Hi, was wondering if you could tell me what a self selecting sample is, what are the disadvantages of such a sample and what are the alternatives? Thanks
I will give you an overview of sampling methods. Sampling methods can be divided into RANDOM and NON-RANDOM methods.
Random sampling means that everyone in the population has an equal chance of appearing in the sample, thus ensuring the sample is unbiased. Simple random sampling involves selecting people through the use of a random number generator on a computer, up to the required number. Alternatively, there is systematic random sampling, where every 10th, 20th etc name on a list (such as the electoral register) can be selected, depending on the sample size needed. Stratified random sampling divides the research population into a number of strata based on what are regarded as the significant variables, such as gender, age, ethnicity or class. Samples are then randomly drawn from each of the strata and combined to form the final sample. Stratified sampling allows the researcher to ensure that all potential target groups within a population are represented in the final sample, and that all variables considered potentially important are covered. Cluster sampling can be used whenever no sampling frame is readily available. It is based on identifying a number of clusters in the population, such as schools or classes within schools, and then selecting individuals from within these clusters. This is quicker than random sampling, but may of course be biased; the pre-selection of clusters means that not every individual has an equal chance of being selected. Multi-stage random sampling involves the selection of a sample through various stages - each stage involves the selection of a sample from the previous sample chosen, until the researcher arrives at a list of individuals. Spatial sampling involves the study of participants at a particular event (such as a demonstration or open air concert), with individuals chosen randomly from those assembled there.
There are, however, times when the researcher might want to select a sample that is NOT representative of the population - in other words, a non-random sample. Accidental sampling involves the selection of all individuals the researcher happens to come into contact with in a given period. Usually here the aim is to obtain qualitative data. Purposive samples involve the selection of people on the basis that they are likely to be relevant to the subject being studied - this means that the sample reflects judgements made by the researcher, which might be open to question. Volunteer sampling is based on people volunteering to be studied (they answer advertisements or leaflets or posters or radio/TV appeals) - clearly this yields useful information (since those involved are committed to the project) though is not remotely representative. Quota sampling resembles stratified sampling - instead of choosing randomly from strata within the population, here the researcher sets a quota precisely outlining the number of people meeting certain criteria that are to be included in the sample. So, for example, it might include the first twenty white women who appear to be under thirty who pass by. This tends to be the basis of national political opinion polls and market research. In the case of snowball sampling, researchers start with very few people, and ask them for recommendations of further people to interview who fit the criteria of the study; when interviewing these people the same procedure is applied and gradually a sample is built up. It is a method often associated with participant observation.
Both snowball and volunteer samples are essentially self-selecting. They are used wherever the researcher would otherwise have great difficulty in obtaining people for their sample. This could be because no lists for a sampling frame are available, or where the research population may be so small that normal sampling methods would not yield the required numbers, or where members of the research population might not want to be identified. Burglars, heroin users, football hooligans, collectors of ancient Greek coins, gay men and members of a Masonic Lodge might fall into this category! This is why the researcher resorts to the use of like-minded or like-situated individuals. These methods have the obvious benefit of creating a sampling frame where other methods might fail to do so. However, they are highly unlikely to provide a representative sample since they are in no way random, and are entirely reliant on personal recommendation or volunteering.
I hope this is helpful.