Assalamualaikum, hello and salam sejahtera everyone.
For this entry, I'll explain more about SAMPLING DESIGN. Saya akan refer buku Research Methodology and Data Analysis, written by Prof Dr Zainudin Awang sebab for me, buku inilah yang paling complete dan senang untuk faham :'3 kalau nak lebih details, boleh beli buku direct dari Prof ya? Hiks.
Well everyone knows that kalau
population adalah total set of items of interest yang kita berminat nak kaji. Subset untuk population pula dipanggil
sample.
The finding from that sample is generalizes on the whole population, and that is why the researcher needs to apply the proper sampling technique to obtain sample yang representation untuk population for our research project.
( Tak boleh lah kalau kita nak suka-suka macam main ambik je any sample lepastu kita buat conclusion daripada sample tu melambangkan population. Tu dah bias tuuuuu ha kenot-kenot )
Based from this book, dinyatakan main reasons for studying sample instead of whole population:
- Cost constraint - cost untuk kaji sample < cost untuk kaji the whole population
- Time constraint - working time with sample < working with whole population
- Manpower constraint - logically, bila sample sikit, manpower yang diperlukan adalah sikit
Factors that should be considered in choosing the appropriate sample design:
- Research Objectives & Research Questions
- Degree of Accuracy Required
- Availability of Resources
- Time constraint
- Scope of Research
- Statistical Analysis required
Okay selepas dah consider kan segalanya yang perlu di consider di atas, barulah kita proceed dengan which sample design should we used?
There are 2 types of sampling design:
Probability Sampling Design
- Commonly used because the sample is selected at random, all element inside the population has equal chances to be selected.
Type of Probability Sampling Design:
- Simple Random Sampling (SRS)
- Applicable for homogeneous population
- Characteristics : gender, socio-economic status, ethnic group, culture, religion
- Allow for equal chances for any element in the population to be selected as sample
- Steps:
a. The definition of target population
b. List of all elements in population (sampling frame)
c. Assignation of numbers to each element of population
d. Obtain the table of random numbers / generate it from the computer
e. Match the random numbers obtained to the numerical list of population
f. Select the selected element as the respondents for data collection
- The respondents selected this way are totally random, hence the problem of biasness does not arise.
- Disadvantages: requires sampling frame, may take lots of time, and high cost.
- Applicable for homogeneous population
- The element which falls on kth number on the list is chosen as a respondent.
- This sampling procedure fixed the increment in order to determine the following respondent.
- Steps:
a. Determine the size of population (N) and required sample size (n)
b. Determine the interval, k=N/n
c. Select one random number between 1 to k.
d. Then the following respondent are K+k, K+(K+k), .........until the nth respondents is selected
- Applicable when the population is not homogeneous, then the researcher has to stratify the element into homogeneous groups.
- The process of selecting a sample that represents each stratum in a population
- The researcher can stratify the heterogeneous population into homogeneous population within a stratum in terms of socio-economic status, ethnic group, culture, religion, marital status, type of house, etc.
- Steps:
a. Specify the strata (within homogeneous, between heterogeneous)
b. Assign the elements according the strata having similar characteristics of interest in study
c. Select sample within each strata using SRS since the population is already uniform.
d. The number of samples from each stratum must be proportionate to the number of units in the stratum.
- Advantages : The estimate obtained from the sample is more accurate & researcher can compare whether different strata perform differently regarding certain variable of interest.
- Applicable when the study covers a large geographical area, the population is not homogeneous, sampling frame is not available.
- The clusters can be : residential area, growth corridor in country, development region
- Steps:
a. Specify the clusters of interest within the population
b. Select randomly a few clusters from the available clusters.
c. The respondents are selected from the selected clusters.
Non-Probability Sampling Design
- The element in the population does not have equal chances of being selected as a sample.
- The result of study is only applicable to the particular sample, do not reflect the whole population
- However, the sampling is very economical & easily performed.
Type of Non-Probability Sampling Design:
- The sample is obtain at convenience.
- The procedure is not random because the respondent is selected because of they happen to be at the right places and at the right time.
- Advantages : least expensive, least time consuming, sampling units are accessible
- Disadvantages : data obtained is not randomly distributed, selection bias, sample is not representative, not suitable for descriptive & causal research.
- The respondents are selected based on the judgement of researcher that they have the required characteristics to be included in the study
- Similar to stratified sampling in that a particular stratum is the focus of the study
- Respondents are selected based on certain characteristics of interest by the researcher.
- Researcher will selects one respondent who fulfills certain characteristics for the study
- After obtaining the required data from the first respondent, the researcher will asks this particular respondent to locate his friends/anyone who possess the same characteristics.
- The second respondent will also help the researcher to locate the next respondent.
- Typically used when : the target population is small, the characteristics of population is unique, no sampling frame.
Well kalian boleh jugak refer
https://www.ma.utexas.edu/users/parker/sampling/srs.htm untuk lebih lanjut tentang type of sampling design ni :3
Semoga dapat input sikit-sikit dari entry ni. Jangan lupa refer buku untuk lebih mendalam occay!
By the way, if ada readers yang interested untuk dapatkan ebook / software AMOS / file tentang Research Methodology dari Prof Dr Zainudin, boleh email ke saya di :
adiyanasuhaimi@gmail.com
Harga? FOC, don't worry. Sharing is caring eit ;) tapi kalau ada sesiapa yang nak share notes or anything yang boleh dijadikan panduan atau apa-apa saja yang berkaitan dengan ilmu statistics, dialu-alukan untuk email saya okay? Kecil tapak tangan, nyiru saya tadahkan. Gittew :p
And kalau ada sesiapa yang berminat untuk dapatkan servis (amboinyah!) or any helps regarding AMOS or SPSS software, boleh lah email jugak heeeee. Servis kita kira kawan-kawan lah hiks.
See you soon!
Byeeeeeee :]