Sample size tells you how many participants you need. Sampling technique tells you how you will pick them – and that choice decides whether your results can be applied beyond the patients you actually studied. In MD and MS theses, the sampling section is often one vague line (“patients were selected randomly”) that does not match what was actually done.
This guide explains the main probability and non-probability methods, which ones suit hospital-based and community studies, how to describe your method accurately, and the errors that examiners and journal reviewers pick up on.
1Population, Sampling Frame and Sample
Three terms need to be clear before you choose a method:
- Target population – the group you want your conclusions to apply to, such as adults with type 2 diabetes in your district.
- Study (accessible) population – the part of that group you can actually reach, such as adults with type 2 diabetes attending your hospital’s medicine OPD during the study period.
- Sampling frame – a list or mechanism from which participants are drawn: a household list, an OPD register, a list of schools, or a ward admission register.
- Sample – the participants who are actually included.
Probability sampling needs a sampling frame, because every unit must have a known chance of selection. If you have no list and simply include patients as they arrive, you are using a non-probability method – which is perfectly acceptable for many theses, provided you call it by its correct name.
Sampling technique and sample size are separate decisions. Calculate the number first – see our guide on sample size calculation or use the sample size calculator – then decide how those participants will be chosen.
2Probability Sampling Methods
In probability sampling, every unit in the sampling frame has a known, non-zero chance of selection, and chance – not the investigator – decides who is included. This allows results to be generalised to the population the frame represents.
Simple random sampling
Every unit has an equal chance of selection. Number every unit in the frame and select using a random number table or computer-generated random numbers, for example with the RAND or RANDBETWEEN function in Excel. Example: selecting a set number of case files from a numbered list of all admissions in a year for a record-based study.
Systematic random sampling
Calculate the sampling interval (k) by dividing the size of the frame (N) by the sample size (n). Pick a random starting point between 1 and k, then select every kth unit. Example: selecting every kth household along a village list after a random start. It is simple to apply in the field, but can give biased results if the list has a repeating pattern that coincides with the interval.
Stratified random sampling
Divide the population into groups (strata) that differ in a characteristic relevant to the outcome – age group, sex, urban/rural residence, school type – then sample randomly within each stratum. Allocation can be equal or proportional to stratum size. This ensures that smaller subgroups are adequately represented.
Cluster sampling
Instead of individuals, randomly select whole groups (clusters) such as villages, schools or anganwadi centres, and study all or a sample of individuals within them. It is practical when no list of individuals exists, but people within a cluster tend to be similar, so cluster samples usually need a larger sample size. This is handled by applying a design effect in the calculation – discuss the value with your statistician.
Multistage sampling
Sampling in two or more stages, for example selecting blocks within a district, then villages within selected blocks, then households within selected villages. Large surveys commonly use this approach. For a thesis, keep the number of stages realistic for one investigator and describe each stage clearly.
3Non-Probability Sampling Methods
In non-probability sampling, selection does not rely on a random process. Results are harder to generalise, but these methods are often the only practical option in hospital settings.
Consecutive sampling
Every eligible patient presenting during the study period is included until the required sample size is reached. This is the most common and usually the most defensible non-probability method for hospital-based theses, because it reduces the investigator’s choice of who gets in. Example: all patients undergoing laparoscopic cholecystectomy who meet the criteria between two dates.
Convenience sampling
Participants are included because they are easy to reach – whoever is available when the investigator is free. It is quick, but the risk of selection bias is high. If you recruited only on your OPD days or only during duty hours, be honest that the sample was a convenience sample and discuss it as a limitation.
Purposive sampling
Participants are chosen deliberately because they have specific characteristics or experience relevant to the question. It is widely used in qualitative research, for example selecting nurses with ICU experience for in-depth interviews. Our post on qualitative vs quantitative research explains where this fits.
Snowball sampling
Existing participants help identify others with the same characteristic. It is useful for hard-to-reach groups, such as people with a stigmatised condition, but samples tend to be connected to each other and may not represent the wider group.
Randomised allocation in a trial is not the same as random sampling. A trial can enrol patients consecutively (non-probability sampling) and then allocate them randomly to groups. Describe both steps separately in your Methods.
4Hospital-Based vs Community Studies
| Setting | Usual sampling methods | Why |
|---|---|---|
| Prospective hospital-based study (OPD/IPD) | Consecutive sampling | No list of future patients exists; including every eligible patient limits selection bias |
| Retrospective record-based study | All eligible records, or simple/systematic random sampling from the record list | A complete list (register or database) is available as the frame |
| Comparison of two hospital groups | Consecutive sampling within each group; matching if planned | Groups recruited in parallel over the same period |
| Community survey (village, ward, district) | Systematic, stratified, cluster or multistage random sampling | Household or population lists allow probability methods |
| School-based study | Cluster sampling of schools or classes; stratified by school type if relevant | Schools and classes form natural clusters |
| Qualitative interviews or focus groups | Purposive or snowball sampling | The aim is depth and relevant experience, not statistical representation |
For retrospective designs, our guide on retrospective vs prospective studies discusses how the choice of design affects data quality.
5How to Describe Sampling in Your Methods
A reader should be able to repeat your selection exactly from your description. Cover these points:
- The name of the sampling technique, stated correctly.
- The sampling frame, if any – which register, list or database.
- For random methods: how random numbers were generated, the sampling interval and random start for systematic sampling, the strata or clusters and how they were chosen.
- The recruitment period and setting.
- How eligibility was checked and when consent was taken.
- What happened to eligible people who declined or were missed.
Template: hospital-based study
“All patients aged [range] with [condition] attending the [OPD/ward] of the Department of [department], [institution], between [start date] and [end date], who fulfilled the inclusion criteria and gave written informed consent, were enrolled consecutively until the calculated sample size of [n] was reached.”
Template: community study
“A [two/three]-stage sampling method was used. In the first stage, [number] [villages/wards] were selected from the list of [frame] by [simple random sampling using computer-generated random numbers]. In the second stage, households in each selected [village] were chosen by systematic random sampling, with a sampling interval calculated as [N/n] and a random start. One eligible [participant] per household was included; if more than one was eligible, [rule for selection].”
Your synopsis, Methods chapter, limitations and any later journal article must describe the same sampling method. If the method changed during the study, explain why in the thesis and ensure the change was approved by the IEC where required.
6Common Sampling Errors in Theses
| Error | Why it is a problem | Correct approach |
|---|---|---|
| Calling consecutive or convenience sampling “random” | No random process was used; examiners will ask how randomisation was done | Name the method you actually used |
| Writing “simple random sampling” with no frame or method | Random selection is impossible without a list | Describe the frame and how random numbers were generated |
| Ignoring clustering in a school or village survey | The sample may be too small for the precision you claimed | Apply a design effect in the sample size and account for clustering in analysis if needed |
| Confusing random sampling with randomisation | They answer different questions – who is studied vs who gets which intervention | Describe selection and allocation separately |
| Generalising hospital findings to the whole population | Hospital patients differ from people in the community | Limit conclusions to similar patients and settings |
| Not recording eligible patients who were missed or refused | Hidden selection bias | Keep a screening log and report the numbers |
Can a colleague repeat your selection from your Methods alone? Is the technique named exactly as it was done? Are your conclusions limited to the population your sample represents?
7Getting Help with Sampling
A sampling method that is correctly chosen and honestly described protects your thesis at the synopsis stage, the IEC and the viva. If you are unsure which method fits your setting, or how to write a multistage design clearly, the PubMedico team can help with your sampling plan and sample size together. Message us on WhatsApp at +91 78787 33004.
❓ Frequently Asked Questions
Quick answers to common questions on this topic
Yes. Consecutive sampling is widely used in hospital-based theses and is generally regarded as the most defensible non-probability method, because every eligible patient during the study period is included. Name it correctly in your Methods, describe the recruitment period and setting, and acknowledge in the limitations that findings apply mainly to similar hospital populations.
Random sampling decides who is selected into the study from a population, using a sampling frame and a random process. Randomisation decides which treatment or group each enrolled participant receives in a trial. A trial may use consecutive sampling for enrolment and still randomise allocation. Report the two steps separately in your Methods.
It depends on the frame available and your area. Systematic random sampling works well when a household list exists. Stratified sampling helps when subgroups such as urban and rural need representation. Cluster or multistage sampling suits large areas, schools or villages, but usually needs a larger sample size because of clustering. Discuss the choice with your guide and statistician.
You can use a printed random number table or computer-generated random numbers, for example the RAND or RANDBETWEEN functions in Microsoft Excel. Number every unit in your sampling frame first, then select the units that match the random numbers. Write in your Methods exactly which method you used, so that the selection can be checked.
Yes. The synopsis should state the sampling technique, sampling frame and recruitment procedure along with the sample size, because the ethics committee and research committee review them together. Whatever you approve in the synopsis should match what you do and later report in the thesis. Any change should be discussed with your guide and, where required, the IEC.
📍 Medical Thesis Help in Your City
PubMedico works with MD, MS, DNB, MDS, PhD and MSc Nursing scholars across India. Find thesis writing, statistical analysis and publication support near your college:
…and 38 more cities across India. See all locations on the home page →