf a sorvey programme, fortunately there is a short-cut "by using a direct approach through the local governmentstructure* Again using the Sudan da4^, and with a slightly increased sample
size, "the arrangement is as follows.23« PPS selection at first and second stages. 120 local authorities, one enumeration area "f E1A) in each local authority and 2,880 households
at the third stage.
Selection chances in unstratified sample?
Any local authority (PPS)s 120 h /2,049»142 (where hQ is number of households in council).
Any EA in this local authority (PPS): h /hg (where hQ is number of households in EA). ■
Any one of 24 households in this EA: 24/h •
Overall selection chance for any rural household: 120x24/2,049 >H2
= 2880/2,049? H2.
Suppose the PPS sample gives 14,400 households in the 120 local
authorities selectedc Then the sampling fraction in each local authority
will bes 2s880/l4,400 x 120/h e.g.- in an EA of 275 households the
number to be selected is 0*2 ie0u436 x 275 * 24- Income stratification
at the third stage remains unchanged-
24« The method requires a rather extensive tut unavoidable exercise in compiling a complete list of local government unite and the approximate numbers of households they contain but this is quicker than a national
cartographic operation*
25* A].though some of the data on local authority sizes could be built up from previous census records it is probably "better to envisage an operation in collaboration with the local authorities. This approach is in any case
needed for survey administration and in preparing for the next census. A
simple questionnaire covering the necessary details is given on Page 9
but it should be understood that the returns received from local authorities will need to be reconciled with any available census records.
E/CN.14/CAS. 10/18
Page 826. Local authorities are normally too large for survey preliminary enumera
tion purposes end would lead to households too widely scattered for
convenience during budget recording* It is for this reason that they are divided into Eis of approximately 100 households with one EA being drawn from each local authority, again with probability proportional to size. Villages are not recommended as suitable units for the second area stage because of
considerable size variation It should also be noted i.iat the demarcation of EAs is a further contribution to preparations for the next population census.
27 • Mapping and all other preparatory field work for the survey are confined to the selected local authorities. Other aspects of the sampla design are similar to those in the samples drawn from e census frams described above and tbe arrangements remain self-weighting as before^.
28* Coverage limit=itionc to exclude nomadic and any other special areas should
be decided beforo the selection of primary area units begins. The point to be borne in mind in this respect is that earlier surveys should oovar as much of the total population as possible but there would be no advantage to be derived from extending operations to parts of 3 country where communications and other difficulties are severe.Urban sample
29. The urban sample can be very similar to that for rural areas and details are not discussed here. It will be seen that the rural sample requires the use of 100—120 enumerators i.e. one to each of the selected penultimate—stage units* For the urban areas about another 60 enumerator are needed. If each deals with 24 households; as in the rural areas, budget recording can take place on lr440 households at any given time. It is suggested that the
ultimate-stage sample should be replaced every three months which means that
5*760 households can be investigated in the course of a yoar.
30a In addition to considerations of economy5 the reasons for sample , replacement along these lines are the less significant- seasonal variations in urban area3, greater dependence on employment income which can be
measured in a short time, and the smaller amount of patience which urban people normally have for lengthy inquiries.
E/CN.14/CAS.10/18
Page 9
Questionnaire for collecting data on sizes of Village and Quarter Council
Province
List of Village/Quarter Councils %
People's Rural/Town Co.uncil Location of People's CounoilSerial Number
Name of Village/
Quarter Council. Location Approximate Number
of households Observations
Prepared Date
\J See paragraph 25.
E/CN.14/CA3.10/18
Page 10
Survey procedures ... ■
31. The methods of selecting enumeration areas (liAs) is described above.
In each of the selected EAs it will be necessary to enumerate all households
using two simple questionnaires (Forms 1 and 2 on pages 17 and 18 ).
32, Form 1 is a visitation record which summarizes data according to the dwelling units covered by.the enumeration. It gives minimal information on the numbers of households and persons in each unit and ia also used as a means of recording basic housing information.
33« Form 2 is concerned with the structure of individual households in each dwelling unit and with crude income data which provide a basi3 for
stratification in the ultimate-stage sample of households. This form consists essentially of a normal census-type listing of household members plus estimates of non-agricultural earning by individual members (wages,
.trading, .handicrafts, etc) and the value of agricultural sales of the
household as a whole in the past year. There are additional questions on property owned and accommodation which are intended as supplementary information and the form also includes a summary of household membership.
34» It should be understood that Form 2 is required to provide only a broad distinction between lower, middle and higher income households. However it should be borne in mind that it uses cash income as an indicator of total income and would therefore be less effective in areas with a high proportion
of subsistence.activity5 this form may need some modification after field
testing.35» In both Forms 1 and 2 a distinction is made between-household members
present and those temporarily away. Similar records are maintainedthroughout the survey to take account of changes in effective household size. It should also be noted that in Form 2 it may be more convenient
for enumerators to record monthly wages and other earnings and then multiply the total by 12^ this would require only a minor modification in the form.
Processing of preliminary enumeration and selection of ultimate-stage household sample
36. The procedure for summarizing the preliminary enumeration records and
selecting the households to be included in the detailed survey of income, consumption and expenditure is outlined in the following notess
(l) Sort the Form 1 records into two packs; urban and rural. Arrange
the forms in each pack in a suitable geographical order.
(2; From Form 1 summarise the total numbers of houses, non-agricultural
and farming households and persons present and away. Enter thefigures for each EA in separate versions of Working Sheet 1 (Page 39) for urban and rural areas'and complete the Census Area sub-totals.
Overall totals for urban and rural areas will also be needed.
Page 11
Compare the figures for numbers of persons with those obtained from a previous population census if possible. Also »«e the sample results to give a national population estimate Whioh itlII serve as a further check on the enumeration and also indicate serve as 8 further check on the n
whether the EA sample is performing'in. a satisfactory ^ two versions of Working Sheet 1 for urban and rural areas
■ provide control and othsr fibres for the remainder of the
processing. Form 1 is set aside for more detailed"analysis at a
later stagec .
(3) Place the Form 2 records in separate files for each EA. The urban
and rural files need to be kept separate.
(4) Insect the forms to see roughly the range of declared earnings plus household sales of agricultural produce ./
(2) 7. This must be done separately for urban and rural areas because the income distributions are different- Divide these
two ranges into suitable classes and show the class limits in
Iparatf versions of Working Sheet 2 (Page 40 *«**?*\*°*
■ areas, In deciding the class limits care must be taken ^^
large proportion of lower-income households falling into the lowest class while at the same time retaining a satisfactory division of
the. higher-income households.
(5) Sort the forms in each EA file into income classes (remembering that the classes are different for urban and rural areas). Count the numbers of households in each class in each E4 ar.d_ complete Working Sheet.2 in the same geographical order as Hosfcing^Sheet 1.
From-the totals for the whole survey .calculate the total income in
each class- on the basis of the class mean. .(6) Then decide the income classification to be used for division of the records into three urban and three rural strata, In urban and rural areas separately the"strata should-be arranged so that each accounts for an approximately equal share of'total income, Stra-i limits can be calculated from the data in working Sheet £.
(7) ''Enter the stratum number of each household in the unidentified box at the top of Form 2 and arrange the forms in each EA file in _
thT-ee strata packs. This does not involve a complete-new--sorting .operation because the forms, are already in the order of the
- original income classes. Count the forms and enter- the numbers m Wooing Sheet 3 (Page 41) which also has to be prepared separately
for urban and rural areas. Entries should be in the same _ geographical order as before- These figures provide the basis for
selecting the "budget sample,(8) The numbers of households to be selected should be equally divided
■between the three strata/ Specific suggestions *n the size of the
household samples .in urban and rural areas are included m" the notes
on sample design but they must be reviewed in the light of travel
and other local conditions. In case of doubt it is safer to ctoose
E/CN.14/CAS.10/18
Page 12the smaller numbers proposed? there is no point in stretching enumerators and supervisors beyond reasonable limits. The ratios between the numbers to be investigated and the corresponding ' overall totals in Working Sheet 3 give the three urban and three
rural strata sampling fractions. The sampling fraction deflators have to be calculated for each EA separately and are given by the average numbers of households per EA in the rural and urban samples divided by the number of households in the EA concerned (120/hQ in the last of the illustrations above). These figures are then applied to the appropriate overall strata sampling fractions to get the three deflated sampling fractions required for each EA, enter them in the spaces provided in "Working Sheet 3.
(9) Apply the deflated sampling fractions to the numbers of households in the corresponding strata in each EA and enter the number of households to be selected in forking Sheet 3-
(10) Select the ultimate-stage households according to the numbers ■ shown in Working Sheet 3 by a simple systematic arrangement. The forms for the selected households should be marked with a "B" and placed as a separate group at the beginning of each EA file,
(11) The M files are then returned to the enumerators in the field who contact the "3" households informing them that they have been selected for detailed budget recording. The object is to obtain' a definite assurance of co-operation at this stage because it is difficult to make substitutions once the recording has started.
In the case of a refusal or any other problem making it impracticable to retain a particular household in the sample the enumerator
should select another household from the same stratum with approximately the same income. It is for this reason that the
complete file of Form 2 records has to be returned to him.
(12) The number of substitutions must of course be small. Supervisors should report details of the changes proposed and the changes should be made only after authority has been given by the head
office.
(13) Enumerators then compile lists of the selected households in duplicate on Working Sheet 4- une copy is returned to the head office together with the file containing the preliminary enumeration records. Information on household members, etc. contained in
these records can be processed separately at a later stage.
(14) An additional operation to select sub-samples from the main
household sample would be required if it were decided to maintain daily records of consumption expenditure, etc. for a small
proportion of households. However no such arrangement is proposed -
in the present notes. . "'
k/cn.i.;/cas. :o/i8
Page 13
The arrangements described, above involve only one main and one subsidiary sorting operation on household forms. The households have to be counted in completing Working Sheets 2 and 3 hut the processing does not involve any additions of
the figures contained in Form 2. Some additions are required in compiling Working Sheet 1 from Form 1 but they are of a minor nature,,
All of the processing is carried out at the EA level which means that the forms can remain in their files Bad there is nc problem in handling the rather large quantity of papers involved- The wb/^le process can be completed very quickly provided there is sufficient clerical labour available.
The best way of ensuring an adequate supply of labour ia to arrange for field supervisors to bring the files for their own areas to the head office, carry out the analysis and then
take the files back to the field again when it is completed.
By participating directly in this part of the processing the supervisors gain a 'Very much better understanding of survey objectives and problems and also a clearer idea of the
economic and social conditions of the households they are investigating*
Detailed records of income, consumption and expenditure
37- The survey arrangement described in this note envisages continuous recording of the activities of a fixed sample of rural households over a
complete year and a sample of urban households which changes quarterly.
It would be very unreasonable to compile records over such a long period without providing for frequent summaries of the results as a means of
assessing progress and identifying recording errors by comparing the survey aggregates with other available economic data,.
38- It is understood that the integration of the results of different
survey rounds in an on-going survey programme will be possible through theintroduction of certain core questions at the household level. Form 3 (Page 19)
has been designed for this purpose= The column headings are in most cases self-explanatory» However, the enumerator's manual should explain these in most detail. This core questionnaire at the ultimate-stage household level should be adapted to the prevailing conditions of any country.
39, Form 4 (Pages 20 - 37) has therefor^ "been designed as a monthly return to
be submitted by enumerators. As previously- indicated material is collected through four visixs to the sample household concerned. The form ia essentially a comprehensive record of receipts and payments and its structure iaexplained by the summary section on the first page. It applies to both urban and rural areas but a version of the form with less agricultural detail
might be more convenient for urban use.
i 10/18.
Page..1.4"
40. Commodity items and their units have not "been shown in the specimen Form 4 because they wiir vary from country to country \J* They have to be pre-printed in the spaces which have "been left "blank and are arranged in a way which
entails some repetition, "but permits easy summarisation of data at regional
office level. Examples' of the working sheets required for this purpose are not included in the present paper but they are very simple and follow the arrangement of Form 4. There is no need for coding of items "because they areidentified "by their standardised positions in the field records and the
working sheets-
*t" ■
41. The general arrangement envisaged is.as follows. At the end of each
month field supervisors bring the Form'4 records compiled by their enumerators
to the regional offices. They will have already scrutinized them incollaboration with the enumerators and they will do so again with the staff
of the regional office. The set of columns at the right of Form 4 whichcontain all necessary indicative information and the monthly totals can then be.out from the remainder of the form. They are then summarised manually within'strata straight onto the working.sheets. Weights are then applied to the strata totals which can then be added to provide a single set of monthly
totals for each geographical area, again keeping urban and rural areasseparate. An example of how the weights are calculated is given below.
42. Regional offices then send the monthly totals to the head office for
amalgamation at the national level.
43. This part of the processing indicates seasonal changes and also provides a continuing check on.the results of the survey so that recording and
other faults can1be identified and corrected quickly.
44. After the summaries have been completed each month the detached parts of,Form 4 are placed in files for individual households. At the end of the year the records for each household are summarised to give annual totals for rural households and quarterly ones for'those in urban areas. These
summaries-are'-the basis for the remainder of,the processing which comprises income, and other -classifications of the data. It is the only part of the processing which takes place after the field work has been completed and can
if necessary be computerized.45. In addition to the. material collected in Form 4 other information can
be obtained from each household. For example it may be desirable to compile information on loans outstanding at the beginning and end of the survey and also, .to.measure the household farms, Additional modules of this kind can
normally be introduced without too much difficulty.1/ ECA has commodity lists for a number of African countries and they
will be made available in a separate document.E/CN.H/CAS. 10/18
Page 15Calculation of strata weights
46. The strata weights referred to above are the same for all geographical
areas (urban and rural separate) covered by the survey unless adjustments have
to be made for missing households. Calculation of the weights is illustrated in the following table which again uses figures relating to the last of the rural sample designs discussed earlier.
Stratum
1 2 3 All strata
Total households in selected EAs
2 400
4 800
7 200
14 400
Number selected at ultimate stage
96O 960 96O 2 880
Sampling
fraction Weight
0.
0.
0.
400 200 133
0.
1.
1.
5
0
5
0.200
47* The weights are obtained by dividing the overall ultimate-stage sampling
fraction (0.200) by the individual ultimate-stage strata sampling fractions.
When applied to the survey results for each stratum, the weights produce
figures which would be obtained (less accurately) from an unstratified sample
with the same number of households as that actually used. Data for the three strata can then be added. The advantage of the arrangement is that there is no manual processing problem arising from any large change in the order of magnitude of the figures. When the data for all urban and rural areas have been nraalgamated, they can of course easily be raised to the level of national estimates if necessary^ most of the results are normally more useful and manageable if presented as household averages.
E/CN.14/GAS. 10/18 Page 16
Questionnaires
Pages Form 1 Visitation record for
preliminary enumeration 17
Form 2 Preliminary enumeration household
record 18
Form 3 Uliimate-stage household record 19 Form 4 Monthly record of household
transactions for ultimate-stage
sample 20
Page 17
Government of Form 1
Household economic surveys Year
Census!
area i
Supervision
area
Enumeration
area
Enumerator
Name: No.
Town or village
House No.
Tota
Jumber of households Non-
farm Farm
ing Total
List of dwelling units
Total persons Pres
ent Away- Total
Description of dwelling uni No. of
rooms
Structure Roof Walls Floor
. 1
Year when built,10/18
Page 18 Fora 2
Governaeat of
Household Eoonoals Survey: Tur
Enuaerator
Supervise?
Visited
Oft
Cheeked
on
CeiMa*
Area Town or Village Address of heuseheld
SA EA House
No.
Ward or Quarter
HOMM- hold No.
Ho.
1 -.. 2 3
4
5 6 7 8 9 10 11 12 13 14 IS
N a m «
RP RA or VIS
R*latioft«M»
to head Sax Aga EdwatloMl
AttatnuMt
OMupailon
now or
last job
States 1. Employer 2* Salfcap.
3. Employet 4. Unaap.
Total (1)
Non-agr.
aarning*
In past
year
Value of agricultural sales In past year Property owned '_
Fields Building*
Near Far Total
Aeeoanodatl on No, of rooa*
Monthly rent Condition of tenanoy If not rented SiMwaryi
Household Members Present
Away
Total
Adults Juveniles Adults Juveniles
persons
M F Total
Crops
Llvs- stoek &
Poultry
Cash erops Cereals Roots Pulses Nuts & seeds Vegetables Fruit*
Other Large animals Saall anlaals etc.
Poultry Livestock i poultry products
; Total (2)
Total of (1) and (2)
OT7-343
For«3 Governmentof HouseholdEconomicSurveyiY«ar Enumerator Swpervlaor
Visited on Chaoked on■
Census Arei Townor Village Addressof hntunhnlri
SAEAHouse No. Wardor Quarter
Houae- holdNo, No. 1 7 3 4 5 6 7 B 9 10 XX 12 13 14 15
NAME
RP RA or Via
Relation ahlpto hxad
SaxAgsMarital atatusEthnic orotipDuration of prtaant raaldanoe
Plaeaof previous realdanca
bdnca~ tlonal attaln- MMt
Mainooeupatlon (Stat*Iffull tiM«student atpresent}
Siaiu* Ifune«p« £seeking work,state Ifnever worked before
whereworking Non- howse- hold •ett- vlty
House hold enter* prise Housing Type
condition: FloorWallaRoofNo.of roo«a
Yearwhen builtConditionof tenancyAnnual rentMeansof lightingSourceof waterQlstanoeofsourse ofwater(kas)Sewage PrimarysohoolSecondarysohoolHealthcentreLocalnarketCoop.storeBusroute
>■ hdCO a• 0>sw ao
H>
\OGOStaple staple
Page 20 Government of
Receipts and payments record Address:
Household economic survey: Year
Census area I
Supervision Area j Enumeration Area I
Stratum i
i
Household j
I
Month jnumerator
Staple
4
~1
CO Eh H
O
H
CO EH Ml
Ph
Farming
income
Sales
Home
consumption of own produce
Crope Livestock
Livestock products Crops
Livestock j
Livestock products
Consumption of own-produced seed.; fodder,eto.
Income from other enterprises Private income
Other receipts
Total receipts Farm operating costs
Other enterprise operating costs
Consumption
Food and drink
Clothing and footwear Gross rent, fuel and power
Furniture, household go*ds and operation Medical care and health services
Transport and communications
Recreation, entertainment, education; etc
Miscellaneous goods and services
Home consumption of own produce Non-consumption current expenditure Other payments
Total payments
Explanation of "balance: Balance
+ or -
1
Page 21
NUMBER OF HOUSEHOLD 1st week No
Present
Adults Males Female Juveniles Males
Females
Awaii
Adults Males Females Total
2nd week Nc
3rd week No
4th week No
Total No
Averago No
!_.
Farming income Saleo
/
Page 22
Total livestock products Hime consumption of own produce
S.. 10/18 Pag,, 23
tie at Milk
Milk pro due ts_
1st -week.
Skins-or hides
2nd week Q. Y
3rd week V
4bh week v
Total
j
Hair ar wool M&ntirp
Total livestock products
Consumption of own- produced seed & fodder
Total own-produced seed and fodder
from other enterprises
Trading resales (ex
cluding own prr-
ductjon)
Y
Sale of goods pro- duced or processed Home consumption of goods produced or processed
Use of gooda produced or processed in house- hold enterprises
V Y V
Charges for services rendered or work done Other "business
receipts
Total income from other enterprises
E/CN,14/CAS.ir.'/l8
Page 24
Private income
Wages and salaries
Property income (rent, interest & dividends)
Social security and social assistance Remittances received,
transfers of goods, cash grants & gifta Insurance & private pension "benefits
Total private income
Other receipts
Sale of enterprise property
Sale of private
property _,
Loans received for enterprise activity Loans received for
private purposes
1st week 2nd week 3rd week 4th week
V T V V
Total V
1
Repayment of loans given
Other non-income receipts
Total other receipts PAYMENTS
Farm operating costs Seed fodder, etc. from own produce
1 , _..
Seed, fodder, etc.
purchased Fertilizer
Insecticides and pesticides
Construction materials
repair
Sub-total fwd
E/CN ,i;
Page 25
Sub-total f"wd
Agricultural services and transport
Rental of equipment and Buildings
Wages and salaries
paid (including pay ments in kind)
Farming taxes less subsidies
Rent on land Interest paid on farm debt
T 0 tal farming_ costs Other enterprise opera-
tins costs
Value of own products used in household
enterprises
Cost of raw materials
& semi-finished goods
"bought for processing etc.
Cost of goods "bought for resale
iiixpenditure on miscel laneous business supplies
Rent and other charges paid in cash
Fuel, electricity,
gas, petrol, water, etc.
Cost of hire & repair of equipment- transport charges p&id & other miscellaneous expenses Repairs to buildings^
maintenance charges
Sub-total fwd
1st week V
—
2nd week V
3rd veok V
4th vroek V
■
Total V
6
E/CN.14/CA3.10/18
Page 263rd week 14th. week
-™E__I__
Sub-total fwd
Wages and salaries paia
(inol. payments in kind)
Business taxes,
licences, accountants' fees & other service charges
Interest paid on business loans
Total other enter prise operating costs
Consumption expenditure
Food and drink Cereals & cereal products
Total cereals &
products
Total meat
. 10/18
week } 2nd V
Pish
Fresh fish & shellfish Fish, dried or salted Canned fish & shellfish
; ! i
Kilk, dairy products and
Total milk, dairy products and eggs Oils and fats
Total oils & fats Fruits & ypgetables
(others than tubers)
E/CN.U/CAS.10/18
Page 20Total vegetables
Total nuts & seeds Pota'tope, tubers and
roots Potatoes
Sweet potatoes Other starchy roots
Starches (pure, dry)
Other tubers & roots Total potatoes, tubers and otter roots
Page 29 •
Coffee (beans or ffround)
Total coffee, tea,
Othpr foods
Cost of processing foodstuffs other than cereals
Total other foods Non-alcoholic drinks
Total non-alcoholic drinks
E/CN.H/CAS.l'v/lb
Page 3^
Alcoholic drinks
__ . ^
Total alcoholic drinks
Tobacco Cigarettes Snuff
Pipe tobacco Other smokables
-Total tobacco Summary1 food & drink
Cereals and products Meat
Pish
Milk, dairy products and eggs
Oils and fats Fruits
Vegetables Pulses
Nuts and seeds
Potatoes, tubers, etc.
Sugar
Coffee, tea, cocoa, etc Other foods
Sub-total fwd
1st week Q
""■■-■—..
V
2nd week
ft
j
V
_-—-^""
V V
3rd week
Q
.-^
4th week
Q
i
1 —1
V
■
— ~.
V
V .
Total
Q
VV
11
Page 31
Sub-total fwd Non-alcoholic drinks Alcoholic drinks
Total food & drink Clothing and footwear
Men & "boys over 15 years
Total men & boys i/cmen & girls over 15 years
Total women & girls Children 15 yrs
Total children
E/CN.14/GAS.10/18
Page 32Summarys clothing and footwear
Men & boys 'Women & girls Children
Total clothing &
footwear
Gross rnnt, fu*l & power House rent
Ea.tes.
Water Charcoal Firewood Kerosene Electricity
Electric batteries Gas
Other fuel & power Total rent, fuel and power
Furniture, houspnold poods & operation
Beds Chairs Tables Cupboards Shelves Mirrors
Pictures & decorations Lamps
Carpets. & rugs O.$her furniture Repairs to furniture
Total furniture
1st week V
,
2nd week V
3rd week V
i
1
4th week V
i
■
,. . .
Total V
--.- .
13 ■
E/CH-14/CAS.10/]-3
Page 33
Sheets Blankets Towels
Table clothes Napkins
Pillow cases Curtains Pillows Mattresses Other linen or bedclothes
Total linen, etc.
Stove, oven Refrigerator Fan
Air cooler, air conditioner
Sewing machine
Other durable equipment Repairs to equipment
Total equipment Pans
Trays Bowls
Dishes, plates Coffee pots Tea pots Cups
Glasses, mugs
Knives, spoons, etc.
Other utensils Repairs to utensils
Total utensils
1st week V
2nd week V
3rd week V
4th week V
Total V
14
E/CN.14/CAS.10/16
Page 34Suit cases, "bags Animal skins Mosquito nets Thermos flasks Torches
Buckets & washing tubs Brushes
Rope and string Matches
Shoe polish
■Washing powder Cleaning powder Other
Total other equipment Cleaning and laundry Cooking and other domestic services
■ Total household services
Summary.;. fur ni tur e, household goods and operation
Furni ture Linen Equipment Utensils
Other equipment Services
Total furniture, household goods <k operation
1st week■
V
■
2nd week V
3rd week Y
4th vrsek V
■ 1
"i
.- ■ .-
-
r
Total V
...
... . ;.■■■■ ....-■-
!
i
_ __
i
E/GN.14/CAS, .0/18
Page 35
I^dical earn and health s^rvic^s
Kedicine
Doctors' fees, etc.
Other medical fees Hospital care, etc.
Total medical care Transport and communi- pations
Private car
Motorcycle, scooter Moped
Bicycle
Spare parts for private vehicles Repair of private vehicles
Petrol, oil, etc.
Train Bus, taxi Boat Airplane
Animal transport Removal & storage of
household goods Postal services
Telephone & telegrams Total transport &
communication
Recreation, pirtprtain—
mpnt, education and cultural services
Television Radio
Musical instruments Sub-total fwd
1st week V
2nd week V
3rd week j 4th week
V V
!
,
Total V
j
i
16
E/CN.14/CAS.10/18 Page 36
Sub-total fwd Sports equipment
Photographic equipment Repairs to equipment Cinemas
Sports
Clubs and unions Weddings
Funerals Births
Seasonal feasts Other recreation arid culture Books
Newspapers & magazines School fees
School equipment Total recreation, education, etc.
1st week V
2nd week V
i
JKisCellanedua goods j
and services
Hair dressing, etc- Other personal services Toilet soap
Perfume
Dry perfume (sandal- wood, etc.-)
Other toilet articles Gold ornaments
Silver ornaments Other ornaments Expenditure in restaurants, cafes
& hotels
Other servinefT n.e»o.
Total miscellaneous expenditure
3rd week V ■•
4th week V
Total V
j
■ ■ ■
■
•■
17
E/CN.M/CAS.IC/Lu
Page 374th week! Total V
Home consumption of own
produce (detailed above)
Livestock
Livestock products Other goods produced
processed
Total home consump
tion of own produce Non-consumption current
expenditure
Other direct taxes
Pension & social
security contributions Insurance premiums
Remittances & gifts (not to household members]
Total non-consumption current expenditure Other payments
Purchases of property Land for agriculture Land for business Land for residence
Vofcioles for Farm equipment Other enterprise
equipment
Construction of buildings (inol. major repairs or
extensions)
Loans given
Investment & purchase of other assets
Total other payments
Pago 38
he o to
ShOet 1 jifiLiiin-: rv; .;-( r,vi tpjS&XS.. ;. ^Q
sheet 2 Declared ircoms ■ Va"
4UWorking
sheet 3 Selection of households 41
Working
shQ9b 4 ^ Ljst of bouEoholds selected "42
Super vision
Area
Enumera tion- Area
Total
I
^-.
Total
Total Census area total
Number of houses
Enumeration Totals
dumber of households Hon-farm
_
Farming Total
Page 39 Working Sheet 1
Number of persons Present
^ ^
Away Total
E/CN.U/CAS; 10/16 Page 40
[forking Sheet 2
Declared Income
Super vision
Area
■ - ■■
jnumera
tion Area
Total
Census area total
Income Class 1
. . ..
Total
National1 total Class mean Total income
Selection of households
E/CN.14/GAS.10/18
Page 41
Working Sheet 3
Super- Trision Area
Enumera tion Area
Available for selection 3 tra turn
■' 1 2 3
Total
Sampling fraction deflator
-■
Deflated sat pling fraction Stratum
1 2 ■3
To be selected
Stratum
1 2 3 •
E/CN.14/CA3.10/1H
Page 42forking Sheet 4
Enumerator
Namei No. Census.
area Sector Zone
T
List of households selected
Reference No.
Home Household Address Stratum
S US OLE-VISIT ECOlToIvIIC 3"Ur Survey design and content
48. Some of the smaller countries and island states -wishing to initiate
survey programmes may find it impracticable to employ many enumerators or to visit individual households over long periods of time. In thesecircumstances a single intensive visit to each of the households in n properly organised sample would "be an attractive solution for investigating household structure, economic activities and levels of living. Th« purpose of this section of the paper is to investigate the feasibility of such an operation*
49. It is clear that a single visit to any reasonably co-operative household can produce a great deal of useful information on its membership, activities
and living conditions. The problem is to know whether these "basic data canbe used as a starting point for establishing a crude but satisfactory set of
household accounts.
50# The suggestion is that such work can be don<* >y a small team of well
trained enumerators provided it is accepted that the figures for private consumption expenditure will be considerably less accurate and detailed than in the normal kind of household budget survey. A draft questionnaire for a single—visit survey is given on pages 53-61 and is discussed in the following notes.Notes on questionnaire
51. The questionnaire is intended to "be applied to f> sample f»f much the
same kind as described elsewhere in this paper i.e. households selected from fully enumerated geographical units. Again an economic stratification
of households would have significant advantages.52. Although many parts of the questionnaire have been based on current and previous surveys the form as a whole has not yet been tested in the
field. The comments which follow therefore aim to provide both a description of the questionnaire content and an indication of special points or weaknesses
which statistical offices might wish to take into account when adapting the questionnaire for experimentation under local conditions. The numbers of
sub-headings "below refer to sections in the questionnaire.53. At the top of the first page, -the information for identification of the sample household is given. The information on census area, SA and EA together with house number and household number may "be coded. The reference to Town or Village, Ward or Quarter and the Address of the sample household will "be useful to the field-personnel for locating the household on the ground.
However, an item that will seem superfluous under the existing conditions of any country should >e omitted from the questionnaire. The house number will come fron the visitation record given in Form 1 in rage ]_y#
E/CN.14/0AS.10/10
Page 441• List of household members
54. The questionnaire "begins with the normal type of census listing to which a few additional items have been added to provide a guide for the more
detailed investigation of ecojiomiii_apt_iyity.._. _...55. First it is necessary to be clear about the household definition.
Essentially this is of a de jure nature because there is oniy a single opportunity, for compiling the list arid changes cannot be recorded as in a budget survey covering a period of some months. The suggestion is therefore
that all persons considered to be full members of the household when regarded as an economic unit should be included. Column 3 provides an indication of whether or not individuals are normally present. The general guideline should perhaps be that anyone wh^ is stated to be not normally present and does not participate significantly in the household economy, or belongs to a permanent household elsewhere, should probably be excluded from the list.
55* The form makes no specific provision for visitors present at the time of the interview and it is probably better to exclude them unless there is an important economic relationship with the household. In'special cases they could be included in the list nf household members. A more complete alternative would be to list all visitors and identify them as such by an additional classification in column 3- This has the advantage of giving the de facto as,well as the de jure population covered by the survey but the additional information does not'appear to be of special value in the kind of
inquiry discussed here.
57. Relationship to head of household, sex, age and level of education (columns 4-7) have the same purposes~as-a:n--a normal census form and need no detailed explanation. The only point to be noted is that the first and last of these items call for the preparation of code lists? it is envisaged that enumerators would code all their own records in the field.
58. Main occupation in column 8 is also straightforward, and also calls for a single code list. However its relationship t^ the last set of questions in
columns 10-13 requires a little thought. For example a person whose main
occupation is blacksmith may also work on the household farms. The relationship ie therefore raxaer loose and could nou be made more precise without specifying secondary occupations which might unduly complicate the form.
59. Colvmn 9 indicates economic status in the main occupation and a simple code such as 1, Employer 2. Self-employed 3- I*nployee 4. Unemployed may be
shown on the form so that entries can be made directly by the enumerator.
Again the question is a guide rather than a precise indicator for the
completion of columns 10-13*
£0. The :.a;;-l four columns .'"a re intended to shew the main sources from which tbe household gets its funds and they are investigated in more detail on
subsequent pages of the questionnaire. No attempt should be made on the first
page to include any estimates of income? these figures have to be obtained
from a very much more careful examination of household activities.
E/CN.14/CAS.10/18
Page 4561. However it is necessary to have a clear idea of what is meant by
household and non-household enterprises from the outset. There seems to be no precise .formal definition of a household enterprise but broadly it concerns
the kind of productive activity which takes place at home or in the immediate surroundings. Such activities include independent farming and livestock production, trading, handicrafts, etc. Some of the household enterprises can employ full-time labour.
62. A non-household enterprise has to be something which can be distinguished separately from normal household activities. If a member of the household
has a wage or salary income he is usually working in a non-household enterprise.
A more difficult case arises where a member owns and operates a business with
separate premises. If such a business is covered by the establishment surveys of the country concerned it can perhaps be regarded as a non-householdenterprise with the owner and any other members working in it having private
incomes from that source. Otherwise it is probably better to treat it as ahousehold enterprise and record its activities in the appropriate part of
the questionnaire.
63. Columns.10-13 should show the economic activities in which each house hold member is engaged. Some members may have no economic activity while others may be engaged in one or more. The questions give a broad indication
not only of the sources of household income but also of the distribution of
household labour. It should be noted that farming is intended to cover bothcrop production and livestock. A separate column has been used for fishing
because it is important in some countries and the industry is organisationally rather different from agricultural activities. The headings should of course be arranged.in the light of local circumstances.64. It should be noted that to permit integration of the results of different
survey- rounds of the on-going programme, this suggested schedule with items on marital status, ethnic group and migration added to it, may serve as the core questionnaire. However, care should be taken to keep the questionnaireas simple as possible so that undue importance is not paid to topics unrelated
to economic activity of the household. It is imperative that great care and
emphasis be given to the items relevant to the economic survey.
2. Housing and amenities
65. This se.ction has been derived from part of a questionnaire currently being used in Kenya. By way of illustration the Kenya coding arrangement has also been indicated. This involves the ticking of appropriate cells
(except in the last part of 2.1 and in 2.2) and then later entering the
corresponding code numbers in the external boxes. An alternative would be to enter the code numbers direct and to omit the provision for the initial ticks.66» Questions in section 2 give general information on housing, water
supplies, lighting and sanitation and the availability of various amenities.
They are placed on the second page of the questionnaire because they are easy to answer and give the enumerator an opportunity to gain the confidence of household respondents.