Академический Документы
Профессиональный Документы
Культура Документы
4 Sampling technique
The size of the population and the area covered was large. In order to best answer the research
random sampling and simple random sampling. The population within the cluster was
heterogenous. Each cluster was a small scale representation of the total population. Cluster
random sampling was used to reduce cost by increasing sampling efficiency. Main difference
between cluster random sampling and stratified random sampling, is that, in cluster random
stratified random sampling, analysis is done to the element within the strata. In stratified
random sampling, a random sample is drawn from each of the strata (Coolican, 1999).
Federal Department of Town and Country Planning Malaysia clusters the states in
Peninsular Malaysia into four zones: Northern, Central, Southern and Eastern. The cluster
Putrajaya F. T
State Education Departments, group the schools under the administration of the District
Education Offices. Cluster random sampling was employed by taking every second district
from the list provided by the State Education Department. This procedure is shown in Table 3-
3.
Table 3-2: Cluster random sampling to select district
Larut/Matang/Selama Pekan
C. Highland
*Source: State Education Department websites (Pulau Pinang, Perak, Pahang & Melaka)
proportion to the number of target population in the states from the selected districts.
Systematic random sampling (every third school in the list selected) was employed to select
the number of schools determined, from the list of school names provided by EMIS data,
Ministry of Education Malaysia. At school level, simple random sampling was used by Head of
Science teachers since the researcher did not have access to the name list of science teachers
Anticipated
Proportion respond
No. of No. of
of sample Sample rate
State target schools
from each size number
population involved
state (additional
40%)
Pulau
584 21.6 34 73 102
Pinang
∑ 𝑠𝑐𝑖𝑒𝑛𝑐𝑒 𝑡𝑒𝑎𝑐ℎ𝑒𝑟𝑠
𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑐𝑖𝑒𝑛𝑐𝑒 𝑡𝑒𝑎𝑐ℎ𝑒𝑟𝑠 𝑖𝑛 𝑒𝑎𝑐ℎ 𝑠𝑐ℎ𝑜𝑜𝑙 =
∑ 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑐ℎ𝑜𝑜𝑙𝑠
2701
= 526
≈ 5 𝑡𝑒𝑎𝑐ℎ𝑒𝑟𝑠
In this study, sample size was determined using the value 2701 that is the number of science
teachers who represent the total target population according to states. Utilizing the Krejcie &
Morgan’s formula (Krejcie & Morgan, 1970), sample size is calculated as below.
𝑋 2 𝑁𝑃(1 − 𝑃)
𝑠= Whereby, s = sample size
𝑑2 (𝑁 − 1) + [𝑋 2 𝑃(1 − 𝑃)]
N = Identified population size (2701)
P = estimated population rating (0.5) as
this magnitude yields maximum sample.
d = degree of maximum accuracy (0.05)
X2= Chi-square value at 1 degree of freedom
(3.841)
3.8412 (2701)(0.5)(1 − 0.5)
𝑠= = 336.45 ≈ 337 𝑡𝑒𝑎𝑐ℎ𝑒𝑟𝑠
0.052 (2701 − 1) + [(3.8412 )(0.5)(1 − 0.5)]
Literature mentioned that the norm for return rate in behaviorial sciences for individual
voluntary respondents using traditional mail was 60± 20 percent (Baruch, 1999). In anticipation
of this norm, 40% extra questionnaires were added to the above sample size. This measure is
important to minimize the impact of not only lost questionnaire but also number of unwilling
The maximum number of science teachers at lower secondary in each school as estimated
was five. However, it was more realistic to assume that a minimum of three science teachers
were teaching science in each school at lower secondary level. Using this value the number
472
𝑇ℎ𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑐ℎ𝑜𝑜𝑙𝑠 𝑖𝑛𝑣𝑜𝑙𝑣𝑒𝑑 𝑖𝑛 𝑡ℎ𝑖𝑠 𝑠𝑡𝑢𝑑𝑦 = = 157.3 ≈ 158 𝑠𝑐ℎ𝑜𝑜𝑙𝑠
3