# Sampling Method

In sampling technique, instead of observing and studying each and every unit of the universe, only a part of it is studied, assuming that it best represents the entire population. It is applicable only to random samples. In short, data collection and sampling technique play an important role in the quantitative research.

Before acquiring knowledge of sampling techniques it is necessary to understand the concept of sampling.

Simply a smaller representation of a large whole is called sample. When the researcher would like to find out something about the population in the specific region, some elements are selected in the population by researcher, selection process of elements is called sampling.

Shortly, when population is relatively large (e.g. All patients in the hospital or all households in the village ) and is physically not accessible, the researcher surves only a sample.

Sampling can be defined as “ the selection of part of an aggregate or totality on the basis of which a judgment or inference about the aggregate or total is made.” The concept like parameter, statistic, sampling errors, precision, variable, sampling frame, sampling design etc. are used in sample and sample design.

## Theory of Sampling

Sampling theory is a study of relationship between samples and population. It is applicable only to random sample. The theory of sampling is known as the methodology of drawing inference of the universe from random sampling. The theory deals with,

1. Statistical Estimation
2. Testing of Hypothesis
3. Statistical Inferences

### Statistical Estimation

Estimating the value of unknown parameter is the main objective of sampling. Point estimate and interval estimate are the two type of estimates. Point estimate is a single estimate in the form of a single figure. On the other side interval estimate has two limit. i.e. lower limit and upper limit within which the parameter value may lie.

### Testing of Hypothesis

Accepting or rejecting a hypothesis is the second objective of the sampling theory. This theory helps us to decide whether an observation obtained in sampling has occurred due to a real one or fluctuations in sampling.

### Statistical Inferences

Theory of sampling helps in making a generalization about the population and determining the accuracy of such generalization.

## Scope of Sampling Method

Sampling is a widely used technique. The sampling technique plays an important role in the field of quantitative research.

In the research of social sciences, commerce and management, education and science, the researcher uses the sampling technique systematically. Also in our daily activities the sampling technique is very important.

For example, when the housewife cooks the rice, she examines very little part of it and then decides it is eatable or not. Likewise, the pathologist also takes the sample of urine or blood for diagnosis of the patient’s disease or illness.

In other words, sample is an instrument for learning about large masses by observing a very few individuals.

## Features of Sampling Method

The main features of sampling method can be described as under:

• Save time & money: The sampling technique saves time and money of the researcher. Compared to census technique, sampling technique is less expensive and less time-consuming.

• Respondent Co-operation: Through sampling technique the researcher achieves greater response and co-operation from respondents.

• Accuracy and Reliability: Sampling technique increases the accuracy of data and if the researcher selects the units of sample carefully, automatically reliability of results highly increases.

• Detailed and Depth study: Sampling will reduce a large number of people. The sample unit is small. Therefore, the researcher can easily elaborate and examine the sample in detail and depth with various dimensions.

• Scientific base and greater suitability: The conclusions derived from the sample study can be checked by other samples. Moreover, according to most surveys, the sampling techniques are found suitable in maximum cases.

## Limitations of Sampling Techniques

The limitations of sampling technique are mentioned below:

1. Requirement skilled person: It is necessary that the researcher is skilled enough for using sampling technique. If the non-skilled researcher performs the sample selection, it can be incorrect leading to sampling error.

2. Less accuracy and reliability: Compared to census technique, the conclusions of sampling techniques are less accurate and reliable. It is more liable to error than the census technique’s conclusions.

3. Misleading conclusions: For using sampling technique it is necessary that extra care is taken by the researcher. Otherwise, the conclusions derived from sampling techniques will become misleading.

## Characteristics of Sampling

For reliable and accurate conclusions from the research it is necessary that the selected sample exhibits certain qualities. These qualities are as follows

1. Every sample should be independent of each other.

2. The homogeneity in selected sample is required. Every element in selected sample should be identical with another element.

3. An ideal sample represents the characteristics of entire population.

4. The selection of sample should be sufficient. If the selected sample size is very small or too large, the researcher can not derive correct conclusion.

## Types of Sampling Method

Basically, Sampling is divided into two types

• Probability/Random Sampling
• Non- probability/Non Random Sampling

When every unit of the total population has an equal probability of being selected for the sample, it is called probability sampling. This technique of sampling is relatively time consuming, complicated and expensive because of a high degree of representativeness.

Probability/Random Sampling

The probability sampling methods are described as under:

### Simple Random Sampling

According to Goode and Hatt, “A random sample is one which is so drawn that the researcher, from all pertinent points of view, has no reason to believe a bias will result.

In other words, the units of the universe must be so arranged that the selection process gives equiprobability of selection to every unit in that universe.” Sometimes random sampling is referred to as Representative Sampling.

The simple random sampling is also known as the method of chance selection. In this sample each possible sample combination or each and every item in the whole population has an independent and equal chance of being included in the sample. While adopting this technique of sampling some precautions have to be taken. The list of all the units that are available for the purpose of population should be well defined.

According to this method, the sample units are selected by number sub-method like Tippet’s tables, pricking blindfolded, Personal identification number (PIN) Lottery Method, Computer etc.

#### Advantages of Simple Random Sampling

1. Very simple and easy method for understanding.
2. Each unit of the population has equal chance for selection.
3. No place for personal bias.

#### Disadvantages of Simple Random Sampling

1. It is an expensive and time-consuming method.
2. Practically it is very difficult to give an equal chance of selection for all the element.
3. The researcher has no control over the selection of the units.

### Stratified Random Sampling

Sampling Techniques In this method, the whole population is divided into homogeneous groups called ‘ strata’ and a sample is drawn from each stratum. It should be noted here that the strata should be homogeneous but between the strata, there should be heterogeneity.

According to research objectives, the researcher will divide the population into homogeneous strata. The homogeneity is based on various criteria’s of the population. e.g. age, gender, religion, caste, class, family type, education, occupation and so on. The stratified random sampling is systematically defined as,

“The method involving dividing the population in homogeneous strata and then selecting simple random samples from each of the stratum.”

The types of stratified sampling are, (a) Proportionate and (b) Disproportionate.

When the researcher draws the number of items proportionately from each stratum, it is called proportionate and if the researcher draws the unequal number of items from each stratum, it is called disproportionate.

#### Advantages of Stratified Random Sampling

1. It is useful for comparing sub-categories/groups
2. Compared to simple random sampling, it can be more precise.
3. The researcher has greater control over the selection of the sample.

#### Disadvantages of Stratified Random Sampling

1. Compared to simple random sampling, it requires more effort.
2. If the stratification is based on non-scientific consideration or biases, automatically the sampling can get biased.
3. It is time-consuming and complex method of determining sample.

### Systematic Sampling

It is also called interval or quasi random sampling. It is a variation of simple random sampling. Systematic sampling is randomly selecting the first respondent and then every K th person after that; ‘K’ is a number termed as sampling interval.

In this method, the population are arranged in such a way that each unit of the universe can be identified by its order. e. g. Muster roll, companies, telephone directory etc. From this, the sample is drawn based on specific interval.

For example, arrange the names of students in MBA class in ascending or descending order on the basis of alphabet. In this order the researcher draws the sample at regular intervals with a random start.

Suppose, in selecting a sample of 40 students out of 400, the population total, viz 400 is divided by 40. Then any random number between 1 to 10 is selected by the researcher. Suppose, the researcher selects the random number 3. Then, item numbered 3, 13, 23, 33, 43, ..…, etc. are selected. This system ensures greater representativeness throughout the population.

1. Simple and easy method for use.
2. It is a rapid method in probability sampling. Further, it eliminates several steps for choosing a sample.

1. This method is used only when the systematic sufficient is available.
2. According to Black and Champion, in this method, each element has no equal chance of being selected.

### Cluster Sampling

According to this method, the total population is divided into clusters and after that the sample is drawn either from the selected clusters or from all clusters.

Small grouping within the population are called clusters. Initial clusters are known as primary sampling units. 1st stage state level, 2nd stage district level, 3rd stage village level and 4th stage household level. In stratified sampling each stratum should be as heterogeneous as population is.

But in cluster sampling all the units should possess all the characteristics of the population.

1. It is an economical method
2. This method is useful for capturing the wide disparity of heterogeneity of the population.

1. Cluster sampling is less precise than random sampling.
2. Within the clusters, there is not as much information as in the sample.

Non Probability/Non Random Sampling

This type of sampling technique does not provide every element in the population with the equal chance of being selected in the sample. The researcher decides which sample element should be chosen. This method of sampling is usually used for qualitative exploratory analysis.

The major forms of non probability sampling are described as follows.

### Convenience Sampling

It is also known as accidental or unsystematic sampling. In this method the researcher interviews that population which is selected neither by judgment norby probability.

Shortly, the researcher studies all those people who accidentally come in his contact during research period or who are available most conveniently.

Generally, the results obtained by this method are unsatisfactory and biased. This method may be used when complete list of the source is not available, when sample unit is not clear and when population/universe is not well defined.

1. For making pilot studies, convenience sampling is useful.
2. It is economical and a quick method of sampling.

1. Mostly, the results, obtained by this method are unsatisfactory and biased.
2. In this method, the respondents maybe those who are vocal.

### Quota Sampling

It is another version of stratified sampling. There is little difference between stratified and quota sampling. In stratified sampling population is divided by strata and then the researcher chooses the respondents, but in quota sampling it works on ‘quota’ fixed by the researcher.

Like stratified sampling in quota sampling, the researcher specifies the groups. The groups are prepared on the basis of sex, age, income, education or other characteristics much like the ‘strata’ in stratified sampling.

1. Compared to other sampling techniques it is an economical method.
2. It is less time consuming sampling method.

1. In this method, there is a possibility of interviewer’s bias influencing the sample selection.
2. It is not possible to estimate the error.
3. Quota sampling is not a representative sampling technique.

### Purposive Sampling

This method is also called judgment sampling method. In this method the researcher selects the sample according to his personal judgment. But, when the researcher is choosing the sample, he omits extreme items and considers only the average items.

Shortly, in this method, the researcher selects only those variables which represent the universe but the selection of these units is based on prior judgment and deliberate thinking.

(1) This method is useful when a small number of sampling units are in the universe. (2) It is useful for solving the urgent problems in business and making public policy decisions.

1. In this method, there is no objective way of evaluating the reliability of sample results.
2. It is not useful for the large universe.

### Snowball Sampling

In this sampling method, the researcher begins his research with few known and available respondents. This technique is used by the researcher when the population is rarely found and not easily accessible.

In this method, when the first respondent is found to the researcher then the researcher can ask respondent to provide contact details of other individuals who fall in the same population. This process is followed by the researcher until adequate number of respondents are interviewed or no more respondents are discovered.

1. This technique is useful in case of rare population.
2. This technique reduces the sample size.
3. Comparatively, it is an economical sampling technique.

• It is very difficult to track the respondents at a time and in time.
• Sometimes this technique’s cost is going to be very high.

## Sample Size

The wrong sample size in the research would make the study difficult. If the researcher chooses small sample size it makes the result less accurate and a larger sample size involves more cost and time. On this background it is necessary that the size of sample is optimum in nature.

But it depends on the following factors:

1. The population size: The larger/ smaller the size of the population, the larger/smaller should be the sample size.

2. Nature of population: For homogeneous population, the small sample size may be sufficient. But for the heterogeneous population, a large size of the sample required.

3. Type of sampling: In stratified or cluster sampling, smaller sample size will be required. But in the case of simple random sampling, large sample size is necessary.

4. The degree of accuracy: When the researcher is highly interested in inaccurate results of the research study, he needs to have a larger sample size.

5. Qualitative or Quantitative Study: This factor also affects the size of the sample. For qualitative research, the size of the sample has no numerical boundaries. But, for quantitative study, the researcher himself can decide the sample size ( e.g. accidental sampling).

## Sampling and Non-sampling Errors

### Sampling Errors

Sampling errors are also known as sampling fluctuations. When the researcher completes his research through sample survey of small proportion of the total population and derives the results, naturally, there would be a certain amount of errors or inaccuracies. Such type of errors are called sampling errors.

There are two types of sampling errors:

• Biased Errors: This type of errors arises due to the researcher’s bias or prejudice in sampling technique of sample selection.

For example, when the researcher adopts the purposive sampling method in place of simple random sampling technique.

As a result of such selection, some fluctuations or errors are bound to arise, which are called biased errors.

• Unbiased Errors: Unbiased errors arise due to chance differences between the members selected in the population and those not included or selected. It is also known as random sampling error.

If the researcher increases the size of the sample automatically the random sampling errors decline.

### Non Sampling Errors

It occurs in any type of survey in research. It includes mistakes and biases. Non sampling errors mainly arise due to vague questionnaire, inappropriate statistical unit, data processing operation errors, faulty interviews, inadequate and inconsistent data specifications, errors from respondents reply etc. It should be noted that the researcher cannot reduce non-sampling errors to zero.

## Summary

• Sample: A smaller representation of a large whole is called a sample.

• Probability Sampling: When every unit of the total population has an equal probability of being selected for the sample, it is called probability sampling.

• Non Probability Sampling: This type of sampling technique does not provide every element in the population with an equal chance of being selected in the sample.

• Sampling Errors: When the researcher completes his research through a sample survey of a small proportion of the total population and derives the results, naturally, there would be a certain amount of errors or inaccuracies. Such type of errors is called sampling errors.

• Non-Sampling Errors: Non sampling errors mainly arise due to vague questionnaire, inappropriate statistical unit, data processing operation errors, faulty interviews, inadequate and inconsistent data specifications, errors from respondents reply etc. It should be noted that the researcher cannot reduce non-sampling errors to zero.