Sampling Methods
How participants are SELECTED from the target population — this drives who the results apply to, not the comparison within the trial.
The strategy for choosing a sample from a target population. Probability sampling gives every member a known, non-zero chance of selection (simple random, systematic, stratified, cluster). Non-probability sampling does not (convenience, quota). Sampling is about who gets in — it mainly affects external validity (generalisability), not internal validity.
Selection runs population → frame → sample — the method sets how representative the sample is.
The funnel from a target population, through the sampling frame you can actually reach, down to the analysed sample — via either probability methods (known chance of selection) or non-probability methods (convenience, quota).
Read top-to-bottom and ask two questions: does the sampling frame really cover the target population (or is a slice missing — under-coverage), and was selection random or self-selected? Probability routes give a sample that mirrors the population; non-probability routes risk a skewed, unrepresentative sample.
Sampling decides external validity — whether the findings transfer to your ED patients. It does not fix the internal comparison: that is the job of randomisation. A daytime convenience sample can be internally valid yet ungeneralisable to night-shift or rural presentations.
Sampling = who gets in→ external validityRandomisation = which arm→ internal validity- Probability = known chance of selection; non-probability = not
A qualitative ED study explores why patients leave before being seen, recruiting a convenience sample of daytime weekday attenders at a single department. That sampling frame is fine for generating themes, but it under-covers night-shift, weekend and ambulance arrivals — the very groups most likely to leave. External-validity limit: conclusions apply to that site and time window, not to all ED attenders. Convenience sampling never makes the result wrong internally — it limits who it generalises to.
- Confusing sampling (selection from the population) with randomisation (allocation between arms).
- Treating a convenience sample as representative — it carries selection bias.
- Ignoring under-coverage: a sampling frame that misses part of the target population.
Quick check
Does the sampling method mainly affect internal or external validity?
Answer: External validity (generalisability) — it determines who the results apply to. The internal comparison is protected by randomisation, not by sampling.
Ready to build your plan? EMF Premium gives you all 40,000+ questions, 20 mocks and 1,215 OSCE stations from £29/month — or a one-off 3- or 6-month pass.