Problems of Sampling in NWFP Pakistan
-
Upload
imranahmadsajid -
Category
Documents
-
view
842 -
download
1
description
Transcript of Problems of Sampling in NWFP Pakistan
Assignment on
Problems of Sampling In North West Frontier Province
BY
IMRAN AHMAD SAJIDM.Phil/PhD-1st semester
Session: 2009
Submitted To:
Dr. Amirzada AsadChairman
Department of Social Work
DEPARTMENT OF SOCIAL WORKUNIVERSITYOF PESHAWAR
ACKNOWLEDGEMENTSAll praises to ALLAH, the most Merciful, Kind, and Beneficent, and source of all
Knowledge, Wisdom within and beyond our comprehension. all respects and possible
tributes goes to our Holly Profit MUHAMMAD (Swal Allaho Alaihy Wasallam), who is
forever guidance and knowledge for all human beings on this earth.
Thanks to Dr. Amirzada Asad, Chairman Social Work department, and the course
instructor, who has contributed enthusiasm, support, sound advice, particularly his
supportive attitude was always a source of motivation for me. He guided me in a polite
and cooperative manner at every step.
I am also in debt to all those writers who has written such informative and thought
provoking books and other material.
Imran Ahmad Sajid
i
SUMMARYSampling is very frequently used in researchers for half the century. Sampling is carried
out in order to save time, energy and cost by studying only a part of a given population.
Probability and non-probability are the major sampling methods. But sampling is not an
easy task to perform and when it is in 3rd world poor countries taking out sampling
procedures is similar to chewing iron bean. In this assignment report some of the
problems faced by the samplers in the province of NWFP are discussed.
ii
TABLE OF CONTENTS
ACKNOWLEDGEMENTS........................................................................................................................... i
SUMMARY............................................................................................................................................. ii
Introduction...........................................................................................................................................1
A. Key Concepts.................................................................................................................................1
I. Element.....................................................................................................................................1
II. Population.................................................................................................................................1
III. Universe.................................................................................................................................1
IV. Sampling Frame.....................................................................................................................2
V. Sample.......................................................................................................................................2
VI. Representative Sample..........................................................................................................2
VII. Sampling Bias.........................................................................................................................2
B. Why Sample...................................................................................................................................2
C. Methods of Sampling.....................................................................................................................3
1. Probability Sampling..................................................................................................................3
1.1. Simple Random Sampling..................................................................................................4
1.2. Systematic Sampling..........................................................................................................4
1.3. Stratified Sampling.............................................................................................................5
1.4. Cluster Sampling/ Area Sampling.......................................................................................6
1.5. Multistage Sampling..........................................................................................................6
2. Non-Probability / Purposive Sampling.......................................................................................6
2.1. Accidental Sampling...........................................................................................................7
2.2. Quota Sampling.................................................................................................................7
2.3. Purposive Sampling............................................................................................................7
2.4. Snowball Sampling.............................................................................................................8
D. Problems of Sampling in NWFP.....................................................................................................8
1. Resources Limitation.................................................................................................................8
2. Inadequate Sampling Frame......................................................................................................9
3. Problem of Language.................................................................................................................9
4. Refusal.....................................................................................................................................10
5. Size of Sample..........................................................................................................................10
6. Problems in Area Sampling......................................................................................................11
7. Problems of Generalization.....................................................................................................11
8. Lack of Training on Research Methods....................................................................................11
E. Conclusion...................................................................................................................................11
Bibliography.........................................................................................................................................13
Books...............................................................................................................................................13
Internet Source................................................................................................................................13
Introduction Last year in 2008, 2 of my friends, Subhan & Kamran, have conducted a research thesis, for
MA degree, on “role of PBM in poverty alleviation”. They went to PBM Head office NWFP in
order to obtain a sampling frame of PBM beneficiaries. They found that the list either was
incomplete and not up-to-date or the addresses of the beneficiaries given in the register
were not accurate. This and lots of other problems the researchers face in NWFP when they
are going to sample their population and collect data. This assignment is dedicated to
finding out the problems associated with the sampling in NWFP. We will clarify some key
concepts used frequently in sampling and then will move on to types of sampling and then
for the problems of sampling in NWFP.
We need to be familiar with some key concepts frequently used in sampling debates.
A. Key Concepts
I. ElementAn element is the smallest unit or part of a population capable of possessing a particular
characteristic.
II. PopulationPopulation is a well defined set of elements.1 It consists of all the units or elements in which
we are interested. For example all the pharmaceutical industries in Peshawar according to
the list maintained by labour directorate, all the hotels in Nathiagali as listed in a tourist
directory etc, constitute populations.2
III. UniverseAll individuals, events, groups or wholes, under investigation are known as the universe. For
instance all the cattle in a village, all the house owners of Peshawar, all the hotels in
Nathiagali, all the Motor Rakshas in Peshawar etc. these are the universes of their
respective populations.3 In actual practice there are gaps between given universe and its
population.
1 Nigel, Gilbert. (2001). Researching Social Life. 2nd Ed. London: Sage Publications. P. 59. 2 Jespal Singh. (2007). Methodology and Techniques of Social Research. New Delhi: Kanishka Publishers. P. 249. 3 Jespal Singh. (2007). Op.Cit. p. 248.
1
IV. Sampling FrameA sampling frame is a list of the members of the population under investigation and is used
to select the sample. 4
V. SampleAs the name implies, a sample is a smaller representative of the larger whole. 5 a sample is
the part of population selected for the study, in the belief that it represents the most, if not
all, the characteristics of the population. If the researcher has no belief in its
representativeness then it is not a sample.
VI. Representative SampleA representative sample is one which looks like the population, from which it was selected,
in all respects that are potentially relevant to study.6 If a sample is not representative of its
population then it will be considered as biased sample.
VII. Sampling BiasSampling bias means the tendency to reject the logic of rational evidence in favour of
presumed beliefs. If a set of figures is such that all its indicators tend to lie in the same
directions, it is biased. 7
B. Why SampleWhen we are going to conduct research we take sample with the following objectives in
mind;
To save time To save energy of the researcher To minimize the expenditure or cost
But remember that it should not be at the cost of reliability or validity of the research
results. Our aim is to get a complete picture of the whole universe by studying only a part of
it. 8
4 Nigel, Gilbert. (2001). Op.Cit. p. 60.5 Iqbal Saif. (1984). Basics of Research Process. D.I.Khan: Sultan Printing Press. P. 99. 6 Russel, K. Schutt. (1999). Investigating Social World: The process & Practice of Research. 2nd Ed. Thousand Oaks: Pine Forge Press. P. 110. 7 Jespal Singh. (2007). Op.Cit. p. 263.8 Iqbal, A. Bhatti. (2007). Elementary Statistics. Lahore: Bhatti Publishers. P. 282.
2
These were some basic concepts which are frequently used in sampling designs. Now how
sample is selected?. There are two methods for selecting a sample. You either select is by
chance or by choice. They are called probability and non-probability sampling respectively. A
detailed discussion is needed on this topic.
C. Methods of SamplingThere are two types of sampling methods viz probability sampling and non-probability or
purposive sampling. Probability Sampling is where every individual element in a population
is chosen at random and has a known non-zero or equal chance of selection. Therefore the
selection process is predetermined and once the units have been selected the goal is to
collect data from them all. 9
In Non-Probability or Purposive Sampling the chance of selection for each element in
population is unknown and for some elements is zero. Probability sampling is non-
judgmental, purposive sampling is judgmental. 10
Each of these methods of sample selection has their respective types or techniques of
sample selection which needs elaboration here.
1. Probability Sampling
Probability is a statistical terminology. It is the proportion of times that a particular outcome
may be expected to occur out of many repetitions of the event. For example we flip a coin,
the probability of getting a head or tail is one time. Usually the probability is expressed as
decimal fraction from 0 to 1. Zero indicates that the event will not occur at all, 1 indicates
that the event will occur certainly. 11
Probability sampling is the process of sample selection in which elements are chosen by
chance method such as flipping coins, drawing number balls from a bowl or throwing a dice
etc. there are several types of probability sampling but all share a common trait, i.e. the
9 Nigel, Gilbert. (2001). Op.Cit. p. 6110 Ibid. 11 Zari Rafiqu. (2007). Research Methods in Social Sciences. Peshawar: Unpublished book. P. 160.
3
selection of units for the sample is carried out by chance procedures and with known
probabilities of selection. 12
The following types are there in probability sampling;
Simple random sampling
Systematic sampling
Stratified sampling
Cluster or area sampling
Multistage sampling
1.1. Simple Random Sampling
Briefly denoted as SRS, simple random sampling is the most common and familiar type of
probability sampling. It is the selection at random. The elements are selected randomly but
note that the selection is not haphazard. It is random. For simplicity we draw sample
through Lottery method, Tippet number method or through computer generated
randomization.
The essential condition for SRS is that you must have a sampling frame. Without sampling
frame, simple random sampling can not be drawn. In lottery method each element in
population is assigned a number and the numbers are written on separate slips or chits. All
the slips are put into a hat or drum and the required amount of ample is drawn from it.13
The tippet number is a table of random numbers developed by a statistician Tippet, through
this table random sample can be drawn. But today the computer generated randomization
is the easiest and the faster method of generating simple random sample.
1.2. Systematic Sampling
Undoubtedly the systematic sampling is the most widely known modification of simple
random sampling. In this method the first element is selected randomly and the rest get
selected automatically.
12 Ibid. 13 Ibid.
4
First the population size “N” is divided by the required amount of sample size “n” to yield
the sampling interval “F”. Then the first element is selected through simple random
sampling-lottery method, tippet number method etc. after the selection of the first element
every Fth element is selected in the sampling frame thereafter.
Example:
Population Size N=150.
Sample Size n=15. Therefore
Sampling Interval=150/15= 10.
Now select the first element randomly which e.g. is 83. Now select every 10 th element after
83. i.e. 93, 103, 113, 123, 133, 143, 3, 13, 23, 33, 43, 53, 63, 73.
Systematic sampling is also called Regular Interval Sampling due to the regular interval
occurring in this method.
1.3. Stratified Sampling
Stratification means making groups within population. The word strata has been derived in
social sciences from Geology which means Layers of rocks or atmosphere. In social sciences
strata means a group of population.
If the population is not homogeneous, randomization will yield biased results. Therefore we
use stratified sampling in which population is first divided into separate sub-populations or
strata. After strata are formed a separate sample is drawn from each strata through
randomization.
For example your population consists of 60 male and 40 female elements and you want to
draw a sample of 10. If you select the sample randomly there is very likely a chance that all
the members included in sample may happen to be all Males or all females. Therefore we
stratify the population and select elements from each strata. i.e. 6 males and 4 females.
The benefit of stratification is that the more you stratify the more you stratify the more you
reduce the bias, the more you randomize the more you reduce the bias.
5
1.4. Cluster Sampling/ Area Sampling
Sometimes it is not feasible to attempt to prepare a list of every person living within a
particular area and from that list to select a sample for study.14 In such cases we use cluster
sampling. In cluster sampling the population is first divided into cluster i.e. units containing
several sample elements. After the clusters have been formed, a sample of clusters is drawn
from total group of clusters. The sample will consist of all the elements contained in the
selected cluster.
Area sampling is a modified form of cluster sampling in which maps are used to form
clusters and then select the sample. Cluster sampling brings down cost and facilitates data
collection. It also helps in minimizing sampling bias.
Like stratification, clustering also improves the sampling design. The difference between
stratification and clustering is simple. In stratified sampling sample is taker from within each
stratum but in cluster sampling all the elements within a selected cluster are included in
sample. Strata is homogeneous, cluster is heterogeneous.
1.5. Multistage Sampling
In multistage sampling the sample is selected in different stages and different types of
sampling are used in different stages. Generally stratification and clustering is done at first
stage and randomization at the second stage. For example we may select National Assembly
circles at the first stage and provincial and district circles at second stage etc.
2. Non-Probability / Purposive Sampling
Non-probability sampling is also called judgmental or purposive sampling. It is a method of
selecting a sample in which the choice of selection of sampling elements depends entirely
on the discretion or judgment of the sampler. The investigator inspects the entire
population and selects a sample of typical units which he considers close to the average
population. This method provides a lot of freedom to the investigator in the inclusion or
14 Iqbal Saif. (1984). Basics of Research Process. D.I.Khan: Sultan Printing Press. P. 104
6
exclusion of a sampling unit. But the disadvantage is that we can not obtain any valid
estimate of risk of error involved.
But it does not mean that non-probability sampling is never appropriate, sometimes it yields
more representative results then probability method. 15
This method has following types;
Accidental sampling
Quota sampling
Purposive sampling
Snowball sampling
This needs a brief elaboration.
2.1. Accidental Sampling
This is also called incidental, convenience, availability or voluntary sampling. These are the
different names used for this sampling technique. In this type of sampling any person who is
available and is a typical of the universe is included in the sample. The most common type is
standing in the public, in railway station, at a super market and asking who ever seems
relevant.
2.2. Quota Sampling
It is the proportionate selection of items, i.e. cases are selected for sample on the basis of a
quota system. The researcher has information about a distribution of certain characteristics
which are believed to be related to the research at hand, within the population it is possible
to establish quota for their inclusion in the sample.
There is an ambiguity that what is the difference between stratified sampling and quota
sampling? This is still unclear even to the experts in this field.
2.3. Purposive Sampling
This is the total selection. You select those elements which fulfills your purpose. This type
allows us to select our sample for study under a purpose, i.e. we have to predetermine that
15 Zari Rafiq. (2007). Research Methods in Social Sciences. Peshawar: Unpublished book. P. 170
7
a particular group is important to us. e.g. “health problems of rural married women”. For
this topic we will select women, who are married, and from rural areas. This is purposive
sampling.
2.4. Snowball Sampling
Snowball or network sampling is used when there is no adequate list to use as a sampling
frame. It is a method for obtaining samples of numerically small groups, such as minorities,
illegal drug users etc. it involves contacting a member of the population to be studied and
asking him/her whether they know anyone else with the required characteristics. The
nominated individuals are interviewed in turn and asked to identify further sample
members. This continues until the desired numbers or when no further member is available. 16
Till this point we have been discussing what sampling is and how sample can be selected but
in NWFP selecting sample is not that easy task. Numerous problems are involved in sample
selection in NWFP.
D. PROBLEMS OF SAMPLING IN NWFP
Sampling is an essential part of quantitative research but it isn’t an easy task. However,
when we talk about sampling in 3rd world poor countries and a poor province of NWFP,
sampling is a much more difficult task and enormous problems are there in sampling
procedures and designs in this province.
We are going to discuss some of them here in the following lines.
1. Resources Limitation
This is the first problem which strikes a sampler in NWFP. The limitation of resources is not
only true for NWFP but for the entire Pakistan and even the entire 3rd world countries. A
sample is required to be selected through an optimum design which is a basic aspect in
sampling. In developed countries a sampler has a variety of sampling designs at his disposal.
He can choose random sampling design, clustering, stratified or any other design which
16 Nigel, Gilbert. (2001). Researching Social Life. 2nd Ed. London: Sage Publications. P. 63
8
seems most optimum to him according to the research. He chooses a sampling design not
due to resources limitation but to the requirement imposed by optimum philosophy.
In NWFP, however, a sampler does not have this freedom of choosing a sampling design. He
has no, or very little, alternative designs at his disposal. His choice of sampling design is
limited by resources available at hand. For example if a sampler chooses a cluster sampling
design here in NWFP, e is driven, not by optimum philosophy, but by the resources
available. He has no proper sampling frame or he lacks other basic information that is why
he resorts to cluster sampling which is more practical in NWFP. 17
So the resources limitation is a basic problem in sampling in NWFP.
2. Inadequate Sampling Frame
This is the second problems faced by the samplers. A sampling frame is a list of all the
names and addresses in a population. For instance voters list in an electoral constituency, a
patwari’s record of landholdings, attendance register in a school,18 customers list of warid
telecom etc. as mentioned at the beginning the sampling frames in NWFP are often non-
existent or inadequate. The census is held every 10 years normally. It is an essential
condition of a good sampling frame that it should be up-to-date. The record of 1998 can not
be used for 2008 or 09 as it becomes out dated but our problem here in NWFP is that the
available sampling frames is out dated and not to date. This is one problem. 2ndly the
names and addresses mentioned are often inappropriate and the duplication of elements
also occurs. 3rdly there is no record of those who has left an area and those who has come
to live here. These are some problems related to sampling frame which exists here in NWFP.
You never get a complete and to date sampling frame and you have to rely upon what is
available. Therefore this inadequacy of sampling frame causes bias in results.
3. Problem of Language
This is the 3rd problem which samplers face when he has chosen a person and then go to
him/her to collect data from him/her. The language of majority people here is Pushto and
17 Slobodan S. Zarkovich. (1960). Some problems of sampling work in underdeveloped countries. In Martin, Bulmer., & Donald, P. Warwick. (Edt. 2000). Social Research in Developing Countries: Surveys and Censuses in the Third World. London: Routledge. P. 10118 Jespal Singh. (2007). Methodology and Techniques of Social Research. New Delhi: Kanishka Publishers. P. 264.
9
Hindko. Only a small number of people understand Urdu language and even much smaller
number understands English language. Our problem is that the research methodology
taught and questionnaires made, are in English language. A sampler prepares questionnaire
or interview schedule and then at the time of interview he has to convert its wording into
Pushto or Hindko language. Now for converting interview schedule from English to
Pushto/Hindko the sampler has to choose those words which are easily understandable to
the interviewee. The sampler must find the shared vocabulary between him and the
interviewee in order to get valid data.
Now this problem of mutual language and shared vocabulary is here in NWFP and this
create some inbuilt bias in the sample selected.
4. Refusal
Refusal is the 4th big problem for the sampler in NWFP. The people in NWFP and Pukhtoons
particularly are conservative in their outlook. They are not easy to be interviewed. People
often refuse to get interviewed due to lack of trust and variety of other reasons. Some of
the respondents may answer some questions but refuse to answer other more personal
questions. Furthermore a male sampler can not collect data from a female respondent in
NWFP.
In case of refusal the sampler will choose another respondent for data collection but this
will create bias in sampling.
5. Size of Sample
Sample size refers to the number of elements covered by a sample. Most of the researchers
have said that a sample should be 15-20% of its said population. But in NWFP even this
number of sample is very difficult to choose for study. The reason is simple. It’s the lack of
resources due to which samplers in NWFP can not collect data from 20% respondents and
sample covered often is 3-5% of its population. There is a lack of such organizations who
support the researchers in their research.
Due to the smaller sample size bias may creep in results and this can not be generalized on
the whole population.
10
6. Problems in Area Sampling
Area sampling is a modified form of cluster sampling in which maps are used for forming
clusters. But the problem is that we don’t have proper maps of most of the areas. The maps
available do not often provide satisfactory information. Further there is no clear
demarcation where one village begins and the other ends. This ambiguity further creates
problem in cluster making.
7. Problems of Generalization
NWFP is a province with diverse ethnic groups dominated by Pukhtoons, followed by
Hindkowans. The chitral and Kalash regions have their own ethnicity and culture much
different from main land NWFP. This lack of homogeneity causes problems for samplers.
The results of one district can not be generalized on the entire province. Even within
Peshawar City, the capital of NWFP, the information gathered from Pukhtoons can not be
generalized on Hindkowans-the 2nd major ethnic group in the City. So this diversity of the
NWFP, cultural and geographical as well, creates problems in sampling.
8. Lack of Training on Research Methods
Although the courses on research methods are offered in universities of the province and a
student is required to produce a research thesis in order to be qualified for an M.A/M.Sc
degree. Yet the practical aspect of sampling is not given due consideration. The sampling
designs and procedures often involve delicate statistical work but students of the social
sciences lack statistical knowledge. Therefore they face a significant amount of problem
when sampling and making inferences.
These were some of the problems which the researchers face here in NWFP.
E. Conclusion
Sampling is an important part of research. Selecting the sample-not only sample but
representative sample-is very necessary for the validity of your data. If your sample do not
represents the opinion of its population then it is biased and not reliable. The NWFP is a
11
poor province of a 3rd world poor country. Resources for samplers are not satisfactory. The
researchers when sampling face numerous problems-some of them have been mentioned
here in the above lines. Despite these problems the researchers are making their efforts and
trying to find out the more representative samples through some new methods. The
methods used in the developed countries are not applicable here. The researchers have to
work hard to and do some creative work in order to overcome the problems of sampling.
Lesser the sampling problems the more representative your sample is, the more
representative your sample is the more valid and reliable your research will be.
12
BIBLIOGRAPHY
BooksAnsari, S. (Edt 1963). Social Research in National Development. Peshawar: Pakistan Academy for Rural Development.
Babbie , E. (2005). The Basics of Social Research. 3rd Ed. Toronto: Thomson Wadsworth Inc.
Bhatti , I. A. (2007). Elementary Statistics. Lahore: Bhatti Publishers.
Gilbert, N. (2001). Researching Social Life. 2nd Ed. London: Sage Publications.
Rafiq , Z. (2007). Research Methods in Social Sciences. Peshawar: Unpublished book.
Saif, I. (1984). Basics of Research Process. D.I.Khan: Sultan Printing Press.
Schutt , R. K. (1999). Investigating Social World: The process & Practice of Research. 2nd Ed. Thousand Oaks: Pine Forge Press.
Singh, J. (2007). Methodology and Techniques of Social Research. New Delhi: Kanishka Publishers.
Singh, Y. K. (2005). Research Methodology. New Delhi: APH Publishing Corporation.
Zarkovich , S. S. (1960). Some problems of sampling work in underdeveloped countries. In Martin, Bulmer., & Donald, P. Warwick. (Edt. 2000). Social Research in Developing Countries: Surveys and Censuses in the Third World. London: Routledge Publishing Inc.
Internet Source
Hunt, N., & Tyrrell, S.(2001). Discuss Sampling Methods. Coventry University. Coventry: UK.
Retrieved 21-07-09 from http://www.coventry.ac.uk/ec/~nhunt/meths/introd.html
Trochim, M.K. W. (2006). Probability Sampling. Retrieved 22 July 2009 from
http://www.socialresearchmethods.net/kb/sampprob.php
13