Definition Of Sample And Probability Sampling Methods

What Is Sampling?

Reasearch Design Methods In conducting research, there are two main ways to collect data from respondents: by census or using sample. Unlike census which takes the data one by one from each member of the population, sampling is simpler way to do. Generally, sampling (or called survey) is a way to collect data from population by selecting some of them. Sampling lets you get data from some members, not all of population members, who meet the requirements . As we can define our research population clearly, then there will be no difficulties in choosing some population members to be our sample - according to several criteria. Only them who have characteristic that can represent population where they below are appropriate to be chosen.

In this course, we are not learning on statistics behind the several sampling techniques presented here, we will discuss in more general matter. However, if you have any questions or comments you can send them to us.

Then, when should we use cencus and when should we use sample?

In general, these conditions favor the use of the sample compared to census:

• When we use limited source of fund to conduct research, we had better using sampling method than cencus. Using sample can help us minimize our budget to collect respondent and for buying souvenir for them.


• When you have a very limited time to conduct research. Cencus obviously will take a very long time in process because you have ask all population member. It will even take longer time, if the respondents are separated in different places.


• Population size is huge. If there are only a few members in population you can conduct a cencus.


• Variance in population is small as each population members tends to have a similiar characteristic. Remember that the main requirement to be a sample is having same characteristic with population that she/he belongs.


• When your research potentially will cause bigger non-sampling error and smaller sampling error. It means that if you enter all population members it will make bigger discrepancy in data so that you are recommended to use survey.


• Survey is held in any conditions that cencus is impossible to conduct. For example, in quality control system in a food product we cannot take all products, open the package and eat all of them. We test them by selecting several of those products.


• Sample is usually used to get more detail answer from respondents.

Now you already know whenever a researcher should use survey or cencus. In common research, there are two type of sampling methods: probability sampling and non-probability sampling.

Probability Sampling

Probability sampling is a sampling technique in which every member of the population is known, or any member of the population has an equal chance to be a sample. Probability sampling consists of:

1. Simple Random Sampling

In this sampling technique, researchers have known populations or have a list/sample frame of all members of the population, then they took a random sample from a list of members of the population.

Example :

From 25 members of the population are already known, researchers want to take five samples. Byusing computer program, we can randomly obtained numbers 2,6,9,12 and 23

2. Systematic Sampling

This technique starts from taking the first sample at random then the next sample is taken systematically by the formula: r, r + i, r + 2i, r + 3i, r + 4i, ....., r + (n – 1)i.

Example:

From 25 members of a population, researchers want to take five samples. The first number, from 1-5, randomly chosen number 3, then the following numbers: (3 +5 =) 8, (3 +5 x2 =) 13, (3 +5 x3 =) 18, and (3 +5 x4) = 23

3. Stratified Sampling

There are several conditions which are suitable for use stratified sampling method: population elements are known, members of a single homogeneous strata and across heterogeneous strata, the differences between the strata clearly visible, the number of population element within each stratum may differ.

There are five steps in conducting a stratified sampling :

1. Determine the characteristics will be used as strata, it can be based on demographic or other characteristics that researcher wants to investigate. It is recommended that the number of strata should be less than six.


2. Divide population elements into each strata.


3. Estimate the value of variance in one stratum and variance between strata.


4. Determine the confidence level required to each strata based on information needed.


5. Specify the sample size for each strata. It can be a proporsional size or disproportional in each strata. For disproportional sample size, take more sample from strata that has higher standard deviation. Finally, select sample randomly from each strata.

Example :

From strata A we get number 2 randomly, from strata B we get 8, from strata C we get 14, from strata D we get 20 and from strata E we get 23. As each stratum has the same number of elements, each one stratum only one sample was taken.

4. Cluster Sampling

Cluster sampling is a probability sampling method that is generally aimed at reducing the cost of fund and time needed in research. This sampling method is often used when the respondents live in different locations. In cluster sampling, the primary sampling unit is no longer the individual element in the population (for example, household using home-phone) but a larger cluster of elements located in proximity to one another (for example, cities). Cluster samples commonly are used when lists of the sample population are not available. There three form of cluster sampling methods: one-stage sampling, two-stage sampling, and multistages sampling.

Here are seven steps for specifying a cluster sample:

1. Determine the degree to which those within one area are likely to be similar to one another or to interact.


2. Decide on the number of units to bee skipped between individual units, based on similarity and interaction.


3. Consider the degree of variance that’s likely to exist from one area to another.


4. Specify a minimum number of clusters that would be still; be large enough to sample the entire region adequately.


5. Divide the total sample size by the minimumnumber of clusters to obtain the number to be within each cluster.


6. Select the first or key unit in each cluster on random basis.


7. Determine the procedure for moving from the key unit to others within cluster, maintaining random selection.

Example of two-stages cluster sampling:

From five provinces A, B, C, D, E randomly chosen B,D and E to represent the population. Within each cluster then randomly chosen one upto two clusters. The resulting sample consists of population elements 7, 18, 20, 21 and 23.