What it means
Before statistics can say anything about a whole, it must fairly choose the part. The simple random sample is the gold standard of choosing: every member of the population has an equal chance of selection.
The method is a lottery conducted with rigor: number every member, draw numbers at random, and measure exactly those drawn, no substitutions for convenience. The NIST engineering statistics handbook treats random sampling as the baseline assumption of measurement science: without it, the mathematics of confidence intervals has no ground to stand on.
The payoff is unbiasedness: the sample's average estimates the population's average without systematic tilt, and the margin of error is computable from the sample size alone. The catch is the frame: you can only sample from a list you actually have, so a customer survey samples known customers, and the ones who left are invisible by construction.
Reality fights the method constantly: non-response, unreachable members, and costs force compromises, and every compromise is a potential bias wearing a lab coat. Variants exist for structure: stratified sampling divides the population into layers first, and cluster sampling picks groups, both trading the simple ideal for practicality.
For a non-finance reader, a simple random sample is tasting the soup by stirring first: every spoonful represents the pot only if every drop had an equal chance of being scooped. The square-root law governs the budget: quadrupling the sample halves the error, so precision is bought at a punishing exchange rate and every survey ends as a negotiation with the finance office.
Design effects humble the neat formulas: stratified and clustered designs gain practicality but inflate errors, and honest reports quote the effective sample size, not the nominal one. Big data has not repealed the lesson: a biased sample of millions loses to a random sample of hundreds, because bias does not average out, it scales.
Audit and quality control live on the same foundation: the warehouse stock check and the transaction test both rest on the spun drum, and both fail when the auditor lets the client pick the items.
In practice
Real-world examples.
Example
A lender surveys 1,000 of 80,000 borrowers, drawn by lottery from the account master file. Each borrower has the same 1.25% chance of selection. The analytics team records the random seed so the draw can be reproduced.
Example
Substituting walk-ins for unreachable respondents is refused, because it measures the lobby, not the book. The branch network offers to help, but the statistician keeps the original draw and chases the selected borrowers instead. The response rate is reported openly.
Example
The mobile-channel estimate lands at 63 percent with a three-point margin the board learns to demand. Directors ask for the margin of error before they discuss the result. The figure now shapes the app budget, not an anecdote.
Formula
Calculation
Each of the N population members has selection probability n/N; the sample mean estimates the population mean with standard error approximately s divided by the square root of n, shrinking with the square root of the sample size. For a proportion p, the standard error is the square root of p x (1 - p) / n, and a 95% margin of error is about 1.96 times that figure.
Worked example with fictional figures. A lender draws 1,000 borrowers from 80,000, so each borrower has a selection probability of 1,000 / 80,000 = 1.25%. In the sample, 63% use the mobile channel.
The standard error is the square root of (0.63 x 0.37 / 1,000) = the square root of 0.000233, which is about 0.0153, or 1.53 points. The 95% margin of error is 1.96 x 1.53 = about 3.0 points, so the estimate is 63% plus or minus 3 points, a range of roughly 60% to 66%.
The square-root law shows the cost of precision. Quadrupling the sample to 4,000 borrowers halves the standard error to about 0.76 points and the margin to about 1.5 points, but costs four times as many interviews. Going from 1,000 to 4,000 responses buys only 1.5 points of extra precision, which is why every survey ends as a negotiation with the finance office.Case study
Seen in the real world.
This case study is fictional and illustrative. A made-up lender wants to know how its 80,000 borrowers really use their accounts, and the analytics team proposes surveying all of them. The statistician counters with a simple random sample of 1,000 drawn from the account master file. The first lesson comes from the frame itself: the master file excludes closed accounts, so the survey can never hear from the borrowers who left angry, and the report carries that boundary on its cover.
The second lesson is the discipline of the draw: when the branch network offers to fill in for unreachable respondents with walk-ins, the statistician refuses, because a substituted sample measures the lobby, not the book. The results land with a computable margin: 63 percent use the mobile channel, plus or minus three points, and the board learns to ask for the error bars before the conclusion. The post-project review becomes company doctrine: the cheapest thousand answers that can support a decision beat the most expensive census that cannot, provided the drum was honestly spun. The statistician's parting slide shows the same survey redone with walk-in substitutions, and the eight-point gap in the mobile estimate is the price of convenience made visible.
Watch out
Common mistakes.
- Confusing random with haphazard; convenience samples feel varied but carry systematic bias that no sample size repairs.
- Ignoring the frame; a perfect draw from a broken list measures only who is on the list, never who is missing.
- Substituting non-respondents; replacing refusals with eager participants converts a probability sample into a poll of the willing.
Questions
People also ask.
What is a simple random sample?
A sample where every member of the population has an equal, independent chance of selection, usually drawn by random numbers from a complete list.
Why is it the statistical gold standard?
It produces unbiased estimates, and the margin of error follows from the sample size alone, so confidence can be computed honestly.
What breaks it in practice?
Incomplete frames, non-response, and convenience substitutions, each of which reintroduces the bias the method exists to remove.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%