Preview

INTERNATIONAL LABOUR OFFICE

METHODS OF
FAMILY LIVING STUDIES

Report prepared for the Seventh International
of Labour Statisticians
(Geneva, September 1949)

GENEVA
1949

Conference

STUDIES AND REPORTS
New Series, No. 17

P U B L I S H E D BY THE INTERNATIONAL LABOUR O F F I C E

Geneva, Switzerland

Published in the United Kingdom for the INTERNATIONAL
by Staples Press Limited, London

LABOUR OFFICE

PRINTED BY " IMPRIMERIES RÉUNIES S.A.", LAUSANNE (SWITZERLAND)

CONTENTS
Page
INTRODUCTION

CHAPTER I : Scope and Timing
Scope
Timing

of Family Living Studies

5
5
7

CHAPTER I I : Sampling Methods

CHAPTER I I I : The Collection of Data
Methods of Collection
The Data to be Collected

12
12
16

CHAPTER IV : Classification of Receipts and Disbursements
Receipts
Disbursements, Outgo, Consumption Expenditure

17
17
20

CHAPTER V : Analysis and Appraisal

of Results

CHAPTER VI : Food and Dietary Analyses
Food Expenditure
Food Consumption

34
34
35

CHAPTER V I I : Special Problems of Family Living Studies in Underdeveloped Territories

CHAPTER V I I I : Farm Family Living Surveys
gations
Farm Family Living Surveys
Special Investigations

43
43
46

and Special

Investi-

CHAPTER I X : Proposed Resolutions

APPENDICES
I : Resolution of the Third International Conference of Labour
Statisticians, Geneva, October 1926 ; Recommendation of the Committee
of Statistical Experts of the International Labour Office, First Session, 12 to 15 December 1933

I I : Extracts from the Reports of the Subcommission on Statistical
Sampling to the United Nations Statistical Commission

INTRODUCTION
The subject of methods of conducting family budget enquiries
was placed on the agenda of the Third International Conference
of Labour Statisticians, held in Geneva in October 1926. In
preparation for this Conference the International Labour Office
prepared a report x which summarised the work done in this
field during the preceding period and presented certain important
questions for the consideration of the delegates. The Conference
adopted a resolution on family budget enquiries which is reproduced in Appendix I.
S u b s e q u e n t l y t h e I n t e r n a t i o n a l L a b o u r Office p u b l i s h e d i n t h e

International Labour Review a series of analyses of individual
family living studies as carried out in the different countries.
I n 1933 an analytical summary of recent surveys was published. 2
In connection with a study of workers' nutrition and social policy
the Office drew up a classification of foodstuffs for use in international comparisons of food consumption and made an analysis
of the materials on food consumption as found especially in recent
family budget studies. 3
In|1940, the Office published a revision of the document on
methods¡ 4 , and between May 1939 and April 1941 published
analytical summaries of recent studies in a series of articles in the
International Labour Review?
All this work was in a sense prehminary to placing the subject
of methods of family living studies again on the agenda of an
International Conference of Labour Statisticians, with a view
to a revision and expansion of the resolution setting international
standards|for conducting studies of family living conditions.
1
I.L.O., Studies and Reports, Series N, No. 9 : Methods of Conducting Family
Budget Enquiries (Geneva, 1926).
2
"Recent Family Budget Enquiries", International Labour Review, Vol. X X V I I I ,
No. 5, Nov. 1933, p . 635.
3
I.L.O., Studies and Reports, Series B, No. 23 (Geneva, 1936), p . 195.
4
I.L.O., Studies and Reports, Series N, No. 23 : Methods of Family
Living
Studies (Geneva, 1940).
6
Vol. X X X I X , Nos. 5 and 6, May a n d June 1939, p p . 662 and 814 ; Vol. X L I ,
No. 6, J u n e 1940, p . 645 ; Vol. X L I I I , No. 4, April 1941, p . 474.

METHODS OF FAMILY LIVING STUDIES

Following the war, the Sixth International Conference of
Labour Statisticians was convoked in Montreal in August 1947.
In connection with a discussion of cost-of-living statistics, the
Conference adopted a resolution, recommending to the Governing
Body of the I.L.O. t h a t the subject of family living studies be
placed on the agenda of an early conference, in the following terms :
The Sixth International Conference of Labour Statisticians,
Recognising the need for new studies of family consumption in order
to furnish up-to-date weights for cost-of-living index numbers as well
as for other reasons, and above all in order to provide for comparisons
of consumption levels with given standards of adequacy, and
Recognising the special need for studies of family living in less
well developed areas where such studies have not been made hitherto,
Requests the Governing Body of the International Labour Office
to place on the agenda of an early International Conference of Labour
Statisticians the subject of the making of family living studies, including
the application of new sampling techniques to the selection of families,
Requests the Governing Body of the International Labour Office
to consider the advisability of referring the special problems of making
such studies in less well developed areas to a technical committee of
specially qualified experts. 1
Accordingly, the Governing Body at its 104th Session in
March 1948 decided to convene the Seventh International Conference of Labour Statisticians late in 1949 and included the subject
of studies of family hving conditions on the agenda in the following
terms :
Methods of family living studies : techniques and methods of family
hving studies, with special reference to the appUcation of sampling
procedures, standard accounting methods, and uniform classifications.
At the same session, the Governing Body authorised the calling
of a preliminary meeting of statistical experts early in 1949 to
define the scope of the topics on the agenda.
The materials available for the present study include the
extensive survey and summaries of methods and results of family
hving studies already mentioned. The studies present the developments in recent surveys and throw light on significant conclusions for future work. Four principal points were brought out :
the need for applying sampling techniques to the selection of
families studied ; the need for systematic accounting concepts
1
1 . L . O . , Studies and Reports, New Series, No. 7 (Part 4) : The Sixth International Conference of Labour Statisticians, Resolution X V I (Geneva, 1948), p . 69.

INTRODUCTION

and procedures to cover the receipts and disbursements of the
families ; the need for standard nomenclature and classification
of the materials ; and the need for systematic analysis of the
results including the application of consumption scales. The
present document will rest largely upon this earlier survey of
methods, which contains appendices and detailed references to
methods in use in different countries.
With regard to food consumption, a summary of the detailed
report prepared by Dr. E. J . Bigwood for the Technical Commission on Nutrition of the Health Organisation of the League of
Nations was included as a separate chapter, " Problems connected
with food consumption with special reference to dietary surveys ",
in Methods of Family Living 8tvdies.
I n preparation for the present discussion, furthermore, a
survey of recent family living studies was undertaken to ascertain
the principal developments since 1938.1 Although much of this period
was a time of war in many countries, a considerable number of family
living studies were undertaken, but except for those in the United
States and India and perhaps one or two other countries, in
which significant developments in sampling techniques may be
noted, these reflect principally the increasing recognition t h a t
such surveys should take account of major changes in consumption
and conditions of living.
I n this connection mention should be made also of the adoption
by the Sixth International Conference of Labour Statisticians,
in Montreal in August 1947, of a resolution recommending " the use
of small sample studies of consumer purchases in the intervals
between the more comprehensive surveys . . . in order to provide
the basis for discovering significant changes in consumption patterns to indicate the need for revisions in the weighting diagrams". 2
The reports of the Subcommission on Statistical Sampling
of the United Nations Statistical Commission dealing with the
subject of the appUcation of samphng for family living studies
are an important source of valuable recommendations. Their
conclusions are reproduced in Appendix I I .

1
The Food and Agriculture Organisation of the United Nations has in preparation a report on nutrition and dietary surveys.
2
Resolution I I , paragraph 14, op. cit., p . 62.

METHODS OF FAMILY LIVING STUDIES

A draft of the present report was submitted to a small group of
statistical experts, at a preparatory meeting authorised by the
Governing Body of the International Labour Office, for a preliminary discussion, with especial reference to the definition of the
questions to be included. The report as presented has been revised
in the light of the discussions at this preparatory meeting, and
the Office would like to express its thanks to the experts who
participated for their most helpful criticisms and suggestions.
It should, however, be pointed out that the Office assumes full
responsibility for the proposals submitted in this report to the
Seventh International Conference of Labour Statisticians.

CHAPTER I
SCOPE AND TIMING OF FAMILY LIVING STUDIES
Two topics are discussed in the present chapter : the scope
and the timing of the studies.
SCOPE

The great majority of official family living studies have been
undertaken in order to furnish a basis for cost-of-living index
numbers and have been based on conditions of family living and
expenditure in urban wage earners' families. During the past
two decades, however, a considerable proportion of surveys have
included, in addition to wage earners' families, families of salaried
employees and civil servants. I n a number of studies, furthermore,
notably the study of consumer purchases in the United States,
the scope has been extended to cover the entire population, rural
as well as urban, and including all income classes. Not a few
studies have included the farm and agricultural population.
Thus there has been a general tendency towards a broadening
of the scope of these studies to cover the entire population as
well as the principal elements in it.
At the same time there has been a shift from an exclusive
concern with bases for cost-of-living studies to a greater appreciation of the data as throwing Ught upon standards or levels
of living. From this point of view, the interest in specific groups
and their special conditions of living has contributed to the increased attention paid to closely defined classes. Each important
sector of the population, for example the farm labourer, the salaried
worker, the miner, may have its special consumption patterns.
Studies of the effect of changing income on consumption patterns,
as, for example, in the case of the families of the unemployed
who have lost their usual source of income, throw fight upon the
problem of maintaining standards of living. Differences between
major areas are often marked, for example, in the United States
between North and South. I n particular, differences between

METHODS OF FAMILY LIVING STUDIES

race groups, between well developed and less well developed
areas, call for study of the differences in living standards.
Furthermore, the analysis of the materials gathered in these
studies throws valuable light upon various problems of national
interest, such as for example, the possibilities of savings in different
income classes, the break-even point, the elasticity of demand
for different items of expenditure, and different habits of consumption in different social and economic groups.
In the resolution of 1926, the scope of family budget studies
was defined in terms of families representative of " large homogeneous sections of the population ". In the detailed specifications,
mention was made of " different income classes within the sections
of the population under consideration ", as well as " different
industries and districts " and " manual and non-manual workers'
families ". Thus, the resolution contemplated a broad coverage
within the " sections " of the population.
The Sixth International Conference of Labour Statisticians,
meeting in August 1947, adopted a resolution on cost-of-living
statistics which specified as possible bases of weighting diagrams
the consumption levels of different economic groups in different
types of communities.1 In practice, separate cost-of-living index
series may be developed for special groups with specific consumption patterns the price movements of which vary from those
for the groups covered in the general price index. The consumption
patterns of these different economic or special groups are determined on the basis of family living studies the data for which
give directly or as a result of analysis the materials for the consumption pattern concerned.
The question is thus raised how and to what extent the resolution of 1926 should be modified to take account of these changes
in point of view.
In the first place, the aim should be to cover the entire
population in order to have as complete a picture as possible
of consumption patterns and standards of living of the population
as a whole.
Secondly, data on consumption patterns and family living
should be collected for each important sector of the population.
The most significant sector, and the one most often the subject
of a family living study, is the group of urban wage earner families.
Other important sectors include the families of salaried employees,
1

Resolution I I , op. cit., p . 60.

SCOPE AND TIMING OF FAMILY LIVING STUDIES

civil servants, farmers and farm labourers, employees in different
industries as well as families belonging to different income classes
and different races or national groups, and those living in particular
areas. The significant subdivisions of course vary from country
to country.
Thirdly, studies of the level of living or consumption patterns
of families of different size and composition or families living in
special circumstances should be undertaken, for example, to throw
light upon changes in standards of living or consumption patterns
in different stages of family development or in families undergoing changes in income per consumption unit.
The following parts of resolution No. I (see Chapter IX) cover
these points x :
The main purposes of the study of family living conditions are to
provide—
(a) the pattern of expenditure for use as a basis for the weighting of
price indices ; a n d

(b) as full information as possible with regard to levels and standards
of living, including data permitting the analysis of income, expenditure, and savings in relation to them (I, l). 2
Family living studies should, in principle, obtain data on consumption
patterns and family living conditions for all important sectors of the
population, including, for instance, different income classes, different
social and economic groups, such as wage earners, salaried employees,
different race and national groups, urban and rural sectors, different
family or household types, and different geographical areas (I, 2).
Information should also be obtained both for industrially developed
regions and for regions which are undeveloped, and for areas where
specific industries and agriculture are predominant (I, 3).
TIMING

Family living studies should be repeated from time to time
as changing circumstances make it desirable and necessary to
measure changes in consumption patterns and conditions of
living.
The 1926 resolution specified that such studies should be
conducted " at intervals generally of not more than ten years ".
The resolution of the 1947 conference recommended t h a t small
sample studies should be undertaken from time to time to determine
when a large-scale survey was needed.
1
The topics of farm family living and special studies of levels of living in
certain particular conditions are discussed below (Chapter VII).
2
The arabic numbers refer to t h e paragraphs of the resolution.

METHODS OF FAMILY LIVING STUDIES

Mention should be made, furthermore, of the principle of
rotation recently adopted in the practice of the United States,
where each year new studies of two or three cities are undertaken
until in the course of time such new studies have been made for
the whole number of cities for which index numbers of cost of
living are calculated. Such a plan provides for a continuous
programme so t h a t the services of the experts and sampling specialists once trained to conduct such enquiries can be utilised
continuously. As a consequence, the series of studies is performed
with a high and uniform degree of competence, in contrast to
the situation where a large staff must be assembled, less well
trained and often inexperienced, to conduct a single large survey
undertaken once in ten years.
I n principle, such surveys should afford a continuous view
of changing conditions. They should therefore cover not only
the normal or usual economic conditions, but also the conditions
associated with the peak of prosperity and with economic depression. I n practice, the major studies have usually been made
during a period intermediate between depression and prosperity.
Special surveys of the unemployed or of persons on relief are
often taken during a depression period. This point is covered by
resolution I as follows :
The major survey should be undertaken at least once in ten years ;
preferably, the major survey should be replaced by progressive surveys
covering small samples at short intervals designed to cover the same
ground as a major survey (I, 9).

So far as the period to be covered by the studies is concerned,
this should in principle be a full year. Only in this way can a
complete picture of all phases of family living be covered, including the different seasons of the year. In practice, however,
the requirement t h a t a full year be covered in the family accounts
raises difficulties : the number of families who can be induced
to keep accounts for a full year is often limited, so t h a t the scope
of the enquiries, where this method is employed, is reduced, and
the techniques of random sampling in t h e selection of families
rendered difficult of application. Where the interview method
is used, the enquiries often cover a full year, except t h a t questions
as to food are sometimes limited to the week immediately prior
t o the enquiry. I n a number of enquiries the data on food are
obtained for four periods of a week or fortnight each, spread
over the year so as to reflect food consumption at each of the

SCOPE AND TIMING OF FAMILY LIVING STUDIES

four seasons. On this point, the problems of samphng are of
especial importance.
The period t o be covered is dealt with in resolution I as follows :
The survey should cover the period of a year, to take into account
seasonal changes in the expenditure pattern ; in the case of recurring
items of expenditure such as food, four quarterly dates should suffice,
but for non-recurring items, such as clothing, periodic information
throughout the year is essential (I, 10).

CHAPTER

SAMPLING METHODS
The application of sound sampling methods to the selection
of families in family living studies is perhaps the major improvement in recent family living investigations. Details of the methods
of selection and samphng applied were given in Methods of Family
Living Studies1, and a summary of the application of sampling
methods in recent studies was given in a note prepared for the
meeting in September 1948 of the Subcommission on Statistical
Sampling of the United Nations Statistical Commission. 2 Appendix
I I contains the recommendations of the Subcommission on this
point.
The resolution of 1926 did not ignore the question, but phrased
its recommendations in general and non-technical terms as follows :
Families representative of large homogeneous sections of the population.
The families should he selected to include a sufficient number from
different income classes within the sections of the population under
consideration, and also from different industries and districts.
These recommendations are so vague t h a t they are well nigh
meaningless.
The elements of the programme for the application of sampling
techniques t o family living studies are given in Appendix I I ,
which contains the recommendations of the Subcommission on
sampling at its first and second sessions. For the purposes of
the present Conference, the most significant points can be summed
u p in the following paragraphs :
(1) In making family living studies the selection of families should
be made in accordance with samphng techniques as recommended in
detail in the reports of the Subcommission on Statistical Samphng of the
United Nations (I, 17 (1)).
(2) Provision should be made for a qualified expert in statistical
sampling to whom should be referred all questions affecting the selection
1
2

Op. cit., p . 17.
U.N. document E / C N . 3/Sub. 1/5.

SAMPLING METHODS

of families and the appUcation of sampling techniques to methods and
results (I, 17 (2)).
(3) Where necessary and desirable, special problems in this field
may be referred, at the instance of the sampling expert associated with
the study, to a Sampling Expert Consultant designated by the International Labour Office or by the U.N. Statistical Office, or to the
Subcommission on Statistical Sampling (I, 17 (3)).
(4) In publishing the results of these studies, a clear exposition
should be given of the sampling methods employed and an analysis of
the sampling errors and other types of error involved (I, 30).
(5) In planning family living studies, consideration should be
given to the desirability of making preliminary or pilot surveys to test
methods, as well as to determine changes since previous studies which
would emphasise the need for future surveys (I, 16).
I t is not considered necessary here to list each topic at length
on which the sampling expert should be consulted, since these
topics are thoroughly covered in the reports of the Subcommission
on Statistical Sampling. 1
The reports of the Subcommission on Statistical Sampling
also touch a number of other questions in which there is a sampling
aspect. Since these questions, however, have other aspects, it
is considered that for purposes of these recommendations the
sampling elements involved can be adequately taken care of by
reference to the qualified sampling expert attached to each study,
when it is a question of the policy or practice to be adopted by the
study, and by the reference of points involving further research
to a body such as the Statistical Section of the I.L.O. or the Statistical Office of the United Nations, where the views of qualified
sampling experts will receive full weight.
At the same time, it is considered not desirable to specify
in too great detail in specific recommendations just what types
of sampling devices should be used in family living studies, since
the sampling expert attached to the study should be thoroughly
familiar with the theoretical requirements and current practices
on all these points.

See Appendix I I .

CHAPTER

III

THE COLLECTION OF DATA
A series of special questions concerned with the collection of
data are considered in the present chapter. For the most part,
these points lead to fairly obvious recommendations. They are,
in general, fully covered in the discussions included in the report
on methods of family living studies already referred to. Some
of them are already taken into account in the resolution of 1926.
These points are reviewed here briefly and recommendations are
made to cover them.
METHODS OE COLLECTION

1. Sources of

Information

The chief source of information is the housewife and family
informants who supply data on income and expenditure, family
composition and other topics. Whenever possible, this information
should be checked with other sources. For example, if the housewife does not know the amount of the income received by the
head of the family, special arrangements may have to be made
to obtain the information directly in supplementary visits. Furthermore, information on income obtained from the families can,
in many cases, be checked by employers' payroll data.
2.

Methods for Obtaining

Data

One of the important questions in connection with taking
family living studies is the method of obtaining information.
Two principal methods are employed : the account book method
and the interview method. In the account book method the
housewife is provided with an account book in which the various
expenditures on different items as well as the various items of
income are entered day by day as they occur. I n the interview
or schedule method, an agent visits the housewife and makes
enquiries in detail in regard to expenditure and income over
a period of time, usually a year, covered by the survey. Information

THE COLLECTION OF DATA

on expenditure on food is often limited to the week just past—
but often this part of the enquiry is repeated so as to cover four
sample weeks during different seasons of the year.
In general, European countries have tended to prefer the
account book method in obtaining data for family living studies ;
the United States and Canada have tended to prefer the interview
method. I n this connection, an important point is that the
application of sampling techniques has been much more generally
applied in the case of the interview method, which favours a much
larger number of cases on account of its small cost per case. The
account book method, on the other hand, sets certain minimum
qualifications on the part of the housewife — namely, ability
and willingness to keep the household accounts — and, in practice,
has led to considerably less use of sampling techniques in the
selection of families. So far as the relative accuracy of these
two methods is concerned, few studies have been made on this
point. Such studies as have been carried through indicate t h a t
both methods require careful preparation and planning of the
account book and the schedule.
On this question no recommendation to express any preference
between these two methods is proposed. There is need for careful
investigation of sources of error and of relative accuracy as well
as of the appropriate scope of the two methods. An important
point is the possibility of applying satisfactory sampling techniques to the selection of families and of calculating sampling
errors in the results.
A resolution to cover the need for study of the question is
proposed as follows :
The Seventh International Conference of Labour Statisticians...
requests the Governing Body of the International Labour Office to
instruct the Office to arrange for a study of the relative advantages
and merits of the different methods of obtaining information in family
living studies, with due regard to the problems of sampling and to the
margin of error in the results (II).
3. Forms and Plans
Special attention should be paid to the development of the
schedules, forms, instructions to agents, instructions for editing
and plans for tabulation so as to obtain the information in the
form required.

METHODS OF FAMILY LIVING STUDIES

4. Selection and Training of Investigators
I n the taking of family living studies, special attention should
be paid to the selection and training of investigators. The individual agents who visit families, or the supervisors or agents
who visit families, or the agents who review the account books
from time to time should be thoroughly trained so as to ensure
the correct filling up of the schedules or account books.
Mention has already been made in the preceding chapter of
the need for having on each study a qualified expert in sampling
techniques in order t h a t sampling problems which are raised in
the course of the survey may be dealt with competently. When
possible, arrangements should be made for giving t h e investigators
a survey of the purpose and requirements of sampling techniques
as connected with family living studies, in order t h a t specific
sampling problems arising in the course of the investigation may
be noted and dealt with more effectively.
One point of particular importance is the need for testing and
standardising the performance of the different investigators in
the survey, in order t h a t full use can be made of special sampling
techniques as explained in Appendix I I , paragraphs 34-37.
5. Supervision

and Methods of Checking

Results

The procedure of editing account books and schedules should
be set in motion at an early stage of the investigation and special
instructions drawn up so as to standardise the procedure as much
as possible. Competent editing and supervision of the materials
as they are obtained helps in the training of the investigators
and tends to produce a satisfactory result.
Special attention must be paid to the possibilities of checking
entries with one another for consistency and with outside sources
of information for accuracy. Consistency between the two sides
of the account in comparing income and expenditure and verifying the sources from which the account is balanced are useful
not only to check the accuracy of the returns but also to throw
light upon the possibility of important omissions or errors.
I n presenting the results of family living studies, emphasis
should be laid upon the accuracy and reliability of the figures.
The tests and checks to which they have been subjected should
be explained in detail. The margin of error of the figures should
be stated as well as the error due to sampling.

THE COLLECTIOK OF DATA

6. The Preliminary

Study

The preliminary or pilot study is an important part of the
plan, especially when a family living study is taken for the first
time or under new conditions.
I n the preliminary survey the different questions can be tested
with a view t o determining the best form of question, the best
way to test the data, and whether a question on the particular
point is useful. Of course, special attention must be paid to the
particular aspects of the particular investigation. The preliminary
survey may often be utilised to determine whether a full scale
survey is needed.
On these several points the following recommendations are
made :
Special studies should be made with a view to testing the relative
merits of the interview method as compared with the account book
method, with special reference to the circumstances of the enquiry and
the applicability of sampling procedure (I, 18).
Since these studies should be made or arranged for by the
International Labour Office a separate resolution on this point
is addressed to the Governing Body (II).
Special attention should be paid to the development of forms and
schedules in advance of the study so as to be able to obtain satisfactory
and comprehensive results on the basis of a well tested and thought
out plan (I, 19).
The investigators and agents charged with the collection of materials
on family living conditions should be selected from qualified personnel
and given appropriate training to fit them for their work (I, 20).
The materials obtained should be subjected to careful verification
and checking in order to ensure that the results are as free from error
as possible (I, 21).
In publishing the results of the studies a clear exposition should
be given of the sampling methods employed and an analysis of the
sampling errors and other types of error involved (I, 30).
In planning family living studies consideration should be given to
the desirability of making preliminary or pilot surveys to test methods,
as well as to determine changes since previous studies which would
emphasise the need for future surveys (I, 16).
H

T H E DATA TO B E COLLECTED

The data to be collected include, besides the information on
amounts and qualities of goods and services purchased and the
prices paid, information on family size and composition, including

METHODS OP FAMILY LIVING STUDIES

age, sex, marital condition and position in the family of each
member of the household, the presence or absence of lodgers, etc.,
information on health, race and social condition, occupation,
employment and unemployment, industry and industrial status.
I n addition, information on the housing conditions under which
the families are living and other data such as the possession of
a vegetable garden, keeping of poultry, etc., which help to throw
light upon the level of living of the family should be obtained.
The basic information on income and expenditure which forms
the core of family living studies, namely, receipts and income,
disbursements and consumption expenditures, must also be obtained
with sufficient detail to permit the verification and classification
into significant rubrics. These last mentioned points are covered
in the next chapter.
The following paragraph of resolution I covers this point :
The particulars asked for should include if possible all the following
items : details of family income, including social insurance contributions ; direct taxation, etc. ; size and age of the family ; number of
earners ; industry, occupation and industrial status of the earners of
the household and whether unemployed or in receipt of relief ; number
of rooms occupied and expenditure, in as much detail as possible, on
housing, fuel and light, clothing, food consumption and expenditure
(including a valuation of garden produce and free food), household
goods, miscellaneous goods, services, and tobacco (I, 8 (1)).
I n addition, where necessary for the broader studies of standards

of living, the following recommendation is made :
The subjects to be covered by these enquiries should include information as to health, nature of living accommodation and amenities,
an analysis of income, extent and nature of savings, and, unless obtained
under the previous recommendation, quantities of food consumed
(as a basis for dietary and nutrition studies) (I, 12).

CHAPTER IV
CLASSIFICATION OF RECEIPTS AND DISBURSEMENTS
The classifications considered in this chapter are those of
receipts and disbursements. The most important element in
receipts is income, while the most important element in disbursements is consumption expenditure : the principal difficulties
of classification are found in these items. But on each side of
the account are balancing items, and it is essential to provide
information t h a t can be classified so t h a t the two sides of the
account will balance.
RECEIPTS

Receipts cover all income received during the year as well
as all sources of cash from which expenditure may be defrayed,
cash derived from increase of liabilities (loans, debts) or from
decrease of assets (conversion of assets into cash).
Income
Income is classified in two general divisions : income from
earnings of various members of the family, and income from
other sources t h a n earnings. Income from earnings is also classified in two divisions : earnings of the head of the household,
and earnings of other members of the family. In general, it is
of interest to note whether the earnings of other members of the
family are the wife's or the children's earnings. In order to make
the classification complete, other members of the family must
also be included, since in some families adult children or other
relatives may contribute earnings.
Obviously, such a classification involves the question of defining the family, since in classifying income and expenditure
the family must be treated as the unit. I n general, only those
persons whose income is paid into the family fund and whose
expenses are defrayed from the family fund should be counted
as part of the family. Adult children or other relatives living
with t h e family m a y be treated, for t h e purposes of the family

METHODS OF FAMILY LIVING STUDIES

living study, as boarders and lodgers, in which case the amount
they pay for board would be included as part of the income of
the family, but neither their earnings nor their expenditure would
be counted in family income or consumption expenditure.
Other income may be divided into various groups :
Interest and dividends ; rents ; net profit derived from owned
home ; pensions, annuities and insurance benefits ; gifts ; assistance
and relief ; profits from boarders and lodgers ; value of home
produce ; profit from business or farm ; and other sources.
Each of these items is more or less well defined, and the classification is useful in showing the amount received from the different
sources. For the most part the families included in the family
living studies of urban wage earners do not have any large income
other t h a n earnings ; but in studies of other groups the income
received from sources other t h a n earnings is of considerable
interest and importance. The definition of these items offers
certain difficulties in four cases. The " net profit derived from
the owned home " is designed to include as part of the income the
equivalent return to the family from the ownership of the home.
This is useful for the purpose of making a fair comparison between
families who pay rent and families who own their own homes.
If an estimated net profit from the owned home is placed on the
income side, an estimated rental value of the home should appear
as " rent " on the expenditure side. On this point different
methods are followed in different studies. 1
The second point for which figures need to be carefully defined
are the net profits from boarders and lodgers. I n this case, the
net profit is the gross income less expenses. I n the case of lodgers,
the money received as room rent can be treated simply as net
income, unless additional expenditure has been incurred because
of the presence of lodgers, such as extra rent for a larger house,
higher bills for heating and lighting, extra payments for service,
etc. I n such cases the extra expenditure incurred by reason of
the lodgers should be deducted. I n the case of boarders, the
extra cost of food and service involved in furnishing board should
be deducted from the gross income. I n either case the item entered
as income should be the net profit.
The third item which requires definition is the value of home
produce. I n this case, the food produced and consumed b y the
family should be entered at its value as an item of income as
1

Cf. Methods of Family

Living Studies,

op. cit., p . 56.

CLASSIFICATION

OF RECEIPTS

AND

DISBURSEMENTS

well as an item of consumption expenditure. I t may be valued
either at its selling value — what it would bring if sold — or at
its purchase value — what it would cost if purchased — at the
local market. I n either case corresponding items should appear
on both sides of the account.
I n the case of studies concerning business men or farmers,
a special rubric " profits from business or farm " should be provided. I n entering such profits, the gross income from the operation of the business or the farm less the expenses of operation
should be calculated ; this may involve an intricate accounting
procedure for these special cases. For example, in most family
living studies of urban wage earners or of salaried employees
this problem is not important.
Other sources may include such items as windfalls, gambling
winnings, etc.
Other Receipts
The term " other receipts " is used to denote sources other
than income from which family living may be defrayed. These
other receipts are, in general, derived from a decrease in assets
or an increase in liabilities. Assets may be converted into cash
and used for consumption expenditure ; money may be borrowed
by mortgage or personal notes ; the cash balance on hand at the
beginning of the year may be depleted during the year, etc. An
increase in liabilities may result from advances on wages or salaries,
from borrowings, or from goods purchased on credit.
I t is obvious t h a t for any careful accounting of family receipts
and disbursements, a complete check must include all sources
from which cash or income is derived as well as the directions in
which expenditure or investments are made. If apparent expenditure exceeds apparent income, then the question must be
raised whether there has been any change in the assets situation
or any increase in borrowings or of goods purchased on credit.
The treatment of this problem differs greatly in different family
living studies. The most complete and most satisfactory investigations pay close attention to the balance of accounts.
I n the special case of goods purchased on credit, t h a t is, goods
purchased without paying the full price the day when the purchase
is made, different methods are used to adjust the family accounts.
I n some enquiries the full purchase price, t h a t is, the cash payment
plus the unpaid balance due, is entered as expenditure. The
amounts still due represent credit by the seller to the buyer and

METHODS OF FAMILY LIVING STUDIES

are treated in effect as a loan. The most satisfactory method
is to enter as book credits, under " other receipts ", the amounts
still due at the end of the enquiry, a procedure which is adopted,
for example, in family living studies in Czechoslovakia, Finland,
Germany and Japan.
I n other studies, only the sums actually paid during the period
of the study for goods purchased on the instalment plan or on
credit are treated as expenditure. This method was followed
in the Swedish and Danish enquiries, and the amounts still due
on such purchases were disregarded. I n other investigations, for
example, those of Brazil, the expenditure includes the full value
of the articles purchased on credit, but no information is given
on the question whether the credit item figures in " other receipts " .
The classification of receipts may thus be summarised as
follows :
Receipts.
Income :
1. Earnings :
Head of household.
Wife.
Children.
2. Other income :
Interest on dividends.
Rents.
Net profit from owned home.
Pensions, annuities and insurance benefits.
Gifts, assistance and relief.
Profits from boarders and lodgers.
Value of home produce.
Profits from business or farm.
Other sources.
Other receipts :
1. Decrease in assets.
2. Increase in liabilities :
Advances on wages and salaries.
Borrowings.
Goods purchased on credit.
DISBURSEMENTS, OUTGO, CONSUMPTION

EXPENDITURE

As already suggested, the data and analysis of consumption
expenditure are the main purpose of family living studies. These
represent, in money terms and in terms of the goods and services

CLASSIFICATION

OF RECEIPTS

AND

DISBURSEMENTS

purchased, the level of living which the family enjoys. Apart
from consumption expenditure, which will be discussed below,
certain items of disbursement and outgo must be considered which
do not relate specifically to the level of living but which, nevertheless, are items of larger or smaller importance in the family budget.
Among these items of outgo, not included in consumption expenditure per se, are income tax, payment of interest on personal
debts, and similar items which do not increase assets or decrease
liabilities, and which do not represent an element in the standard
of living.
Personal income tax means a reduction of the income available
for living and should not be included in consumption expenditure.
Since it has to be treated separately, the best plan is to place
it as an item of outgo separate from consumption expenditure.
In the same way, interest on personal debts reduces the amount
available for current consumption expenditure ; the debts may,
for example, relate to items incurred for the maintenance of the
level of living in some previous year. They do not, therefore,
properly belong with the consumption expenditure, which measures
the current level of living.
Apart from these special items of outgo, certain other items
of disbursement from family funds must be considered, including,
particularly, repayment of debts, savings and the investment of
surplus funds. These also do not constitute an element in the
current level of living. (They may represent repayment of debts
incurred to maintain the standard of living in a previous year,
or be saved up to cover the possibility of use in maintaining the
standard in a future period.) These disbursements — savings
or repayment of debts or the investment of surplus cash —
constitute an increase of assets or a decrease of liabilities and
thus correspond, on the disbursement side, to the items representing
a decrease of assets or an increase of liabilities on the receipts side
of the account. Thus, each side of the income statement carries
items which balance the account. If outgo exceeds income, the
receipts side shows a net item of decrease in assets or increase
in liabilities ; if income exceeds outgo, the disbursements side
shows a net item of increase in assets or decrease in liabilities.
In the case of a single family, there would be only one such net
entry on one side or the other. I n the case of a group of families in the family living survey, it may be expected t h a t some
families will show net savings and other families will show net
excess of expenditure over income, so t h a t in the group both n e t

METHODS OF FAMILY LIVING STUDIES

increases in savings and net increases in liabilities will be found.
Apart from this, there is the question of errors and omissions
on one side or the other which interfere with the balancing of
the accounts. If these throw the balances too far out of line and
the errors cannot be corrected, the usual practice is to discard
the particular family budget from the tabulation. From this
point of view, therefore, the balancing of the budget is a test
which reveals important errors and omissions.
Consumption

Expenditure

The classification of consumption expenditure is of great
importance in measuring and throwing light upon the level of
living. I n general, consumption expenditure may be analysed
into four principal groups — food, housing, clothing and miscellaneous. The first three of these correspond to the principal vital
needs of nutrition, shelter, and clothing, while the last mentioned
includes a series of items many of which are characteristic of a
high standard of living. The principle of classification is basically
the functional one, but the classification helps also in making
possible a differentiation according to the levels of living.
Food.
I n general, food should be interpreted to include beverages,
but to exclude tobacco. 1 The details of the method for classifying
foodstuffs so as to throw most light upon their significance will
be discussed in the next chapter, which is devoted to food and
dietary surveys.
Housing.
So far as the group of housing or shelter is concerned, four
subgroups are provided : rent proper ; fuel and light and similar
services ; expenditure on furniture, household appliances, etc. ;
and household operation, including repairs and upkeep.
So far as rent proper is concerned, the desirable definition is
to exclude all kinds of services, heat and light, etc., and to limit
the rent item to the amount paid for the bare premises. I n practice,
however, the actual monthly rent payments are usually entered,
which may be either the amount paid for the bare premises or
1
I n some family living studies, tobacco is included with food. But tobacco
cannot be considered as possessing any nutritive value and, therefore, does n o t
properly belong in this category.

CLASSIFICATION

RECEIPTS

AND

DISBURSEMENTS

may include charges for heat and light and even, if the premises
are rented furnished, payments for the use of furniture. On the
whole, the most satisfactory rule would seem to be to limit rent
to the payment for unfurnished and unheated premises and to
place payments for other items under corresponding rubrics.
Fuel and light and similar services connected with the dwelling
should be distinguished as a group. The cost of refrigeration can
best be included here, since electricity or gas used for refrigeration can scarcely be isolated in the accounts. (Fuel used for
cooking purposes should, if possible, be given with food expenditure
but, in the absence of details, in practice it is usually included
with other fuel.)
Expenditure on furniture, household appliances, etc., may best
be included under the concept of housing in a broad definition
to cover the function of shelter. I n most cases, family living
studies include here only current expenditure on these items, it
being assumed t h a t this current expenditure corresponds roughly
to an amortisation or depreciation of the inventory of furniture,
furnishings and equipment. On an average, it may be expected
that a family when first constituted will purchase furniture and
equipment to last over a long period. If the family living study
is based on a good sample, it might be assumed t h a t new families
would be represented in normal proportions and, if so, the total
amount paid for furniture for all families might be assumed to
represent the normal expenditure on furniture for the group of
families as a whole. An alternative arrangement might be to
regard each family as having a capital investment of furniture
and to use for this item an estimated annual value, including
an allowance for interest on the investment, depreciation, etc.,
and to treat purchases of new furniture as capital investments.
The importance of the item, however, does not warrant this amount
of accounting trouble.
The fourth group of housing expenditures includes upkeep
and operation. This is sometimes, if not usually, included with
the preceding entry.
Clothing.
The third major group is clothing. No special problem is
involved here. The principal question, whether to include here
the cleaning and pressing of clothing, should be decided in favour
of its inclusion, as the most logical arrangement.

METHODS OF FAMILY LIVING STUDIES

Miscellaneous.
All other items of consumption expenditure are placed under
the heading " miscellaneous ". These items, however, form such a
large proportion of the total consumption expenditure t h a t they
need to be subdivided into appropriate groups. The following
more or less obvious subdivisions are proposed.
Medical care. Include here all expenditure for doctors, dentists,
hospital fees, and medical or sanitary articles. Contributions to
sickness insurance schemes, though they might logically, so far
as they insure against doctors' bills, etc., be placed here, should
properly be grouped with the insurance items, which insure
principally against loss of income.
Personal care. This includes barbers and hairdressers, toilet
soap, and all other items of personal care.
Insurance.
This group includes contributions to obligatory
or voluntary schemes of insurance against unemployment, sickness
and accidents. On the other hand, contributions to life or endowment insurance should properly be regarded principally as savings.
The same is true of contributions to old-age pensions. I t should
be noted t h a t the contributions to general social insurance schemes
will usually be deducted by employers before the receipt of wages
by the employees. There may be a tendency, therefore, for these
items to be omitted in the statement of income if the amounts
received are reported instead of the amounts earned.
Contributions, etc. Trade union dues and church contributions
belong to this category.
Education and reading. In this group belongs expenditure
on school fees and supplies, books and newspapers and similar
items.
Recreation. Expenditure on motion picture shows, theatres,
concerts, radios, musical instruments, betting, sports goods, toys,
pets, cameras and similar items belongs here.
Tobacco. Expenditure on tobacco should be given as an
independent item under " miscellaneous ". To include it in the
food group, as is done in certain countries on the ground of an
analogy with food ministering to taste, cannot be justified from
the point of view of nutrient values.
Transport and travelling.
This group should comprise all
expenditure on travelling. Cost of bicycles, motor cycles and

CLASSIFICATION

OF BECEIPTS

AND

DISBURSEMENTS

automobiles should, in principle, be calculated on a yearly basis
including, on the one hand, the cost of interest and depreciation
and, on the other hand, costs of operation. However, in most
family living studies only actual payments during the period for
automobiles, motor cycles and bicycles are included, i.e., purchase price paid, taxes and licences, all forms of automobile insurance, and repairs and fuel. A separate subgroup for automobiles
may be recommended if any considerable proportion of families
in the survey own their own motor cars.
Gifts and charities. Current expenditure on these two items
are included here. Gifts exchanged between members of the
family should not, of course, be counted.
Servants.
The simplest procedure is to include here only
expenditure on wages for servants and daily workers, etc., and
to give expenditure on board, lodging, washing, etc., in their
respective groups. Since most enquiries relate to families which
do not keep servants, the problem is of practical importance only
in the relatively well-to-do families. Only where families keeping
servants form a numerically important group is it justified to undertake special calculations of expenditure on servants' board, lodging,
etc., and to deduct these sums from family expenditure under the
respective headings and add them to expenditure for servants
and daily workers.
Other items. As suggested above, all items which account
for more than 10 per cent, of the total miscellaneous expenditure
should be shown as special groups.
From the point of view of international comparisons, it would
be desirable to reach agreement as to the content of the different
groups, or at least as to certain rubrics, either groups or subgroups, for which comparability can be established.
The classification of disbursements is then as follows :
Disbursements.
Outgo :
1. Consumption expenditure :
Food.
Housing :
Rent.
. Fuel.
Furniture, furnishings, etc.
Household operation.

METHODS OF FAMILY LIVING STUDIES

Clothing.
Miscellaneous:
Medical care.
Personal care.
Insurance.
Contributions, etc.
Education and reading.
Recreation.
Tobacco.
Transport and travelling.
Gifts and charities.
Servants.
Other items.
2. Non-consumption expenditure :
Income tax.
Interest on personal debts.
Other disbursements :
1. Increase in assets :
Savings (life insurance).
Other investments.
2. Decrease in liabilities :
Payment of debts due to others.
The Conference may wish to incorporate these points in one
or more resolutions specifying in detail the exact classification,
or it may prefer to draw specifications in broad outlines, leaving
the details to be filled in by the different countries. On the one
hand, the more uniformity in classification and grouping can be
achieved the more significant the results will be from the point
of view of comparability ; on the other hand, it may be difficult
to obtain agreement on specific details. The following paragraphs
of resolution I cover these points :
Where possible, a balance of receipts and disbursements should be
established on a uniform basis for each family so as to permit an appraisal
of the margin of error in the individual account (I, 22).
A distinction should be made between income and other receipts
and special attention should be paid to the sources from which an excess
of outgo over income is met (I, 13 (1)).
Income should be divided into earnings and other income, and these
two groups should be shown with suitable subdivisions (I, 13 (2)).
In classifying disbursements, special attention should be paid to
savings and other investments and to repayment of debts in cases
where total income exceeds outgo (I, 14 (])).
Items of non-consumption outgo (income tax and interest on
personal debts) should be separated from the items of consumption
expenditure (I, 14 (2)).
The classification of consumption expenditure should provide for
four main groups, food, housing, clothing, and miscellaneous (I, 8 (2)).

CLASSIFICATION

OF RECEIPTS

AND

DISBURSEMENTS

The group of housing should show, separately, at least the two
subgroups, rent, and fuel and light (I, 8 (3)).
If?.
The group of miscellaneous should show, separately, each major
item or subgroup comprising as much as 10 per cent, of this group as
a whole (I, 8 (4)).
It is recommended that under miscellaneous the following subgroups be shown : medical care, personal care, insurance, contributions,
etc., education and reading, recreation, tobacco, transport and travelling, gifts and charities, servants, and other items (I, 8 (5)).
Where feasible, attention should be paid to collecting data
on the inventory of assets and liabilities.

CHAPTER V
ANALYSIS AND APPRAISAL OF RESULTS
The most important part of the family living study is the
analysis of results to make the materials gathered useful and
meaningful. This requires not only t h a t the data be collected
according to the best definitions and methods but also t h a t the
materials be tabulated and analysed to give t h e most significant
results.
Since the major purpose is t o throw light upon standards
or levels of living in different groups, an analysis of the data is
essential t h a t will define the different groups and levels. For
example, the data should be classified according to family size
and composition in order to show, or to eliminate, the different
effects which fam ly size and composition have upon family
expenditure and consumption. I n pursuance of this purpose,
the data can be classified to show expenditure or consumption
per consumption unit—a device for summarising differences in
family size and composition in terms of a scale in which the
requirements of different family members are set equal to percentages of t h a t of the adult male unit.
The principal use for an economic consumption scale is to
classify families according to consumption levels or according to
the relation between income and requirements. 1 A classification
of families according to income per family is useful in a preHminary
way to group families according to the economic levels indicated
by the earnings of the chief breadwinner. But obviously, a given
money income has different meanings according to the size of
the family it must support. One purpose of an economic expenditure scale is to make possible the grouping of families according
to income per consumption unit, so as to give a clearer picture
of the amounts available in terms of relative requirements.
I n a considerable number of recent family living studies the
data are classified in this way. A considerable variety of economic
1
Cf. on this subject, R. M. WOODBTJBY : " Economic Consumption Scales and
their Uses ", Journal of the American Statistical Association, Dec. 1947.

ANALYSIS AND APPRAISAL OF RESULTS

scales is utilised. In some of the recent family living studies,
notably in those in the United States, a considerable degree of
complexity of detail is applied in the calculation of these economic
scales. The scales, in general, are based upon actual experience,
showing the relative expenditures in the country or in the different
communities on the different items of the budget, notably food,
clothing, and miscellaneous items, as they vary in families of
different sizes. The later studies have tended to give greater
importance to the younger members of the family, perhaps partly
because of the increased attention paid to such matters as vitamins
and minerals in food values, which tend, in general, to raise
the relative cost of food where there are children in the family.
Close attention to the details of the calculation of economic scales
will lead to the conclusion t h a t price differences in different communities may affect the scale values, and that price changes
affecting some items more than others may also produce slight
changes in relative scales. In principle, therefore, it would seem
difficult to conclude t h a t any single scale is theoretically valid
for all communities or all circumstances. The conclusion, therefore, seems warranted t h a t the principal use of economic scales
is to make possible the regrouping of families according to effective
economic levels measured in terms of income (or consumption
expenditure) per consumption unit. From this point of view,
the economic consumption scale to be adopted should be the one
most closely related to and adapted to the particular country
concerned. 1
One important question is whether these classifications should
be based upon income per consumption unit, or on consumption
expenditure per consumption unit. I t is clear t h a t consumption
expenditure represents a much closer approximation to the cost
of a given level of living than income. Consumption expenditure
per consumption unit, therefore, is perhaps the best index of
the actual level of living. But if it is a question of comparing
family resources with family requirements, income per consumption
unit is used as an alternative basis for classification. In practice,
a decision is usually made in each study as to which of these two
1
Any attempt to equate given amounts of income per consumption unit in
one country with a corresponding amount in another country, or to provide t h a t
the limits of the classes of income per consumption unit correspond from country
t o country, would meet with extraordinary difficulties. The same is true of classes
arranged according to income per family. I t is difficult to say t h a t a group of
families, e.g., in the United States, with incomes of $2,000 to $3,000 per family,
corresponds t o a group with incomes of £500 to £750 in the United Kingdom.

METHODS OF FAMILY LIVING STUDIES

bases of classification will be adopted. The majority of family
living studies which provide for an analysis of this type provide
a classification according to consumption expenditure per consumption unit.
On these points involving the use of a consumption scale
two principal questions are raised—whether the Conference will
wish to adopt and recommend any particular consumption scale l,
and what type of scale should be used for an economic classification. In general, no recommendation for any specific scale
of expenditure on consumption is proposed. It is recommended,
however, that each study should state the scale used and, in
addition, give the sex and age composition of the families covered
in its survey.2 If such details are available, it is possible to
convert results based on one scale to results based on a different
scale, so that differences between scales in different family living
studies can be eliminated as a source of non-comparability in the
results obtained.
As for the second point, preference should be given to the
classification according to consumption expenditure per consumption unit as a basis for separating families according to different
levels of living.3
Another classification according to economic status is often
used, namely, classification according to income per family. If
income is regarded as a clue to social position, so that income
per family (irrespective of size) is considered a better index than
income per consumption unit, the basic economic classification
may be based on income per family. In practice, most studies
do utilise this classification, with or without an additional classification according to income or consumption expenditure per
consumption unit.
Analysis according to family type and size is an important
basis of classification. In many studies the group of families
selected is limited in such a way that only families of a given
size and composition are included, e.g., families of five persons
1
The resolutions of 1926 proposed " in addition t o averages per family and
for families classified according to size, the data expressed in terms of some more
precise unit of consumption, for example the consumption of an adult male, should
be given. For reducing data for families of different sizes to terms of a common
unit, it is desirable, where suitable scales showing the relative consumption of persons of different age and sex are available, to apply one scale to food commodities
a n d the second scale to other items." No specific scale was recommended.
2
See below, Chapter VI.
3
There are many details on the different methods of calculating the scales
a n d expenditure per unit which it is not considered necessary to discuss here.

ANALYSIS AND APPRAISAL OF RESULTS

consisting of husband and wife and three children under sixteen
years of age. The advantage of this close limitation is t h a t the
results are much clearer in focus and are not subject to the large
variations which arise from differences in family size and type.
The majority of studies, however, include more than one family
type. In a number of cases, families with boarders and lodgers or
broken families are excluded from the study. Comparisons between the levels of living between families of the same size and
type make it possible to study variations in the content of living
in relation to expenditure without resorting to consumption unit
analysis. They also make possible a more direct comparison
between countries for purposes of the study of international
differences.
Various other classifications are useful from the point of view
of analysis, e.g., kind of work, whether manual or non-manual
workers, employment, unemployment, etc.
The content of living, or the goods and services which are
purchased to maintain the family at a given level, is also a significant basis for analysis. For this purpose, as already discussed,
the expenditure can be classified into major groups and percentages
of expenditure on the different groups shown, so t h a t in different
countries and different regions variations in the percentage spent
on food, for example, can be shown. I n particular, it is found
t h a t a high proportion of expenditure on food is associated with
low standards of living and a large percentage of expenditure on
miscellaneous items with a high standard of living. For the purpose of such comparisons, the basis upon which percentages are
calculated should, of course, be uniform. The logical basis for
this purpose is the total consumption expenditure. Income is
a less satisfactory basis, since it is subject to heavy deductions
for income tax, which varies from country to country and from
group to group in such a way as to distort the percentages spent
on a n y given item.
I t is of great importance to be able to show not merely the
amounts expended but also the items purchased. For certain
purposes of analysis, the possession of a certain item—radio,
telephone, automobile—gives an important clue to the living
habits of the family. The presence of conveniences, such as
running water, hot water and washing machines, represents an import a n t element in the comfort and efficiency of living. I n particular,
analysis carried to the point where it is possible to appraise the
level of living, t o distinguish, e.g., between families of adequate

METHODS OF FAMILY LIVING STUDIES

living levels and those where living must be characterised as
below a satisfactory level, offers interesting possibilities. This
might involve an appraisal b y experts of the mode of living as a
whole or in particular phases. So far as food is concerned, such
appraisal may be given by nutrition experts and be in terms
of the adequacy of the nutrition. This is discussed in the next
chapter.
So far as housing is concerned, a housing expert might establish
a set of criteria of adequacy of housing t h a t could be applied
to the results of a family living study. The same may be said of
clothing. As t o health, insurance, and various specific items
included in miscellaneous expenditure, an appraisal by experts
might yield interesting conclusions when placed in relation to
income or other social or economic factors.
An important aspect of the analysis of family living studies
is the analysis of variability. I n the publication of results, not
only averages should be shown but also standard deviations and
the distributions around the mean. This is extremely important
in the appraisal of the errors to be attached t o t h e different results.
Comparatively little has been done up to the present in applying
statistical techniques to the analysis of results of family living
studies. Emphasis should be laid also upon the desirability of
publishing in full, a t least for a small sample of families, the
detailed data obtained in the enquiries. By utilising such materials
the statistician can often derive conclusions beyond those which
are presented in the reports of the family living study itself.
With regard to the analysis of the materials b y the use of advanced
statistical methods much remains to be done. With the application of such methods to the materials, and b y making the
materials available to qualified research workers, it is possible
greatly t o extend t h e value and significance of t h e conclusions.
I n this connection reference may be made to the important work
of Professor R. G. D. Allen in this field, as well as to a number of
special analyses based upon the abundant material available in
t h e wide family living studies made b y the United States.
The following recommendations are made to cover these
points :
Each study should state the consumption scales utilised in its analyses, with full details of the method of their construction (I, 31 (1)).
Each study should publish the sex and age composition of the
survey population, in order to permit the conversion of the results from
one scale to another (I, 31 (2)).

ANALYSIS AND APPRAISAL OF RESULTS

For economic analyses details obtained from the families should
be classified according to income per family (I, 23 (a)).
In order to eliminate the effects of differences in the size and composition of families, consideration should be given to the possibility of
classifying the details obtained from the families also according to
consumption expenditure per consumption unit (I, 23 (b)).
Provision should be made for an analysis of the materials on levels
of living, to shot? variations by income for families of different size and
composition, or by income per consumption unit, for important social
and economic status groups and other significant classes (I, 23 (c)).
Consumption analysis should be provided in terms of specific goods
and services which throw significant light upon levels of living, as,
e.g., the telephone, automobile, radio, running water, etc. (1,24).
Wherever possible, an attempt should be made to interpret the data
obtained in terms of norms and standards in order to permit an appraisal
of the level of living, with especial reference to food, housing, health,
etc. (I, 28).
In analysing the results, it is desirable to show not only averages
but also distributions, so as to permit the calculation of standard
deviations and errors (I, 25).
Consideration should be given to the possibility of utilising the
materials for special studies, for example on savings, the break-even
point (that is, where income equals outgo), and the elasticity of consumption (I, 29).
Attention should be paid to the desirability of publishing data for a
selection of individual families, with details of income and consumption
expenditure, in order that they may be available for such further use
as experts can make of them for international comparisons and for
other scientific purposes (I, 32).

CHAPTER VI
FOOD AND DIETARY ANALYSES
An important section of family living studies is t h a t dealing
with food expenditure and consumption.
FOOD

EXPENDITURE

So far as food expenditure is concerned, among other important
points family living studies should show the proportion of consumption expenditure which goes on food. This is useful not
only for an analysis of the results of the study itself, but also
for international comparisons, to show differences between different
groups in various classes.
For this purpose, the definition and coverage of food items
should be uniform as well as the definition of consumption expenditure. The latter has been discussed in detail in a preceding
chapter. As to the former, the principal points to note are, first,
t h a t expenditure on tobacco should be excluded from food expenditure, since tobacco has no food value and, therefore, is not part
of food expenditure. Secondly, though alcoholic beverages belong
with food, for comparative purposes most analyses of food expenditure should exclude expenditure on alcoholic beverages. This
is advisable for two reasons : primarily, because information as to
expenditure on alcoholic beverages is apt to be inaccurate or
incomplete, and secondly, from a dietary standpoint the calorie
values of alcoholic beverages are not fully utilised in the body
and it is not clear how much effective food value they represent. 1
With these exclusions, therefore, the data on food expenditure
are relatively homogeneous and comparable.
A classification of food expenditure according to food groups
is found in many surveys and shows the direction of such expenditure. Such a classification may furnish, for example, expen1
Of. E. J. BIOWOOD : Guiding Principles for Studies on the Nutrition of Populations (Series League of Nations Publications, III. Health, 1939, III. 1, p. 1).

FOOD AND DIETARY ANALYSES

diture on cereals, meats and fish, fats and oils, dairy products
and eggs, vegetables and fruit, miscellaneous items, and meals
outside the house.
The amount of expenditure on food per consumption unit
is also of interest. Its principal value lies in comparisons either
with similar expenditure for other groups or with the minimum
expenditure necessary to maintain a satisfactory food standard.
If the minimum necessary for maintaining a satisfactory food
standard can be determined, then the fact t h a t in a given family
less than this minimum is expended—due regard being paid to
family size and composition—would be a basis for concluding
t h a t the food consumption of the family is in all probability
inadequate.
F O O D CONSUMPTION

Not only should data on food expenditure be obtained, b u t
also data on quantities of different foods consumed. The problems
of obtaining satisfactory information on this point are summarised
in the report Methods of Family Living Studies already cited,
and more complete details are given in Dr. E . J . Bigwood's Guiding
Principles for Studies on the Nutrition of Populations, mentioned
above. The Nutrition Division of the Food and Agriculture
Organisation of the United Nations is publishing a report, Dietary
Surveys—Their Methods and Interpretations.
The difficulties in the way of obtaining satisfactory data on
food consumption and of analysing the data are very considerable.
The technical problem should be placed in the hands of a dietary
expert. Methods in the collection of the data vary between
estimating quantities of various foods consumed during an immediately preceding period, recording quantities consumed derived
from quantities purchased and changes in food inventory, and
weighing meticulously the food eaten by families as a whole or
by individual members. The periods for which detailed studies
are undertaken are commonly less than the full period of the
survey and are often limited to a week or to three or four weeks
at different seasons of the year.
Once the data have been collected, the food consumption of
the families can be appraised in terms of the different nutrient
elements. This requires the conversion of the quantities of food,
with the use of data on food composition available in food composition tables, into corresponding values of calories, proteins, minerals—iron, calcium, phosphorus, etc.—and vitamins A, thiamin

METHODS OF FAMILY LIVING STUDIES

and riboflavin, C, D, * etc. The quantities consumed expressed
in this way can then be compared with the requirements for each
of these items as established by the nutrition experts, usually
in the form of dietary standards or recommended dietary allowances.
For the entire survey population, or for different subgroups,
or for each family, the quantities of each food element consumed
can be calculated, as a whole or on a per capita basis, and compared
with the requirements of each food element calculated in the
samé way for the group as a whole or on a per capita basis ; the
comparison gives a measurement of the adequacy of the diet in
respect of the specific nutrient element expressed as a percentage
of intake to requirements. These percentage figures offer a basis
for comparisons between famines, groups, or survey populations
in regard to the adequacy of diet. I n this operation the use of
consumption scales is a useful aid. These scales are derived
from requirement figures for persons of different sex and age in
terms of those of the adult male, and are used to estimate household or group size in terms of nutrition units. The diet of any
two families or groups, etc., can then be compared directly for
each nutrient.
I t should be emphasised t h a t separate consumption scales
must be applied to each element, calories, proteins, iron, calcium,
phosphorus, vitamin A, thiamin, riboflavin, etc., since the scale
showing relative requirements in calories for children of different
ages in terms of the adult male is not the same as a scale for iron,
etc. The advance in knowledge of nutrition has brought the
abandonment of the single calorie scale as applied to the requirements for all food elements. Both the scales and the interpretation
of the results in terms of the requirements of the adult male are
matters for the nutrition expert.
I t is scarcely the province of the Conference to adopt specific
scales for different purposes, since the development of such scales
is t h e function of nutrition speciahsts. Scales for recommended
dietary allowances for calories and specific nutrients from which
consumption scales are derived have been established. International
agreement on some of these scales has been sought and partially
realised. There has been discussion of the establishment of recommended dietary allowances on scientific bases of physiological
requirements for various areas, types of population, etc. The
Food and Agriculture Organisation and t h e World Health Organisation are at present working on this project. So far as agree-

FOOD AND DIETARY ANALYSES

ment or a consensus on specific scales or recommendations has
been reached, the nutrition expert associated with the study
could adopt them. Scales of recommended allowances, and hence
consumption scales, are subject with increasing knowledge to
alteration, and the interpretation of the requirements of the unit
—the adult male—is also subject to change. A change in the
amount regarded as the requirement of t h e ' adult male by the
nutrition experts may involve a change in the whole scale ; alternatively, the scale may be maintained, provided t h a t the same
percentage of increase in requirements is applied throughout to
persons of all ages. The interpretation of an average consumption
of a given vitamin per adult male unit is thus a highly technical
matter for the nutrition expert.
In view of the fact that recommended dietary allowances,
and also consumption scales derived from them, vary from study
to study and from time to time, it is essential t h a t the basic data
of the survey should be reported in full, i.e., the food consumption
and nutrient intake levels fer capita and the age-sex composition
of the group surveyed. In addition, the requirements used for
assessing the adequacy of the diets and for establishing consumption scales for reporting intake levels on a nutrition unit basis
must be included. With this information, the results of a study
can be converted from one scale to another so t h a t the results
of one study can be compared with those of another on the basis
of the same scale and interpreted on the basis of the same requirements.
For purposes of comparing food consumption in different
surveys it is useful to classify the foodstuffs. For this purpose
a classification of foodstuffs has been drawn up by the International
Labour Office designed to facilitate international comparisons of
food expenditure and consumption 1 for use in family living studies.
1
The classification given in the text is intended as an expansible one, to be
developed wherever, the foods consumed in a given country require further subdivisions. The division and subdivisions are intended to be as significant as possible
of the food values concerned. A classification, for example, of amounts expended
in a group of families as a whole on different foods and food groups parallel to data
on quantities purchased analysed in detail for nutritive values for a sample group
of families might furnish a basis for useful conclusions. The classification in the
text is not intended to be a substitute for the detailed classification required for
dietary analyses.
A more detailed classification designed especially for international comparisons
of food consumption for use in dietary analyses is being prepared by the Food
and Agriculture Organisation.
See also Food Composition Tables for International Use (prepared by Charlotte
CHATMELD) (F.A.O. N O . 4 8 / T c h / 8 , 6 J u l y 1948.)

METHODS OF FAMILY LIVING STUDIES

I. Cereals :
(1) Bread:
(a) Wheat.
(b) Rye.
(c) Maize.
(d) Other.
(2) Cakes, biscuits, pastry.
(3) Flour :
(a) Wheat flour.
(b) Rye flour.
(c) Other flour.
(4) Other cereals :
(a) Macaroni.
(b) Rice.
(c) All other.
II. Meat, fish, etc. :
(1) Beef, fresh (including frozen and chilled).
(2) Pork, fresh.
(3) Mutton, fresh.
(4) Veal, fresh.
(5) Ham and bacon.
(6) Sausages, pork butcher's meat.
(7) Preserved meat (salted, smoked, canned or dried).
(8) Fish and other sea food :
(a) Fresh fish.
(b) Salted, smoked, canned, dried fish.
(c) Oysters, lobsters, crabs, etc.
(9) Poultry, game, other meat, etc. :
(a) Poultry.
(b) Game.
(c) Minced meat.
(d) Horse.
(e) Liver, kidney, sweetbread, etc.
(f) Other (snails, frogs, etc.).
III. Fats, margarine, etc. (except butter) :
(1) Animal fats.
(2) Margarine.
(3) Vegetable fats and oils.
IV. Milk products and eggs :
(1) Whole milk :
(a) Cows'.
(b) Other.
(2) Skimmed milk.
(3) Other milk :
(a) Condensed milk.
(b) Dried milk.
(c) Buttermilk, whey.
(4) Cream.

FOOD AND DIETARY ANALYSES

(5) Cheese.
(6) Butter.
(7) Eggs.
Vegetables and fruits :
(1) Dried peas, beans, lentils, etc.
(2) Potatoes.
(3) Fresh vegetables :
(i) Leafy, green, and yellow vegetables.
(ii) Root (other than yellow) vegetables.
(iii) Tomatoes (including tomato juice).
(iv) Other fresh vegetables.
(4) Fresh fruit :
(i) Citrus fruit (including citrus fruit juice).
(ii) Other fruit.
(5) Preserved vegetables and fruit.
(6) Nuts.
VI. Miscellaneous foods :
(1) Sugar, etc. :
(a) Sugar.
(b) Syrup.
(c) Honey.
(d) Jams, marmalade.
(e) Sweets.
(2) Condiments :
(a) Salt.
(b) Other condiments.
(3) Tea.
Coffee :
(i) Beans.
(ii) Substitutes.
(5) Cocoa and chocolate.
(6) Other non-alcoholic beverages (mineral waters, sweet cider
etc.).
VII. Alcoholic beverages :
(1) Wine.
(2) Beer.
(3) Distilled beverages.
(4) Other.
VIII. Other and unclassifiable foods.
IX. Meals taken outside the home.
I t should be noted t h a t the purpose of the classification is to
group together the foods according to their general character,
and, in particular — so far as a classification can do this — according to the kind of food values supplied. Full light on food values,
of course, can be thrown only by analyses in terms of calories

METHODS OF FAMILY LIVING STUDIES

and the specific elements in the diet. Apart from that, however,
a classification to show the chief sources of calories—cereals
and potatoes, sugar—and sources of minerals and vitamins,
in particular milk, milk products and eggs, fruit, vegetables, etc.,
offers a useful indication of the quality or adequacy of the diet.
Among the uses to which these materials can be put is to
show the relation between expenditure on foods or on specific
items of food, or the quantities consumed, and the level of income
or consumption expenditure. The analysis may show, for example,
t h a t certain foodstuffs are consumed in larger amounts as the
income of the family increases.
On these points the following recommendations are m a d e :
The survey should be designed to throw as much light as possible
on dietary and nutrition problems (I, 26 (1)).
I t is recommended that, as resources and facilities permit, a dietary
and nutrition survey, with a dietary and nutrition expert as consultant,
be made in order to throw light upon the character of food consumption
and the adequacy of nutrition (I, 26 (2)).
It is recommended that the diet be assessed by comparing intake
with requirements and use be made of the technique of consumption
scales for each of the different nutritional elements (I, 26 (3)).
I n publishing the results each study should give the basic
d a t a on food consumption and intake levels, and the age and
sex composition of its survey population as well as the requirement
levels and consumption scales utilised, in order to permit recalculation of the results concerning nutrition in terms comparable
with the results of other studies (I, 26 (5), 31 (1) and (2) )}
The consumption scales used should be those adopted by an international organisation, where such action has been taken by the appropriate international bodies concerned with nutrition (I, 26 (4)).
Provision should be made for the appropriate analysis of food data
in terms of family composition, levels of living, income or consumption
expenditure levels and other social and economic conditions (I, 27 (1)).
I t is recommended that food expenditure and consumption should
be classified according to the standard international list of foodstuffs
(I, 27 (2)).
Attention should be paid not only to average results, but to the
variations as between different families or groups, so as to provide a
complete picture not only of average adequacy of diets, but also of the
proportion of families with specific deficiencies (I, 27 (3)).

This is a summary of the paragraphs referred to, not a transcript.

CHAPTER VII
SPECIAL PROBLEMS OF FAMILY LIVING
STUDIES m UNDERDEVELOPED TERRITORIES
This subject was raised in a resolution adopted a t the Sixth
International Conference of Labour Statisticians, which suggested
that, in considering this question, t h e special problems of taking
family living studies in such territories might be referred to a
special committee of qualified experts.
Comparatively little has been done in family living studies
of this type. 1 Among the best of them is the extensive study
of living conditions of coolie labour in Netherlands Indonesia.
Such studies present special difficulties. Most of them are concerned with rural living, sometimes under such special conditions
as on plantations ; the detail applicable to urban family living
studies does not fit these conditions. Special arrangements may
exist for food purchases or food preparation. There may be
special arrangements for housing. The expenditure on miscellaneous items, recreation, etc., may be peculiar to the locality.
I n planning such surveys, it is necessary to take these special
conditions into account. Not only must the schedule or, the
account book be adapted to these special conditions, b u t also
practically all the details of the family living study plan must be
reviewed and revised in the light of these problems. The selection
of individuals or families who are to be included in the study
constitutes another specific problem.
The first question for consideration by the Conference is whether
resolutions might now be adopted b y the Conference which would
present a consensus of opinion about the way to conduct such
studies. I n the absence of such a consensus, the Conference
might prefer to place the whole subject in the hands of a special
committee of qualified experts to prepare a detailed report on
the question with recommendations to be considered by a sub1

Cf. for example, U.N. DEPARTMENT o r SOCIAL A I T A I R S : Miscellaneous

Ana-

lyses of Field Enquiries into the Living Conditions of Selected Social Groups (Miscellany N o . 1, July 1948 ; Miscellany No. 2, F e b . 1949).

METHODS OP FAMILY LIVING STUDIES

sequent conference. If such a committee were established, it
might provide the means of furnishing technical assistance and
advice concerning such studies and in this way contribute to
improving the technique and methods of this type of study.
As to scope, the technical committee would concern itself primarily with taking family living studies in tropical countries or less
well developed areas, or in countries where no previous family
living study had been undertaken.
If such a committee were set up, its personnel should include
persons who have dealt with such studies in tropical areas, together with an expert or experts in family living studies in general
and an expert in sampling methods. With such a committee,
progress in developing suitable techniques might be made more
rapidly t h a n if each country when it faces the question of conducting such a study prepares its plan without the benefit of the
experience of other countries. With a view to the more rapid
development of satisfactory studies in such areas, it is therefore
recommended t h a t the Conference should adopt a resolution in
this sense.
The resolution might read as follows :
The Seventh International Conference of Labour Statisticians...
requests the Governing Body of the International Labour Office to
set up a special committee of qualified technical experts to study the
problem of making family living studies in less well developed areas
with a view to developing plans for such studies and creating a body
of experience which could be drawn upon to assist countries to carry
through such plans, with due regard to the special problems of sampling,
and the particular objectives sought in these surveys (III).

CHAPTER VIII
FARM FAMILY LIVING SURVEYS AND SPECIAL
INVESTIGATIONS
The present chapter deals with the special problems concerned
with farm family living surveys and surveys of special groups
of the population.
FARM FAMILY LIVING SURVEYS

Though previous reports have dealt mainly with family living
studies of wage earners in urban areas, it is clear that family
living studies should not be limited to these, but should also
provide for similar studies for rural conditions. In fact, a number
of family living studies do include rural as well as urban areas,
and a considerable number of surveys are devoted exclusively
to farm families.1
In taking farm family living surveys, two special problems
must be provided for. One problem is the determination of
income in the case where the family lives on and operates the
farm ; the other is concerned with how to evaluate the food
produced and consumed by the farm family. The latter question,
of course, applies also to any family which produces a part of
its food supply from its own small garden.
The general questions of sampling and the selection of families
apply as much to farm family living studies as to those of urban
areas. The problems of sampling, however, are perhaps more
intricate. On the one hand, the problem is to select a representative sample of families of urban workers employed in different
industries and occupations ; on the other hand, the farm families
must be selected to represent the different kinds of farms operating
under different conditions of soil, the kind of farming practised,
the different proportions of woodland or of grazing land, and of
1
See, for example, " An International Survey of Recent Family Living Studies :
I V (Families of Farmers a n d Agricultural Workers) ", International Labour Review.
Vol. X L I I I , No. 4, April 1941, p . 474.

METHODS OE FAMILY LIVING STUDIES

different kinds of crops. The proximity of t h e farm t o an urban
market may help to determine the character of the farm operation,
whether for example gardening, poultry raising, milk supply, etc.
Each of these different kinds of land utilisation may need to be
taken into account in the sampling operation, and the different
conditions in these different types of farm may affect the problems of farm accounts and of food consumption and valuation.
So far as farm accounts are concerned, the basic requirement
for farm family living studies is to be able to determine the net
income received by the farm family. Where the farm family
owns or operates the farm business, it is essential t h a t the information be sufficiently detailed to determine net profit. In general,
the usual practice is to consider the total profit from farm operations as the income of the farm family, irrespective of whether
it should be considered as remuneration for the labour of the
head of the family or of members of his family or as return on
capital invested. The important point is the net income the farm
operator receives. But it is essential t h a t all expenses incidental
to the operation of the farm should be deducted from total receipts,
so as to leave the net profit. Among these expenses, for example,
would be purchase of equipment, hire of farm labour, interest
paid on capital borrowed for farm operations, purchase of fertilisers, seeds and all the numerous expenses incidental to the
operation of the farm. I t is essential t h a t adequate accounting
of these expenses be made, since otherwise the total income would
be in excess of the amount available for family maintenance.
I t is equally clear t h a t attention must be paid to changes in
inventories and in the purchase or sale of farm assets, if a true
picture of farm income is to be obtained.
An alternative method would be to consider solely the farm
family consumption expenditure and renounce the task of determining the farm family income. This is, however, not an acceptable solution if it is desired to be able to compare incomes of farm
owners and operators with those of farm labourers or with those
of urban workers.
I t follows t h a t family living studies in farming areas will
distinguish, in general, two classes : families of farm labourers,
and families of farm owners or farm tenants. I n the case of the
former, income is constituted principally by wages, while in the
case of the latter, farm profits must be taken into the account.
In practice, farm family living studies commonly cover both
groups, for the reason t h a t in effect the conditions of living of

FARM FAMILY LIVING AND SPECIAL GROUP SURVEYS

the farm tenants, and even of farm owners on marginal land, is
frequently on the same level as, or but little higher than, that
of farm labourers. This point of course is one to be determined
by the results of the investigation.
In certain countries special groups are found, for example,
in Sweden, where an arrangement exists by which a farmer is
hired for the year or the season and given a house and a plot of
ground as part of his remuneration. The variety of leases according to which the tenant farmer will share profits with the owner
of the land is an important point with reference to any survey
of agricultural conditions and of the conditions of family living
in rural areas. In the present connection, it is important to be
able to show the conditions of living of families of these different
types in the different areas.
The second important special problem in family living studies
in farm areas is the determination of the amounts and value of
the food produced on the farm and consumed by the family.1
This is important from the point of view of comparing the results
of family living studies of rural with those of urban families.
In rural areas a considerable proportion of the food supply may
be actually grown on the land belonging to or rented by a farmer
or provided for the use of the farm labourer. Different methods
for accounting for this are followed in different surveys. In
principle, the value of the food consumed should be entered on
both sides of the family account, since it constitutes both income
in kind and expenditure in kind. The question of the valuation
of food has received a considerable amount of attention. In
general, two methods are followed : (1) to enter the farm value
or the estimated price which might be obtained by selling the
food at the farm ; (2) to enter the value at the price which the
farmer would have to pay if he purchased the food at the local
market. Either method is defensible. In principle, the latter
is preferable, since the purchase value corresponds more closely
to the situation in comparable urban families.2

1
The same arguments apply also to other elements in consumption, such as
fuel, housing, transport, electricity, telephones, etc., which are derived from t h e
farm or involve a separation between farm accounts and the accounts of the family
for consumption expenditure : food is singled out for discussion as the principal
a n d most general element.
8
Any detailed comparisons between local rural prices and urban prices must
take into account special local conditions, such as nearness to urban markets,
tourist markets, etc., as affecting prices of local food products.

METHODS OF FAMILY LIVING STUDIES

The following paragraphs of resolution I cover these points :
In surveys of farm living conditions, every effort should be made
to obtain the income and consumption pattern of the farm families,
that is, distinguishing the income and the production for home consumption of the farm family from the receipts and expenditure of the
farm operations (I, 4 (2)).
In accounting for food produced and consumed by the farm family
in farm family living studies, the value of such food should be estimated
and entered on both sides of the account, preferably according to the
cost of items in the local market (I, 4 (3)).
SPECIAL INVESTIGATIONS

Among the special types of family living surveys which should
be considered are surveys of family living conditions among the
unemployed 1 , among workers experiencing a change in income
level, or among workers in a particular social or race group. Other
special groups can easily be cited, such as persons engaged in
a particular occupation or industry, where special circumstances
are sufficiently important to be worth studying in relation to
income and standards of living.
I n the first place, any special problems of sampling must be
considered. The question is whether the population groups under
study set any special problem for the selection of a true random
sample of families for study. This is, of course, for the sampling
expert attached to the study to determine.
The special question which needs to be considered in the case
of these special investigations is whether the character of the
groups is such as t o affect the collection or analysis of information
on certain important aspects of the survey. For example, a survey
of family living among the families of the unemployed encounters
special problems in respect of incomes. The incomes of such families
are low and irregular, and include in many cases incomes drawn
from relief or from unemployment insurance payments. The
method of selection of the famines as determined by the sampling
procedure must be such as to give a representative sample for
the purposes of the survey. The amount and character of the
income of these families are entirely different from those of the
families of employed workers.
One point of interest of such a survey is to show the effect
1
See, for example, " An International Survey of Recent Family Living
Studies : I I I (Families of Unemployed Workers) ", International Labour Beview,
Vol. X L I , No. 6, J u n e 1940, p. 645.

FARM FAMILY LIVING AND SPECIAL GROUP SURVEYS

of loss of income upon the consumption pattern. I t is of special
interest to be able to compare present with previous income
received b y the family so as to be able to throw light upon the
changes in expenditure both as to actual amount and as t o percentages. Of particular interest is the way the elasticity of the
different items of expenditure varies with the loss of income.
Certain items, for example, such as rent, may show little reduction. Expenditure on food may show a considerable reduction,
but not below a certain minimum. Expenditure on clothing
may show sharp reductions or be for practical purposes postponed.
Expenditure on miscellaneous items may similarly show postponement. Such a survey of family living conditions in these
families may therefore offer particularly interesting materials for
analysis.
The Conference may wish to adopt the following recommendation :
From time to time, as resources and facilities permit, special studies
of family living among special groups of the population should be made,
such as, for example, the families of the unemployed, families living
under conditions of varying income or on relief, in order to throw light
upon the problem of maintaining standards of living in these circumstances (I, 15).

CHAPTER IX
PROPOSED RESOLUTIONS
I n the present chapter, the various recommendations proposed
in the different chapters are brought together. Resolution I
includes the elements pertaining to statistical standards for taking
family living studies, supplementing the resolution adopted in
1926. I n addition, two resolutions addressed to the Governing
Body of the Office touching the subject of future work in this
field are given separately.

I
The Seventh International Conference of Labour Statisticians,
Having been convened by the Governing Body of the International Labour Office, and having met in Geneva from 26 September to 8 October 1949,
Recognising the importance of comprehensive studies of
family living conditions, not only as a basis for cost-of-living
index numbers, but also as furnishing basic information on levels
and standards of living and for changes which are taking place
in them, information which is required both for purposes of policy
making and for appraisal of policy with especial reference to
raising standards of living, and
Considering t h a t in order to improve the statistical methods
of each country as well as to permit their international comparability, international standards for the taking of studies of
family living conditions, based on the best methods which have
been established by the experience of the different countries,
should be set up b y international agreement as a general objective
for the development of statistics on these topics,
Adopts this
day of October 1949 the
following resolution :
GENERAL OBJECTIVES

1. The main purposes of the study of family living conditions are to provide—
(a) the pattern of expenditure for use as a basis for the weighting
of price indices ; and

PROPOSED RESOLUTIONS

(b)

as full information as possible with regard to levels and
standards of living, including data permitting the analysis
of income, expenditure, and savings in relation to them.
SCOPE

2. Family living studies should, in principle, obtain data on
consumption patterns and family conditions for all important
sectors of the population, including, for instance, different income
classes, different social and economic groups, such as wage earners,
salaried employees, different race and national groups, urban and
rural sectors, different family or household types and different
geographical areas.
3. Information should also be obtained both for industrially
developed regions and for regions which are undeveloped, and
for areas in which specific industries and agriculture are predominant.
4. (1) I n particular, information should be collected in
respect of farm families—families of farmers, farm tenants and
farm workers.
(2) I n surveys of farm living conditions, every effort should
be made to obtain the income and consumption pattern of the farm
families, t h a t is, distinguishing the income and the production
for home consumption of the farm family from the receipts and
expenditure of the farm operations.
(3) I n accounting for food produced and consumed by the
farm family in family living studies, the value of such food should
be estimated and entered on both sides of t h e account, preferably
according to the cost of items in the local market.
5. The selection of sectors in respect of which the information
should be segregated should be determined by each country in the
light of national requirements.
PATTERNS OE E X P E N D I T U R E

6. The scope of the enquiries relating to the pattern of expenditure will be determined b y the scope of the price indices to be
based on the results of the enquiries.
7. I n order to provide a basis for the calculation of price
indices, on the lines indicated in paragraph 5 of resolution I I of
the Sixth International Conference of Labour Statisticians, information will be required in accordance with paragraphs 2-4 (1) above.
8. (1) The particulars asked for should include if possible
all the following items : details of family income, including social
insurance contributions ; direct taxation, etc. ; size and age of
the family ; number of earners ; industry, occupation and industrial

METHODS OF FAMILY LIVING STUDIES

status of the earners of the household and whether unemployed
or in receipt of relief ; number of rooms occupied and expenditure
in as much detail as possible, on housing, fuel and light, clothing,
food consumption and expenditure (including a valuation of
garden produce and free food), household goods, miscellaneous
goods, services, and tobacco.
(2) The classification of consumption expenditure should
provide for four main groups, food, housing, clothing and miscellaneous.
(3) The group of housing should show, separately, at least
the two subgroups, rent, and fuel and light.
(4) The group of miscellaneous should show, separately, each
major item or subgroup comprising as much as 10 per cent, of
this group as a whole.
(5) I t is recommended t h a t under miscellaneous the following
subgroups should be shown : medical care, personal care, insurance, contributions, etc., education and reading, recreation,
tobacco, transport and travelling, gifts and charities, servants,
and other items.
9. The major survey should be undertaken at least once in
ten years ; preferably, the major survey should be replaced by
progressive surveys covering small samples at short intervals
designed to cover the same ground as a major survey.
10. The survey should cover a period of a year, to take into
account seasonal changes in the expenditure pattern ; in the
case of recurring items of expenditure such as food, four quarterly
dates should suffice, b u t for non-recurring items, such as clothing,
periodic information throughout the year is essential.

MEASUREMENT OF L I V I N G STANDARDS

11. Additional information which is required for the measurement of living standards should be obtained either in conjunction
with the surveys of expenditure patterns, or on a rotation plan
b y subjects or b y sectors of the community.
12. The subjects to be covered by these enquiries should
include information as to health, nature of living accommodation
and amenities, an analysis of income, extent and nature of savings,
and, unless obtained under the previous recommendation, quantities of food consumed (as a basis for dietary and nutrition
studies).
13. (1) A distinction should be made between income and
other receipts and special attention should be paid to the sources
from which an excess of outgo over income is met.
(2) Income should be divided into earnings and other income,
and these two groups should be shown with suitable subdivisions.

PROPOSED RESOLUTIONS

14. (1) I n classifying disbursements, special attention should
be paid to savings and other investments and to repayment of
debts in cases where total income exceeds outgo.
(2) Items of non-consumption outgo (income t a x and interest
on personal debts) should be separated from the items of consumption expenditure. .
15. From time to time, as resources and facilities permit,
special studies of family living among special groups of the population should be made, such as, for example, the families of the
unemployed, families living under conditions of varying income
or on relief, in order to throw light upon the problem of maintaining the standards of living in these circumstances.

P L A N OP E N Q U I R I E S

16. I n planning family living studies, consideration should
be given to the desirability of making preliminary or pilot surveys
to test methods, as well as to determine changes since previous
studies which would emphasise the need for future surveys.
17. (1) I n making family living studies the selection of families
should be made in accordance with sampling techniques as recommended in detail in the reports of the Subcommission on Statistical
Sampling of the United Nations.
(2) Provision should be made for a qualified expert on statistical sampling to whom should be referred all questions affecting
the selection of families and the application of sampling techniques
to methods and results.
(3) Where necessary and desirable, special problems in this
field may be referred, at the instance of the sampling expert
associated with the study, to a Sampling Expert Consultant
designated b y the International Labour Office or b y the United
Nations Statistical Office, or to the Subcommission on Statistical
Sampling.
18. Special studies should be made with a view to testing
the relative merits of the interview method as compared with
the account book method, with special reference to the circumstances of the enquiry and the applicability of sampling procedure.
19. Special attention should be paid to the development
of forms and schedules in advance of the study so as to be able
to obtain satisfactory and comprehensive results on the basis
of a weil tested and well thought out plan.
20. The investigators and agents charged with the collection
of materials on family living conditions should be selected from
qualified personnel and given appropriate training to fit them
for their work.

METHODS OF FAMILY LIVING STUDIES

21. The materials obtained should be subjected to careful
verification and checking in order to ensure t h a t the results are
as free from error as possible.
22. Where possible, a balance of receipts and disbursements
should be established on a uniform basis for each family so as to
permit an appraisal of the margin of error in the individual
account.

ANALYSIS AND PRESENTATION OF R E S U L T S OF E N Q U I R I E S

23. For the purposes of measuring levels and standards of
living the basis of classification should include the following,
wherever practicable :
(a)
(b)

(c)

for economic analysis details obtained from the families
should be classified according to income per family ;
in order to eliminate the effects of differences in the size
and composition of families, consideration should be given
to the possibility of classifying the details obtained from
the families also according to consumption expenditure per
consumption unit ;
provision should be made for an analysis of the materials
on levels of living, to show variations b y income for families
of different size and composition, or by income per consumption unit, for important social and economic status
groups and other significant classes.

24. Consumption analysis should be provided in terms of
specific goods and services which throw significant light upon
levels of living, as, e.g., the telephone, automobile, radio, running
water, etc.
25. I n analysing the results, it is desirable t o show not only
averages but also distributions, so as to permit the calculation
of standard deviations and errors.
26. (1) The survey should be designed to throw as much
light as possible on dietary and nutrition problems.
(2) I t is recommended that, as resources and facilities permit,
a dietary and nutrition survey, with a dietary and nutrition
expert as consultant, be made in order to throw light upon the
character of food consumption and the adequacy of nutrition.
(3) I t is recommended t h a t the diet be assessed by comparing
intake with requirements and use be made of the technique of
consumption scales for each of the different nutritional elements.
(4) The consumption scales used should be those adopted
b y an international organisation, where such action has been
taken by the appropriate international bodies concerned with
nutrition.

PROPOSED RESOLUTIONS

(5) In publishing the results each study should give the basic
data on food consumption and intake levels upon which the
conclusions as to diet are based.
27. (1) Provision should be made for the appropriate analysis
of food data in terms of family composition, levels of living,
income or consumption expenditure levels and other social and
economic conditions.
(2) I t is recommended that food expenditure and consumption
should be classified according to the standard international list
of foodstuffs.
(3) Attention should be paid not only to average results,
but to the variations as between different families or groups,
so as t o provide a complete picture not only of average adequacy
of diets, but also of the proportion of families with specific deficiencies.
28. Wherever possible, an attempt should be made to interpret the data obtained in terms of norms and standards in order
to permit an appraisal of the level of living, with especial reference
to food, housing, health, etc.
29. Consideration should be given to the possibility of utilising the materials for special studies, for example on savings, the
break-even point (that is, where income equals outgo) and the
elasticity of consumption.
30. I n publishing the results of the studies a clear exposition
should be given of the sampling methods employed and an analysis of the sampling errors and other types of error involved.
31. (1) Each study should state the consumption scales
utilised in its analysis, with full details of the method of their
construction.
(2) Each study should publish the sex and age composition
of the survey population, in order to permit the conversion of
the results from one scale to another.
32. Attention should be paid to the desirability of publishing
data for a selection of individual families, with details of income
and consumption expenditure, in order t h a t they may be available
for such further use as experts can make of them for international
comparisons and for other scientific purposes.
II
The Seventh International Conference of Labour Statisticians,
Having been convened by the Governing Body of the International Labour Office, and having met at Geneva from 26 September to 8 October 1949,
Recognising the importance of reaching agreement on the best
method of collecting data for family living studies on t h e basis
of adequate and sound tests,

METHODS OP FAMILY LIVING STUDIES

Requests the Governing Body of the International Labour
Office to instruct the Office to arrange for a study of the relative
advantages and merits of the different methods of obtaining
information in family living studies, with due regard to the problems of sampling and to the margin of error in the results.
Ill
The Seventh International Conference of Labour Statisticians,
Having been convened by the Governing Body of the International Labour Office, and having met at Geneva from 26 September to 8 October 1949,
Recognising the need for making family living studies in less
well developed areas, and the importance of adopting suitable
methods for these studies,
Requests the Governing Body of the International Labour
Office to set up a special committee of qualified technical experts
to study the problem of making family living studies in less well
developed areas with a view to developing plans for such studies
and creating a body of experience which could be drawn upon
to assist countries to carry through such plans, with due regard
to the special problems of sampling, and the particular objectives
sought in these surveys.

APPENDICES

APPENDIX I

A. RESOLUTION OF THE THIRD INTERNATIONAL CONFERENCE
OF LABOUR STATISTICIANS, GENEVA, OCTOBER 1926

METHODS OF FAMILY BUDGET ENQUIRIES

(1) In order to provide adequate information with regard to actual
standards of living, enquiries should be conducted at intervals generally
of not more than ten years into the income, expenditure, and conditions
of living of families representative of large homogeneous sections of
the population.
The families should be selected to include a sufficient number from
different income classes within the sections of the population under
consideration, and also from different industries and districts. It is
preferable to ask for less detailed information rather than to reduce
the number of families covered by the enquiry.
(2) In order that the results of the enquiries may be as complete
and comparable as possible, information should be based on daily
records of income and expenditure kept by a member of the family
for a period of twelve months.
Where it would be impracticable to obtain annual records, every
effort should be made to secure from as large a number of families as
possible budgets covering at least four periods of not less than a week,
one in each quarter, or two periods of at least a fortnight in different
seasons of the year. These records would supply adequate information
regarding items of expenditure which recur daily. In the case of items
of income and expenditure (clothing, furniture, etc.) which recur only
at considerable intervals, annual records kept by a smaller number
of families or information on which annual estimates could be based
should be sufficient.
Special forms or account books should be distributed to the families
for their use in making these records of income and expenditure. It is
desirable that competent persons should be appointed to visit the
families during the course of the enquiry and advise them regarding
the keeping of the records.
(3) The records should show the district in which the family
resides, the industry and occupation to which members of the family
1
For a discussion of the change of terminology from " family budget enquiries "
to " family living enquiries ", see Methods of Family Living Studies, op. cit., p . 2.

METHODS OF FAMILY LIVING STUDIES

who are gainfully occupied belong, and the composition of the household, including the sex of each member and the age of juvenile members.
They should show in sufficient detail the nature of the housing accommodation.
Information should be given for each important item of income
and expenditure. By income should be understood earnings in money
and kind, sums received from boarders and lodgers, from insurance
funds, pensions, and investments, together with income from allotments
or in the form of gifts. The chief items of income and expenditure
should be specified on the forms, or account books, distributed. In the
case of items of expenditure, the quantity purchased, as well as the
cost, should be recorded wherever practicable. Where a family inclndes
boarders, lodgers, or domestic servants, information should be given
to show the proportion they represent in the family consumption. In
the calculation of expenditure, the cost of maintenance of domestic
servants should be added to their money wages.
(4) In compiling the results of an enquiry, if the number of budgets
secured is adequate, separate averages should be given for important
districts and industries. Families of manual and non-manual workers
should be shown separately. Averages calculated for different income
groups should be shown.
In addition to averages per family and for families classified according to size, the data expressed in terms of some more precise unit of
consumption, for example the consumption of an adult male, should
be given. For reducing data for families of different size to terms of a
common unit, it is desirable, where suitable scales showing the relative
consumption of persons of different age and sex are available, to apply
one scale to food commodities and a second scale to other items.
The tabulated results should show the average income from different
sources, and the quantity of an expenditure on each of the chief commodities consumed. Separate information should be given for each item
of income and of expenditure which constitutes on the average not less
than 1 per cent, of total income or expenditure. Other items of interest
may be shown separately as desired. The items of expenditure should
be classified into the following groups : food, clothing, housing accommodation, fuel and light, furniture and furnishings, and miscellaneous.
It is desirable that payments of direct taxation should be given as a
separate item. I t is also desirable to determine the nutritive value of
the principal articles of food together with the cost of a thousand
calories of each commodity.
Where the sole object of an enquiry is to provide weights for the
calculation of cost-of-living index numbers, satisfactory results may be
obtained from a less detailed investigation than that indicated above.
Information regarding district, industry, composition of family, and
expenditure upon each of the several items to be included in the index
numbers only is necessary. It is desirable, however, to have, in addition,
information as to quantities.
The results will be tabulated to meet the requirements of the series
of cost-of-living index numbers which it is proposed to publish in each
country. Thus, if separate series of cost-of-living index numbers are
to be published for a number of districts or categories of workers, the
result of the family budget enquiries will be tabulated separately for
each district or category ; if a single national index is to be calculated,
the results may be shown in the form of general averages only.

APPENDICES

B. RECOMMENDATION OF THE COMMITTEE OF STATISTICAL
EXPERTS OF THE INTERNATIONAL LABOUR OFFICE,
FIRST SESSION, 12 TO 15 DECEMBER 1933
The Committee of Statistical Experts emphasises the fact that
statistical enquiries into international comparisons of the cost of living
and the purchasing power of wages cannot be undertaken without
information as to the family budgets of the working classes. In view
of the fact that in many countries no family budget enquiries have been
undertaken since a period now more or less remote, which fact substantially impedes the Office in its investigations, the Committee draws
special attention to the resolutions adopted by the Second and Third
International Conferences of Labour Statisticians, which recommended
that all countries should conduct enquiries into family budgets at
regular intervals and on a uniform, or in any case comparable, basis.
In order to increase the value of these enquiries, it is desirable that
they should be carried out simultaneously, and at periods as near to
each other as possible.

APPENDIX I I

EXTRACTS FROM THE REPORTS OF THE SUBCOMMISSION
ON STATISTICAL SAMPLING TO THE UNITED NATIONS
STATISTICAL COMMISSION
REPORT OF THE FIRST SESSION, 22 TO 27 SEPTEMBER

1947

Chapter VI : Sampling Methods in Family Budget Enquiries
51. The International Labour Office is concerned with methods
of conducting family budget surveys and has requested the advice of
the Subcommission on the matter. Many countries are considering
undertaking large-scale surveys for determining the weights for costof-living indices and for studying changes in patterns of consumption
for other purposes. A complete count is impracticable in family budgets
and similar surveys. It is of vital importance that sound sampling
methods should be used in this field.
52. In the view of the Subcommission, some very important
recommendations can be made in regard to sampling for the study of
family budgets. Such sampling should be conducted by a procedure
that will give small bias and calculable errors of sampling. In budget
studies, the definition of the universe amounts to defining areas as well
as to defining a particular type of family (i.e., by income, number of
children, type of worker, occupation, etc.), and there is great difficulty
in laying down specifications by which a particular family, dissociated
from its domicile, can be recognised as in or out of the universe.
53. Once the definitions are clarified, there are certain recognised
rules of sampling that should be followed for selecting the households
that are to be surveyed. These rules must be determined to suit the
requirements of each specific case. The theoretical basis for several
possible plans has been worked out and the statistician will apply
whichever plan appears to be most appropriate. Among sampling
methods which provide estimates of error, some are more efficient than
others. The theory of experimental design aids in devising that way
of laying out the survey which will secure the information desired at
minimum cost. The physical materials available, such as maps and
lists, and previous information will determine the optimum design.
54. The procedure of selection and estimation, if properly carried
out, will permit an evaluation of the sampling error. But evaluation
of the sampling error is impossible unless there is complete control of
the operation through all of the stages of enumeration and tabulation.
For this reason and because it is possible only to define some of the
problems and procedures by examples, it is necessary that the planning
and execution of the survey be conducted under the guidance of a
competent mathematical statistician. This point has been previously
emphasised in the general recommendations (Chapter III).

APPENDICES

55. Surveys are subject to many different types of error other
than the sampling error. The method of interpenetrating networks of
samples, as already indicated in Chapter I I I , is particularly useful for
evaluating errors arising from different sources.
56. The Subcommission, considering that the proper use of statistics is only possible with knowledge of the errors in the information,
strongly recommends that publications of family budget survey results
be accompanied by statements of the errors to which they are subject.
When this is not possible because of the lack of an appropriate sample
design for the respondents, a statement to this effect is essential.
57. The Subcommission recommends also that where it is feasible
to carry out the survey, possibly on a smaller scale, but with periodic
repetition, it is advantageous to do so. In any case it is most inadvisable to conduct a single large survey without preliminary preparation
in the form of pilot studies, through which knowledge necessary for the
design of an efficient sample can be gained, questionnaires tried out, and
interviewers trained.
58. In addition to the problem of selection of respondents, there
are other aspects of family budget surveys in which the use of sampling
is the only means by which effective decisions may be made, such as
determining the frequency of the family budget surveys. Such matters
can only be decided after properly designed and interpreted experiments
have been made, and the Subcommission considers that the making and
immediate publication of such experiments is to be strongly encouraged.
Chapter III : General Recommendations
13. Sampling can be used with great advantage in speed and
economy in tabulating and processing the material collected by a
complete enumeration or even by a sample survey. By the use of
interpenetrating networks of samples it is possible when desired to
study simultaneously variations in results arising from different interviewers, different ways of presenting the questions, different systems
of training interviewers, etc., and to eliminate such variations in the
comparisons required. Through the proper use of sampling procedures,
interpenetrating samples can be used also for the above purposes even
in a complete enumeration.
14. For these reasons, the Subcommission makes the following
recommendations :
(a) It is advisable to consider the desirability of carrying out a
sample survey in conjunction with any attempted complete census
(especially in the fields of agricultural and population enquiries) with
a view, (i) to assessing the margin of error, comparative speed, cost
and convenience of organisation, and (ii) to obtaining supplementary
information. The cost of such sample surveys will usually be relatively
very small.
(b) I t is desirable to make a sample survey instead of attempting
a complete enumeration whenever adequate funds, physical facilities,
or personnel of sufficient ability are inadequate for a complete enumeration.
(c) I t is sometimes desirable to use the same basic sampling
structure for different types of statistics. It is also sometimes possible
to collect information required in different enquiries simultaneously,
although the extent to which this can be done is limited by the demands
that can be made on the interviewers and respondents, and by the

METHODS O F FAMILY LIVING STUDIES

fact that different types of surveys may require different interviewers
or different technical qualification.
(d) It is often advantageous to carry out a series of repeated
sample surveys at short intervals along with or even instead of a complete
census at long intervals. In such a series of repeated surveys, it will
usually be possible to make appreciable improvements in the sampling
technique and thus reduce over-all costs as well as to obtain more
detailed information and information of better quality with the progress
of time.
(e) All these recommendations are subject to this most important
provision : a sample survey should be carried out only under the technical guidance of professional statisticians not only with adequate
knowledge of sampling theory but also with actual experience in sampling practice, and with the help of a properly trained field and computing
staff.
REPORT OF THE SECOND SESSION, 30 AUGUST
TO 11 SEPTEMBER 1948

Chapter V : Consideration of Sampling Methods
in Family Budget Enquiries
24. The Subcommission considered a request from the International
Labour Office regarding the application of sampling in family budget
studies. Some general recommendations were made at the first session
of the Subcommission, and more specific information was requested on
present methods. An account of such methods was presented by the
International Labour Office (E/CN3/Sub.l/5).
25. The subject was discussed at the second session in the light
of this information. In view of the complexity of the subject, and the
varying economic and social conditions in the different countries, the
Subcommission considered it was not in a position to make detailed
recommendations without considerable further investigation, which it
was unable to undertake at their second session. However, in the
course of the discussion a number of general points emerged. The
Subcommission considered that those were worth emphasising, since,
although they may be well known to the most advanced workers in the
subject, it was apparent, from the information submitted, that they
had been insufficiently taken into account in many recent surveys.
26. Any family budget enquiry which is designed to supply quantitative regional and national estimates, covering either all families or
families of a particular type, must be based on a proper random sample.
The information in E/CN3/Sub.l/5 gave evidence that methods of
selecting families or households in many surveys were such that no
attempt can have been made to fulfil this condition.
27. The Subcommission is aware that the failure to take random
samples is in many cases due to the inherent difficulty of collecting the
required information, thus leading to the selection of such families as
have promised their co-operation, often families with whom the field
investigators are already acquainted in some other connection.
28. Although such a method of selection is very unlikely to give
reliable results, the situation will not be improved simply by taking a
random sample, if this results, as it has done, in lack of response from
a large proportion of the households selected for the sample. The

APPENDICES

practice of substitution, by which a non-respondent family is replaced
by another, either drawn at random from the original frame, or selected
by judgment or specification, does not resolve the difficulty.
29. The main problem therefore appears to be that of improving
methods of collecting the information, so that surveys can be undertaken on a proper random sample with confidence that the amount
of non-response will be sufficiently small so as not seriously to invalidate
the results.
30. Improvement of methods of collecting the information can
only result from research and experiment, much of which must be
carried out in all or many of the countries concerned, since a method
which may be eminently suitable for one set of conditions may be
entirely impracticable in another.
31. The Subcommission is aware that a considerable amount of
such research has already been undertaken in certain countries. They
did not, however, have opportunity to review the research as a whole.
Subject to such review, it appeared to the Subcommission that research
might be directed along the following main lines :
(a) Comparison of the interview and account book methods.
I t may be noted here that the account book method cannot be applied
in all circumstances—it is useless, for example, where illiteracy is
widespread—and is in any circumstances likely to result in a much
greater degree of non-response, or of failure to complete the records
over the required period.
(b) Length of period for which data are sought from individual
families, and whether the period should be continuous or in separated
parts, e.g., one week in each quarter.
(c) Simplification of questionnaires, with due regard to the
information required.
(d) Subdivision of the questionnaire, some items being collected
from one set of families, others from another set. One possible method
is to submit a short list of basic questions to all families in the sample,
and to subdivide the total schedule in suitable subgroups of questions
for different sub-samples of the main sample. (This is an application
of phase sampling; see " Recommendations concerning the preparation
of reports of sample surveys ".) The set of questions must be so designed
that all the required inter-relations are covered.
(e) Improvement of the technique of interviewing. In this connection the Subcommission wishes to emphasise the importance of
endeavouring to secure the co-operation of all members of the family.
It frequently happens, for example, that the wife is ignorant of some or
all the expenditures of her husband.
(f) Tests of different classes of interviewers, and investigation of
differences between individual interviewers.
(g) The effect of inducements, such as remuneration and of propaganda, etc., on the degree of co-operation.
(h) Tests of the alternatives of asking for income as well as
expenditure, and striking a balance, or of asking for expenditure alone ;
the latter may be superior where there are motives for concealing
certain sources of income.
32. Investigation of all the above points must be mainly by means
of proper comparative tests of alternative procedures. For this purpose
the employment of the techniques developed in modern experimental
design is essential.

METHODS OF FAMILY LIVING STUDIES

33. Different methods of collecting the information, different
questionnaires, etc., cannot be tested on the same family, and the type
of experimental design required is therefore analogous to that used in
agricultural field trials, in which each plot receives one of the experimental treatments, families being here equivalent to plots. As a simple
example, alternative ways of comparing the account book method with
the questionnaire method, with or without the introduction of different
investigators, may be considered.
34. If the whole test is carried out by a single investigator pairs of
families may be selected, the two members of each pair being chosen so
as to be as alike as possible. The account book method is then assigned
at random to one member of each pair, the questionnaire method being
assigned to the other member of the pair. This is exactly analogous to
a randomised block design for two treatments, and the accuracy of the
comparison can be estimated, and its significance tested, by the analysis
of variance technique appropriate to this design.
35. This type of design can easily be extended to include more
than two alternatives. With three investigators, for instance, sets of
six families can be selected, and the six combinations of the two
methods and three investigators assigned at random to the six families
of each set. This is a simple example of pictorial design, the two factors
being methods and investigators, enabling not only the differences
between methods and between investigators to be assessed but also any
differential performance between the investigators in the different
methods ; e.g., there may be little difference between investigators
when the account book method is used, but one of the investigators
may be outstanding with the questionnaire method.
36. Different investigators may, of course, be deliberately associated with different methods, e.g., investigator A may be used on the
account book method, and investigator B on the questionnaire method,
but in this case it must be recognised that any observed difference may
be due either to investigators or methods, or a combination of both.
In general, therefore, it is best to design the test so that the effects of
investigators and of methods can be separately assessed.
37. An alternative to the formation of sets of which the members
are as alike as possible is the use of interpenetrating samples, one such
sample being assigned to each of the method-investigator combinations
which it is designed to test. This procedure (which is not in fact
formally very different from the other) has the advantage that no
preliminary information need be collected about the families. Furthermore, lack of response will not complicate the analysis of the results
to the same extent as in the randomised block method, where lack of
response in one of a set, for instance, requires the use of the " missing
plot technique " if the remainder of the results of that set are to be
utilised. On the other hand if there is full response the use of sets will
result in greater precision since the inter-set variance is eliminated from
the results.
38. In the above discussion nothing has been said as to the way
in which the families on which the tests are to be carried out are to be
selected. Clearly they must be reasonably representative of the types
of family on which the final survey will be conducted. It is useless,
for example, to test out methods on middle class families if data are
required for working class families. On the other hand if over-all estimates for the population are not required there is no absolute need for

APPENDICES

t h e test material to be selected strictly a t random — it m a y , for instance,
be considered advisable, a t any rate in the initial stages, where m a n y
different methods are under investigation, t o utilise families whose
co-operation can be assured, e.g., families contacted through some social
welfare organisation.
39. Even if contact is made through a social welfare organisation,
random selection should be practised within this limited population.
The organisation should in fact be asked to provide a list of all families
with which it has contacts. The statistician should select the sample
families from this list, which can t h e n be approached b y t h e organisation.
If t h e selection of sample families is left with the organisation, a much
less representative sample of these families with which t h e y have
contacts is likely t o be obtained.
40. Clearly if contact is made through some social welfare organisation, even with t h e above precautions, the results will not necessarily
be fully valid for t h e more general population : in particular, the amount
of non-response is likely to be far less than will be experienced with a
random sample.
41. Consequently a final test of the most promising procedures
should be made on a fully random sample. This m a y often form p a r t
of the full-scale survey, using interpenetrating samples. I n a n y case
it is always advisable t o include tests between investigators in any fullscale survey.
42. However well comparative tests of the above t y p e are planned
a n d executed there are certain types of error which will not be determined with certainty b y their aid : if, for example, all families t e n d t o
minimise certain types of expenditure, e.g., on alcoholic beverages,
betting, etc., no comparative tests will ascertain the full magnitude of
such errors. I t is therefore of the utmost importance t o t a k e advantage
of all possible available external checks, and it m a y be well worth
while to determine the amount of expenditure on different classes of
goods and services, independently b y a survey of retail sales, etc. This
can only be done on a regional basis if all major transfers of goods and
services across the regional boundary are t a k e n into account, and
consequently requires most careful planning and attention to detail.
43. Surveys of retail sales of this type can be combined with
surveys of retail prices which are in any case required for the construction of cost-of-living indices. The evidence presented to the Subcommission indicated t h a t in this respect also the principles of random sampling
had in m a n y cases been inadequately adhered to.
44. Finally the Subcommission wished t o express its conviction
t h a t continuing enquiry b y means of small-scale successive surveys
was likely t o be much more fruitful in this field t h a n large-scale surveys
only undertaken a t long intervals. Successive surveys at frequent
intervals not only provide much more up-to-date information, together
with information on trends, but also enable a proper survey organisation
of skilled investigators t o be built u p and maintained — a most important
point in this field where all enquiries are so difficult. Once the problems
of collecting the required information have been solved, t h e use of
modern sampling methods will enable all necessary accuracy t o be
obtained with quite small samples.