|Year : 2019 | Volume
| Issue : 3 | Page : 157-163
Types of sampling in research
Department of Cardiology, AIIMS, New Delhi, India
|Date of Submission||01-Nov-2019|
|Date of Decision||20-Nov-2019|
|Date of Acceptance||28-Nov-2019|
|Date of Web Publication||20-Dec-2019|
Dr. Pooja Bhardwaj
AIIMS, New Delhi
Source of Support: None, Conflict of Interest: None
Sampling is one of the most important factors which determines the accuracy of a study. This article review the sampling techniques used in research including Probability sampling techniques, which include simple random sampling, systematic random sampling and stratified random sampling and Non-probability sampling, which include quota sampling, self-selection sampling, convenience sampling, snowball sampling and purposive sampling.
Keywords: Sampling, statistics, methods, cluster, snowball
|How to cite this article:|
Bhardwaj P. Types of sampling in research. J Pract Cardiovasc Sci 2019;5:157-63
| Introduction to Research|| |
Research in common language means to search for knowledge.
Research is made up of two words – Re + cerchier derived from old French recherchier meaning to search.
Definition of research
D. Slesinger and M. Stephenson in the Encyclopaedia of the Social Sciences define research as “the manipulation of things, concepts or symbols for the purpose of generalising to extend, correct or verify knowledge, whether that knowledge aids in construction of theory or in the practice of an art.”
According to Clifford Woody, research comprises defining and redefining problems; formulating hypothesis or suggested solutions; collecting, organizing, and evaluating the data; making deductions and reaching conclusions; and at last carefully testing the conclusions to determine whether they fit the formulating hypothesis.
Research can be taken as the contribution to the existing bundle of knowledge, making it more advanced.
The main objective of research is to know or to find out the answers to questions in a scientific way.
Some of the general objectives of research are as follows:
- To know about a subject or to find out something new in that – exploratory or formulative research
- To know about the subject in depth, for example, the characteristics, nature of a particular group, or individual-descriptive research
- To correlate the association of some particulars with something else – diagnostic research.
There are different types of research; some of them are listed below:
- Descriptive and analytical
- Applied and fundamental
- Quantitative and qualitative
- Conceptual and empirical
- Other types include clinical, historical, and conclusion oriented.
There are different steps which provide a useful procedural guideline regarding the research process, some of the steps are as follows:
- Formulating the research problem
- Extensive literature survey
- Hypothesis developing
- Preparing research design
- Determining the sample size
- Collecting the data
- Execution of the project
- Analysis of data
- Hypothesis testing
- Generalization and interpretation
- Preparation of report or presentation of the results.
According to the above steps, we have to prepare the research design and determine the sample size to carry out a complete research. Hence, we will discuss in detail about the different types of sampling or the sample designs.
| What Is Sampling|| |
Sampling is defined as a procedure to select a sample from individual or from a large group of population for certain kind of research purpose. There are different advantages and disadvantages of sampling. We would be thinking sometimes that – Why there is a need of sampling? the answer is as it is too expensive and too time consuming to survey a whole population in a research study, we use sampling [Figure 1].,,,,
Advantages and disadvantages of sampling
- Saves time and money and gives faster results as the sample size is smaller than the whole population
- Sampling gives more accurate results as it is performed by trained and experienced investigators
- When there is large population, sampling is the best way
- Sampling enables to estimate the sampling errors. Hence, it assists in getting information concerning to some characteristics of the population
- Study of samples requires less space and equipment as they are small in size
- When there is limited resources, sampling is best.
The main disadvantage of the sampling is chances of bias. But, seeing so many of advantages, sampling is the best way to proceed in a research.
Types of sampling
Before we discuss about the different kinds of sampling, let us discuss about what the word sample mean.
In research term, a sample is a group of people, objects, or items that are taken from a large population for measurement. So, to get the accurate results, sampling is done.,,,,
For example, if we have to check all the chips in a factory made are good or not, it is very difficult to check each chip, so to check, we will be taking a random chip and check for its accurate taste, shape, and size.
Hence, sampling is an important tool in research, when the population size is large. Based on this, we have divided it into two types: (1) probability (2) and nonprobability [Figure 2].
These two types of sampling are further divided into the following subtypes:
In this type of sampling, there is a known probability of each member of the population of being selected in the sample. When population is highly homogenous, there are high chances of each member of being selected in a sample. For example, in a bag full of rice, if we want to pick some rice, there are high chances of each rice grain of being selected in a sample. Hence, the sample collected will be a representative of the whole rice bag.
For such a study, the population serves as relatively a homogenous group as every member of the population is the target respondent of the research [Figure 3].
|Figure 3: Sampling: Example of probability, Probability to be a sample of all members is equal in this population.|
Click here to view
Simple random sampling
In this type of sampling, the members of the sample are selected randomly and purely by chance. Hence, the quality of the sample is not affected as every member has an equal chance of being selected in the sample.
This type of sampling is best for population which is highly homogenous.
There are two different ways in which this type of sampling is carried out:
Lottery method/envelope method
In this method, we assign unique numbers to each member or element of the population, say in a population of 100 members, we give number from 1 to 100 to the members on a paper and keep it in a box. Then, we will take out any chit, and the number on that chit is a random sample.
However, in this method, when the population size is larger, it is difficult to write the name of every number on the chits. Hence, another method is used, i.e., random number table (which will be discussed later).
Another example given is an envelope method, say we want to select dilated cardiomyopathy patients DCM patients for yoga in a research project. The details of the 100 patients will be in each envelope and any one will be selected randomly. Hence, here, the chances of all patients to be selected as a sample are equal.
Random number table method
There are different number of random tables available, for example, Fisher's and yates tables and Tippets random number.
Here also, first we assign numbers to the population. If we have population of 20 and we have to choose five samples from this, we have to choose five random numbers from the table. For example, we choose – 12, 19, 01, 08, and 15. Hence, members of these numbers will be selected as the sample.
Types of simple random sampling
In the above section, we have discussed about the methods of doing simple random sampling. In this section, we will be discussing about the types of simple random sampling.
There are two types of simple random sampling:
- Simple random sampling with replacement (SRSWR)
- Simple random sampling without replacement (SRSWOR).
Simple random sampling with replacement
Selecting “n” number of units out of “N” units one by one in such a way that at each stage of selection, the sample each unit has equal chance of being selected, i.e., 1/N.
Simple random sampling without replacement
Selecting “n” number of units out of “N” one by one at any stage of selecting a sample in such a way that anyone of the left units have the probability of being selected as a sample, i.e., 1/N.
For example, if we want to know the number of turtles in a pond of a village, so if we are catching turtles from water, measure them, and return them to water, there are high chances that we choose the same turtle, this is SRSWR. However, if we take out the turtle from the water and don't return it without taking the next, it becomes SRSWOR.
Stratified random sampling
In this type, the population is first divided into subgroups called strata on the basis of similarities and then from each group or strata, the members are selected randomly.
Here, the purpose is to address the issue of less homogeneity of the population and to make a true representative sample.
For example, in a school of 1000 students, if we want to know how many of them will choose medical as their career, asking each student is difficult. Hence, as inquiring the whole class is difficult, we will ask few grades and from them, we will choose samples.,,,,
For example, consider the following number of students in the class:
Grade No. of students “n”
- Grade – 6 – 50
- Grade – 7 – 50
- Grade – 8 – 100
- Grade – 9 – 100
- Grade – 10 – 200
- Grade – 11 – 200
- Grade – 12 – 300
Now calculate the sample of each grade using the following formula:
Stratified sample: n6= 100/1000 × 50 = 5, n7= 100/1000 × 50 = 5…. and so on.
So, in this, from each grade, five samples will be selected, and these will be selected according to the simple random method.
This type of sample is also called random quota sampling.
There should be classification on the basis of age, socioeconomics, nationality, religion, and other such classifications.
Detailed steps to select stratified random sample:
- First, we will target the audience
- Then, we will recognize the stratification variables which should match with the research objective and then will figure out the number of strata to be used
- After gathering the information of stratification variables, we will create a frame on this basis for all elements in target audience
- The whole population is then divided into different strata which will be unique and different from each but should cover each and every element/member of population. But, each member should be in one strata only
- Now, we will assign random, unique number to each element
- Then divide the number of samples to be taken with the total number of population into number of people in that group
- The number now what we got is the number of samples to be selected for that particular strata. Here, we will use the simple random technique.
Types of stratified sampling
There are two types – (a)-Proportionate stratified random sampling – in this type, the sample size is directly proportional to the entire population of strata, i.e., each strata sample has the same sampling fraction. (b) When the sample size is not proportional.
Examples – in a medical college of 1000 students doing postgraduation (PG), there are five different branches of doing PG and we want to study the reading pattern of all the students. Hence, it is highly difficult for us to go and ask every PG student. So, here, we will divide the class according to the subjects and then according to the formulation, we will count each number of samples to be taken from each stratum.
In another example – if in a study a researcher wants to study which sex, male or female, is predominantly affected by heart failure and what are the causes behind that. He/she will divide the given population into two groups – one male and then female. According to the stratified formula, the number of males and females to be selected from each strata will be counted and then the members in sample with simple random method will be selected.
From 1000 people, 700 males and 300 females, according to which if we want to choose 100 people, then 70 males should be selected and 30 females should be selected, and this selection will be random.
Importance of this sampling
- The main advantage of this sampling is that it gives better accuracy in results as compared to other sampling methods
- It is very easy to teach and easy to grasp by the trainees
- Even smaller sample sizes can also give good results using strata
- We can divide the large population into different subgroups/strata according to our need.
When to use stratified random sampling
- When we want to focus on a particular strata from the given population data
- When we want to establish relationship between two strata
- When it is difficult to contact/access the sample population, this method is best as samples are easily involved in research with this method
- As the elements of samples are chosen from some specific strata, the accuracy of statistical results is higher than that of simple random sampling.
Systematic sampling is an advanced form of simple random sampling, in which we need complete data about the population.
In this, a member is selected after a fixed interval. The member thus selected will be known as the Kth element.
Steps to form/select the sample using systematic sampling:
- First develop a well-defined structural population to start on sampling aspect
- Figure out the ideal size of sample
- After deciding the sample size, assign number to every member of sample
- Then, the interval of the sample is decided.
For example, we want to select a total of ten patients from a group of forty, then the Kth element will be selected by dividing 40/10 = 4, so every 4th patient will be taken for sampling – 4, 8, 12, 16, 20, 24, 28, 32, 36, and 40.
Types of systematic sampling
Linear systematic sampling
A list is made in a sequential manner of the whole population list. Decide the sample size and find the sampling interval by formula: K = N/n, where K is the Kth element, N is the whole population, and n = number of samples. Now, choose random number between 1 and K and then to the number what we got add K to that to get the next sample.
Circular systematic sampling
In this, first, we will determine sample interval and then select number nearest to N/n. For example, if N = 17 and n = 4, then k is taken as 4 not 5 and then start selecting randomly between 1 to N, skip K units each time when we select the next unit until we get n units. In this type, there will be N number of samples unlike K samples in linear systematic sampling method.
| Advantages|| |
- It is very easy to create, conduct, and analyze the sample
- Risk factor is very minimal
- As there is even distribution of members to form a sample, systematic sampling is beneficial when there are diverse members of population.
In cluster sampling, various segments of a population are treated as cluster, and members from each cluster are selected randomly.
Cluster sampling and stratified sampling are different from each other.
In stratified sampling, the researcher is dividing the population into subgroups on the basis of age, sex, profession, etc., but in cluster sampling, we are selecting randomly from already-existing or naturally occurring groups/cluster, for example, towns within a district and families within a society.
For example, in a city, if we want to know the list of individuals affected by HBsAg, here it is difficult to find, but if we search area wise, we may get better results. Here, the area acts as a cluster and the individuals will be treated as sampling unit.
In this method, first, we make clusters according to our need and then we select sample according to simple random sampling/systematic sampling.
As the name suggests, it contains many stages and hence called multistage sampling.
In this, each cluster of samples is further divided into smaller clusters and the members are selected from each smaller cluster randomly. It is a complex form of cluster sampling
Naturally, groups in a population selected as cluster
Each cluster is divided into smaller cluster
Then, from each smaller cluster, members are selected randomly.
Nonprobability sampling is a type of sampling where each member of the population does not have known probability of being selected in the sample. For example, to study the impact of child labor on children, the researcher will search and interview only the children who are subjected to child labor.
It is of the following types:
In this type of sampling, according to the purpose of the study, the members for a sample are selected. It is also called deliberate sampling. It is also called judgmental sampling.
For example, to study the impact of yoga on DCM patients, only the DCM patients can be the best respondents for this study; every member of heart disease is not the best respondent for this study. Hence, the researcher deliberately selects only the DCM patients as respondents for this study.
When to use/execute judgmental sampling:
- When the number of people is less in the population and the researcher knows that the target population fulfill his/her demands, in that case, the judgmental sampling is the best sample
- When there is a need to filter the samples chosen by other sample methods, this sampling method is best as it depends on the researcher's knowledge and experience.
Another example of this type of sampling is if a researcher wants to know how many patients of depression are doing particular yoga postures and meditation, he/she will select those patients who he/she thinks will give 100% feedback.
Advantage of judgmental sampling
- As selection of the sampling is done by experienced researcher, there will be no hurdles and thus selecting the sample becomes convenient
- As the samples selected will be good respondents for that particular study, almost we will get the real-time results, as members will have appropriate knowledge and they understand the subject well
- A researcher can produce desired results as he/she can directly communicate with the target audience.
Selecting the members of a sample on the basis of their convenient accessibility is called convenience sampling. In this, only those members are selected who are easily accessible to the researcher.
In this sampling, the available data are used without any further additional requirements.
This is used in pilot testing more commonly.
The participants/samples are selected which are easier to recruit for the study.
Some of the examples for this type of sampling are:
- Different challenges/games at the shopping malls on different festivals
- In a study, a researcher wants to know how many people in a particular area know about dengue, so the researcher will ask questionnaire to the people present and who knows something about dengue will participate in it.
Even the researcher can use the different social networking sites by putting his/her questions on them and interested people will join.
Advantage of convenience sampling
- Very easy to implement and inexpensive to create samples
- Useful for pilot studies and for hypothesis generations
- In a very short duration of time, we can collect data.
Chances of high sampling error.
Also known as chain sampling or sequential sampling, it is used where one respondent identifies other respondents (from his/her friends or relatives or known-to). This kind of sampling is adopted in situ ations where it is difficult to identify the members in a sample.
For example, a researcher wants to study problems faced by the migrants in an area. So, he/she will start from one and that migrant will give him/her the information about the other migrant and so it makes a chain and in this way, sample goes on growing like a snowball and the researcher continues this method until the required sample size is achieved.
When to use snowball sampling:
Snowball sampling totally depends on referrals. In this, the population is unknown and rare, due to which it is highly difficult to find the samples/participants.
Just as snowball increases on adding more snow, samples increase in this technique until we collect enough data to analyze. Hence, it is named snowball sampling.
Types of snowball sampling
There are three types:
- Linear snowball sampling: In this, the collection of samples starts from collecting data from one and then that individual tells about the other and so in this way, a chain is formed and it continues till we get enough number of individuals to analyze.For example, a researcher performing study on Crohn's disease needs to find out the people suffering from Crohn's disease which is difficult, so he/she asks one patient and gets information about the other patient suffering from Crohn's disease and this way, a chain is formed and the researcher will continue to take the patients till enough data he/she collects.
- Exponential nondiscrimination snowball sampling: In this, one individual will be giving information about more than one individual and those individuals in turn will be giving information about the others and in this way, with more and more referrals, the chain is formed and we collect data.
For example, to collect data regarding Diabetic mellitus from an area, we find an individual who is suffering from Diabetic mellitus. So from him, there are high chances that we will get some information about other people he may know suffering from Diabetic mellitus.
- Exponential discrimination snowball sampling: In this type of snowball sampling, one patient gives multiple referrals, but the recruitment will be done only for one patient on the basis of the nature and type of the research study.
For example, if we take the above same example, if that one patient tells us about another five patients who are suffering from the same disease, now if the study researcher wants patients only below 40 years of age and who have much controlled sugar, then he/she selects the patients according to this.
In the following areas, snowball sampling can be applied:
- Medical records: There are many rare diseases which are yet to be researched and there could be restricted number of individuals suffering from such rare disease. Some of the examples of such disease are mad cow disease, Alice in Wonderland, water allergy, laughing death, pica, and Moebius syndrome. Hence, with this kind of sampling, the people affected with such disease can be traced and research could be done
- Social research: In this, we take as many participants as much possible
- Cases of discord: In cases of disputes and act of terrorism, rights violation, we will choose people who are witness for that or people who are affected by that.
Advantage of snowball sampling
- Can collect samples very quickly
- It is cost-effective.
Disadvantages of snowball sampling
- High chances of sampling bias and margin of error
- If no one cooperates, it is difficult to find the samples.
In this kind of sampling, members are selected on the basis of some specific characteristics chosen by the researcher. These specific characteristics serve as a quota for selection of the members of the sample.
In this type of sampling, we gather representative data from a group. It is similar to stratified random sampling which is a type of probability sampling. The only difference between both is that in stratified random sampling, the elements of sample are chosen randomly, but in quota sampling, it is not so.
The number of participants is taken in specific category in well-planned manner; for example, 100 males and 100 females.
It is of two types – controlled quota sampling in which there are limitations to the choice of the researcher. The other type is uncontrolled quota sampling in which there are no limitations, and samples are selected according to the convenience of the researcher.
In this type of nonprobability sampling, the researcher will select the samples according to his/her ease/convenience. This is also similar to convenience sampling with little change.
In this, the researcher first picks up a group of people for research, does it for some time period, collects samples, gives results, and once the research completes, he/she will move on to the next group of people. Hence, in this way, a researcher will fine tune his/her research work with the help of this sampling, and he/she gets chance to work with multiple sampling.
In many of the researches, the techniques used, the data analyzed, and conclusion given by researcher will either come under null hypothesis or disapproving it and accepting the alternative hypothesis.
Null hypothesis is denoted by H0, and there is no significant difference in the variables, whereas alternative hypothesis is denoted by H1, which is opposite to null hypothesis where there is some relationship between the two variables.
However, consecutively, the 3rd option is available, that is, here the researcher, will either come under null hypothesis or if he disapproves it, he accepts the alternative hypothesis.
For example, for advertising the hospital, we distribute leaflets telling about the hospital and its facilities, once the camp organized for checking blood sugar and blood pressure (BP) as free, people will come and do their checkups. Many of the people will just see the leaflet and will move, but some of them will come and check for GRBS and BP. In this case, some might be only checking and going, and there will be another group of people who will check and want to show results to doctor and consult them. Hence, this group of people will provide conclusive results for showing the reports to doctor.
- In this, there are different options to sample size and sampling schedule
- Sampling schedule depends on the nature of research, if we are not able to get conclusive results with one sample, then we will go to next
- This is not time-consuming and also very little effort is required.
The samples obtained cannot be randomized, and we cannot represent the whole population by this.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Elfil M, Negida A. Sampling methods in clinical research; an educational review. Emergency. 2017;5.
Shorten A, Moorley C. Selecting the sample. Evid Based Nurs 2014;17:32-3.
Martínez-Mesa J, González-Chica DA, Duquia RP, Bonamigo RR, Bastos JL. Sampling: how to select participants in my research study?. Anais brasileiros de dermatologia. 2016;91:326-30.
Teddlie C, Yu F. Mixed methods sampling: A typology with examples. J Mix Methods Res 2007;1:77-100.
Cochran WG. Sampling Techniques. 3rd
ed., Vol. 98. New York: Wiley and Sons; 1977. p. 259-61.
Guba EG, Lincoln YS. Competing paradigms in qualitative research. Handbook of Qualitative Research. 1994. p. 105.
Joseph F. Hair Jr. William C. Black Barry J. Babin Rolph E. Anderson Multivariate Data Analysis 7th edition. Pearson Education Limited; England 2014.
Saunders MN, Saunders M, Lewis P, Thornhill A. Research Methods for Business Students. 5th
Edition, Pearson Education, Essex 2011.
Hendlin YH, Vora M, Elias J, Ling PM. Financial conflicts of interest and stance on tobacco harm reduction: A systematic review. Am J Public Health 2019;109:e1-8.
[Figure 1], [Figure 2], [Figure 3]
|This article has been cited by|
||Exercise countermeasure preferences of three male astronauts, a preliminary qualitative study
| ||J.M. Laws, C. Bruce-Martin, N. Caplan, R. Meroni, A. Winnard |
| ||Acta Astronautica. 2022; |
|[Pubmed] | [DOI]|
||Pediatric nurse-patient communication practices at Pentecost Hospital, Madina: A qualitative study
| ||Evans Osei Appiah, Stella Appiah, Samuel Kontoha, Sarah Mensah, Dorothy Baffour Awuah, Awube Menlah, Michael Baidoo |
| ||International Journal of Nursing Sciences. 2022; |
|[Pubmed] | [DOI]|
||Stumble on information or misinformation? Examining the interplay of incidental news exposure, narcissism, and new media literacy in misinformation engagement
| ||Xizhu Xiao, Yan Su |
| ||Internet Research. 2022; |
|[Pubmed] | [DOI]|
||An empirical study to prioritize the determinants of corporate sustainability performance using analytic hierarchy process
| ||Hoshiar Mal, Manishkumar Varma, Niraj Kumar Vishvakarma |
| ||Measuring Business Excellence. 2022; |
|[Pubmed] | [DOI]|
||The Impact of Context Awareness and Ubiquity on Mobile Government Service Adoption
| ||Isaac Kofi Mensah, Deborah Simon Mwakapesa, Antonio Garcia-Cabot |
| ||Mobile Information Systems. 2022; 2022: 1 |
|[Pubmed] | [DOI]|
||Youth Perspectives on Barriers and Opportunities for the Development of a Peer Support Model to Promote Mental Health and Prevent Suicide
| ||Jackie Libon, Jarlo Alganion, Carla Hilario |
| ||Western Journal of Nursing Research. 2022; : 0193945922 |
|[Pubmed] | [DOI]|
||Assessing the effectiveness of compliance inspection in ensuring the quality of ICT products and services: a case of the compliance department at ICASA
| ||Rachel Molatelo Ramahlo, Anton M. Pillay, Jeremiah Madzimure |
| ||EUREKA: Social and Humanities. 2022; (2): 15 |
|[Pubmed] | [DOI]|
||Effects of COVID-19 Pandemic on the Practice of Community Pharmacy in Kisumu County
| ||Nancy K. Nyambega, Esbon N. Wambugu, Michael N. Walekhwa, Filex O. Otieno |
| ||European Journal of Medical and Health Sciences. 2022; 4(4): 39 |
|[Pubmed] | [DOI]|
||Experiences of involuntary job loss and health during the economic crisis in Portugal
| ||Gloria Macassa,Carina Rodrigues,Henrique Barros,Anneli Marttila |
| ||Porto Biomedical Journal. 2021; 6(1): e121 |
|[Pubmed] | [DOI]|
||Dynamic Distributed and Parallel Machine Learning algorithms for big data mining processing
| ||Laouni Djafri |
| ||Data Technologies and Applications. 2021; ahead-of-p(ahead-of-p) |
|[Pubmed] | [DOI]|
||Work interruptions and missed nursing care: A necessary evil or an opportunity? The role of nursesí sense of controllability
| ||Nasra Abdelhadi, Anat Drach-Zahavy, Einav Srulovici |
| ||Nursing Open. 2021; |
|[Pubmed] | [DOI]|
||Motivational Strategies for Stroke Rehabilitation: A Descriptive Cross-Sectional Study
| ||Kazuaki Oyake,Makoto Suzuki,Yohei Otaka,Satoshi Tanaka |
| ||Frontiers in Neurology. 2020; 11 |
|[Pubmed] | [DOI]|