Sampling

 

Sampling

In other words sample is a part of a thing that acts as a specimen or an example for that thing. For example, before launching a new soft-drink in market the company wants to test consumer feedback for the product. The company may set up temporary vendors at an amusement park and let the consumers try the samples of soft drink to collect their feedback. Each of those soft drinks will be called as a ‘sample’. Sampling is the most important step in the direction of carrying out research, once the hypothesis and objectives of research are understood. Sampling is a vital procedure in quantitative research, wherein the researcher first identifies the population to be studied. However studying each and every item or member of the entire population is not only bulky and costly, but also wasteful of time. This collective body of items that can be studied in lieu of studying the entire population is called a sample. Sampling is the method or technique that is used to draw out a sample, which reflects the qualities possessed by the population. Thus ‘sampling’ may be defined as a method of picking out a representative sample from the population to be studied, by using a definite technique. Technique is an essential thing while doing sampling because the sample that is taken must be appropriate in size as well as features, to be suitable for drawing inferences that can be generalized to the whole population. The first step in sampling is to determine the population to be studied. Then next step is to ascertain the qualities of the population that the researcher wants to study. On the basis of the qualities to be studied and the size of the population, the researcher can decide the appropriate proportional size of the sample. The qualities to be studied will also give the parameters for choosing a sample from the population.

‘Socio-legal’ : Sampling is a step that has a bigger role in quantitative research than purely doctrinal research. Such a research in legal field is often called as ‘socio-legal’ research because the researcher examines the execution of legal principles in society. For example a sociolegal researcher wants to study the level of awareness of consumer rights among educated people in a city in Maharashtra, say Pune. Since Pune is a big city, he divides it into different areas and then proceeds to determine the number of people he will approach for data collection in each of those areas. He first finds out the latest census information about population in Pune and finds out a number that would proportionately represent the population. The researcher in this example can also randomly choose the respondents for his questionnaires, like the members of his family or his friends, neighbours, colleagues, etc. However that will not establish the credibility of his research, because respondents chosen according to the researcher’s personal wish cannot yield results that can be called illustrative of the whole population. There are certain techniques that have been developed by researchers over time.

Important Terms There are some key terms that are associated with sampling. These are discussed below in detail:

Universe

 The first step in sampling is to identify and locate the ‘population’ to be studied. Population, also called as ‘universe’, is the entire collection of people on whom the study is to be conducted. For example a study is proposed to assess the access to government provided amenities like electricity, water, etc. to people living in suburbs in NCR India. The researcher proposes that the results of the study will be determined according to the responses of the people receiving the amenities. Thus the ‘population’ or ‘universe’ of the study will be the persons living in suburbs in and around NCR of India. Let us further suppose the researcher wants to conduct the research limited to a particular area of NCR, then the ‘population’ will be the residents of that area. Not only the factors but also the probability of obtaining data from the respondents is to be taken into consideration. For example a study is proposed to assess the level of drug abuse among teenage school-going students in Noida. The ‘universe’ will have to include only teenagers, who go to schools in Noida. But in this study it may be predicted by the researcher that obtaining honest and genuine responses from school children is tricky, the researcher will have to expand the population size to include school authorities, parents of the school teenagers. For authentication of the study the researcher may also include counselors who treat teenagers undergoing treatment for substance abuse. While making a choice of population for the study it is important not only to narrow down the respondents on the basis of aspects to be studies, but also by making speculations about the authenticity of responses that shall be collected.

Sample:

Distinguishing qualities: Sample once the population is fixed the next step is to carve out a fixed portion out of the population for purposes of the study. Sample is drawn out of the universe using sampling techniques. The most important characteristic of a sample is that it should have all the distinguishing qualities of the universe. The sample must be chosen in such a manner that it consists of all the desired characteristics to be studied. If care is not taken to ensure that the sample is not consisting of all the characteristics to be studied, then the results obtained may not be illustrative.

 Adequacy of the Sample : Proportion in choosing the sample ensures maximum accuracy of the study results. Thus adequacy of the sample is another important characteristic of the sample. All the units selected to be included in the sample must be independent of each other’s presence. That is to say, the inclusion of one unit in the sample must not be dependent on inclusion of another unit.

Sampling Units:

Sampling Units Each entity or person or thing which forms the entire universe is called as sampling unit. It is the most basis thing in the universe from which data is to be collected. For example in a study proposed for assessing the violation of human rights among hand-rickshaw pullers in the city of Kolkata, each of the rickshaw-puller is the ‘sampling unit’. Herein the universe will be the entire body of rickshaw pullers in Kolkata

Sampling elements: In some studies more than one sample is drawn out of the universe for making a sound research. In such cases each body of units is called as ‘unit’ and the entities or persons from whom data is collected are called as ‘sampling elements’. For example a Dish TV company wants to conduct a study to gather feedback from families that have subscribed to the Dish connection. The universe is located in a particular area composed of different societies. Each family who have subscribed to the connection is the sampling element. Group of families located in one society will be the sampling units.

Sampling Trait

Each of these factors or characteristics that govern the process of sampling, are called as ‘sampling traits’. Sampling traits may be ‘qualitative’ or ‘quantitative’ depending on the nature and requirement of the study. Qualitative traits are the unchangeable features, e.g. religion of persons, gender of persons, etc. These traits cannot be categorised into a range or scale. Quantitative traits are varying, like income of family, crime rate in an area, pollution level, etc. In research, quantitative traits are also called ‘variables’ as they change and can more easily be divided into range.

Target Population

 All the units present in the universe cannot be the target of the study. As has already been mentioned above, the researcher has to choose to include in his sample only those units that mark the characteristics to be studied by the researcher. Thus, in the example of study of drug abuse among school children, the teenage students of school are the population, while only students who have suffered from or are suffering from the drug abuse problem are included in the ‘target population.

Sample Size

 Deciding the size of a sample is a major concern for a researcher. Size of the sample is the total number of sampling units that the researcher will include in the sample. The sample size cannot be too huge or  too small.

Biased Sample

Even after taking utmost care it is possible that a sample chosen by the researcher represents some characteristics of the population more than the others. Such a sample is called as a biased sample. It is important for the researcher to be aware and make sure that his sample is not biased to avoid sampling errors as well as authenticate his research

Sampling and Non-Sampling

Error No human efforts can be wholly flawless and without errors. Research is also bound to be ridden by some mistakes, small and big. It is a customary practice to mention in the research the loopholes in the results of the results. It shall make the research honest and also serves as a disclaimer for the reader to not treat the results whole and sole analysis on that study. The loopholes in the research may be as a result of wrongly taken sample or due to other technical obstacles. The errors in the research that are caused due to sampling are called sampling errors; while those errors that are caused due to other than sampling faults are called non-sampling errors. While sampling errors can be predicted quite precisely as they can be calculated, non-sampling errors can only be instinctively guessed by the researcher.

Sampling errors can be avoided by being cautious in choosing the sampling technique. Non-sampling errors are the errors in results that arise as a result of pre or post sampling processes. Although non-sampling errors are not connected with the process of sampling, yet all steps in research are closely connected with each other and influence one another. Non-sampling errors occur at stages like research design, data collection, data analysis, etc. Thus, errors that occur without corresponding to the sampling process are called non-sampling errors. Together sampling and non- sampling errors gives an imperfect sample, and therefore, a faulty study result.

Purpose of Sampling:

Accuracy of Results, Time efficient, Cost effective, Convenience

Classification of Sampling: There are mainly three kinds of sampling. Let us understand these kinds as follows:

1.      Probability Sampling:

Equal chance  : Where the sample is chosen in such a manner that all the elements present in the universe have an equal chance of being represented in the sample, then it is called as ‘Probability Sampling’. The sampling techniques that come under ‘probability sampling’ are used in the cases where population is homogeneous.

2.      Non-probability Sampling:

In ‘Non-probability sampling’, all the units do not stand a chance to be included in the sample. Non-probability sampling does not guarantee representativeness. It is also called as ‘decisive sampling’ or ‘purposive sampling’ as the basis of sampling is the free will of the researcher. Purposive sampling is used where the size of the universe is unknown or indefinable.

3.      Mixed Sampling:

There are some sampling techniques which do not fall under the above two mentioned categories strictly. These techniques display some characteristics of a ‘probability sampling’ and some characteristics of a ‘non-probability sampling’. Such sampling techniques may be called as ‘mixed sampling.

Sampling Techniques

Pick units: As we have understood above, sampling means to pick units from the universe to form a sample (or samples, depending on the study) for conducting research. Sampling can be done using some techniques that have been developed over time by researchers. The various techniques that are known and used widely have been discussed as follows:

1.      Simple Random

 Chosen randomly :  Sampling as the name suggests ‘simple random sampling’ refers to sampling done in a simple manner where sampling units are chosen randomly. In simple random sampling there is no procedure followed for sampling, thus it is called ‘simple’. Also units are selected to be in the sample in a random fashion. There is no systematic choosing. Simple random sampling falls under the category of ‘probability sampling. In , probability sampling requires that complete list of units in the universe must be known.

i.                    Lottery

Lottery means where lots are blindly picked, and it is a matter of chance that which lot gets picked. Lottery is the simplest way of conducting sampling. In this method a number is given to all the units in the universe. All these numbers are then written down on small pieces of paper, which are then put together. Whichever number appears in the picking, are the units to be included in the sample. The example  in which the names of 25 employees out of 250 are chosen out.

ii.                  Tippet’s Table

While the lottery method was popularly used for a long time for sampling, various scholars pointed out a fact that even though lottery method ensured a random way of sampling. These researchers have come up with various tables consisting of random numbers. Of these, the table formed by a researcher and scholar named Tippet, is most widely used in social researches. Tippet has formed a table of 10,400 numbers having 4 digits. The method of using this table is to first assign numbers to the complete list of units in the universe and the randomly select any number in the Tippet’s table. Thereafter go on selecting the units from the list as per the numbers given in the table. A portion of Tippet’s table is reproduced below to provide an understanding of how the table works:

 2952         7979    5911    9525    7203

6641          3563   3170    1545    5256

3992          9792    5624   4167    1396   

6008          8125    1300    2693    2456

The advantages of using ‘simple random sampling’ are: a. It is hassle-free method of sampling population is homogeneous. b. There is no chance of personal bias of the researcher to influence sampling.

The following are the disadvantages of using ‘simple random sampling’: a. It cannot be used in heterogeneous population. b. It does not make use of any special and particular circumstances that may be present in a population.

2.      Interval Sampling

A form of random sampling in which participants at uniformly separated points are selected for study and the starting point for selection is arbitrary.  This kind of sampling may be characterised by its systematic nature of uncertainty. Interval sampling is random in the sense that there is no basis for deciding the units to be chosen, yet it follows a systematic format of choosing the uncertain units. Example   every 10^th name of list.

3.      Stratified Sampling

Stratified random sampling is a method of sampling that involves the division of a population into smaller sub-groups known as strata. In stratified random sampling, or stratification, the strata are formed based on members' shared attributes or characteristics such as income or educational attainment. The universe to be studied by the researcher is not always homogeneous. Heterogeneous population is often formed in such a way that it can be divided into different strata of homogeneous population. Stratified Sampling is helpful for doing drawing samples out of such a population. First the population is divided into different strata or layers and then samples are drawn out of each stratum. For example for a study of 1,000 persons, the population consists of persons belonging to four different religions in this manner: 400 people in Religion A, 300 people in Religion B, 200 people in Religion C and 100 people in Religion D. the researcher decides to create a sample of 200 people, that is 20% of the population. Now for the final sample to proportionately represent each stratum, the researcher must draw out 20% of sample from each stratum as well. Thus, there will be 80 persons from Religion A, 60 persons from Religion B, 40 persons from Religion C and 20 persons from Religion D.

Purposive Sampling (Characterstics)

Purposive sampling, also known as judgmental, selective, or subjective sampling, is a form of non-probability sampling in which researchers rely on their own judgment when choosing members of the population to participate in their surveys. Purposive sampling is also known as ‘Judgment Sampling’, as it relies entirely on the wish and judgment of the researcher. This is the purest form of Non- probability Sampling. No unit in the universe stands any chance of being included in the sample except the ones that the researcher himself/herself chooses.

Convenience Sampling

Convenience sampling is a type of non-probability sampling that involves the sample being drawn from that part of the population that is close to hand. This type of sampling is most useful for pilot testing. It is only a matter of chance that a unit may be convenient for the researcher to sample and others are not. The most suitable example is the feedback surveys conducted for any product in the market. A basic example of a convenience sampling method is when companies distribute their promotional pamphlets and ask questions at a mall or on a crowded street with randomly selected participants.

 Cluster Sampling:

The Cluster sampling involves drawing samples from smaller clusters that the population is divided into. Cluster sampling is a probability sampling method in which you divide a population into clusters, such as districts or schools, and then randomly select some of these clusters as your sample. The University has various departments, which can be considered each as a cluster.

 Sequential Sampling

 The term sequential sampling describes any method of sampling that reads an ordered frame of N sampling units and selects the sample with specified probabilities or specified expectations. For example for a study on access to human rights for the LGBT community in India the researcher may not be able to define the universe to draw out a sample, owing to the repressed state they live in India.

Quota Sampling

Quota sampling is a very useful method of sampling where a large body of persons is to be studied. In quota sampling the population is divided into different categories on the basis of some characteristics, and selection of units in the sampling is done according to the proportion that group represents in the entire population. For quota sampling the researcher must first define the characteristics on the basis of which the population shall be divided into groups. The researcher must have knowledge about the proportion that each characteristic group possesses in the population. The sample drawn from the universe would proportionately represent the characteristics in the population. For example quota sampling can be used in a study of pre-teen and teenaged children of imprisoned parents in the state of Bihar. The universe is divided into boys and girls, and the researcher finds out that there are 750 boys and 500 girls. The researcher decides to draw a sample of 250 children. The researcher further divides the universe into age groups. Let us say the composition of the universe is the following

 Gender                Below 6 yrs             6 to 12 yrs           13 to 19 yrs

(Age Group I)     (Age Group II)     (Age Group III)                                                                                                                                                         Total

 Total Boys                  250                              300                  200                              750

 Girls                           150                              200                  150                              500

 Total                           400                             500                  350                               1250

Multi-stage Sampling

 Multi stage sampling, as the name suggests, is sampling carried out in multiple stages. Different techniques at each stage may also be used. For example, for a study on the crime rate in India, the country is divided into different zones, North, West, South and East. This is the first stage of sampling wherein stratified sampling is used, each zone being a stratum. The states in each zone serve as clusters, so the second stage of sampling is cluster sampling. Finally samples from each state are drawn out using purposive sampling. This is a simple example to illustrate the method of doing multi-stage sampling.

 Multi-phase Sampling

 Multi-phase sampling is quite similar to multi-stage sampling, barring some technical differences. The procedure for carrying out sampling is similar, but in multi-phase sampling, the aim is not to create a final sample. Study is done continually in various phases. Unlike multi-stage sampling, each sample is first studied as a sample, before further drawing sample out of it. An advantage of doing this sampling is that in-depth investigation is possible.

Volunteer Sampling

 Volunteer sampling is close to the convenience sampling, as in this type of sampling also the researcher chooses the respondents as per convenience. The only difference is that in this sampling, the researcher himself is a volunteer for the sample; that is to say, the researcher himself participates in the research as a sample. However, it is not considered an objective form of sampling, as personal bias of the researcher has access into the data collection. Also, representativeness of the sample is very questionable. This type of sampling is only done in very small scale researches where empirical verifiability can be set free, so as to make way for qualitative conclusions.