10Min Research - 33. Understanding and Performing Simple Random Sampling in Social Sciences

TL;DR

Sampling is necessary because studying every population element is usually impractical.

Briefing Cornell Notes

Briefing

Simple random sampling is a probability sampling method built on one rule: every element in the population must have an equal and known chance of being selected. That requirement matters because it’s what makes the sample selection defensible for social science research—without it, results can be biased simply because some people (or units) were more likely to be chosen than others.

The transcript frames sampling as necessary because researchers rarely can study an entire population—defined as every individual, event, or object that could be included in the study. Sampling is the process of drawing a smaller set (the sample) from that population. Within sampling, probability sampling is distinguished from non-probability sampling by whether selection chances can be specified. In probability sampling, each element has a chance of selection; in non-probability sampling, selection chances aren’t known or aren’t tied to each element. The focus then narrows to probability sampling techniques, especially simple random sampling, alongside stratified random sampling and systematic sampling.

For simple random sampling, two conditions must be met. First, the chance must be equal: each element has the same probability of being selected. Second, the chance must be known: the probability can be calculated. The transcript stresses a common misconception: researchers can’t perform simple random sampling by “picking” obvious candidates (like selecting the first person seen or choosing based on memory). Instead, the method requires a complete sampling frame—a full list of all elements in the population with enough details to access them. In the class example, the population is 10 students, and the researcher needs a sample of 3. If the students are listed with identifiers (student ID, name, email, phone, or address), then random numbers can be generated from 1 to 10 to select which three students enter the sample.

The transcript illustrates this with a random number generator configured to produce three unique selections without replacement. If each of the 10 students has a 1/10 chance, that translates to a 10% probability per student. More generally, if the population has 5 elements, each has a 1/5 chance, or 20%. Because the probabilities are specified and equal, the selection process is “random” in the statistical sense.

The method is then extended to a more realistic social science scenario: studying servant leadership and organizational performance in higher education. Here, the population might include academic staff, administrative staff, or both. The key operational step remains the same: compile a complete list of the relevant staff (for example, staff numbered 1 to 2000). If the target sample size for perception-based studies is around 200 to 500, and the researcher aims for about 300, the transcript recommends accounting for nonresponse. Rather than sending questionnaires to only 300 people, it suggests generating a larger initial selection—such as 600—so that even with a 50% response rate, the final usable sample approaches the intended size. The selection itself still relies on random-number selection from the full list, preserving equal and known selection chances.

Cornell Notes

Simple random sampling selects a subset from a population so that every element has an equal and known probability of selection. That requires a complete sampling frame: a full list of all eligible elements with enough identifiers to contact or access them. Selection is done using random numbers (e.g., from 1 to N) mapped to the list, typically without replacement, so the chosen units are not influenced by researcher preference. The probability for each element is calculable—for a population of 10, each element has a 1/10 (10%) chance; for a population of 5, each has a 1/5 (20%) chance. In practice, researchers may oversample to handle nonresponse, selecting more people than the minimum target sample size while still using random selection from the full list.

What makes simple random sampling “simple,” and what two conditions must be satisfied?

It’s “simple” because selection is based on equal, known probabilities for every element. The method requires (1) equal chance—each element has the same probability of being selected—and (2) known chance—researchers can calculate that probability (for example, 1/10 = 10% when there are 10 elements).

Why can’t researchers just pick convenient or memorable participants and call it simple random sampling?

Convenience or memory-based selection breaks the equal-and-known-chance requirement. If someone is chosen because they were “seen first” or because their name stood out, the probability of selection differs across elements, so the selection mechanism is no longer random in the statistical sense.

What operational ingredient is required before random numbers can be used?

A complete list (sampling frame) of all population elements, with enough details to access them. In the class example, the 10 students must be listed with identifiers such as student ID, name, and contact information so that random-number outputs can be mapped back to specific students.

How does the transcript connect random-number generation to equal probability?

Random numbers are generated over the index range of the population (e.g., 1 to 10). If three students are needed, the generator produces three selections (configured without replacement). Because each index corresponds to an element with probability 1/N, each element’s chance is equal and calculable—like 10% when N=10.

How does the method scale to a higher-education study with a large staff population?

The same logic applies: create a full list of eligible staff (e.g., academic and/or administrative staff numbered 1 to 2000). Then randomly select the required number of staff using random numbers mapped to the list. If the target sample size is 300 but response rates are uncertain, the transcript recommends selecting more (e.g., 600) to compensate for nonresponse while still keeping selection random.

Review Questions

What are the two requirements (in terms of probability) that define simple random sampling?
Why is a complete sampling frame necessary, and what kinds of details must it contain?
In a population of N elements, how do you compute each element’s selection probability in simple random sampling?

Key Points

1
Sampling is necessary because studying every population element is usually impractical.
2
Probability sampling differs from non-probability sampling by whether selection chances can be specified for each element.
3
Simple random sampling requires equal and known selection probabilities for every population element.
4
A complete sampling frame (full list of eligible elements with identifiers) is required before random selection can be performed.
5
Random-number generation mapped to the sampling frame is used to select the sample without researcher bias.
6
For studies with expected nonresponse, researchers can oversample (e.g., select 600 to target 300 usable responses at ~50% response rate) while keeping the selection random.

Highlights

Simple random sampling hinges on equal and known selection chances for every element in the population.

Calling a selection “random” isn’t enough—researchers must avoid convenience picking and instead use random-number selection from a full list.

If there are 10 elements, each has a 1/10 (10%) chance of selection; with 5 elements, each has a 1/5 (20%) chance.

In higher-education research, the same method works: build a complete staff list, then randomly select respondents and oversample to manage nonresponse.