PROBABILITY SAMPLING

Having chosen a suitable sampling frame and established the actual sample size required, you need to select the most appropriate sampling technique to obtain a representative sample. The basic principle of probability sampling is that elements are randomly selected in a population. This ensures that bias is avoided in the identification of the elements. It is an efficient method of selecting elements which may have varied characteristics, as the process allows for a fair representation of this variability. It also means the laws of probability and statistics apply, allowing us to make certain inferences.

FIVE MAIN TECHNIQUES THAT CAN BE USED TO SELECT A PROBABILITY SAMPLE

1. SIMPLE RANDOM SAMPLING

Simple random sampling (sometimes called just random sampling) involves you selecting the sample at random from the sampling frame. In this approach, all elements are given equal chance of being included in the sample. No one from the population is excluded from the pool. Random Sampling could be implemented through:

* The lottery method. In the lottery method of sampling, the names of the entire population are written on pieces of paper and a given number of names (or sample elements) are drawn at random. * Consulting the table of random numbers. If you already have a list of names of the population from which you wish to draw your samples, you can: 1. Number them.

2. Get a table of random numbers.

3. Close your eyes move your pencil over the table, and then point. See where your pencil stops. 4. The number where your pencil stopped will be your first sample. 5. Select the second sample.

6. Proceed until you are able to get the total number of sample you need.

A basic requirement for this method is a sampling frame for the total population. If there are 4,000 persons in the list, all names should be included in the sampling frame. It is also necessary to consider the sample size. Sample size can be computed using the Lynch et al. (1974) Formula. The formula gives a 95% reliability in estimating the sample size. It gives you the opportunity to adjust sample size according to your resource capability. You may estimate how much you can feasibly accept by adjusting the sampling error (d) that you are willing to commit. The sampling error (.025, .05, and .10) simply means the level of error you are willing to consider.

The Lynch et al. formula is:

n=NZ²p(1-p)Nd²+Z²p(1-p)

Where:

Z = 1.96 (the value of the normal variable for a reliability level of 0.95. This means having a 95% reliability in obtaining the sample size.) p = .50 (the proportion of getting a good sample)

1-p = .50 (the proportion of getting a poor sample)

d = .025 or .05 or .10 (your choice of sampling error)

N = population size

n = sample size

Given the above formula:

If: N = 3,000

d = .05

Then:

n= 30001.962x .50(1-.50)3000.052+1.96² x .50(1-.50)

= 30003.8416x .253000.0025+3.8416 x .25

= 11,524.8 x .257.5+ .9604

= 2882.28.46

= 340.56 or 341

The advantage of applying this formula is that the bigger the population, the smaller the sample size.

2. SYSTEMATIC SAMPLING

This is sampling after every regular interval. This can be undertaken if the features of the population normally characterize what would be applied to simple random sampling. The key here is to determine the population and prepare the sampling frame. Number each one. 1. Determine the sample size.

2. Determine the interval.

Interval (I) = N (population size)n (sample size)

Randomly select the first number. When you have exhausted the numbers in the list, continue counting from the beginning of the list. Another way is to pick another number as the next random start and then select the elements in the interval until all samples are determined.

3. STRATIFIED SAMPLING

Stratified sampling is a modification of random...