## Quota Sampling

##### Introduction

There are two types of sampling; probability sampling and non-probability sampling. Of the two, probability sampling is the best, since there is a higher likelihood of obtaining a representative sample, and sampling error can be taken into account. In this report, we will discuss exclusively about quota sampling.  This is a non-probability sampling. By definition, this is a sampling technique which aims at obtaining representative data from a sample.

Let start with the definition of some terms. Sample, is simple terms means a subsection, subgroup, or portion of the population. Sampling, is the process of acquiring a representative portion of population and population is the entire elements under investigation.

There are a number of reasons as to why investigators may opt to use a sample instead of the entire population, despite sampling error. First, it has an economical advantage since fewer resources are used than when a census is carried out (Levy & Lemeshow, 2013). The sampling has timelessness factor, where the information required can be obtained faster than when a complete enumeration of objects is carried out.

Further, the researcher sort to use sampling when the population has infinite and when some of the population units/objects are inaccessible. Lastly, but not least, when the assessment destructs the objects under investigation (Levy & Lemeshow, 2013). Not forgetting, that sometimes a carefully selected sample may represent the population well than a sloppy conducted census. That is, the sample will truly mirror the population from which it was drawn from (Hu et al., 2017).

##### Literature review

As stated earlier, quota sampling is a nonprobability sampling. A nonprobability sampling in accordance with (Trochim & Donnelly, 2014) is a technique that is based on the judgement of the researcher. The collected sample has the same proportion of the mother population with identical traits and the targeted phenomenon. More importantly, the final sample obtained should have the composition criteria of the quota set. Since this paper discusses one of the nonprobability sampling, it is imperative to highlight some principles of nonprobability sampling.

Principle of quota sampling

In this type of sampling, the quantity of each sampling is find out in advance, however investigators look to find in each quota. For each type, the quotas are of fixed numbers, which the investigators would like to add in the sample. Let consider an example in which an investigator would like to have a sample in which the proportion of each gender by different age levels are approximately the same and the sample size is 1000. Let the population proportion is given in Table 1 and the quota of individual type of sampling unit is the sample size multiplied by the population proportion and the results are shown in table 2. (Chang-Tai Chao)

Table 1 Population Proportion

Table 2 Quotas

Principles of nonprobability sampling.

Quota sampling falls under the class of non-probability sampling. Sampling involves the selection of a lot of the population being considered. In probability sampling each element in the population has a known nonzero chance of being selected through the utilization of an irregular determination system, for example, basic arbitrary testing. Non-probability examining does not include known nonzero probabilities of choice. (Paul J. Lavarakas) Or maybe, subjective techniques are utilized to choose which components ought to be incorporated into the specimen.

In non-likelihood inspecting the populace may not be very much characterized. Non-probability examining is frequently isolated into three classifications: purposive inspecting, comfort testing, and share examining. Share examining has a few similitudes to stratified inspecting. The essential thought of standard testing is to set an objective number of finished meetings with particular subgroups of the number of inhabitants in intrigue.

The testing strategy then continues utilizing a nonrandom choice component until the coveted number of finished meetings is acquired for every subgroup. A typical illustration is to set half of the meetings with guys and half with females in an irregular digit dialing phone talk with study. A specimen of phone numbers is discharged to the questioners for calling. Toward the begin of the overview, one grown-up is haphazardly chosen from a specimen family unit. It is by and large more hard to get interviews with guys.

So for instance, if the aggregate sought number of meetings is 1,000 (500 guys and 500 females), and meetings with 500 females are gotten before meetings with 500 guys, then no further meetings would be directed with females and just guys would be arbitrarily chosen and met until the objective of 500 guys is come to. Females in those specimen family units would have a zero likelihood of choice.

Additionally, on the grounds that the 500 female meetings were in all probability gotten at before call endeavors, before the example phone numbers were altogether worked by the questioners, females living in harder-to-achieve family units are more averse to be incorporated into the specimen of 500 females. Shares are frequently in light of more than one trademark. For instance, a standard specimen may have questioner appointed shares for age by sex by business status classes.

For a given example family unit the questioner may request the rarest gathering in the first place, and if an individual from that gathering is available in the family unit, that individual will be met. On the off chance that an individual from the rarest gathering is absent in the family, then a person in one of the other uncommon gatherings will be chosen. Once the quantities for the uncommon gatherings are filled, the questioner will move to filling the amounts for the more typical gatherings.

In that regard are both hypothetical and down to earth justification for using this kind of examining in research. In hypothetical reasons, this progress of information testing permits the utilization of subjective, blended research techniques and furthermore quantitative strategies. (Paul J. Lavarakas) By the time, when quantitative research is received, this inspecting procedure is oftentimes if not generally seen as sub-par compared to likelihood examining. This system helps in accomplishing objectivity of the specialist.

For instance, the analyst may utilize portion testing for intentionally protecting the incorporation of an uncommon segment of the populace. In any case, it ought to be specified that the fragment won’t not vary altogether contrast from the parent populace. The most common example of the limitation of this type of approach pre-election surveys to anticipate the consequences of the 1948 U.S. presidential decision. The field questioners were offered shares to fill in view of qualities, for example, age, sexual orientation, race, level of urban city, and financial status.

The questioners were sans then to fill the standards with no likelihood examining system set up. This subjective choice technique brought about an inclination for Republicans will probably be met inside the quantity bunches than Democrats. This brought about the example containing an excessive number of Republicans and bringing on the pre-race surveys to inaccurately foresee Thomas Dewey (the Republican applicant) as the victor.

Share inspecting is infrequently utilized as a part of conjunction with region likelihood testing of family units. Zone likelihood examining systems are utilized to choose essential testing units and fragments. For each specimen portion (e.g., city obstruct) the questioner is told to begin at a side of the section and continue around the fragment, reaching lodging units until a particular number of meetings are finished in the portion.

A noteworthy issue with quota sampling is the introduction of the unknown sampling biases estimates in survey. On account of the 1948 U.S. presidential decision, the testing inclination was related with an excessive number of Republicans being chosen. Another issue with quantity examining is that the inspecting technique regularly brings about a lower reaction rate than would be accomplished in a likelihood test.

Most amount tests quit endeavoring to finish interviews with dynamic example families once the standards have been met. In the event that a lot of test is dynamic at the time the amounts are shut, then the reaction rate will be low. (Paul J. Lavarakas)

However, some practical reasons of using quota sampling are that; this approach gives the investigator exploratory power to determine whether the problem exists when no theory exists that such problem exists. Therefore, the researcher may opt only to use sample that he/she thinks will portray the problem, which saves resources and time.

Ethical approach is important when deciding the sampling technique, as it may be unnecessary to subject fewer participants for an in-depth examination. Above all, it is imperative to determine whether quota sampling is a better sampling method to be used with the designed research strategy.

In accordance with (“Definition of ‘Quota Sampling’ – The Economic Times,” 2017) quota sampling is simple, yet effective approach a researcher can gather information in the initial phase of research. As earlier pointed out, quota sampling is ideal when the researcher wants to investigate whether a certain problem exists. As an example, there might be a notion that people in Iowa are against the death sentence.

A researcher may be interested in assessing the situation. However, the researcher may want to include people from Muslim religion. The researcher sets the minimum percentage of Muslims to be included in the sample, say 5%. Through this, the researcher is able to compare the views of people, including from the set quota.

Is quota sampling adequate method as an alternative to probability sampling?

A research was performed to assess whether quota sampling is an alternative to probability sampling (Yang & Banamah, 2014). Although probability sampling remains popular among many researchers and businesses. The research was driven by the fact that collecting data have become expensive and the response rate decreased.

Therefore, the researcher thought that quota sampling as a nonprobability sampling can be used as an alternative to probability sampling. The representativeness of quota sampling with probability sampling was compared as well as the assessment of the between the two sampling techniques using the survey topic and the response rate as factors (Yang, & Banamah, 2014).

It was established that the survey topic greatly influences the response rate, and response rate correlates with sample mean. Nonetheless, the researcher concluded these for that reason; quota sampling is not an alternative to probability sampling (Yang, & Banamah, 2014).

Types of quota sampling

Generally, there are two types of quota sampling:

• Controlled quota sampling (proportional quota sampling)

The researcher, in this case, first impose the restrictions or the limit number of subjects to include. That is the researcher set a quota of each category. This is done to ensure that the major population or characteristics are well represented (Sharan, 2009).  For instance, it may be known that the number of female workers in a certain bank is 60% and 40% are males. The researcher may be interested in 100 sample people. Therefore, he samples 60 females and 40 males, and when 40 males are obtained, even if a potential and informed male come he will not sample them. This technique helps in limiting the researcher’s choices ((“Quota sampling – Research Methodology,” 2017))

• Uncontrolled quota sampling (Nonproportional quota sampling)

In this case, the researcher in not limited in any way and thus selects the subjects as he/she wishes. This is considered as a sample of convenience, or simply it is less restrictive. The researcher just specifies the minimum number in each category, and one does not need to know the proportion. This sampling technique is more of a convenient sampling since the researcher can opt to use any number of participants so long as they meet the minimum requirements (“Quota sampling – Research Methodology,” 2017).

When to use quota sampling

There are scenario or situations that call for the researcher to use quota sampling technique. First, is when the researcher or the organization do not have sufficient time and resources to carry out stratified sampling. Therefore, due to effective and ease of conducting, the researcher uses this technique. Second, this sampling method is used when the accuracy of the information is not important.

For instance, when the researcher is conducting a pilot survey, or when there is no adequate information about the research topic. Also, when the researcher is just interested to study some subsections or subgroups, he sorts to use this nonprobability sampling. That is, when the interest is limited to certain traits, this method is ideal. If the main interest of the researcher is to determine the relationship between groups, then the ideal method is quota sample. Also, when the main aim is to compare the groups.

The sampling frame may be unavailable. Notably, mostly, if not all, of probability sampling, required. In the absence, of the list of all the subjects, the researcher uses this technique. That is, since the probability sampling the researcher should know the selection likelihood of each unit in the population.  Lastly, the research accuracy is not important and also the results are not for generalization purposes.

Thus the researcher can use quota sampling technique. This method results should not be used for generalization purposes since the units selected may sometimes not truly represent the parent population.  Therefore, the researcher can purposefully select quota sampling to investigate certain predefined traits.

Steps of creating quota sampling

There are three acceptable steps in creating quota sampling:

1. Categorize or stratify the population.

First, the researcher divides the population into mutually exclusive groups. For instance, a researcher can stratify by gender, race, ethnicity, career, courses taken, religion and so on.

• Computing quota for each stratum.

The researcher needs to determine the number of subjects to be used in each category. As illustrated earlier, the proportion of each category can be determined depending on the population proportion. Also, the researcher may set purposively a quota to a sample that might have a small likelihood of being sampled (as discussed in the Muslims case). In some cases, the researcher may not involve proportions when he just selects conveniently.

• Continue selecting subjects until quotas for each stratum are obtained.

The researcher continues with selection until the exactly required number is obtained. When the quota of one group is attained, the researcher does not select any other subject even if he obtains a viable sample. However, he/she continues to select items for the other groups that quota has not been attained.

Difference between quota sampling and stratified sampling

Notably, the two sampling techniques are different, although they might be confused. First, in quota sampling the individuals are not selected randomly, that is why it is a nonprobability sampling. The samples are selected based on the researchers’ convenience. On the other hand, the stratified sampling, subjects are randomly selected into the sample (Sharan, 2009).

All the units or subjects in the population have an equal likelihood of being included in the final sample. In quota sampling, there are no call-back for particular samples. This implies that when a subject is missed, the researcher goes on and uses the available subjects. On the other hand, STRICTLY, there is a call-back, since if there are no call-backs the sampling will not be different from quota sample (sample of convenience) (Sharan, 2009).

Last but not least, since quota sampling is based on the judgement of the researcher it is full of biased and the most common type is a non-sampling error. Stratified sampling is not subjected to biasedness since the elements are randomly selected the only common error is sampling error.

This sampling technique is helpful when the researcher is unable to obtain a probability sample, but still, the researcher is aiming at obtaining a representative sample. This technique is adopted as a non-probability sampling that is closely equivalent to stratified sampling. Second, this technique is easier and quicker to carry, unlike probability sampling like stratified sampling (Trochim & Donnelly, 2014).

For this reason, this sampling method is adopted by masters’ students in dissertations where the target populations can be divided into categories. When quotas are used, they ensure that some groups are not overrepresented or other underrepresented. That is the main reason why in the computation of minimum units to be included from each category (quota) the population proportion is useful (Trochim & Donnelly, 2014).

A good example is as illustrated earlier, where a population consist 60% female and 40% male. These proportions were used in obtaining a sample of both categories, out of 100 sample required 60 females were selected, and 40 males were selected. Also, when quota sampling is used, there is stratification, which is ideal when the researcher wants to compare the two strata.

A good example is; in an organization, the researcher may select male and female and compares their average. The averages will determine which gender gets higher pay. Lastly, this sampling method is not dependent on the sampling frame. This implies that it can be used when a list of all subjects is absent, quota sampling can still be used. Therefore, it remains the most appropriate sampling technique when sampling frame is not there.

There are limitations of quota sampling. First, since the sample are not obtained through probabilistic means, it is impossible to estimate the sampling error. It is possible for subjects to be sampled based on the researcher’s convenience and resource usage consideration. This leads to sampling bias.  Results from such data cannot be used for generalization purposes, or cannot be used in inferential statistics. In particular, the results are only recommended for the sample used and not the entire population.

Therefore, limiting generalization. Another limitation is that the population should be able to be dived into two distinct groups. Otherwise, the process of data collection will not be possible. As examples, indicated, the population should have mutually exclusive strata; like gender, male and female (if there are no transgender issues). When the sample requirements are required to be compared, more categories are added.

A good example is, when the researcher is interested in assessing how male and female supporting death sentence changes depending on the religious such as the Muslims and Christians. Notably, the subject should be in one stratum and here there are four criteria. Therefore, the researcher needs to collect data from “male Muslims,” “Female Muslims,” “male Christians” and “Female Christians.”

Evidently, more subjects need to be sampled, increasing the cost and time of carrying out the survey. Further, the researcher needs to understand the population well, so that he/she can make the right stratification, calculate quota, and lastly keep on sampling to meet the quota specifications. Without proper knowledge, one is not able to use quota sampling as he/she might to obtain a representative sample.

Although the sample obtained may be proportional to the population, some of the characteristics may not be disproportional. Incompetency of the researcher or lack of proper skills may increase the biasedness of the sample since the subjects’ collection is based on the researcher’s judgement (Sharan, 2009). Thus, is of great importance that before, using quota sampling researcher should have adequate skills to collect representative samples.

##### Conclusion

There are a number of important aspects of quota sampling highlighted in this paper.  For instance, it has been indicated that this technique is a non-probability sampling, where the selected samples represent certain characteristics. Two types of quota sampling; controlled and uncontrolled sampling were also adequately discussed. When this sampling method should be used was also highlighted among another important aspect. As highlighted, the three steps are used in designing data collection using quota sampling method.

There are important factors that have been pointed out, in the literature of this paper, such as; the question whether quota sampling is a good alternative to probability sampling. Based, in the previous study by prominent scholars, it was established that this method is not is not an alternative to probability sampling.

A clarification was made on the underlying difference between stratified sampling, despite the two having some similarities like stratifying the population into mutually exclusive first. To support some claims or issues brought forward, different examples were given designed to clarify, how the quota sampling will work.

Some of the reasons for using quota sampling brought forward include, the researcher only selecting groups that only interests him. Investigate some traits or characteristics that have not been investigated before, or there is no preexisting literature and when the researcher is interested in comparing the groups of interest.

Despite this being a nonprobability sampling, it has proven to be helpful, especially when the sampling frame is absent. That is, all probability sampling techniques are dependent on the availability of the list of all population subjects. Therefore, the only option is resorting nonprobability sampling. This is also the ideal method when the researcher has less time and limited resources, as it is considered an alternative to stratified sampling.

It is advisable to use this method when the researcher has a deep understanding of the steps and process of data collection. Also, the researcher should avoid so much biasedness so that he can have a representative sample. Lastly, the findings obtained using data obtained via this technique should not be used for inferential purposes; rather they should make conclusion about the sample used.

References

Chang-Tai Chao, “Quota Sampling” (Incomplete information provided about the journal paper)

Definition of ‘Quota Sampling’ – The Economic Times. (2017). The Economic Times. Retrieved 23 March 2017, from http://economictimes.indiatimes.com/definition/quota-sampling

Hu, X. F., Young, K., & Chan, H. M. (2017). Background Bivariate random-effects models represent a widely accepted and recommended approach for meta-analysis of test accuracy studies. Standard likelihood methods routinely used for inference are prone to several drawbacks. Small sample size can give rise to unreliable inferential conclusions and convergence issues make the approach unappealing. This paper suggests a different methodology to… BMC Medical Research Methodology, 17(1), 1-12. https://www.infona.pl/resource/bwmeta1.element.springer-a4c4ef0e-2be0-30d3-9d87-a2c2297724c4

Levy, P. S., & Lemeshow, S. (2013). Sampling of populations: methods and applications. John Wiley & Sons. From https://books.google.com/books?hl=en&lr=&id=XU9ZmLe5k1IC&oi=fnd&pg=PT14&dq=Sampling+of+populations:+methods+and+applications.&ots=oaeRJfQxUr&sig=oUVgwHhuzBCm1bjAZIEHR-uSHyQ

Paul J. Lavarakas, “Encyclopedia of Survey Research methods,” Volume 1 & 2

Quota sampling – Research Methodology. (2017). Research Methodology. Retrieved 23 March 2017, from http://research-methodology.net/sampling-in-primary-data-collection/quota-sampling/

Sharan, (2009). Quota Sampling. Slideshare.net. Retrieved 23 March 2017, from https://www.slideshare.net/sumanto123/quota-sampling

Trochim, M. K., & Donnelly, J. P. (2014). Nonprobability sampling. The Research Methods Knowledge Base website. From https://books.google.co.ke/books?id=0yxBBAAAQBAJ&pg=PA106&dq=Non+Probability+sampling.+The+Research+Methods+Knowledge+Base+website&hl=en&sa=X&redir_esc=y#v=onepage&q=Non%20Probability%20sampling.%20The%20Research%20Methods%20Knowledge%20Base%20website&f=false

Yang, K., & Banamah, A. (2014). Quota sampling as an alternative to probability sampling? An experimental study. Sociological Research Online, 19(1), 29. doi: http://www.socresonline.org.uk/19/1/29.html

Paul J. Lavarakas, “Encyclopedia of Survey Research methods,” Volume 1 & 2