An approach to selecting appropriate farm survey techniques for less-developed areas: An example from Sarawak

An approach to selecting appropriate farm survey techniques for less-developed areas: An example from Sarawak

AN APPROACH TO SELECTING APPROPRIATE FARM SURVEY TECHNIQUES FOR LESS-DEVELOPED AREAS: AN EXAMPLE FROM SARAWAK J. R. BEST Agricultural Extension and ...

519KB Sizes 0 Downloads 84 Views

AN APPROACH TO SELECTING APPROPRIATE FARM SURVEY TECHNIQUES FOR LESS-DEVELOPED AREAS: AN EXAMPLE FROM SARAWAK J. R. BEST Agricultural

Extension

and Rural

Dewlopment Reading, Berks, (Received:

Cmtre, Great

17 January,

University Britain

of‘ Reading,

London

Road,

1978)

SUMMARY

A crucial aspect of the planning of farm surveys carried out for administrative or policy-making purposes in less-developed areas is that of identifying the appropriate technique for collecting any given item of data. Thispaperputsforward an approach to the problem based on a close assessment of the way in which farmers themselves conceive the variables of their production process. Such an assessment indicates, in turn, where rapid techniques, such as the ‘single-shot’ questionnaire, may be usedfor data collection and where, on the other hand, more intensive methods are necessary. The approach offers no generalprescriptionsfor survey design: on the contrary, it is a means of ensuring that survey design is closely adapted to the nature of theparticular farming system under study and it may lead to different ‘mixes’ of survey techniques being used to investigate different farming systems within the same country.

INTRODUCTION

Farm surveys are undertaken for a number of different purposes in developing countries and take a variety of forms. The agricultural census aims to provide ‘baseline’ data for administrative and policy-making purposes. Typically it uses a ‘single-shot’ questionnaire to collect data on farm population, output and land use from a large number of respondent households. The farm management survey tends to have the somewhat different aim of collecting detailed input-output data which are then used either to derive production standards for use in advisory work or to build farm models (of varying degrees of sophistication) for planning purposes. The farm

management

with

regular

result

in

survey visits

data

often

to respondents

involves

intensive

as a means

survey

of ensuring

of a sample

of households

that

recall

faulty

does

not

being lost: the methodology and possible policy application of the 211

Agricultural Administration Printed in Great Britain

(5) (1978)-o

Applied

Science

Publishers

Ltd,

England,

1978

212

J. R. BEST

farm management survey are comprehensively covered by Collinson.’ The village survey differs from both of the two types outlined above in that it consists of an intensive survey of a rural community as a whole rather than that of a sample of individual households. The village survey has to date been a device more used by academic researchers than by policy-makers except in the Indian subcontinent where it is a well-established aspect of the work of the Indian Agro-Economic Research Centres and where it has even featured in the Agricultural Census of India (Lipton and Moore).’ In view of the increasing influence of a holistic, and crossdisciplinary, view of rural development, it may be predicted that variants of the village survey will come to be more widely used for policy purposes. A common feature of all three types of investigation outlined above (and also of hybrid types falling between these categories) is that they attempt to record quantitative data relating to particular farming systems. The problems which this exercise poses to the investigator (lack of written records or of standard measures, the large proportion of farm output which is home-consumed) need no detailed enumeration here. What may be said is that these problems have received less attention in the literature than have the more theoretically interesting questions of sample design (Hunt,3 Sukhatme4) and that, further, this neglect has resulted in serious mistakes being repeatedly made in survey fieldwork. The most common of such mistakes is the over-use of the single-shot questionnaire, but in contrast to this, and equally unthoughtful, is the tendency to reject all but objective measurement techniques (crop sampling, chain-and-compass measurement of land area) in the belief that ‘illiterate’ farmers are unable to communicate with investigators in quantitative terms. The contribution of this paper is to present a method of selecting appropriate survey techniques which are based on a framework for understanding the terms in which, and the extent to which, discussion about the farming system potential survey respondents operate can usefully be carried out with them. The approach was devised by the writer for planning farm surveys in Sarawak, Malaysia. Its use there was found to avoid some of the waste and frustration which comes from devoting resources to the collection of data which prove to be unreliable: more positively, it was valuable as a means of obtaining calculations of the logistics and costs of mounting surveys of varying degrees of scope and detail.

A CONCEPTUAL

FRAMEWORK:

THE

FARMER’S

AWARENESS

OF HIS ENVIRONMENT

The methodology here advanced is based on the assertion that the peasant or subsistence farmer has necessarily a high degree of awareness of the variable factors which are important to his decision-making process. However, his ability to transmit this awareness to an outside investigator is limited by: (1)

his conceptualisation of information

FARM

(2) (3)

SURVEY

TECHNIQUES

FOR

his ability to recall information his view of the sensitivity of information, to an outsider.

LESS-DEVELOPED

AREAS

213

affecting his willingness to impart it

The term conceptualisation is used to denote the precision with which a farmer can express information. Thus, for example, he may have a clear quantitative view of a particular piece of information which he expresses in terms identical to those used by an investigator (quantity and value of marketed output); he may have a clear view expressed in terms peculiar to his own culture (yield expressed in number of baskets); he may have a view sufficiently precise for purposes of his own decision-making but which he does not find it possible to express in units of measurement (whether ‘enough’ of a subsistence crop has been planted or harvested); finally, he may not have mentally registered the information and thus may be in no position to discuss it no matter how interested or persistent an investigator might be. Ortiz,’ in a perceptive study of decision-making among Colombian Indian peasants, discusses the nature of their conceptualisation of production variables and relates the precision of this to the role which each variable has to play in the decision-making process. Thus, Ortiz suggests, in planting a subsistence crop of maniac, farmers will continue until ‘they can see they have enough . . . . It is the size and quality of the stem of maniac which he is planting that will make (the farmer) decide whether or not it will produce and how many tubers it is likely to bear. He will plant more tubers if the stems he is using are not of good quality, if rains do not hamper his work, if he is not exhausted, and so on’. The farmer’s recall of information is affected to an important extent by degree of conceptualisation. It is also, of course, determined crucially by the function of memory. Thus yearly output of a crop which is harvested within a short period (e.g. rice) is likely to be recalled better than that of a crop (e.g. a perennial such as coconut or rubber) which is constantly under harvest. A rather rare event such as the slaughter of a pig is much more likely to be recalled than is, say, the sale or consumption of eggs. It should perhaps be added at this point that, while the recall of illiterate farmers may be good, its value from the point of view of an investigator is often limited by the farmer’s limited awareness of the investigator’s calendar. Thus the farmer may date the events on his farm either somewhat circuitously to other events of the farming year (particularly in Sarawak to the hill-rice cultivation cycle) or, alternatively, to even more momentous events within the community. Collinson discusses in some detail the whole question of farmer recall, including the related problem of ‘memory bias’, in relation to decisions as to what frequency of survey visit is necessary to record accurately various types of data. The question of sensitivity of information provides a problem overlying those created by conceptualisation and recall. In any culture there are aspects of individual and household affairs which will not readily be divulged. The very nature of such ‘sensitive’ information often means that it is both conceived and recalled

214

J. R.

BEST

accurately but that it is inaccessible to the outsider (and often to other members of the community also) for different reasons. The most obvious example of this type of information is that on credit and indebtedness, but it may also be found that data on sales of cash crops are withheld out of fear that they will attract tax and that even ownership of consumer goods may not be admitted for a similar reason. On the other hand, the anxiety of a family not to reveal its poverty may lead it to overstate such things as subsistence crop stocks or frequency of consumption of certain types of food. In discussing the secretiveness of the Land Dayak farmers in Sarawak among which he did fieldwork, Gedde@ suggests that an important reason for this is ‘suspicion of the motives of a supposedly disinterested investigator’ and implies that the most the investigator can hope for is to be regarded as a harmless eccentric. The field worker connected with government is rarely lucky enough to be viewed with even this degree of detachment and often has to devote some attention to disclaiming connection with any public bodies against which grievances may have arisen in the past.

DESIGN

OF SURVEY:

PARkIPATING

IN

THE

FARMER’S

AWARENESS

In designing any particular survey, then, it is suggested that it will be found useful to identify the position of each item of data it is proposed to collect on .a notional conceptualisation spectrum and a recall spectrum. Such a spectrum ranks items of data along a continuum from ‘good’ to ‘poor’ and also incorporates ‘boundaries’ at major points of discontinuity on the spectrum: for example where conceptualisation becomes too poor for a particular variable to form a topic of casual conversation, or where recall ceases to extend over the farming year atid is reduced to a shorter period. Specimen conceptualisation and recall spectrums for one farming system in Sarawak are shown in Table 1. The particular system under consideration is based on forest-fallow cultivation of hill-rice as a major subsistence crop (none of which is usually marketed) with rubber and pepper (Piper nigrum) as major cash crops (the whole output of which is marketed). A wide range of minor subsistence crops is also grown, a number of these as intercrops with hill-rice (maize, gourds, various types of vegetable) and collection of edible jungle produce (bamboo shoots, ferns, wild fruit) is also important. Most households have a small number of pigs (typically two to ten) and poultry (ten to twenty). Cultivation on recent rice-fallow land of tapioca (Manihot palmata) for pig feeding is an important subsidiary activity. * The drawing up of a conceptualisation and recall spectrum requires: (1) (2)

A general knowledge of the main~features of the farming system. Some understanding of the way in which farmers conceive of and discuss their activities.

The knowledge at (1) above need be no more profound than that implied in the

FARM

SURVEY

TECHNIQUES

FOR

TABLE CONCEPTUALISATION

AND

RECALL

LESS-DEVELOPED

1

SPECTRUMS FOR HILL-RICE

FARMERS, SARAWAK

CONCEPTUALISATION

RECALL GOOD

Type

-

of data

Method of Investigation

market prices inventory of title land inventory of major permanent crops ’ inventory of livestock “yield of cash crops yield of major subsistence crop (rice) area under major subsistence crop (rice)

‘single-shot’ questionnaire

GOOD Type of data

-

inventory of title land inventory of major permanent crops yield of cash crops with limited-period harvest market prices of cash crops yield of major subsistence crop (rice) area under major subsistence crop (rice)

-

(1)

-

single-period investigation (questionnaire plus observation) BOUNDARY (2) 1

-

inventory crops

-

inventory non-title

-

yield of cash crops with continuous harvest’ (especially rubber) inventory of livestock“ labour input to major activities

BOUNDARY 1

-

to major periodic reporting BOUNDARY 1

labour input to minor (routine) activities yield of minor crops (fruit, maize, vegetables, etc.) inventory of minor crops yield of jungle produce food consumption rates

(3)

-

periodic observation and measurement -

POOR ’ Items appearing of investigation

-

-

inventory of minor permanent crops (e.g. fruit) inventory of (uncropped) non-title land labour input activities

215

AREAS

lower on recall spectrum for these is that indicated

of minor (e.g. fruit)

permanent

of (uncropped) land

-

labour input to minor (routine) activities yield of minor crops inventory of minor crops yield of jungle produce

-

food

consumption

rates

POOR than on conceptualisation spectrum: the appropriate by their position on the recall spectrum.

method

thumb-nail sketch in the preceding paragraph: it may be gained from fairly superficial observations or even from discussion with a field officer or extension worker familiar with the system. The understanding at (2) requires an empathy based on a certain amount of discussion with farmers and would need to draw on the experience of a relatively aware and articulate field officer. The crucial importance of the ‘spectrum’ concept from the practical point of view is that, once conceptualisation and recall spectrums have been drawn up for a particular situation, they enable an assessment to be made of the limitations in that

J. R. BEST

situation of the various survey methods available. Thus, in the example under consideration, data on the conceptualisation and recall spectrums lying above Boundary (1) in Table 1 (i.e. at the ‘good’ end of the spectrum) can usually be collected by means of a ‘single-shot’ questionnaire (single-visit interview with prepared questionnaire). Normally neither problems of conceptualisation nor of recall over one farming year would seriously interrupt the ‘flow’ of an interview covering these topics. The main difficulty here is that some of the items tend to be reported in terms of traditional or rule-of-thumb measures (sheets of rubber, bags of pepper, baskets of rice) and it is necessary to establish reliable conversion factors for these where possible. Where such measures are found not to be consistent-and therefore not amenable to direct conversion to standard measures-the single-shot questionnaire becomes inappropriate. Between Boundaries (1) and (2) in Table 1 are data which cannot be collected by a formal questionnaire-type interview, either because their recall is a somewhat lengthy process or because they are conceptualised in terms which are found not to yield consistent or reliable conversion factors. In this example, both items of data between Boundaries (1) and (2) show the first of these characteristics (i.e. recall is lengthy) and the recording of an inventory of a household’s fallow land faces, in addition, the second difficulty (i.e. unsuitable conversion factors). Data of this type must be collected by what has been termed the ‘single-period observation’, involving a stay in the community long enough to allow objective measurements to be made of variables found not to be amenable to direct reporting based on interviews with individuals. Below Boundary (2) lie data which must be recorded either within a short period of their occurrence or even as they actually occur. The items between Boundaries (2) and (3) have a greater ‘time-tolerance’ than those below Boundary (3), it is suggested, because they tend to be reasonably well-conceptualised at the time they occur, although recalled accurately for only a very limited period. In Sarawak it has been found possible to collect such data by means of a respondent’s reporting to an enumerator at visits by the enumerator no less frequent than once weekly. Items below Boundary (3) present greater difficulties of recording because they tend to be poorly conceptualised, even as they occur, and correspondingly poorly recalled. To collect data of this type requires at least daily reporting or even direct observation and measurement. The decision as to whether to commit resources to such an exercise should, it is suggested, be informed by a critical assessment of whether the data involved are indeed important to the purposes of the survey. While the chart of the two spectrums in Table 1 shows a number of items in similar positions on both, it will be noted that some items appear higher on the conceptualisation spectrum than they do on the recall spectrum (but not vice versa). The implication of this is that such items (indicated by the footnote) must be collected by the method of investigation appropriate to their position on the recall spectrum since, obviously, no matter how well conceptualised data may be, the ease with

FARM

SURVEY

TECHNIQUES

FOR

LESS-DEVELOPED

AREAS

217

which they are reported depends ultimately on the facility with which they are recalled. Thus, for example, livestock numbers present no difficulties of conceptualisation but they cannot be recorded by a single-shot questionnaire with a whole year as its reference period because recall is poor: for an accurate record, periodic reporting is necessary. It is necessarily less easy to make an assessment of the ‘sensitivity’ of a particular item of data than to assessit in terms of conceptualisation or recall. Sensitivity varies between cultures, it is suggested, both in terms of the items of information which it most markedly affects and also in terms of the intensity with which it operates. It is necessary, therefore, for the investigator to make some assessment of both these variables as a preliminary to deciding how much time and care will be necessary in any given situation, first, to convince his respondents of the aims of the survey and, secondly, to establish a good degree of rapport with them. This will in turn assist a decision on the extent to which a survey can rely on a ‘single-shot’ questionnaire approach and to what extent repeated visits or one extended visit will be necessary in order to establish confidence and rapport. The axiom that ‘land holdings increase in direct proportion to the length of time an investigator remains in a village’ is useful to bear in mind when making a decision of this type.

CONCLUSIONS

The strength of the methodology here put forward is that it enables a flexible approach to be adopted in survey design. In Sarawak this approach has even led to different techniques being used in different villages within the same survey programme. Conceptualisation and recall spectrums drawn up for quite different farming systems to that discussed here might look very different from the examples in the table, but this only reflects the value of the exercise of constructing themnamely to identify the survey technique, or mix of techniques, which is appropriate to the farming system and culture being studied. REFERENCES

1. COLLINSON, M. P., Farm management in peasant agriculture, London, Praeger, 1972. 2. LIPTON, M. and MOORE, M. P., The methodology of village studies in less-developed countries, IDS Discussion Paper No. 10, Brighton, Institute of Development Studies at the University of Sussex, 1972. 3. HUNT, K. E., Agricultural statistics for developing countries, Oxford, Institute of Agrarian Affairs, 1969. 4. SUKHATME, P. V., Sampling theory of suroeys with applications, New Dehli, Indian Society of Agricultural Statistics, 1954. 5. ORTIZ, S., The structure of decision making among indians of Colombia. In: (Firth, R. (Ed.)). Themes in economic anthropology, London, Tavistock, 1967. 6. GEDDES, W. R., The Land Dayaks of Sarawak, Colonial Research Studies, No. 14, London, HMSO, 1954.