In most surveys, access to the entire population is nearly impossible. However, the results from a survey with a carefully selected sample will reflect extremely close results as if provided by the population.
Sampling method, therefore, is an essential part of the research process. If you have surveyed using an appropriate sampling technique, you can be confident that your results will be generalized to the population in question. If the sample were biased in any way, for example, in a survey for older and younger people, if the selection technique gives older people more of a chance of selection than younger people, it will be inadvisable to make generalizations from the findings.
There are essentially two types of sampling methods: probability and non-probability sampling.
1. Probability Sampling Methods:
Probability or random sampling gives all members of the population a known chance of being selected for inclusion in the sample and this does not depend upon previous events in the selection process. In other words, the selection of individuals does not affect the chance of anyone else in the population being selected.
Many statistical techniques assume that a sample was selected on a random basis. There are four basics types of random sampling techniques.
1.1 Simple Random Sampling:
This is the ideal choice as it is a perfect random method. Using this method, individuals are randomly selected from a list of the population and every individual has an equal chance of selection. This method ideal, but if it cannot be adopted, then any one of the following alternatives may be chosen if there is any shortfall in accuracy.
1.2 Systematic Sampling
Systematic sampling is a frequently used variant of simple random sampling. When performing systematic sampling, every kth element from the list is selected (this is referred to as the sample interval) from a randomly selected starting point. For example, if we have a listed population of 6000 members and wish to draw a sample of 200, we would select every 30th (6000 divided by 200) person from the list. In practice, we would randomly select a number between 1 and 30 to act as a starting point.
The one potential problem with this method of sampling concerns the arrangement of elements in the list. If the list is arranged in any kind of order e.g. if every 30th house is smaller than the others from which the sample is being recruited, there is a possibility that the sample produced could be seriously biased.
1.3 Stratified Sampling:
Stratified sampling is a variant of simple random and systematic methods and is used when there are a number of distinct sub-groups, within each of which it is required that there is full representation. A stratified sample is constructed by classifying the population in subpopulations (or strata), based on some well-known characteristics of the population, such as age, gender or socio-economic status, usually by random or systematic sampling methods.
Stratified sampling methods also come in two types – proportionate and disproportionate.
In proportionate sampling, the strata sample sizes are made proportional to the strata population sizes. For example, if the first strata is made up of males, then as there are around 50% of males in the Karnataka population, the male strata will need to represent around 50% of the total sample.
In disproportionate methods, the strata are not sampled according to the population sizes, but higher proportions are selected from some groups and not others. This technique is typically used in a number of distinct situations.
The costs of collecting data may differ from sub-group to sub-group.
We may require more cases in some groups if estimations of population values are likely to be harder to make, i.e., the larger the sample size (up to certain limits), the more accurate any estimations are likely to be.
We expect different response rates from different groups of people. Therefore, the less co-operative groups may be over-sampled to compensate.
1.4 Cluster or Multi-stage Sampling:
Cluster sampling is a frequently-used and usually more practical, random sampling method. It is particularly useful in situations for which no list of the elements within a population is available and therefore cannot be selected directly. As this form of sampling is conducted by randomly selecting sub-groups of the population, possibly in several stages, it should produce results equivalent to a simple random sample.
The sample is generally done by first sampling at the higher levels, e.g., randomly sampled countries, then within these postcodes, then within the households, until the final stage is reached, at which point the sampling is done in a simple random manner, e.g., sampling people within the selected households. The ‘levels’ in question are defined by sub-groups into which it is appropriate to subdivide your population.
Cluster samples are generally used if:
- No list of the population exists.
- Well-defined clusters, which will often be geographic areas exist.
- A reasonable estimate of the number of elements in each level of clustering can be made.
- Often the total sample size must be fairly large to enable cluster sampling to be used effectively.
2. Non-probability Sampling Methods:
Non-probability sampling procedures are much less desirable, as they will almost certainly contain sampling biases. Unfortunately, in some circumstances such methods are unavoidable.
In a market research context, the most frequently adopted form o non-probability sampling is known as quota sampling. In some ways, this is similar to cluster sampling in that it requires the definition of key-sub-groups. The main difference lies in the fact that quotas (i.e., the number of people to be surveyed) within subgroups are set beforehand (e.g. 25% of 16-24 years old, 30% of 25-34 year old, 20% of 35-55 year old, 25% of 56 year old) usually proportions are set to match population distributions. Interviewers then select respondents according to these criteria rather than at random criteria rather than at random. Grouping is subjective. It means that only about a proportion of the population has a chance of being selected in a typical quota sampling strategy.
If you are forced into using a non-random method, you must be extremely careful when concluding. You should always be honest about the sampling technique used and that a non-random approach will probably mean the biases are present in the data. To convert the sample to be representative of the true population, you may want to use weighting techniques.
The importance of sampling should not be underestimated, as it determines to whom the results of your research will be applicable. It is important, therefore to give full consideration to the sampling strategy to be used and to select the most appropriate. Only when you have no choice should a non-random method be used.
All too often, researchers succumb to the temptation of generalizing their results to a much broader range of people than those from whom the data was originally gathered. This is poor practice and you should always aim to adopt an appropriate sampling technique. It is better to take the advice of experts.