Simple Random Sampling in Research: Definition, Formula, Examples

  • Simple random sampling is a sampling strategy in which every member of the population has an equal and independent chance of being selected.
  • It requires a complete sampling frame, which is its single biggest practical obstacle.
  • The probability that any individual is selected is n divided by N.
  • It is the benchmark technique, because most statistical formulas assume the sample was drawn this way.

What is simple random sampling in research?

Simple random sampling is a probability sampling technique in which every member of a population has the same chance of being chosen, and every possible combination of members has the same chance of forming the sample.

Two conditions define it.

  • Equal probability: no member is more likely to be selected than any other.
  • Independence: selecting 1 member does not change the chance of selecting any other.

 

The second condition is what makes the technique simple in the technical sense. Other probability methods, such as stratified or cluster sampling, also give known selection probabilities, but they impose structure that breaks independence.

Simple random sampling types

The technique varies in 2 ways: whether selected members are returned to the pool, and how the draw is physically carried out.

Sampling without replacement

A member selected once is removed from the pool and cannot be chosen again. This is the standard approach in research, because interviewing the same person twice adds no information. Selection probabilities rise slightly with each draw, since the pool shrinks.

Sampling with replacement

A selected member is returned to the pool and could be chosen again. Probabilities stay constant across draws, which keeps the mathematics clean, so it appears in statistical theory and in bootstrap resampling far more often than in fieldwork.

The lottery method

Every member is assigned a number, the numbers are mixed in a container, and the required quantity is drawn by hand. It is transparent and easy to demonstrate, which is why it survives in teaching and in public draws, but it is impractical beyond a few hundred members.

The random number method

Members are numbered, then a random number table or software generator produces the selections. This is the standard modern approach, and it is the only practical option for populations in the thousands.

Why are random samples used?

Random sampling has multiple advantages over non-probability sampling methods:

  • It removes selection bias: chance decides who is included, so the researcher’s preferences cannot shape the sample.
  • It allows you to calculate inferential statistics: confidence intervals, p-values, and margins of error all assume random selection.
  • You can calculate sampling error: you can quantify how far your estimate is likely to sit from the truth.
  • It produces representative samples on average: across repeated samples, estimates center on the true population value.
  • Peer reviewers, regulators, and examiners accept a documented random draw in a way they do not accept researcher judgment.

 

Note the phrase on average. A single random sample can still misrepresent the population by chance, which is sampling error rather than bias. The difference matters: sampling error shrinks as the sample grows, while bias does not.

Limitations of simple random sampling

  • A complete sampling frame is required: for many populations no such list exists.
  • Cost and time: selected members may be scattered across many locations.
  • No subgroup guarantee: a small subgroup can be missed entirely by chance.
  • Non-response undoes it: a random draw with a 30 percent response rate is no longer a random sample of anything.

Simple random sampling formula

Three formulas cover most undergraduate work.

Probability of selection

p = n / N

With a population of 2,000 and a sample of 200, every member has a 200 divided by 2,000 chance of selection, which is 0.1 or 10 percent.

Sample size for a proportion

Cochran’s formula gives the sample size needed for a target margin of error.

n0 = (Z² × p × q) / e²

Symbol Meaning Common value
Z Z-score for the confidence level 1.96 at 95 percent
P Expected proportion 0.5 when unknown
Q 1 minus p 0.5 when p is 0.5
E Margin of error 0.05 for plus or minus 5 percent

 

Substituting those values gives n0 = (3.8416 × 0.25) divided by 0.0025, which is 384.16, rounded up to 385. That is where the familiar figure of roughly 385 respondents comes from.

Finite population correction

When the population is small, the required sample shrinks.

n = n0 / (1 + ((n0 – 1) / N))

For a population of 2,000, the corrected figure is 384.16 divided by 1.1916, which is 322.4, rounded up to 323. You need 323 rather than 385.

Slovin’s formula is a simplified alternative that appears in many undergraduate projects.

n = N / (1 + N × e²)

With N of 2,000 and e of 0.05, this gives 2,000 divided by 6, which is 333.3, rounded up to 334. It is easier to apply but less flexible, because the confidence level is fixed at 95 percent.

Random sampling generator

Nobody draws numbers from a hat any more. These are the practical options.

Tool How to use it Best for
Excel or Sheets RAND function, then sort Most student projects
Python random.sample on a list Large or automated draws
R The sample function Statistical workflows
SPSS Select Cases, random sample Coursework using SPSS
Online generators Enter a range and a quantity Quick one-off draws

 

Drawing a sample in Excel

  • List every member of the sampling frame in column A.
  • Enter the RAND function in cell B2 and fill it down the whole list.
  • Copy column B, then paste it back as values so the numbers stop recalculating.
  • Sort the sheet by column B, smallest to largest.
  • Take the top n rows as your sample.

 

The paste-as-values step matters. The RAND function recalculates every time the sheet changes, so a draw that is not frozen will keep on rearranging itself, making sampling more confusing.

Whichever tool you use, save the generated list.

Random sampling vs systematic sampling

Simple random sampling makes an independent draw for every member selected. Systematic sampling makes 1 random decision, the starting point, then takes every kth member of an ordered list.

Feature Simple random sampling Systematic sampling
Random draws 1 per member selected 1 in total
Ordered list required No Yes
Independence of selections Yes No
Speed in the field Slower Faster
Main risk Sampling error alone Periodicity in the list

 

Independence is the practical advantage of simple random sampling. Because draws are independent, standard variance formulas apply directly, and confidence intervals need no adjustment.

Systematic sampling is a lot more convenient. Sampling every 20th item off a production line needs no numbering of the population, while a simple random draw of the same items would.

Both are probability techniques, and both give every member a known chance of selection. The difference is how many random decisions the researcher makes.

 

Random sampling vs stratified sampling

Stratified sampling divides the population into subgroups first, then draws a random sample inside each one. Simple random sampling skips that step and draws from the whole population at once.

Feature Simple random sampling Stratified sampling
Population divided first No Yes, into strata
Subgroup representation Left to chance Guaranteed
Information needed A list of the population A list plus subgroup data
Precision at the same size Lower Higher
Complexity Minimal Moderate

 

Stratified sampling is more precise whenever the strata genuinely differ on the outcome, because variation between subgroups is removed from the sampling error. When subgroups do not differ, the extra of creating strata doesn’t give you much benefit.

Simple random sampling remains the better choice when you have no subgroup information or when the population is fairly uniform with regard to variables that matter.

 

Frequently asked questions

What is the difference between simple random sampling and random assignment?

Random sampling decides who takes part in a study. Random assignment decides which condition each participant experiences once they have joined.

Random sampling supports generalizing results to a population; random assignment supports causal claims about the treatment, i.e., internal validity. A laboratory experiment often has random assignment without random sampling, which is why its findings can have high internal validity and still be hard to generalize.

Do you need a sampling frame for simple random sampling?

Yes, and this is the technique’s binding constraint. Without a complete list you cannot number the population, and without numbering you cannot draw at random.

An incomplete frame produces coverage error. If your list of a cancer patients omits those with recent diagnoses, no amount of randomness will make up for them, and the sample is biased before the draw begins.

Can a simple random sample still be unrepresentative?

Yes. Randomness guarantees fairness in the process, not balance in any single result. A random draw of 100 students could return 80 women from a population that is 60 percent women.

This is sampling error rather than bias, and it behaves differently: it is calculable, it is equally likely to run in either direction, and it shrinks as the sample grows. If a specific subgroup balance is essential, stratified sampling is the technique that guarantees it.

How large should a simple random sample be?

Sample size follows from the precision you need, not from the technique. Use Cochran’s formula for a target margin of error, then apply the finite population correction if the population is small.

Is simple random sampling always the best technique?

No. Stratified sampling is more precise when subgroups differ. Cluster sampling is cheaper when the population is geographically spread. Systematic sampling is faster in the field. Simple random sampling wins on simplicity and on the fact that every standard statistical formula assumes it.

 

Summarize this Blog with AI

Comment

There are no comment yet.

TOP