Business research methods and applications

¶ … sampling methods that can be used in quantitative research. The first is simple random sampling. This method allows the researcher to get a truly representative sample of the population, free from any bias. Whether simple random sampling is easy or not depends to some extent on the nature of the population. Simple random sampling does not involve targeting any sub-grouping within the population, but it does deliver a more globally representative sample.

Another method is stratified sampling, wherein the researcher targets specific sub-groups. An example of this would be using demographics to break down groups. Marketers commonly use stratified sampling to understand how specific demographics respond to their products. A company might use simple random sampling, but if it wants to test a specific ad campaign for effectiveness in a target demo, then it would use stratified sampling in order to ensure that all members of the sample are within that demo. The advantage is that you have a more targeted population, in particular one that should be more meaningful to the study. The downside is that there is no feedback gained from anybody outside of the strata, which reduces the ability of the results to be replicated universally.

Systematic sampling is another method. This is when the researcher samples every nth person. The approach is fairly easy, and would seem to deliver a random sample, but that assumes that the population is distributed randomly. If the population is not distributed randomly, the researcher will not have a random sample. This method is especially problematic if it is not blind -- for example surveying passerby on the street -- because it invites biases to creep into the selection process.

2. Random sampling is the best method by far, because if done blind it is not going to have any built-in biases. This is used when it is a reasonable to select randomly from the entire population. When that is not reasonable, then simple random sampling will not even be possible. Arguably, if possible, random sampling as the most value.

Stratified sampling is used when a researcher wants to investigate specific sub-groups, but not others. As noted above, this is common with marketers who wish to understand the characteristics of certain demographics.

Systematic sampling is often used for things like surveying passersby, such as on the street or in malls. It is also used as a proxy for random sampling, for example if you have a list of names you use systematic sampling to gain a sample from that list. What this shows is that each sampling method exists for certain situations where it is better to use each

3. Each of these approaches has its strengths and weaknesses. Simple random sampling delivers the sample with the least amount of bias. It can, however, be difficult to work with because it requires that you have access to the entire population, which in many cases is simply not possible. So it may not be the most practical approach, even though statistically it is superior. Simple random sampling, in using the entire population, may also be seen as less refined in terms of the feedback it gives. Consider a survey about marketing a new pop star. This is a product that only appeals to a few demographics -- there might be a fair bit of noise in a random sampling of the general population. It is not necessary to ask people in their 60s about the appeal of someone they may not have heard of. In such a situation, stratified sampling would be a better technique.

Stratified sampling allows the population to be broken down into smaller, more meaningful populations, which should in turn improve the quality of feedback that the survey delivers. It may have less statistical strength, but it may be more applicable. The key to making stratified sampling work is to make the right choices about what strata should be sampled -- as a more refined technique it needs to be used better in order to be more effective.