Chapter Eight i
3. Sampling
methods.
Systematic sampling
Simple random
sampling
Stratified random
sampling
Cluster sampling
Multi-stage
random sampling
Quota sampling
4. Sample
size
When collecting some kinds of information it is important to choose the sample
(usually the people from whom you are going to obtain information) that will provide
the most accurate information. If statistically valid information is required, rather than
"a pretty good idea", it is best to get assistance with sampling methods.
Every person/house/seedling, etc. is given a number. Then every fifth, tenth, etc.
person/house/seedling is chosen for the sample, until the required sample size is obtained.
Where records or lists of people/households/seedlings exist, a certain number of them can
be chosen using a random sampling method. Assign each sample a number. Put all
the assigned numbers of the people/households/seedlings in a basket and pick (without
looking!) one by one, from the basket until the desired sample size is obtained. Random
sampling methods are used to reduce bias.
Groups or strata of the population of people/households/seedlings are separated (for
example people with land and landless people/large households and small households/
fruit tree species of seedlings and fuelwood species). Then each group/strata is treated as'
a separate case, and a sample established for each group/strata.
People/households/seedlings are chosen in groups or clusters and not on an individual
basis. For example, a particularly dry area with poor growing conditions might provide
one "cluster", while an area with rich soil and higher rainfall might provide another
"cluster". Within each "cluster" a random sampling method is used.
Samples are selected using simple random sampling, but at different times or stages.
For example, one stage may be 100 farms. A random sample would be chosen from
these 100 (it would be 15). The next stage would be seedlings planted. On these 15
farms there are 15,000 seedlings planted. A sample of seedlings would be 750 (5%),
or 50 seedlings from each of the 15 households. Another sampling method (every 10th
seedling in the field) can be used for each farm surveyed, so that there will be as little
bias as possible in choosing which seedling to "survey".
A certain number of samples (people/households/seedlings) or quota are required.
The person taking the information goes out looking for information, and stops when
the quota is reached. For example, going to the market and questioning people who are
willing to talk until the necessary quota has been completed. This method relies on
personal judgement, such as who is willing to talk and who is at the market
The information can thus be biased.
The following tables can help you to decide the sample size that is needed.
Total Sample Suggested Sample Percentage
100 15 15%
200 20 10%
500 50 10%
1000 50 5%
Toolbox Page 89
I I