Researchers choose to conduct studies on samples rather than taking whole populations. Inferences drawn from the sample will be generalized to the population from which it is taken.
Choosing the right sample size is essential since too large may be unwanted and expensive, and too small will lead to errors and affect the Research’s effectiveness.
Let’s make this understanding painless with four simple questions.
- Is sample size Estimated?
Sample size estimation is done for the primary hypothesis/ Outcome of the Research, which is tested in the primary analysis. However, sometimes we will have more than one hypothesis, and the sample size may vary for all these hypotheses. In such cases, it is calculated for the primary outcome as discussed, and all other secondary outcomes will be tested for statistical significance in a secondary analysis.
- How large sample size should be?
Practically sample size is estimated, keeping Power as 80% (80% certain of identifying statistically significant outcome) and p-value as 0.05 (5% chance of obtaining the result as extreme as one observed). As we increase the Power beyond 80% or reduce the p-value from 0.05 the sample size will increase and vice versa. A 10 % increase in the final sample size can be done to overcome the dropouts, Incomplete records, and other research-related limitations.
- What affects the sample size in Research?
Study design (Descriptive, Analytical, Interventional, Diagnostic accuracy studies), Outcome measures (primary outcome), Power of the study (usually taken as 80%), the set p-value (0.05 % usually) will affect the final calculated sample size. So, all these should be considered before estimating the sample size.
4.How to estimate the sample size?
To estimate the sample size, we need to have either one of these: percentage or proportion, Mean, Standard deviation, sensitivity, specificity, Odds ratio, relative risk, or effect size of the primary hypothesis/ outcome being tested. These values can be taken either from the available literature or pilot studies if it is novel or inadequate or if no literature is available.
Various formulas for estimating sample size for a research study vary depending upon the study design.
Let’s understand with a straightforward example:
Eg: 1. A local health department wishes to estimate the prevalence of tuberculosis among children under five years of age in its locality. How many children should be included in the sample so that the prevalence may be estimated to 5 percentage points of the true value with 95% confidence, if it is known that the actual rate is unlikely to exceed 20%?
- Anticipated population proportion (P) -20%
- Confidence level- 95%
- Absolute precision (error, d) – 5 %
It shows that for P=0.20 and d=0.05 a sample size of 246 would be needed.
Suppose it is impractical, with respect to time and money, to study 246 children. In that case, the investigators should lower their Confidence to 90% or increase the absolute error to 10%, wherein the sample size will be reduced to 173.
Explanation:
- Here in this example, we knew the prevalence of Tuberculosis wouldn’t be more than 20%, a proportion or percentage taken as a reference to calculate the Sample size.
- It was directly taken from the existing literature if not available; a pilot study can be conducted to ascertain the same.
Take home message:
- Ideally, sample size estimation should be done at the start of the study (protocol writing stage) to overcome any limitations.
- For young researcher it may be cumbersome to calculate the sample size. It would be a wiser decision to take the help of a statistician.
- The authors should give a detailed explanation of sample size estimation while publishing your Research, increasing the manuscript’s acceptability.
- You can learn more detailed aspects from our discord channel on sample size. Join via this link https://discord.gg/ptkapQhu.
- You can now calculate your own sample size at reap coGuide, Link to access https://reap.coguide.in/protocol
“We need to be sure about getting the right answer without exposing too many or too few participants to getting the answer.”
References:
- Andrade C. Sample size and its importance in Research. Indian J Psychol Med 2020;42:102-3.
- Chander NG. Sample size estimation. J Indian Prosthodont Soc 2017;17:217-8.
3.Lwanga, Stephen Kaggwa, Lemeshow, Stanley & World Health Organization. (1991). Sample size determination in health studies : a practical manual / S. K. Lwanga and S. Lemeshow. World Health Organization.
Blog distribution by
Dr. JV Shiva Priya
Mentor coGuide