Embedded Files
POPULATION
In scientific research, the population does not necessarily refer to all people (or animals) in the world, in a country, or even in a particular city or area. The term population refers to the entire group of research interest from which a sample is drawn and to which the researcher will seek to generalise (apply) the results of their investigation. For example a researcher may wish to study the effectiveness of various cognitive tasks on patients living with alzheimer's disease. In this instance the population (or population of interest) is people living with alzheimer's.
SAMPLE
Using the example above we can see how it would be unfeasible or even impossible for us to conduct an experiment that included every single person who has alzheimer's disease. So instead of testing the entire population, we use what is known as a sample of that population. Essentially we are taking a smaller section of the population and testing them on the assumption that our experiment is well enough designed that the results may be relevant to the entire population.
There are a number of ways we can sample from our population, but ideally we want a sample that is large enough to reflect the general characteristics of the population. If achieved this is known as a "representative sample", If not, it might be considered a "biased sample" For example, I could conduct an experiment using my class, but it is unlikely that my class represents the age range, the gender spread or even the cultural diversity that is present at the school. This biased sample may have an impact on my final results and restrict my ability to generalise my findings to the wider population.
One way we can reduce sample bias is to ensure that we are sampling a large enough number of participants to allow the sample to be representative. The law of large numbers suggests that as sample size increases, the attributes (characteristics) of the sample more closely reflect the attributes of the population from which the sample was drawn. For example, the more people who are selected for an experiment, the more likely it is that they will reflect and therefore be representative of the population.
Larger samples also minimise the likelihood of an unexpected sampling error resulting in a sample which does not represent its population well and would therefore make it difficult to apply the results to that population. For example if I have a population 100 people and a sample of 1, it is impossible for that 1 person to represent all of the traits present in the population. If I sampled all 100, then my sample perfectly reflects the population. As populations are generally MUCH larger than this, it is impractical to test all members, so we need to find a good balance between generating a truly representative sample and the time it takes to experiment on that number of participants.
The two main sampling types that we look at in this study design are random sampling and stratified sampling.
Random sampling involves a process by which all members of the population of interest have an equal chance of being selected in the sample. This might be achieved through a random number generator, picking names out of a hat or any other means that is genuinely random. This method reduces bias and coupled with a large sample size, increases the likelihood that the sample is representative of the population of interest
Stratified sampling involves dividing the population of interest into sections (or strata) based on a specific characteristic before randomly sampling from those strata. For example, a representative sample of this school would include a good spread of students from across all year levels, so I might choose to randomly sample 20 students from each year level. In this scenario I have stratified my population into year levels and then sampled from those strata.
Sampling techniques that fall beyond that may be considered to be "convenient". A convenience sample is exactly as it sounds: it was used because of its convenience to the experimenter. This includes using all members of a class that I teach simply because they are easily accessible. Similarly, asking for volunteers is convenient, even asking random people at a busy city intersection might be considered convenient as it fails to take into account those living in the country. Generally these samples are NOT representative of the population of interest, and if they are it was a matter of luck and not good experimental design. This means that the results fo that experiment cannot be generalised to the wider population as the sample does not reflect the wider population.
ALLOCATION
Once you have collected a sample using one of the above techniques the last step is to allocate your participants to either the control or the experimental condition. In most cases you will only have these two groups and the process should follow a random procedure. By that we mean that all members of the population have an equal chance of being in the control or the experimental group. This could be done in a range of different ways, but
Page updated
Report abuse