Jump to Navigation

Surveys as a Research Method: The Backbone of Quantitative Research is Within Your Organization's Reach

Research Sample

It is usually impossible for a researcher to directly measure the attitudes or behaviors of every person in the population they are studying.  Instead, they take a sample of that population for the purpose of conducting research.

Let's return to the Metamorphosis Project example.  We were interested in the level of neighborhood belonging of residents of South Los Angeles.  But we knew we could not survey every single resident in that entire population, so we went through a procedure to get a sample – a smaller but representative subset – that could be used in the place of measuring the entire population.

There are two main types of samples – Probability (or Representative) samples, and Non-Probability (or Non-Representative) samples.

Probability or Representative samples are the best sample you can have because they accurately “represent” important aspects of the population from which they are selected.  There are several different types of probability samples, but in short, each member of the population has an equal chance of being selected for the study. 

Returning to the Metamorphosis Project example, because the survey firm we worked with randomly generated telephone numbers of local community members to interview, our survey is considered representative, because each community member pretty much had an equal chance of being contacted and taking part in the research.  If a probability sample is collected, the results can be reported as “generalizable” to the population – this means that your findings within this smaller subset of individuals should hold true for the entire population, even though you did not speak to everyone in that population.  There are several different types of probability samples – see our additional resources if you are interested in learning more about these different strategies.

Non-Probability or Non-Representative samples do not provide the same types of generalizable results, but often due to a lack of time, resources or expertise, these samples are used.  They can still provide valuable insight into your population, but you must be extremely careful not to make claims that your survey definitely “represents” the population in question, because you are not sure how much your sample really represented the broader community.  For instance, you might ask every person who comes into the office of your organization to fill out a survey about problems in the community.  This is not a random sample – only people who enter your office have an opportunity to fill it out – but it still can provide some valuable information that might inform your work or lead you to ask new questions.  Once again, there are a number of different types of probability samples – see our additional resources for more information.

Keep in mind the resources and capabilities of your population in question.  Is it likely that your research population might be illiterate?  If so,opt for a method that would not require them to read.  Is it likely that your research population might not have a telephone?  If so, try to go door-to-door or survey them in a public or private space where they are likely to gather.  Is it likely that your research population does not speak English?  If this is the case, make sure your materials are available in their native language.  Does the population have internet access? If so, you might consider an online survey – there are several free and low-cost services that can be used.  Is there an available list of the population in question – either a list internal to your organization or part of the public record?  If this is the case, you could call them directly, mail or E-mail out the survey, or try to conduct the interviews in person.
Can you get participatory and enlist the help of community members in the process of sampling? They can help administer surveys at your office, at community events, or at other communication asset spots in the neighborhood.

Syndicate content