Sampling Techniques

When the target population has been identified one has to consider how a sample is going to be formed. Before looking at sampling techniques let us first define bias and sampling frame.

Bias is the tendency of a pattern of errors to influence data in unrepresentative way

The main types of bias are:

• Selection bias
• Structure and wording bias
• Interviewer bias
• Recording bias

Sampling frame is the structure which supports the identification of each member of a population.

Example 1.6:

Customers can be identified from company records
Employees can be identified from personnel records.

Three categories of sampling techniques are normally used in business and commerce. These are: 

1. Random Sampling
   
2. Quasi – random sampling
   
3. Non – random sampling

RANDOM SAMPLING

There are two types:

Simple Random Sampling

This ensures that each member of the population has equal chance of being selected for the sample. Normally it requires the use of random sampling numbers.

Advantages	Disadvantages
Unbiased General acceptance by the layman that the method is fair.	Need for a population listing Each chosen subject need to be located and questioned Certain significant attributes may be under or over represented.

Stratified Sampling

Procedure

1. The population has to be stratified by identifying significant attributes for the current investigation and partitioning the population into groups which each have a unique combination of these levels.
   
2. The proportion of the population lying in each partition should be calculated.
   
3. The total sample size has to be split up into the properties calculated above.
   
4. Sample random is performed on each partition using the sample sizes identified in (3)
   
5. The results are then combined to obtain the required stratified sample.

Advantages	Disadvantages
The sample itself is free from bias	Extensive sampling frame is necessary Significant attributes can be subjectively selected Increased costs (due to extra time and manpower)

QUASI – RANDOM SAMPLING

This is often used when random sampling is either not possible or too expensive to consider. 

There are two methods that are commonly used:

Systematic Sampling

This method can be used where the population is listed or some of it is physically in evidence. e.g. a row of houses, items coming from a production line. This technique should be used where the population is homogeneous. The technique involves choosing a starting point at random and then choose the rest in a regular pattern.

Advantages	Disadvantages
Easy to use Can be used where no sampling frame exists	Bias can occur where there are recurring sets in the population.

Multi-Stage Sampling

This is normally employed where the population is spread over a wide geographical area.

Procedure

1. Split the area into regions
   
2. Randomly select a small number of regions
   
3. Select sub-samples from three regions proportional to the size of the area.

Advantages: - Less time and manpower is needed
Disadvantages: Possibility of bias
   Not truly random

NON – RANDOM SAMPLING

These techniques are employed when the above techniques are not possible or practical. There are two well-known techniques in this case.

(a) Cluster Sampling
This is employed where no sampling exists for a population which is distributed over some geographical area. One has to select one or more geographical areas and sample all members of the target population.
Advantages:  An alternative to multi-stage sampling
   Cheaper than other methods
 Disadvantages: Selection bias
    Not random

(b) Quota Sampling
The method involves the use of a team of interviews each with a set number (quota) of subjects to interview.
Advantages:  No non response
   Low cost and convenient
Diadvantages: Selection bias
   Not random
   Interviewer bias

Sample Size
There is no formula for calculating the sample size. Factors involved in determining the sample size are:
• Money and time available
• Aims of the survey
• Degree of precision required
• Number of sub-samples required