A purposive sample is one made up of people with specific traits. For example, you may want to study only people who are between 30 and 50. Or, you could have a very specific list of traits that you are looking at - Men between 30 and 50 who have lived overseas for at least five years and have recovered from a drug problem. Obviously, if the sampling is too limited, it may make it difficult to find a large enough sample. However, by limiting the traits of the sample you are not trying to generalize to a universal population; your target population is the sample as you have defined it. Very often, purposive sampling is used in order to address a concern about a certain group of people. So, in a sample of children of refugees to the US in the IB program, the researcher may be looking at how to help this particular population cope with stressors better.

When trying to find a purposive sample, often researchers make use of social media, newspaper ads or even bulletin boards in the student lounge at the university. As a result, the volunteers come to the researcher. This is a self-selected sample. The main strength of this type of sampling is that the participants are usually motivated to take part in the study. The main limitation is that you may only get a certain type of person replying to the ads - resulting in sampling bias. For example, you only get students who hang out in the student lounge - eliminating introverts. Or, you only get people who use Facebook - which says something about the way they interact with technology.

Snowball sampling is also known as network sampling. This type of sample asks participants to recommend other participants. This is usually done when the type of people you are looking to study would not normally respond to an ad in the paper. For example, I am looking to study CEOs who used to have a drug problem. Gay men "in the closet" in a fundamentalist religious home. Women teachers who were abused as children. By using a snowball sample, people that the researcher has studied help him/her to find people. This assumes that people know people who have similar traits - and this is often the case. In addition, it helps with rapport. In studying such sensitive topics, a researcher has to build trust with the participant. However, someone has already shared their story with the researcher and recommends to a friend that the researcher is trustworthy and the research worthwhile, this goes a long way toward establishing rapport, even before the participant meets the researcher.

Quota sampling is very similar to a stratified sample. Both types of samples want to have the sample represent the demographic distribution of the larger population. So, for example, if in our school 60% of the students are female and 40% are male, then both samples want to have the same distribution so that it reflects the population. However, the difference is a philosophical one. Stratified sampling is used in quantitative research where random allocation is important. So, if I want to have a sample of 50 students in my sample, I would first get as many volunteers as possible. Let's say that I get 90 girls and 72 boys. Now, I have to randomly choose 30 girls and 20 boys in order to make the percentages work. In quota sampling, I simply accept participants until I have reached the assigned quota. So, the first 30 girls that sign up. The first 20 boys. Random sampling is not important. The advantage of this method is that it is representative of the population. The disadvantage is that since you are only taking the first participants to sign up for the study, you may experience sampling bias.