Sampling: What It Is and Why It Is Used
Chapter -Six
Syllabus topic 5, "Research Methods"
Pages 385 to 387 of 543
In one line
Sampling is studying a part in order to say something about the whole, and it works only if the part was chosen in a way that gives every member of the whole a known chance of being in it.
In the wording a student can write in an exam: a sample is a subset of a population selected for study, and sampling is the process of selecting it; the population or universe is the entire set of units about which conclusions are to be drawn, the sampling frame is the list from which the sample is actually drawn, and the sampling unit is the individual element selected; sampling is used because studying an entire population is usually impossible in time, cost and access, and it is valid only where the method of selection permits an inference from the sample to the population.
The vocabulary, which must be exact
Population or universe. Every unit the conclusion is to be about. Every household in three settlements; every accused produced for remand in one court in six weeks; every college affiliated to one University.
Sampling frame. The list from which the sample is drawn. This is the term students omit and it is where most error enters, because the frame is rarely the population. The voters' list is not the population of adults; the clinic's register is not the population of people with legal problems.
Sampling unit. What is selected: a household, a person, a file, a sitting.
Sample size. How many.
Sampling error. The difference between the sample result and the true population value, arising because a sample is not the whole. It can be estimated for probability samples and cannot for others.
And non-sampling error, which is everything else: a bad question, a mistaken entry, an interviewer effect. It is usually larger than sampling error and no increase in sample size reduces it.
Why sample at all
Cost. A hundred and fifty schedules is a fortnight; fifteen thousand is not.
Time. A study that takes three years answers a question nobody now has.
Access. Some populations cannot be reached in full.
And accuracy, which is the surprising one. A well drawn sample studied carefully can be more accurate than a census done badly, because the effort available per unit is greater. A hundred careful interviews beat a thousand rushed ones.
When not to sample
When the population is small enough to take entirely. Chapter 990: eleven colleges, eighty-four register entries, six weeks of remand sittings. A census removes sampling error and every objection about representativeness, and a student should always ask this question first.
When the units are very unequal. Where one unit dominates, a sample that misses it is worthless. If a study of legal aid work in a district covers ten clinics of which one handles most of the caseload, a sample of three that excludes the big one describes nothing.
Sampling: What It Is and Why It Is Used
And when the question is about a particular unit, which is a case study, chapter 980.
The frame problem, worked
This is the part that produces most bad student research, so it is worth an example.
The population. Adults in one settlement.
The frames available. A ration card list, a voters' list, the households a local body knows about, or the researcher walking every lane.
What each omits. The ration card list omits those without cards, who are likely to be the poorest and most recently arrived. The voters' list omits those not registered, who are likely to be migrants. A local body's list omits whoever it does not recognise.
And notice the direction. In a study about legal awareness or legal aid, every one of those omissions removes precisely the people the study is about.
What to do. Choose the frame deliberately, say which one was used, and state in the limitations what it omits, chapter 1280. A frame is never perfect; concealing which one was used is what makes it fatal.
Sample size
There is no single right number, and a student should resist the idea that there is.
What it depends on. How varied the population is; how precise the answer must be; how the results will be broken down, since analysing by sex and by settlement together needs enough in each cell; and what is feasible.
A practical rule for a student. Decide the smallest group you will want to report on separately, and make sure there are enough of those. A study of ninety people reported as one number is fine; the same ninety split into twelve categories is nine numbers of no value each.
And honesty beats size. Forty respondents with a stated frame, a stated method and a stated refusal rate is research; four hundred chosen by convenience is not, chapter 870.
A worked example
Designing the sample for the awareness study of chapter 970.
Population. All adult residents of three named settlements in one taluk.
Frame. No list exists that covers them, so the frame is constructed: the researcher walks each lane and numbers every dwelling. That is a real frame, it is disclosed, and it omits only people with no dwelling, which is stated as a limitation.
Sampling unit. The household, with one adult respondent per household.
Method. Systematic: every fifth dwelling from a random start, chapter 1010.
Size. Ninety-six approached, chosen because two weeks allow about eight a day.
Non-response. Twelve refusals recorded, giving eighty-four completed, and a response rate that is reported.
Sampling: What It Is and Why It Is Used
The claim the study may then make. About adults in those three settlements, at that time, subject to the frame's omission and the refusal rate.
And the claim it may not make. About the taluk, the district or the State. Writing that sentence into the dissertation is what makes the rest of it usable, chapter 1280.
Quick revision
Vocabulary: population or universe, every unit the conclusion is about; sampling frame, the list actually drawn from, which is rarely the population and is where most error enters; sampling unit; sample size; sampling error, estimable only for probability samples; and non-sampling error, which is usually larger and is not reduced by a bigger sample.
Why sample: cost, time, access, and accuracy, since effort per unit is greater.
When not to: when the population is small enough for a census; when units are very unequal and one dominates; and when the question is about a particular unit, which is a case study.
The frame problem: every available frame omits somebody, and in studies of awareness and legal aid the omissions remove exactly the people the study is about. Choose deliberately, disclose the frame, and state what it omits.
Size: no single right number; it depends on variation, required precision, the breakdowns wanted and feasibility. Honesty beats size.
Test yourself
1. Define population, sampling frame and sampling unit, and say why the frame matters most. The population or universe is the entire set of units the conclusions are to be about; the sampling frame is the list from which the sample is actually drawn; the sampling unit is the individual element selected. The frame matters most because it is rarely identical to the population, and whoever it omits is silently excluded from the study.
2. Give two situations in which a researcher should not sample. Where the population is small enough to study entirely, since a census removes sampling error and every objection about representativeness; and where the units are very unequal, so that a sample missing the dominant unit describes nothing.
3. Distinguish sampling error from non-sampling error. Sampling error is the difference between the sample result and the true population value, arising because a sample is not the whole; it can be estimated for probability samples. Non-sampling error is everything else, such as a badly drafted question, a mistaken entry or an interviewer effect; it is usually larger and is not reduced by increasing the sample size.
4. What practical rule should govern sample size for a student? Decide the smallest group that will be reported on separately and ensure there are enough units in it, because a sample adequate for a single overall figure becomes useless once it is broken into many categories.
The rest of this subject
These notes are cut from the University's printed syllabus. Open the syllabus itself, or the past papers, for the same subject.