Professional Documents
Culture Documents
net/publication/233988740
CITATIONS READS
2 893
4 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Mohd Fadzil Abdul Rashid on 03 June 2014.
Abstract:
Systematic sampling is among the popular method of probability techniques. It is easier to
implement the technique compared to other approaches. However, it is found that, it is quite
difficult to select the samples (households) based on the skip intervals for larger coverage areas.
Thus, a modified approach has been introduced for implementing the systematic sampling. This
paper describes the modified approach of implementing the systematic sampling for the
migration behavioural survey in the Klang Valley. Besides that, problems of conducting the
survey and the result are discussed.
1.0 Introduction
This paper is written based on experiences of conducting the migration behavioural survey in the
Klang Valley by implementing a modified approach for the systematic sampling technique. The
survey was conducted in two periods of times that is during the months of September to November
2007 and March to May 2008. Hence, this paper is an attempt to describe the overall process of
conducting the survey. It includes introduction to the survey, implementation of the systematic
sampling using a modified approach, problems of conducting the survey, and results thereof.
The purpose of the survey is to obtain actual information about phenomenon of migration distribution
in the Klang Valley. It includes trends and distribution of migration within the areas of the Klang
Valley, characteristics of migration distribution in terms of economic, physical and social aspects, and
relative importance of migration decision-selectivity factors. Generally, the information from the
survey is to formulate a new methodology (or new model) for spatial modelling of migration potential
in the Klang Valley. The model is called as the MGP model (migration potential model). The
direction for formulating the model is the most significant area in migration studies at present. In
Malaysia, this formulation is a new dimension for internal migration studies.
Table 1 shows the areas of survey by the number of samples size. It shown that, about 10 areas have
been selected for the survey. The selection of the areas is based on the multi-stages cluster sampling.
Rashid [6] has described in details on the application of the multi-stages cluster sampling and
determining the samples size.
The total sample size of 400 households as shown in Table 1 were based on 5% precision level. The
samples size can be reduced to 204 samples by using 7% level of precision. Thus, the minimum and
the maximum number of the samples size for the survey is in the range of 204 to 400 samples.
Table 1: The areas of survey and samples size
The survey was conducted by using the questionnaires. In order to meet the objectives of the survey,
the questionnaire was organised into five sections: (A) background of respondents, (B) background of
respondents that moving out to Klang Valley, (C) migration within the Klang Valley, (D)
characteristics of migration distribution, and (E) general opinions and suggestions.
Note that, in the survey, the respondents were not required to answer all the questions (all the
sections), but only answer the sections which were related to them on the basis of status of migration.
The face-to-face interview method was applied for administering the questionnaires. Thus, in the
survey, the enumerators met the respondents directly at their home.
The target respondents were the heads of households. Note that, the family members of the
households also can be the participants in the situation when the head of household was not available.
In addition, the households were limited to whom staying in residential areas of urban. Thus,
households of hotel, hostel, squatters or slum areas and others were excluded in the survey.
As mentioned earlier, the systematic sampling technique was applied for implementing the migration
behavioural survey in the Klang Valley. In order to conduct the survey easily, a modified approach
has been introduced to implement the systematic sampling technique. Hence, this part is an attempt to
describe the implementation of the modified approach for the systematic sampling. Besides that,
overview of the systematic sampling and overall process of conducting the survey are also discussed.
The systematic sampling is one of the probability sampling techniques. See discussion such as in
Cooper and Schindler [3], Bordens and Abbott [1], McBurney and White [5], Bryman and Bell [2],
and Ghauri and Gronhaug [4]. According to Bordens and Abbott [1], this technique is the popular
technique that is often used in conjunction with the stratified sampling. Currently, this technique is
2
widely used in social sciences surveys. It is because this technique can be applied without fully
completing sampling frame or list. The samples can be selected or determined during the survey
process. According to Cooper and Schindler [3], this is the major advantage of the systematic
sampling technique. It is easier to instruct enumerators (interviewers) to choose the dwelling units
listed based on every kth element or skip interval. The kth element is determined by dividing the
sample size into the population size (dwelling units/ households) to obtain skip interval pattern
applied to the sampling frame. See discussion in Cooper and Schindler [3] and McBurney and White
[5]. For example, if the sample size is 46 of 488 dwelling units, then if starting point is 2, so the
samples would included such as the 2nd, 13th, 24th (488/46, or skip interval = 11). The steps to draw a
systematic sample as follows:
As mentioned above, the survey was conducted in the two periods of times that is during the months
of September to November 2007 and March to May 2008. The survey was delayed from November
2007 to March 2008 because of lack of budget to conduct the survey, especially to pay for the
enumerators.
The overall flows of implementing the systematic sampling technique for the survey can be illustrated
as in Figure 1.
Multi-stage
cluster sampling
10 Areas
Breakdown 10 areas
into sub-areas
Grouping enumerators
by sub-areas
Draw samples
Answered
questionnaires
Figure 1 indicates the implementation of the systematic sampling for conducting the survey in the 10
areas of survey. However, it is likely to mention that, there is a modified approach that has been
introduced to implement the systematic sampling for the survey. Detailed description on the
implementation of the modified approach is in next part.
3
3.3 The Implementation of the Modified Approach of Systematic Sampling
The modified approach is by breaking down the 10 areas of survey (see Table 1) into sub-areas (see
Figure 3.1). The introduction of the modified approach for the survey was based on two main reasons:
a) to handle the difficulty of applying the systematic sampling technique for the areas of
the survey due to larger coverage areas, and
b) to save valuable time and budget.
An example of the sub-areas for Section USJ 13 is shown as in Figure 2. Refers Figure 2, USJ 13 has
been divided into five sub-areas. By blocking the areas into sub-areas, it is found that, the
implementation of the survey is much easier and it can save time and money. This is because, based
on the modified approach, the enumerators were allocated in the different sub-areas. Thus, the
implementations of the survey for the sub-areas would be done in concurrently.
The enumerators were put alone or allowed to be in pairs depending on the size of the coverage of
sub-areas (or households). The numbers of the enumerators were only limited up to two persons for
every sub-area. Furthermore, the enumerators which were allocated to a specific sub-area were
responsible to complete the number of the samples (households) in their sub-area based on the skip
intervals that have been determined. Table 2 shows the result of skip intervals calculation for every
area of the survey.
The steps to apply the skip intervals for selecting the households for every area have been described
as in Part 3.1. By referring to Table 2, to select the households in USJ 13, for example, it involved: (a)
select the first element randomly (between 1 to 9) in the sub-areas, if the starting point is 2, so the
samples would included such as the 2nd, 19th, 36th (skip interval = 17), and (b) draw the next samples
by choosing every 17 entry (dwelling units) until obtained the total 59 samples size. The same process
was done for others 9 areas of the survey as shown in Table 2.
4
Table 2: Calculation of the skip interval for every area
Based on the modified approach, the total households for every area of the survey are the summation
of their sub-areas.
In order to achieve the desired numbers of households for every area and to follow the correct way of
applying the systematic sampling technique, two steps/ procedures have been suggested to the
modified approach. The two procedures are elaborated below:
a. starting point of selecting the households for every sub-area should be at the first
dwell unit (or which is the dwell unit becomes the first available dwell unit). It is to
easier for counting the skip intervals (or to choose the households),
b. selecting the households for every sub-area should be started from opposition/
different directions (i.e. left side and right side). It is to easier for counting the skip
intervals crossing the sub-areas boundaries (if needed). In this case, a little adjustment
on the boundaries of the sub-areas was done.
The two procedures for selecting the households for every sub-area that have been described above
can be conceptualised as in Figure 3.
3
1
Figure 3 indicates the conceptualisation of applying the two procedures of the modified approach for
selecting the households in the survey. It showed that, the households for every sub-area (i.e. sub-
areas 3 and 4) are selected from the earlier dwell unit. At the same time, for the sub-areas 3 and 4, the
5
selections of the households are started from the opposition directions. Based on these two
procedures, all the households in the areas have the same chance of being selected for the survey
samples. Thus, the introduction of the modified approach for the systematic sampling is might be
acceptable. It is found that, Cooper and Schindler [3] have recognised such kind of the approach. In
addition, it is found that, the modified approach would avoid subtle biases in the implementation of
the systematic sampling technique. See Cooper and Schindler [3] for further explanation on subtle
biases issue in the systematic sampling.
As commonly happens in a household survey, the survey faced with some problems. The problems
and the ways to resolve them are elaborated as follows:
a) if the selected respondents (head of households) are not available, for instance, not at
home, do not cooperate, then the enumerators were instructed to choose the nearest
dwelling unit for replacing the unavailable respondents, and
b) if most of the households in smaller areas (sections) such as USJ18 Subang Jaya,
USJ13 Subang Jaya, and Section 12 Shah Alam did not give full cooperation; then the
enumerators were instructed to get the maximum number of samples in the sections
as they can.
The two problems caused a failure for the enumerators to get the actual sample size for the several
areas of the survey. In the survey, the problems were considered as the constraints of the survey and it
is commonly happened to other surveys. Due to limited time and budget the survey is not repeated.
Based on the difficulty, only about 364 households of the total 400 households were collected in the
survey. The distribution of the collected samples by sections is shown as in Table 3.
% Out of Target
Areas of Study Number
Number
Taman Universiti Bangi 46 100.0
Section 15 Bandar Baru Bangi 29 100.0
Taman Semenyih Jaya Semenyih 34 100.0
USJ 4 Subang Jaya 56 100.0
USJ 18 Subang Jaya 29 76.3
USJ 13 Subang Jaya 49 83.1
Section U18 Shah Alam 15 100.0
Section 12 Shah Alam 3 20.0
Section 3 Shah Alam 29 88.0
Section U10 Shah Alam 74 100.0
Total collected samples N = 364 91.0
Three main reasons for the total 364 households are good enough for the survey are as follows:
a) the total 364 households are in the range of the minimum and the maximum of the
sample size (refers Part 2.1),
b) the total 364 households are more than 90 percent of 400 (refer Table 3), and
6
c) the selection of the samples is by using the probability sampling techniques, so that,
the results based on the collected data can be inferred to the entire of the population
of the research, that is, all households in the Klang Valley.
6.0 Conclusion
This paper has discussed in details on the process of applying the systematic sampling technique using
modified approach for the migration behavioural survey in the Klang Valley. It includes the
procedures for implementing the modified approach in the systematic sampling. Through
implementation of the modified approach, the survey is much easier and practical to be conducted and
save valuable time and budget. More importantly, the survey has succeeded to collect the appropriate
numbers of the samples with the correct application of the concept of the systematic sampling so that
it contributes to representative data for the inference to be made from the required population of the
study.
1. Mohd Fadzil Abdul Rashid is Doctorate Student (Demography) at Centre for Statistical Studies, Faculty of Computing
and Mathematical Sciences, Universiti Teknologi MARA (UiTM) Shah Alam, Selangor Malaysia.
2. Ishak Ab. Ghani is a Chartered Statistician of the Royal Statistical Society (U.K.) and Professor (Dr) at Centre for
Statistical Studies, Faculty of Computing and Mathematical Sciences, Universiti Teknologi MARA (UiTM) Shah Alam,
Selangor Malaysia
REFERENCES
[1] Bordens, K. S., and Abbott, B. B. (2005). Research Design and Methods: A Process
Approach, McGraw Hill, New York.
[2] Bryman, A., and Bell, E. (2003). Business Research Methods, Oxford University Press, New
York.
[3] Cooper, D. R., and Schindler, P. S. (2006). Business Research Methods, McGraw-Hill, New
York.
[4] Ghauri, P., and Gronhaug, K. (2005). Research Methods in Business Studies: A Practical
Guide, Prentice Hall, United Kingdom.
[5] McBurney, D. H., and White, T. L. (2004). Research Methods, Thomson Wadsworth, United
Kingdom.
[6] Rashid, M.F.A. (2009). Spatial Modelling for Distribution of Migration Potential in the
Klang Valley (3rd Draft Ph.D Thesis), Centre for Statistical Studies, Faculty of Computing
and Mathematical Sciences, University of Technology MARA (UiTM), Shah Alam.