# Sampling Methods

There are two types of sampling techniques: Probability sampling is a random selection that allows you to draw strong statistical inferences about the entire group. Non-probability sampling involves non-random selection based on convenience or other criteria that allow you to easily collect data. It would normally be impractical to study an entire population, for example in a questionnaire survey.

Sampling is a method that allows researchers to obtain information about a population based on the results of a subset of the population without having to study each individual. Reducing the number of people in a study reduces cost and effort and can make it easier to obtain high-quality information, but this must be balanced against a large enough sample size with enough power to see a real association. (Sample size calculation is covered in Section 1B (Statistics) of the DFPH syllabus.)

## Probability Sampling

## 1. Simple Sampling

In this case, each individual is selected entirely at random and each member of the population has an equal chance or probability of being selected. One way to get a random sample is to assign each person in a population a number and then use a table of random numbers to decide which people to include.1

For example, if you have a sampling frame of 1000 people, numbered 0 through 999 , use groups of three digits from the random number table to choose your sample. So if the first three numbers in the random number table were 094, select the person labeled "94" and so on.

As with all probability sampling, simple random sampling allows sampling error to be calculated and reduces selection bias. A particular advantage is that it is the simplest probability sampling method. A disadvantage of simple random sampling is that it may not select enough people with your trait of interest, especially if that trait is rare.

It can also be difficult to define a full sampling frame and inconvenient to contact them, especially when different forms of contact are required (email, phone, post) and your sampling units are spread over a wide geographic area.

## 2. Systematic Sampling

Individuals are selected from the sample frame at regular intervals. The intervals are chosen to ensure an appropriate sample size. If you need a sample size n from a population of size x, you must select each x/n to sample. For example, if you want a sample size of 100 out of a population of 1,000, select each 1,000/100 = 10th member of the sampling frame.

Systematic sampling is often more convenient than simple sampling and easy to use. However, it can also introduce bias, such as when there are underlying patterns in the order of people in the sampling frame, such that the sampling technique matches the periodicity of the underlying pattern.

As a hypothetical example, if a group of students were asked for their opinion on university facilities, but the central list of all students in the Registrar's Office was organized in such a way that the gender of the students alternated between male and female, the choice of a uniform interval (e.g .every 20 students) would result in a sample of all males or all females. While the distortion in this example is obvious and should be easily corrected, this may not always be the case.

## 3. Stratified sampling

In this method, the population is first divided into subgroups (or strata), each with a similar characteristic. It is used when we can reasonably expect the measure of interest to vary between different subgroups and we want to ensure that all subgroups are represented.

For example, in a stroke study, we can stratify the population by gender to ensure that there is equal representation of men and women. The study sample is then obtained by taking equal sample sizes from each stratum. With stratified samples, it may also be appropriate to choose unequal sample sizes from each stratum.

For example, in a study of nurses' health outcomes, if there are three hospitals in a district, each with a different number of nurses (hospital A has 500 nurses, hospital B has 1,000, and hospital C has 2,000 ), then it would be appropriate for the Select sample numbers from each hospital proportionally (for example, 10 from hospital A, 20 from hospital B, and 40 from hospital C).

This warrants a more realistic and accurate estimate of health outcomes for nurses across the county, whereas a simple random sample would overrepresent nurses from hospitals A and B. The fact that the sample was stratified must be taken into account. in the analysis stage.

Stratified sampling improves the accuracy and representativeness of results by reducing sampling bias. However, it requires knowledge of the relevant characteristics of the sampling frame (details of which are not always available), and it can be difficult to decide which characteristics to stratify.

## 4. Cluster sampling

In a cluster sample, subgroups of the population are used as the sampling unit, rather than individuals. The population is divided into subgroups called clusters, which are randomly selected for inclusion in the study. Clusters are usually already defined, for example individual GP practices or cities could be identified as clusters. In the case of the one-stage cluster sample, all members of the selected clusters are included in the study.

Two-level cluster sampling then randomly selects a sample of individuals from each cluster for inclusion. The grouping must be taken into account in the analysis. The annual General Household Survey in England is a good example of a cluster sample (single stage). All members of selected households (clusters) are included in the survey.1

Cluster sampling can be more efficient than simple random sampling, especially when a study is conducted over a large geographic region. For example, it is easier to contact many people in a few GP practices than it is to contact a few people in many different GP practices. Disadvantages include a higher risk of bias if the selected clusters are not representative of the population, leading to higher sampling error.

## Sampling without probability

## 1. Convenience sampling

Convenience sampling is perhaps the simplest sampling method, as participants are selected based on availability and willingness to participate. Useful results can be obtained, but the results are prone to significant bias, as those who participate voluntarily may differ from those who choose to (voluntary bias) and the sample may not be representative of other characteristics such as age is. or sex. Note: Voluntary bias is a risk of all non-probability sampling methods.

## 2. Quota Sample

Market researchers often use this sampling method. Interviewers are given a quota of subjects of a certain type that they are trying to recruit. For example, an interviewer may be instructed to select 20 adult males, 20 adult females, 10 adolescent girls, and 10 adolescent boys to be interviewed about what they are watching on television. Ideally, the ratios chosen would proportionally represent the characteristics of the underlying population.

While this has the benefit of being relatively straightforward and potentially representative, the sample selected may not be representative of other characteristics that were not considered (a consequence of the non-random nature of the sample). two

## 3. Judgmental (or intentional) sampling

This technique, also known as selective or subjective sampling, relies on the researcher's judgment in choosing who to invite to participate. Researchers can implicitly choose a **"representative"** sample that fits their needs, or target individuals with specific characteristics. The media often use this approach when collecting public opinion and in qualitative research.

Judgmental sampling has the advantage of being inexpensive and profitable while yielding a variety of responses (particularly useful in qualitative research). However, in addition to being biased by volunteers, it is also prone to error in judgment on the part of the investigator, and the results, while potentially comprehensive, are not necessarily representative.

## 4. Snowball sampling

This method is often used in the social sciences when studying hard-to-reach groups. Existing subjects are asked to name other subjects they know so that the sample grows like a rolling snowball. For example, if you are conducting a survey of high-risk behaviors among people who inject drugs, participants may be asked to nominate other users for the survey.

Snowball sampling can be effective when identifying a sampling frame is difficult. However, when choosing friends and acquaintances of people already studied, there is a significant risk of selection bias (choosing a large number of people with similar characteristics or viewpoints to the originally identified person).

## Sampling Bias

There are five important potential sources of bias to consider when choosing a sample, regardless of the method used. Sampling bias can be introduced when:1

A previously agreed sampling rule deviates from this

People in hard-to-reach groups are skipped.

For example, the selected people are replaced by others if they are difficult to reach.

There are low response rates.

A stale list is used as an example frame (e.g. when excluding people who recently moved to an area).

Give your opinion if have any.