Sampling techniques- terms

Reasons to Use Snowball Sampling

- when you are trying to reach populations that are hard to find or are otherwise inaccessible. - Finding subjects who would normally be reluctant to participate; may be more willing participate because they trust the person who referred them.

Observation unit

Element from which information is collected; the actual unit that we include in our sample. The term often refers to individuals (people), but it could be families, companies, cities, countries, etc. (whatever you are studying). Observation units are sometimes referred to as subjects or participants.

EPSEM

Equal Probability of Selection Method. The fundamental principle of probability sampling is that a sample is likely to be representative if it is selected through a method in which each unit or case in the population of interest has an equal probability of being selected for the sample.

4. Cluster sampling

Involves random sampling in which each sampling unit (at least until the final stage) is a collection or cluster of elements, usually based on geography. Samples are selected in stages. The process of selecting a sample in a large city might begin with randomly selecting voting precincts, then blocks within precincts, then studying all of the households on the selected blocks.

Study population

It is unlikely that you will have an accurate listing of the entire population, and even if you did, you may not be able to reach them all, so some make a distinction between the population you would like to generalize to (the theoretical population), and the population that will be accessible to you (the study population). The study population is the aggregation of elements from which the sample is actually collected.

Error in sampling

Made up of two components (systematic error and random error). Systematic error or bias There is a flaw in the sampling procedure so that not all members of the population have an equal and independent chance of being included in the sample. Random error Sampling variability; occurs due to chance factors. Even when everything is done right in sampling, random variation can be expected (the sample will not perfectly match the population, simply by chance).

Probability sampling methods

One can specify for each sampling unit of a population the probability that it will be included in the sample. (The probability of being selected is known for each member of the population.) These methods select their samples using random selection techniques.

3. Purposive/judgment sampling

Researchers select participants with a purpose in mind. They are seeking people with certain characteristics. The researcher uses his or her own judgment about whether a person fits the criteria. A person's chance of being selected depends on the subjective judgment of the researcher. They might be looking for college students, or white females between the ages of 30 and 40, or parents with children under 6. They size up those passing by, and try to stop anyone who looks like they fit the category of interest to ask if they would be willing to answer a few questions.

1. Convenience sampling (also known as available subjects sampling)

Researchers select whatever sampling units are available to them, often the closest live subjects, as when a professor uses students in an introductory class as survey respondents.

4. Snowball sampling

Sample that starts out small, and becomes bigger and bigger as it goes along. This is conducted in stages. 1. identify and study a few people having the characteristics of interest 2. use those people to help you identify others with the characteristic who might participate 3. study those subjects and ask them to identify others.

5. Expert sampling

Selecting a sample based on their expertise in some area.

6. Modal instance sampling

Subjects are selected because they fit the image of the "typical" case (i.e. the typical voter, the typical college student, the typical stay-at-home mother, the typical church goer, etc.)

2. Quota sampling

The chief aim is to select a sample that has certain characteristics (generally to reflect characteristics of the population in the same proportion as they are found in the population). • Researcher decides which characteristics are relevant to their study (i.e. Republicans and Democrats for a study of voting behavior; blacks, whites, Asians, Native Americans, and Hispanics for a study of racial and ethnic relations). • Researcher determines each group's representation in the population • Uses representation to set a quota for each group; determines quotas • Researcher finds people/sampling units with the characteristics of interest, continuing until all of the quotas are filled.

1. Simple random sampling

The most basic probability sampling method. Each individual unit has an equal (and known) chance of being selected. You need— • a list of all sampling units in the population • a system for selecting units from the list that will guarantee that each unit has an equal chance of being selected - A common procedure is to assign numbers to each person or sampling unit (so that selection cannot be based on names, labels, or any other identifier) and then pick # at random. (They could be drawn from a fishbowl or hat, or selected using random number table or program.)

Population

The total group or the total collection of entities that the researcher is interested in; the entity that the researcher wants to understand. The total collection of entities (people, housing units, schools, etc.) about which information is desired. It is the group that we want to generalize to.

Nonprobability sampling methods

There is no way to specify the probability that each unit has of being selected, and there is no assurance that every unit has even a chance of being selected. - The disadvantage of these methods is that, since the probability of a unit being selected is not known, the investigator cannot claim that the sample is representative of the population. This greatly limits generalizability beyond the specific sample used in their study.

7. Heterogeneity sampling

This involves selecting for diversity. - Researchers are trying to maximize the variety represented in their sample

Sampling

the process of selecting units of observation (participants) for a research project. - The goal of this is to select a sample in such a way that descriptions of the characteristics of the sample (statistics) accurately reflect the parameters of the population from which the sample was drawn.

WHEN A NONPROBABILITY SAMPLE IS ADEQUATE

- If the researcher is not interested in generalizing the results beyond the sample - when a population cannot be easily defined - when a list of the population is unavailable o (i.e. there isn't a list of drug addicts or homeless people in this country; making it difficult to come up with a method of using pure probability sampling to study those segments of the population). - for exploratory research - for pretesting the methodology to be used for a larger study (that uses probability sampling).

Reason to Use Cluster Sampling

- When collecting data from a population that is spread across a wide geographic area There are concerns, though, because there is a possibility of error at each stage in the selection process, rather than in a single stage, as is the case with the simple random sample.

Sampling frame

A list of sampling units from which the sample, or some stage of the sample, is selected. It is a list of all of the members of the population from which a sample is drawn; the pool from which the sample is obtained. (Actually, it is not always a list. It could be directions that spell out in detail the procedures to be used to select the sample in a way that will assure representativeness.)

Representative

A sample is representative if it accurately reflects/reproduces the important characteristics of the population from which it was drawn.

Sample

A subset of the population, selected to provide us with information about the population. - A subgroup of the population which is observed and measured; the results are used to draw conclusions about the population.

Sampling unit

An element considered for selection in some stage of sampling ( it could be an individual, a household, a school, a census tract, or whatever).

2. Systematic sampling

This type of sampling is not random, but is often considered close enough for practical purposes. Researchers take a complete list of the population and sample every kth sampling unit on the list, beginning at some randomly selected point. The value of k will be determined by the size of the population and the desired sample size. If you have a population of 10,000, and you want use a sample size of 500, you divide 10,000 by 500, to come up with 20. You randomly select a starting point (#346 on your list, for example), and you start with that unit, and select every 20th unit after that, going back to the beginning of the list when you reach the end.

3. Stratified sampling

You divide (stratify) the population into sublists, based on some trait that you feel is relevant. Then you select a random sample from each of the sublists, with the number of cases selected from each sublist being proportional to their representation in the population. An alternative method would involve weighting cases according to their representation in the population. If you were interested in making sure that the full range of attitudes of a minority group are represented, but, based on the sample size, a stratified sample of the type I just described would include only a few individuals from that minority group, you might want to select a larger sample from that group, and weight the results to match their representation in the population.