A sampling is a collection of items to the elements form a population or universe. Hence a sample is only a portion or subset of the universe or population. 50 out of 5,000
Population or universe refers to the entire group of people, events or things of interest that the researcher wishes to investigate.
Making a census study of the entire universes is not possible on account limitation of time and money. So need to go for sampling.
- Choosing the sample units (how are to be surveyed)
- Choosing the sample size (how many to be surveyed)
- Choosing the sampling procedure (how to ensure that those who are to be interviewed are included in the sample)
- Choosing the media (how to reach respondents in the sample? Through mail survey, email, personal interview, or telephone interview.)
The Sampling Process
Step 1 Define the population
Population must be defined in terms of elements, sampling units, extent and time.
- Conclusions cannot be drawn concerning a population until the nature of the units.
- A research project is required to define the population.
Step 2 Specify the sampling frame
A sampling frame is a means of accounting for all elements in the population.
Step 3 Specify the sampling unit
The sampling unit is the basic units containing
Step 4 Selection the sampling method
The sampling method is the way the sample units are selected.
Step 5 Determination of the sample size
The number of elements of the population to be sampled is chosen.
Step 6 Specify sampling plan
The sampling plan involves the specification of how each of the decisions made thus far is to be implemented
Step 7 Select the sample
It is the actual selection of sample elements. It requires a substantial amount of office and field work.
Principles of Sampling
These are crucial to keep in mind while determining the sample size.
- In a majority cases of sampling, there will be a difference between the sample statistic and the true population
- The greater the sample size, the more accurate will be the estimate of the true population mean.
- The greater difference in the variable under study in a population for a given sample size, the greater will be difference between the sample and population mean.
Types of Sampling
- Probability Sampling
- Non-probability Sampling
- Probability Sampling:
In probability sampling, every element in the population has a known non-zero probability of being selected. This probability is attained through some mechanical operation of randomization.
Four types of probability samples can be identified.
- Simple Random Sampling: Simple random sampling is the purest form of probability sampling and commonly used in research.
- Lottery is also a method of selecting a simple random sample.
- May resort to the use of random table.
- There are numerous computer programmers that generate random numbers.
This method eliminates the possibility that the sample is biased by preferences of the person selecting sample.
- Systematic Sampling: Systematic sampling can increase the sample’s representativeness when the population elements can be ordered in some pattern with regards to the characteristics being investigated.
This sampling method involves the random selection of the first item and then selection of a sample item at every Kth interval.
K = Size of Population/Size of Sample required = N/n
To draw systematic sample, the researcher needs to follow the following procedure:
- List the total of units in the population.
- Decide the sample size.
- Calculate the sampling ration (K = total population size divided by size of the desired sample).
- Draw a sample choosing every Kth
- Stratified Sampling: This sampling method is when we have to select samples from a heterogeneous population such as male and female, or employed or unemployed. Here the population is divided un sub-groups or strata and a simple random sample is taken from each such sub-group. Three key questions which have to be addressed.
- The bases of stratification
- The number of strata
- Sample size within strata.
- Cluster Sampling: Populations may be too large to allow cost effective random sampling or even systematic sampling. Cluster sampling is best used when these sampling are not possible. These clusters might be schools, colleges, industries or even geographical regions. This method is also known as multistage cluster sampling technique. It may not as statistically efficient as random sampling.
- Non-probability Sampling
Non-probability sampling is described as those samples, which are not determined by chance, but rather by personal convenience, or judgement of the researcher.
There is thus the potential for bias in non-probability sampling.
There is growing recognition that non-random samples can credibly represent populations, given that selection is done with the goal of representativeness in mind.
There are four methods of non-probability sampling:
- Convenience Sampling: It refers to samples selected not by judgement or probability by because3the elements in a fraction of the population can be reached conveniently. These samples are called “accidental”, “men-in-thestreet” or “haphazard” sample. It is valid in exploratory research or in the pretest phase of a study. This method is quick, convenient and less expensive.
- Purposive or Judgement Sampling: A purposive sample is one which is selected by the researcher subjectively using his or her judgement. This selection of the sample is deliberate and purposive. Sample representativeness is highly dependent upon the good judgement of the researcher. Probably the most valid usage of judgmental sampling is to obtain expert opinion.
- Quota Sampling: This method of sampling is a restricted judgement sampling technique in which the first stage consists of developing control categories or quotas of the population. Second stage consists of sample elements being selected on convenience or judgement. Selection is normally left to the discretion of the researcher. In this method, the population is divided into a number of segments and the researcher arbitrarily selects a quota of sample items from each segment.
- Snowball Sampling: Snowball is a special non-probability used when the desired sample characteristic is rare. This sample is widely used in applications where respondents are difficult to identify and are best located through referral networks. Hence this sampling is also9 known as “chain referral sampling or network sampling”. An initial group of individual is discovered and then subsequent respondents, possessing similar characteristics, are identified based on referral provided by the initial respondents.
Particularly used in drug culture, teenage gang activities power elites, community relations, political activities, insider trading.
The procedures followed in snowball sampling are as follows:
- Make contact with one or two cases in the population.
- Ask these cases to identify further cases.
- Ask these new cases to identify further new cases.
- Stop when either no new cases are given or the sample is as large as is manageable.
Sampling versus Non-sampling Errors
A sampling error is the error, which is made in selecting samples that are not representative of the population. This error is the result of chance. Increasing the size of the sample can reduce sampling error. The error can be completely eliminated by increasing the sample to include every item in the population.
Non-sampling error, as the name suggests, it everything else (besides the sampling error) that can inject inaccuracies and bias into the results of a study.
These errors include, but not limited to:
- Inaccurate reporting by the respondents (biased guess, inaccurate memory, poor recall etc.).
- Actual lying by the respondents.
- Poor sampling design e.g. the inability to locate proper respondents due to poor instruction, poor maps, non-existence address and so on.
- Misinterpretation of question due to ambiguous wording.
- Respondents terminating their participation in the data gathering.
- Failure of the interviewers to follow instruction, which leads to their leading respondents giving non-verbal clues and recording errors.
- Coding and/or editing error.
Lipstein (1975) offers some broad guidelines for minimizing non-sampling errors in surverys:
- Keep the sample survey as easy to execute as possible.
- Use the smallest sample consistent with study objectives.
- Restrict the questionnaire to data essential to the main issue.
- Pre-test the questionnaire.
- Make an effort to minimize participant’s fatigue.
- Whenever possible, for future guidelines, rotate key questions to discover when respondent fatigue begins
- Establish procedures for keeping both respondent and interviewer involved in the study.
- Do not ask respondents questions they really cannot answer.
- Do not ask the interviewer to do the impossible, such as requests encourage sloppy work and cheating.