Probability And Non-Probability Sampling Techniques: Advantages, Disadvantages And Examples

1. Explain and distinguish between probability sampling and non-probability sampling techniques, illustrating your answer with their advantages, disadvantages and examples.

2.Discuss the Elements of Population and the importance of sample size

3. State and explain probability sampling and non-probability sampling

4. State and explain three sampling techniques for each the probability sampling and non-probability sampling 

The Elements of Population and Importance of Sample Size

Sampling is the procedure of obtaining data used in statistical analysis from the representative part of a population (Best and Kahn 2016). The population from which the researcher wants to generalize the outcome is the target population that includes all items or people he wishes to understand. The samples are selected from sample units and further a sampling frame is drawn from the list of potential respondents like, the list of student, hotels and telephone directory (Ross 2017). Generally, the sample represents the larger population and the researcher believes that entire population is applicable to the conclusion drawn from the sample result. In research, two general approaches for sampling are used, which are as follows:

• Probability sampling- when all the elements in the targeted population like persons or households have equal opportunity of being selected in the sample through random selection (Cressie 2015).
• Non-probability sampling- when the elements present in the group are chosen based on their availability and it does not involve random selection (Ruppert 2014).

In the current scenario, statistics is widely used in the business world to make quick decision for maximizing profit. The consumer market is rapidly increasing with the population size growing da by day (Rohatgi and Saleh 2015). In such a large population interviewing each individual would require lots of time besides being very expensive. However, if sub-set or sample is identified and analyzed from the population representing the group. Then a reliable outcome can be obtained for the entire population.

Population is the group of targeted people or individual that is accessible for meeting a well-defined set of eligibility criteria (Quenouille 2014). Population is aggregate or totality of observations on which the research is concerned. The vital factor of choosing a sample is that the population must be clearly definite and accurately identified to get the desired result (DiMaggio 2013). The specific types of population are as follows:

• Target population- it is the group of people meeting the criteria.
• Subject population- it refers to the group of people taking part in the study.
• Stratum or Strata- it is the mutually exclusive population segment established by single or more characteristics (Johnson and Wichern 2014).

Sample is the sub-set of the large population, which is selected for the study, refers to appendix III. It is the crucial element of research methodology. For example, from the population of 600 students only 300 students are selected as target population and only 50 students are further selected as sample (Peck, Olsen and Devore 2015). It is very important for the researcher to determine the size of the sample before selecting the sampling technique. Determination of sample size is importance due to the following reasons:

• Lower cost.
• More accurate result or outcome.
• The population elements are easily available.
• Speed of data collection is very high.
• Less field time
• Studying the entire population is impossible

A good sample represents the overall population and it should be appropriately sized, unbiased, and random in nature (Ott and Longnecker 2015). The sample size can be ascertained by the following formula:

   N= N/1+Ne2

Where, n is the sample size,

N is the population size

E is the margin of error, and

1 is the constant value.

In probability sampling, the samples are generally selected with a known probability and each unit present in the population have more or less same opportunity of being chosen. For larger population survey, probability sampling is important to generalize the population characteristics (Denscombe 2014).  Probability sampling creates the overall sampling procedure to be free from biasness and uses less reliance compare to the human jugdement.

For example, from a group of 12 students only 3 students are requires as sample. First, a random number is assigned to each element given in the data. After assigning, the number of elements in each row is marked with highest assigned number. Like 66 from first row, 82 in second and 73 from the third row.  The selected sample is further analyzed (Bryman and Bell 2015). The number is assigned randomly to each elements present in the population so there is equal opportunity of being chosen, thus making the procedure more probabilistic.

Sampling Techniques

Advantages:

• Reliable and accurate
• Involves lesser judgment
• Less time consuming
• Easily computed
• Unbiased in nature

Disadvantages:

• Chances of selecting the same class of samples only
• Monotonous and redundant

In this sampling technique, the odds of any element of being selected cannot be calculated for a sample (Ritchie, Lewis and Elam 2013). This sampling generally depends on the researcher’s subjective judgment and not any random selections refer to appendix III.

For example, when a media person takes an interview of the individual on street or the researcher takes the opinion of subject matter experts for opinions to know the general populace thought about the issue helps in saving time and resources though the statistics analysis is not accurate (Bernard, Wutich and Ryan 2016). Non-probability sampling is often used in generating qualitative data.

Advantages:

• Cost effective
• More manageable
• Low time consumption
• Population elements are not necessary
• Sampling skills not required

  Disadvantages:

• Biased in nature
• Unreliable in nature

 In probability sampling the population parameter of estimated sampling is unbiased in nature and a complete list of population elements is required whereas in non-probability sampling the parameters are biased, as the population elements are not considered (Levy and Lemeshow 2013). Probability sampling quantitative data is required and each unit are identified and valued. Whereas, details like activities, habits and traits of the population are required to estimate non-probability sampling. Probability sampling is generally requires lot of effort and time than non-probability sampling refer to appendix I. The accuracy and reliability of sample outcome is more in probability sampling technique as it is computed with confidence after properly analyzing the representative data (Siegel 2016). On the other hand, non-probability sampling is less reliable and inaccurate. The sampling error can be known through statistical measures in probability sampling but through non-probability sampling the true measures for ascertaining the sampling error is unknown.

This is entirely random method of obtaining the element and the method is applied when the population size can be easily available, precise and homogeneous. All the subsets of sample frame are allowed alike opportunity (Hair 2015). The sampling schemes may be used in two forms, with or without replacement. Lottery system or random numbers are used to determine the units to be selected.  In this method, the population subset is randomly achieved without practical knowledge or logic. For example: researcher selects 20 units of students from a batch of 70 , then he can apply any criteria for randomly choosing the students, like students obtaining highest marks, pupils above 15 year of age or students from second row (Mertler and Reinhart 2016).

This is a special way of random sampling technique, where the first sample unit is randomly taken from a population and the other sample units are taken at fixed intervals. Except the first sample units complete randomization is lacking from the other units (Ritchie, Lewis and Elam 2013). For instance, if the production department of an organization wants review the level of quality of the goods from the sample, then the first selected sample units if is good then all the remaining goods in the population sample is also considered to be good.

This type of sampling is more efficient statistically as it more improvised technique than the simple random and systematic sampling techniques. In stratified sampling technique, the population is divided into classified set known as stratum or strata. A stratum has units that have homogeneous attributes, as the attribute of one stratum is different from the other strata. For example: to measure the living standards of people in any city the individuals are first segmented into different classes like the upper, middle and lower class. Then they are further divided into various attributes like education, income and job nature (Bernard, Wutich and Ryan 2016). If the proportion of respondents used is same or alike in the sampling technique then it is known as proportional strata sampling method. The variance is equal to zero or near between the two strata or else it is disproportional stratified sampling.

Probability Sampling

This is the widely used and popular of all non-probability sampling techniques. In convenience sampling, all the samples are only chosen when the researcher can easily access the sample, according to their convenience (Peck, Olsen and Devore 2015). The subjects are selected because they can easily be recruited. This sampling technique is considered cheapest, least time consuming and easiest.

This kind of non-probability sampling technique is used when the researcher makes sure that the elements are proportionately or equally represented. The dependent traits are considered based on quota (Hair 2015). For instance, if researcher requires equal representations and the basis of quota is the level of college year. For a sample size of 95, the researcher must only select 25 students each from 1^st, 2^nd, 3rd and 4^th year respectively. Moreover, the quota should be based on gender, age, education, religion, race and socioeconomic status.

Snowball sampling is applicable when the targeted population is hard to reach or hidden. These include targeted population like drug addicts, individuals with fatal disease and homeless people. This kind of non-probability sampling is generally practised when the size of the population is small (Rohatgi and Saleh 2015). In snowball sampling, the researcher usually asks the subject for identifying another adequate subject to meet the research criteria.

Such kind of sampling is easier to use and prior information about the population is not required but the method can be impracticable if the sampling frame is large. There is independent and equal chance for every element to be selected and the degree of representativeness is quiet high but it can be tedious and time consuming in nature.

In this kind of sampling method, the appropriate sampling frame can be easily identified and easily selected but the given sample may be biased in nature targeted population hidden periodicity aligns with the selection. Not all elements are given equal chance. Systematic sampling is less random in nature than simple random sampling.

The sampling technique is applicable when the population size is heterogeneous in nature and includes various groups related to the subject. This probability sampling provides higher statistical efficiency and easy to be computed but the sampling technique is very expensive and time-consuming (Bernard, Wutich and Ryan 2016). Stratified sampling can sometimes lead to classification error as prior knowledge of population distribution and composition is required.

This kind of sampling approach is only applicable when the people or individuals of the population groups are convenient to sample. This method is usually convenient and applied for known population groups but the findings are not reliable ((Ritchie, Lewis and Elam 2013). There are problems of accuracy, as the researcher cannot generalize the findings.

Just as stratified sampling, first, the population is divided into commonly exclusive sub-groups depending on the quotas related to attitudes, demographics and behaviours. Quota sampling is practiced when stratum is available and it is impossible to follow stratified sampling probability technique. However, quota samplings is difficult to generalize and dependent mostly on the subjective decisions. The sampling only reflects the targeted population based on quota so there is possibility of sample being selected to be bias in nature (Johnson and Wichern 2014). Quota sampling includes specific subgroups in the desired proportions, which can easily be managed and quickly evaluated. Quota sampling requires several researchers as not everyone can be equally competent therefore; the outcome can fluctuate and not be uniform.

Non-Probability Sampling

Snowball sampling is useful in research that is qualitative in nature. When the population is hidden or gaining access is quiet, difficult then snowball-sampling technique is used. The target market is well defined, unique and small in form but compiling the list of sample units is quiet difficult. However, in this kind of sampling techniques bias is generally present (Cressie 2015). Snowball sampling technique has limited scope as other can be recruited as a subjects as referred by the researcher. Such kind of sampling is often termed as network sampling.

Conclusion:

Thus, from the report it can be concluded that using sampling techniques in business saves money and time. This is more practical and logical approach. In the current scenario, the use of sampling technique is very useful in collection of data in business and various industries.  The business owners can use different sampling techniques to determine the current market niches, satisfaction level and products. Using an adequate sampling strategy and appropriate sample size can help in increasing the business and industry. Furthermore, required measures taken to decrease the measurement and sampling errors would yield reliable and valid information. In case when the population size is very large than applying different sampling techniques the necessary information and data can be collected.

The probability and non-probability sampling techniques used in the businesses and industry also have some challenges, which is crucial for the business decision. It is recommended for the researcher to complete the sampling research within the stipulated time and not for a long duration of time. The sample may change or any event can occur that can change the probable outcome. Moreover, it is also recommended for the businesses are to expand the customer base while sampling to find new market niches to increase the business prospect. Sampling the market population beforehand helps in new product development of the businesses and the potential customers are ascertained. Service Industry can also benefit from sampling techniques by using customer sampling through surveys and ascertaining the satisfaction level of the customers. 

References:

Bernard, H.R., Wutich, A. and Ryan, G.W., 2016. Analyzing qualitative data: Systematic approaches. SAGE publications.

Best, J.W. and Kahn, J.V., 2016. Research in education. Pearson Education India.

Bryman, A. and Bell, E., 2015. Business research methods. Oxford University Press, USA.

Cressie, N., 2015. Statistics for spatial data. John Wiley & Sons.

Denscombe, M., 2014. The good research guide: for small-scale social research projects. McGraw-Hill Education (UK).

DiMaggio, C., 2013. Introduction. In SAS for Epidemiologists(pp. 1-5). Springer New York.

Hair, J.F., 2015. Essentials of business research methods. ME Sharpe.

Johnson, R.A. and Wichern, D.W., 2014. Applied multivariate statistical analysis (Vol. 4). New Jersey: Prentice-Hall.

Levy, P.S. and Lemeshow, S., 2013. Sampling of populations: methods and applications. John Wiley & Sons.

Mertler, C.A. and Reinhart, R.V., 2016. Advanced and multivariate statistical methods: Practical application and interpretation. Routledge.

Ott, R.L. and Longnecker, M.T., 2015. An introduction to statistical methods and data analysis. Nelson Education.

Peck, R., Olsen, C. and Devore, J.L., 2015. Introduction to statistics and data analysis. Cengage Learning.

Quenouille, M.H., 2014. Introductory statistics. Elsevier.

Ritchie, J., Lewis, J. and Elam, R.G., 2013. Selecting samples. Qualitative research practice: A guide for social science students and researchers, p.111.

Rohatgi, V.K. and Saleh, A.M.E., 2015. An introduction to probability and statistics. John Wiley & Sons.

Ross, S.M., 2017. Introductory statistics. Academic Press.

Ruppert, D., 2014. Statistics and finance: An introduction. Springer.

Siegel, A., 2016. Practical business statistics. Academic Press.