“How Many Respondents Do I Need?” – A Practical, Evidence‑Based Answer for Research Clients
“How many respondents do I need?” is one of the most common questions clients ask at the start of a research project-and one of the easiest to oversimplify.
While it’s tempting to look for a universal rule or standard number, research methodology literature is clear: there is no single correct sample size for all studies. Instead, sample size should be determined by the purpose of the research, the level of precision required, and how the results will be used (Cochran, 1977; Krejcie & Morgan, 1970).
At Veridata Insights, we help clients move from fixed numbers to fit‑for‑purpose sample sizes-balancing statistical rigor with real‑world constraints.
Why There Is No Universal Answer
Sample size is not a standalone decision. According to Cochran, determining an appropriate sample requires considering both statistical requirements (such as variability and confidence) and practical constraints (Cochran, 1977).
Key factors include:
- The research objective
- The desired confidence level and margin of error
- The structure of the population
- The intended analysis (overall trends vs. subgroups)
- Budget, time, and operational feasibility
A study designed to generate exploratory insights does not require the same sample size as one informing high‑stakes commercial or policy decisions.
Start With the Research Question, Not the Number
Methodologists consistently emphasize that sample size should be driven by research intent, not convention. The first question should always be:
What decision will this data support?
For example:
- Exploratory research aims to identify patterns or generate hypotheses
- Confirmatory research aims to estimate magnitude or test differences
- Tracking research aims to detect change over time
Research at different stages requires different levels of statistical confidence-and therefore different sample sizes.
What Academic Research Tells Us About Sample Size
Confidence Levels and Margins of Error
In quantitative survey research, sample size is commonly calculated to achieve a desired confidence level and margin of error. Krejcie and Morgan (1970) demonstrated that, for large populations, a sample of approximately 384–385 respondents is sufficient to estimate population proportions with a 95% confidence level and ±5% margin of error.
This finding underpins many standard sample size tables still used today. However, these estimates assume:
- Random sampling
- A single, overall estimate
- No segmentation
Once additional requirements are introduced, larger samples become necessary.
Segmentation Changes Everything
One of the most common misunderstandings clients encounter is assuming that total sample size applies equally to all cuts of the data.
If results are needed by:
- Region
- Customer type
- Industry
- Demographic group
then each subgroup must meet minimum sample requirements. A total sample of 400 may be statistically sound overall but insufficient when split into multiple segments.
From a statistical perspective, subgroup analysis effectively creates multiple smaller studies within one dataset-each with its own precision requirements (Cochran, 1977).
Different Methods, Different Expectations
Sample size norms also vary by research method.
Qualitative Research
Qualitative research prioritizes depth and meaning, not statistical inference. Academic literature emphasizes the concept of data saturation-the point at which no new themes emerge-as the guiding principle rather than numerical thresholds.
In practice:
- 15–30 in‑depth interviews often provide sufficient insight for focused qualitative studies
- The quality and relevance of participants matters more than volume
Quantitative Research
Quantitative studies rely on numerical evidence and statistical testing. Sample sizes must be large enough to:
- Detect meaningful differences
- Reduce sampling error
- Support generalization to a wider population
Smaller differences or more complex analyses require larger samples.
Common Misconceptions About Respondent Numbers
“Bigger Is Always Better”
Academic research shows diminishing returns beyond certain sample sizes. Once acceptable precision is achieved, additional respondents may add cost without materially improving insight (Cochran, 1977).
“We Just Need a Standard Number”
Krejcie and Morgan’s work is often misused as a universal rule, rather than as a baseline under specific assumptions.
“Online Surveys Remove Sample Size Limits”
While technology reduces data‑collection friction, it does not change the statistical principles that govern inference, bias, and precision.
Practical Guidelines (With Context)
While every study should be designed individually, the following evidence‑informed ranges can help frame early conversations:
- Exploratory qualitative studies: 15–30 participants
- Directional quantitative insight: 100–200 respondents
- Statistically robust surveys: ~385+ respondents
- Segmented or high‑stakes analysis: Custom‑calculated samples
These figures align with established methodological research but must always be adapted to the specific use case.
From “How Many?” to “How Confident?”
Academic and applied research both point to the same conclusion: sample size is ultimately about managing uncertainty.
A more useful framing than “How many respondents do I need?” is:
- “How precise do these results need to be?”
- “What risk am I willing to accept if I’m wrong?”
At Veridata Insights, we work with clients to answer those questions first-then design sample sizes that are both methodologically sound and commercially practical.
There is no magic number when it comes to respondents. The right sample size depends on the research objective, analytical ambition, and decision context.
By grounding research design in established academic principles while remaining sensitive to real‑world constraints, organizations can invest confidently in research that delivers insight, not just data.
Connect with Veridata Insights today to learn more.
