120

SOME FIELD PROBLEMS IN RURAL SOCIAL RESEARCH

TrITK PARTICULAR REFERENCE TO ^HANA

x
P.A. Twumasi

•*•• I^t r odu ctjLon

This article is concerned with highlighting some of the perti~
nent field problems and issues commonly encountered by fieldworkers
in carrying out. rural social research*'  'The--specific intention is ,
to draw attention to some' of the .particularities and problems of
rural social research and to suggest some field strategies to help
to lower field error-and to increase reliability  and validity
measurements in collected field data*

II*  The Research Rationale

In Social Research the scientist uses tlie scientific method to

discover patterns of social forms and relationships*  An organised
and a systematic method is used to seek information into the social
composition, living arrangements, activities and views of a group of
people*

The .social scientist assumes the existence of the existing social.,
situation*  That there is in existence a well: defined social structure
and the people, within this well defined social structure, do not act
in quixotic way*  Rules of social behaviour exist*  People relate and
interact in a meaningful way to maintain a form of social cohesion*

Assuming all these positions the social scientist develops a

methodology with a built in significant level of objectivity criteria,
to study the nature of the existing social structure*  He uses the
scientific method to study the social phenomena in order to present

P.A-  Twumasi is Senior Lecturer in the Department of Sociology,
University of Ghana, Legon*

121

an intelligible image of the system.  In th; research operation,

the serious social scientist is dogma free except his pre-

occupation with his basic assumption that through rigorous

scientific.methodology he will be able to discover patterns of

f 

*

social interaction and "meanings", social participants give to
their social situations and relationship.

How do these people define the nature of their existing social

reality?  What meanings do they glive to social action?  What is the

nature of the relationship between the research worker and his

respondents?  What method dan help to increase the efficiency level

of the operational tools of measurement?  What methods in sampling

can be designed to increase the response rate in rural community

studies?

Ill: Field Experience

In Ghana, according to the recent census figures, about 70 per

cent' of the people live in rural and outlying settlements.  A few

of the population (30 /o) live in the urban towns and cities*  A

significant majority of the people therefore, share traditional ways

of life and work in traditional based Institutions.  Essentially,

these institutions are different'and the mode of life has its own

situational logic.  Kinship plays an important role in their social

relationships.  Many of the people are subsistence farmers*  They work

on kinship farms.and return home late in the evening.  A significant

proportion of them consists of old people and the very young*  Many

of them are illiterates.  They are exposed to a different style of

life.*  They share a different belief system and their cultural ideas

are quite different from what exist in the industrialized social

systems.

122

Experiences acquired from three social surveys Within the

period 1()72"~7^ can throw some light on the postulate that there
is a remarkable difference between the urban response rate, among

literate respondents end the rural response rate involving illi-

terate respondentso

In 19?2, the University of Ghana Medical School Community

Medicine, asked two research scientists, S.O. Boateng, ISSER and

P*A» Twumasi, Department of Sociology, to conduct a social survey

into '"Housing Conditions and Utilisation of Health Services with

Particular Reference to the Population of Achimota Village,

Adabraka and Tesano in Accra"*  The Study was conducted during the

long—vacation period from June to September 1972•

In pursuit of the research problem it was necessary to

indicate the nature cf the housing conditions and to show whether

there is any difference in health behaviour pattern with particular

reference to the population in Achimota Village, Adabr;aka and Tesano?

At Achimota Village rwe selected a homogenous group: of people at
Achimota Kopevi Village•  They belong to a particular ethnic group*

Most of them were illiterates, unemployed and those employed were

mainly of the self"employed variety*. Adabraka residents were fairly

mixed ~ all types of people were.found- in this locality*. . A signifi-

cant proportion of them were educated and employed in formal organiza-

tion of work*  The other survey, area was inhabited by, people from the

professional class, lecturers, doctors, .lawyers-and other high admini-

strative personnel*  They represent a significant proportion of the

resident who live at Tesano*  It is a residential area*

52 medical students were selected and trained in field methods,

problems involved in interviewing, measures of reliability and

validity in data collection, establishment of field report' and other

related field issues*  The important point in this training phase, was

123

to instil in them field skills and techniques, to acquire similar

orientation in probing questions an^ to develop methods to help

increase the response rate in the collection of relevant data*

We took into consideration the nature of the field situation*

Those who were familiar with the area were selected to work in that

area*  The language question was also noted*  The field assistants were

put into groups of three*  The interviewers were conversant with the

particular predominant language spoken in the area-  They were also

introduced to the psychology of interviewing techniques, how to

establish field rapport, ask the right questions at the appropriate

times and measures used to increase field response rate*

Daily checks were made*  The collected questionnaire schedules were

checked and edited*  Incomplete frames were returned and refilled*  Field

assistants were continually encouraged to go to the selected houses at

appropriate times suitable to the respondents*

The sampling scheme adopted for the study was probabilistic*  The

essential argument in probability sampling as argued by Kish  is that we

can specify for each element of the population the probability that each

element will have the chance to be included in the sampling design*

Each individual within the selected population universes (i*e* Achimota

village, Tesano and Adabraka) had non zero chance of being included in

the sampling design*  What is necessary in probability sampling theory

is that "for each element and its combination there must be some

specifiable probability that it will be included"  (Selltiz, p*5>13)*

We adhered to this principle in probability sampling because it

is the only approach, in sampling methodology, that makes possible

representative sampling design*  This method enables the designer to

#heck error in an organized and systematic way*  It makes it possible

for the designer to estimate the extent to which the collected data based

124

on  the  estimated  sample,are  likely  to  be  different  if  he  were  to

study*  the  entire  population. 

In • using; tho-'--probability  'sampling

frame, we  expect  the  estimated  sampling  mean  to  differ  or  to  differ

insignificantly  from  the  expec^d  siirvey  population  value*

•In,••• order  to  lower5 variance 'within  .eacb  stratum,  of  the  survey
universe,  the  sampling  scheme  adopted  for  the  study  ?was  d ^ t i f ie
sampling  method*.-  We began  this  scheme, by  systematic  sampling  procedure

.vrith--lt-''^"aihGl:q9i.''S-tart, for  each  selected  stratum.

As.said  earlier,  the  areas  were  stratified.  There  were  differences
frbwever.  But  these  differences  whereas  allowing  us  to  obtain  more  infer*"
ination  did  not  contribute  significantly  to  the  sampling  error  of  the
population  mean«•••  In  fact  it  can  be  shown  that  differences  between  strata

m e an  in  the  population  do-not  contribute  to  the  sampling  error  of  the
estimate  of  the.population  jnean*  Sampling  error  of  the  estimate  of  the
population  mean  comes  from  variations  amen^; sampling  units.that  are  in
the  same  stratum.  Hence  through  stratification  the  investigator  can  get
homogenous; sampling  units  to  lower  Sampling  variance*

Also  the  nature  of  the  sample  size  was  taken  into;.consideration*

Wherj  a  stratum  showed  more  variability  than  other  areas,  a  larger  sample
size was taken,->meahing that a stratum of less variability got a, smaller

'. 

." 

•• 

i'

.. 

-

*

i

"

-• 

.

.

.

.

- 

.

•

•

•

•

"

•

'

•

• 

'

.

.

.

sample size*  This method enabled us to explain variability in a more
me aningfulr way•

As indicated in Table 1, the sampling figures for the three areas

were as follows: Achimota village nJO respondentsj Adabraka l,f800

respondents and Tesano 3?0 respondents*' The corresponding figures show

the response and noh response rates estimated in percentages*  The ndh

response rate is higher in Achimota village (I3.0°/o); this village it

must be remembered, is inhabited mainly by rural oriented, illiterate

population*

125

Table 1

Response and I'lon Response Rates at Achimota

Village, Tesano and Adabraka

Response

Non Response

Rates:

Response rate

None Response

Rate

Total

Achimota
Village

Abs.

0/
/o

Tesano

Adabraka

Abs

°/o.

Abs

°/o

3?0

86.ii

365

93.5

1760

.  60

^0

13-6
100

5
370

100

1800

?#

2-5
100

See B.O. Boateng and. P.A.: Twumasi, Community Health Report No«6
University of Ghana Medical School, Department of Community
Health, Accra,-19^2-

In both. Tesano and Adabraka, the inhabitants are fairly well

educated and work in formal institutional establishment.  The

response rates were significantly hi^h.  What then accounts for

the high non response rate in Achimota village.

Before we attempt to explain the differences in the response

rate it is equally important .-to, lo.e>k at similar differences in

response rates in two other social "stir-veys •

In 19?3 t*16 Population Dynamics Programme of the University of
Ghana "(in conjunction with the University of North Carolina, Chapel

Hill, U.S.A.) agree'd to finance a study jinto some traditional

attitudes towards health, disease an^.. family planning in four

selected Ghanaian Communities, The 'Principal Investigators were

Drs* G.K. Nukunya and P.A. Twuraasi (both of the University of Ghana)*

We selected two urban'areas and two relatively isolated rural

communities.

126

(-0  Nsawan "~ It is one of the principal towns in the

Eastern Kegi-on^-witli a population of 25,528,
according to the recent  census figures*

(ii)  Dobo^o - It is.in the Eastern Region*  It is situated

on-.-the Nsaxam~Ab'liri road*  It is a small
farming  community whose  inhabitants are mainly
subsistence.. farmers*  According to the census
•figures its population: was 278.  It is a rural
settlement*

In thd-Volta Region J#e sel acted. Dzelukofe  and- Abor*

(iii)  Dzelukofe T shares the characteristics of an urban town.
It is about. 130 kilometres ifrom...Accra, with: a
a "popul at i oil - of'. 5,153  ]?e ople •

(if)  Abor  "• is a small-.-fommunity.. By the • standard defini"

tion, it ;is a rural in both sociological and
demographic.terms,.' with a population, of 3 ^ 3^
inhabitants.  It is ethnically a community..
The  inhabitants represent  9&*!?8o/o of its people.

The preceding Table 2 shows  clearly the response and non response

rates.  The response rate is higher in the urban areas than in the
rural.settings (see Table•2)

Table 2

jSesponse and Non Response Rates in Four Communities?

'Hsawam, Dob or o^ Dzelukofe • and Ab or

Response and
Non Response  Rates:

Response  rate
Non  Response
— rate
Total

Nsawam

Abs

69&

/o
99a

6

0.9

700

100.00

Abs

55

15
7b

Doboro

Dzelukofe

•  0/
/o
78*3

21.7

Abs

19?

7

Abor
Abs
17£

/o

. 96-5

>5

X5

O/
/o
92.1

7.9

100.00

200

100*00

19Q-

100.00

See Dr* G*K. Nukunya and Dr. P*A. Twumasi, Traditional  Attitudes

Towards Health Disease and Family^Planning in Four. Selected
Ghanaian Communities, Legon| Population Dynamics Programme
Study. June, 1974*

127

The other social research was conducted in June 19*^4-, by a

group of principal investigators headed by Professor N*Q* Addo,

Director of Population Dynamics Programme, University of Ghana.

The purpose of the survey was to study "The" Impact of Tourism on

^ocial Life in Ghana"«  This research was commissioned by the

Ghana Tourist Control Board*

The sample areas included both urban centres and rural

communities*  All the regional capitals wereiinc3;uded;v^^^;'d^Biplo*.

Some villages were also included*  The villages were selected, one

from each region,., to- act as-control to determine if there was a

difference between rural "-urban analysis of tourist behaviour pattern*

The. field assistants vwery®. University of Ghana students*  They

were given an intensive orientation'-course into field psychology,

acquisition of .research techniques and skills to give th%m ^meaningful

insight into the nuances of fieldwork.  They were distributed at the

end of their training to the selected areas;to interview and to assist

the respective respondents to.':.£ill- the questionnaire schedules*

During the fieldwork, many of the field workers complained about

the difficulties met in contacting both'rural and urban respondents

in certain areas*. Recalls' were made*  Substitute samples were framed*

Interviewers £&"Entered the field•  They were closely supervised*

In the final analysis there was some improvement in the urban response

rates where as the difficulties in reaching "some of the respondents in

rural areas persisted*  Part of the reason was that some rural res-

pondents were hot found in their-'usual place "'of residence*  This fact

was borne by data from Paga in the 'Upper Region, Ewhia in Ashanti

Region and Kato in the Brong""Ahaf o Region*  These communities are

rural, inhabited mainly by farmers*

128

The- data cited In the last few pages clearly throw some

light on^ the nature of response rates in rural and urban studies*

The comparative relationship between rural and urban response rate#S

is established*  It will be more meaningful to get more data from

other research studies to make definitive statements*  The interest**

ing point about this pnalysis is"'tha-.it future research workers skou3,ei

meaningfully take into, consideration  •.the nature of this tentative

observed relationship in order to improve the qualify and the quantity

of data in rural field  research-  This preposition is significant

because in Ghana and in many of the developing countries in Africa a

significant proportion of the .-.people Hire in rural and outlying

settlements*  A minority reside in. the urban arieas*

A possible explanation is ecological*  In'the urban situation^

we have large dense permanent settlements*  The buildings are well

defined*' The area maps are relatively well outlined*  The numbering
of the houses is relatively much •hotter outlined; than- found in rural
environment.  Secondly the rural population exhibit-a different style

of"life*  They are mainly subsistence farmers, they have different

work culture and habits.  They go to their farms during the best

part of the .day, seme return late in the evening, others never return

during the day.  Some may choose to sleep in the farming houses or

settlement especially during the planting and harvesting seasons*

Sonre go-on long ways,, on footpath, to attend to kinship business

anel-funeral arrangements.  Many rural folks have been noticed to shy

away from embarrasing  questions.  Wrong identity of the interviewed*

may a&io play a part to increase the non response rate*  As indicated

by one of the field interviewers, "some of us were mistaken to be tax

collectors"•

129

In Small"Scale Social Surveys, unlike  census studies where

the mass media and other government publicity media assist to

prepare  the local inhabitants  to remain in their homos, we tend

to get poor publicity.  So the non response rates in small scale

social  surveys tend to be much higher than  found in census studies*

The travelling  •arrangements of the rural folks, work habits

and pure sensitivities about answering  certain research  questions

do account  for  the higher non response rate in rural studies*  The

educated urban!te  it must be pointed  out, shares a similar  orienta"

tion with the field interviewer.  He understands the. interviewer,

tends to co~operate because he knows the meaning  of these reeearch

studies*  Questions about sex and other related  sensitive matters •

are freely discussed.

Different norms and  cultural idiosyncracies  can effectively

account  for the discrepancies between the non response rates in'-,the

two cultural settings.  These indications show  clearly that in

carrying  out field research in rural cultural settings, the field

scientist must  of necessity use  culturally relevant field methods

in the collection and validation of field material*

As pointed out by Kish  two types of errors can be  encountered

in all field surveys.  These errors are  (l) sampling errors and  (2)

non sampling errors*  The  interaction  of these  two errors tend to

produce total research error*

Put diagramatically-the  position is that of a right arigled

triangle, in which AB  "  BC '/+ AC ••" That is to say the sq«  on the

• 

• 

•

2 

2 

2 

hypothenuse  is equal to AG  +;BC'.  It means there is a functional

2 

2

relationship between  sampling and non ,sampling errors, to produce

interactively "Total Error"*

ng Error

,. 

Non" sampling  error

To reduce errors therefore  in social research the two legs, AC

and BC must be critically  controlled  through rigorous  scientific

methodology with a'view  to lower errors.  On the AC leg, is sampling error*

Inv-Sampling design^ a widely'accepted  model  cor.binea the/variable-error  and

the bias .into the total error* 

"

To reduce error in sampling, a meaningful representative  sampling

must be designed-  Kish. mentions three points in this regard  (l) that

the true value must be uniquely defined  (2) that the  true value must

be defined  in such a manner  that the pur-noses o'7 the survey are met

and  (3) where it is possible to do so consistently with the first two"

criteria, the true value should be defined  in .terras of operations which

can actually be carried through*

These measures if neglected  can affect sampling frames as well

as the non sarapling measurements*  For some, items the  true values can

be obtained relatively  easily but for  others difficult  to obtain*

The non sampling errors can emerge from non" coverage, non response^

errors in observation, framing of the questionnaire and asking sensitive .

131

questions without establishing a meaningful field rapport*  Errors

of non observation can also result from failure to obtain informa""

tion from certain segment of the survey population*

It is in this sphere that we can distinguish between two sources

of non response error*  These are non coverage and non response*  The

former means there is a failure on the part of the designer to include

some meaningful units of the defined population area in the actual

designing of the sampling frame*

The non response refers to the failure on the part of the field

interviewer to get information on some respondents, originally included

in the samplar frame*  It could be due to the non coverage of area unit,

that is to seeing coverage errors as a result of incomplete listing*

Incomplete listing is usually an outcome of inaccessibility and diffi~

culties in mapping the area, sufficiently.  Non response rate will

increase sampling error by decreasing the effectiveness of the calcula-

ted sample size and the non sampling error*

As pointed out.also by Som  sampling bias may arise from inadequate

or "faulty conduct of the specified probability sample or from methods

of estimation of the universe values*"  This may be due to wrong

selection procedures and partial or incomplete enumeration of the

selected units.  In sampling, the researcher rnust include diverse

elements in the proportions in which they occur in the actual population*

Size of sample alone is no proof that the estimate will be accurate*

Hsin Pao Yangr  opines that "a small sample cross-checked by various

5

methods may under certain circumstances produce more- accurate and

reliable information than a large one"; if meaningfully selected, such

a small size is more economical and efficient to handle than a larger one*

132

Definitions of operational variables must be clear and

distinct*  Vague concepts, •unclear definitions and improper

application of theory will also blur the focus of research*

Inherent in social research is the ability to develop viable

methods to collect valid and reliable data*  Field methods must be

carefully designed and selected to suit each empirical social

situation*  To ascertain views and opinions about social phenomena,

there is the need to reflect constantly on the issues of reliability

and validity of tools of measurements*

Questionnaire construction, interviewing methods, field rapport

and the possibility cf designing a strategy to increase reliability

and validity indexes must h»~given^fc field consideration*

The argument is that the rural population live in a different

social environment*  The style of life, their mannerism, work habits

and value orientation must be carefully noted, evaluated and assessed

before planning rural social surveys*  Unlike census sueh surveys are

riot given adequate governmental publicity*  Thus if inadequate prepara-

tions are made, errors can emerge to lower the response rates and these

affect reliability and validity of measurement*  Field procedures must

be guided by its propriety and fruitfulness*  The need is to seek

sedulously respondents who are acute observers of the social situation*

Informants can be used or a small number of such individuals can be

brought together in a discussion.  This method may be more useful in

many instances than the formal questionnaire method especially when

dealing with illiterate homogenous population*

IV.  Field Strategy:

The initial problem is to select an appropriate research topic*

The selection of a research topic needs some thought*  Do the rural

folks have answers to the research problem?  Do they often shy away

from discussing the essence of the topic?  What methods can be used to

133

extract the field material?  The ability to perceive in some brute

experiences the occasion for a problem, and especially a problem

whose solution has a bearing on the situation,is a good starting

point in any field research*  The ability of the social scientist

is sharpened if he reads what has been dene in the field of his

interest, ;o throw some light in the area of his operational research,

and to know about the sensitivity of the people in the area, to learn

about relevant techniques already in use in cdllection of field data,

and the methods used, to establish rapport with the respondents in the

field situation*  Secondly after the selection process is finalised

there is the need to specify the crucial research concepts and variables

to be used in the research process*  These concepts must be defined

empirically and translated into local languages appropriately*  Such

testable indicators must have empirical based validity and reliability

measurement criteria.  The prevailing social mood must be assessed?

The empirical definition must be relevant to the existing social mood*

One way to achieve a meaningful empirical definition is to pretest the

social indicators or the operational variables in- the area, of research,

to learn or to discover "how the people in the social situation" define

and give meaning to the concepts, under consideration*

Thirdly the methodological tool for' data collection must receive

some considerable thought*  In the rural soc.'-c-ty/ the majority of the

people are illiterates, cannot read and write in'the official English

language*  Sensitive questions must be carefully framed to avoid mis~

understanding and preferably only asked when the necessary field rapport

has been established between interviewer and the interviewee*  In making

the decision about the particular field technique, the researcher must

keep in mind the nature of the social situation, the types of people

and the nature of the field problem*

134

The most predominant techniques used hov;ever are the field

interviewing technique, properly constructed and meaningfully

translated questionnaire, the participant, observation, the

structured observation technique and the panel discussion method*

It is important to use more than one method in collecting field

data-  Then in entering the field, the researcher must give a

proper account of himself*  He introduces himself to the power

structure of the community, the legitimate chief, his elders,

other prominent leaders of the community etc. in order to gain a

legitimate entry into the community.  The aim of the project must

be given together with its applied implications*  If these intro-

ductory discussions are clarified then unnecessary suspicion will

be discarded*  This formality can give the field researcher the

passport to enter into the chosen community*

Then he must settle down to do some serious fieldwork*  His

life style and general approach.tc the,fieldwork must be meaningful

to t.he people in the localities. .He must be aware that he is dealing

with a gemienschaft society-.  They act in. a friendly manner*  Kinship

relationship plays an important part in the day to day activity of the

people*  The researcher must be careful not to offend any.person in

this situation*  He must be fair and objective in his relationship

and social interaction*  Any word discussed with a particular household

will be known by others in other households*  If he has the use of a

motor vehicle, for example, he may be asked to give "lifts", to. people

in the village who may urgently need such an assistance*  If he does

a favour to c-.ie section of the community, he must be. prepared to 4x> like'

wise to othcr.3 in order to maintain good field relationship and rapport*

In all collected field data, the question of reliability and

validity issues should also .be determinecl»  Reliability refers to

consistency of field answers, i*e* consistent dataf are the answers

reliable and how relevant are the answers in respect to the research

135

problem^  Hie"latter question addresses itself to the issue of

validity*  It is precisely in this direction that at least two

methodological' tools must be used to check for consistency and

reliability of the field data*  Furthermore- the interviewer must

have a built~in checks, by introducing, certain questions which can

help him to detect errors' In 'data;';"collection in a systematic afid in

an organised way«  For example in the area o.f age of respondents,

specific life incidences must be "referred to*  This will give an

illitefcat.e respondent a focus to remember the age in which he was

born etc*  Also through lengthy conversations, an implicit con""

tradictory statement can be discovered and corrected*  In other

words if formal questionnaires are used among illiterate folks, the

researcher must remember to check his answers by supplementing data

collection process" with informal interviews to discover the mood of

the situation, and to correct ambiguities and other irregularities*

This is the position of the researcher who goes to the field

in a rural area*  It represents the commitment on the part of the

researcher to actively collect  ' reliable and valid data*  It means

precisely that the researcher must be intimately acquainted with the

aspirations of the people must understand the language of the people,

the meaning of their life styles and must learn to view the social

world of his respondents from the way they structure such experiences

and view their experiences*  It means that the researcher must be

humble enough to mix with the people he is studying*  This is so

because being from a different social background he should be extremely

careful not to impose his values and orientation on the social situation*

Many of the research workers are usually recruited from the University

population*  The point to be remembered in this regard is that the

observer is in a relatively different social position when he goes to

the field*  He must not act in any way different to reflect that he

looks down upon the people he is observing*  He must in all humility

136

learn tc accept and to co!1$Lect data meaningful in the social

environment he finds himself*

REFERENCES

1.  Kish Leslie, 

1967  :  Survey,Sampling  (New York:1

John Wiley.)

2.  Kish L., 

1967  :  op. cit*,  p. 509-

3»  Selltiz Claire, 1967, et all. Research Methods"in Social Relations,
(New York: Holt, Rinehart and Winston*)
pp. 513-515.

4»  Som, R#K« 

19^3  *  A Manual of Sampling Techniques (London,

)

5«  Yang Hsin Pao  197l  :  Fact Finding, with Rural People^
(FAO Publi cations, Rome) p• 35•