Is there any way of figuring out the minimum sample required in the pilot study which will in turn help in finalizing the final sample for the study?
According to Connelly (2008), extant literature suggests that a pilot study sample should be 10% of the sample projected for the larger parent study. However, Hertzog (2008) cautions that this s not a simple or straight forward issue to resolve because these types of studies are influenced by many factors. Nevertheless, Isaac and Michael (1995) suggested 10 – 30 participants; Hill (1998) suggested 10 to 30 participants for pilots in survey research; Julious (2005) in the medical field, and van Belle (2002) suggested 12; Treece and Treece (1982) suggested 10% of the project sample size. I would say that 10 would be a minimum, and 30 might be considered in your project sample size is expected to be 300.
Refs.
Connelly, L. M. (2008). Pilot studies. Medsurg Nursing, 17(6), 411-2.
Hertzog, M.A. (2008). Considerations in determining sample size for pilot studies. Research in Nursing & Health, 31,180-191.
Hill, R. (1998). What sample size is “enough” in internet survey research? Interpersonal Computing and Technology: An Electronic Journal for the 21st Century, 6(3-4).
Isaac, S., & Michael, W. B. (1995). Handbook in research and evaluation. San Diego, CA: Educational and Industrial Testing Services.
Julious, S. A. (2005). Sample size of 12 per group rule of thumb for a pilot study. Pharmaceutical Statistics, 4, 287-291.
Treece, E. W., & Treece, J. W. (1982). Elements of research in nursing (3rd ed.). St. Louis, MO:Mosby.
van Belle, G. (2002). Statistical rules of thumb. New York: John Wiley.
Hi Manu,
I find Lehana Thebane's HRM-733 Class Notes to be very helpful when approaching sample size calculations:
https://fammedmcmaster.ca/research/files/sample-size-calculations
That's simple: chose n as large as possible so that the experiment is just feasable (money, time, effort). A (serious) sample size calculation can be done when you already *have* the information from such a pilot study. The pilot study is a pilot study - using whatever is available.
According to Connelly (2008), extant literature suggests that a pilot study sample should be 10% of the sample projected for the larger parent study. However, Hertzog (2008) cautions that this s not a simple or straight forward issue to resolve because these types of studies are influenced by many factors. Nevertheless, Isaac and Michael (1995) suggested 10 – 30 participants; Hill (1998) suggested 10 to 30 participants for pilots in survey research; Julious (2005) in the medical field, and van Belle (2002) suggested 12; Treece and Treece (1982) suggested 10% of the project sample size. I would say that 10 would be a minimum, and 30 might be considered in your project sample size is expected to be 300.
Refs.
Connelly, L. M. (2008). Pilot studies. Medsurg Nursing, 17(6), 411-2.
Hertzog, M.A. (2008). Considerations in determining sample size for pilot studies. Research in Nursing & Health, 31,180-191.
Hill, R. (1998). What sample size is “enough” in internet survey research? Interpersonal Computing and Technology: An Electronic Journal for the 21st Century, 6(3-4).
Isaac, S., & Michael, W. B. (1995). Handbook in research and evaluation. San Diego, CA: Educational and Industrial Testing Services.
Julious, S. A. (2005). Sample size of 12 per group rule of thumb for a pilot study. Pharmaceutical Statistics, 4, 287-291.
Treece, E. W., & Treece, J. W. (1982). Elements of research in nursing (3rd ed.). St. Louis, MO:Mosby.
van Belle, G. (2002). Statistical rules of thumb. New York: John Wiley.
I have to admit, I don't get it. When the size of the "parent" study is known, estimable, projected, - then it must be known what effects to expect, what models are resonable, what difficulties to solve, what can be lost with wrong conclusions and what is won by correct conclusions, and, eventually, how large a sample must be to answer the question(s) keeping the desired cost/benefit ratio (via the error-rates).
If all this is is true (i.e., the size of the final study is or can be projected) - why on earth should one do a pilot study?
It might be that the data doesn't turn out to be as useful as hoped anyway, but then an exit/stop condition for the study can be defined. The study then is not a "pilot study", it is the real "parent study", but possibly stoped (e.g. for ethical reasons) before the full sample size is acquired.
A pilot study is what one would call a run through or dress rehearsal of the actual study. A pilot is a good way to troubleshoot any equipment problems, familiarize the team with the procedures, and too generally see if the experiment design has any potential flaws.
I agree, plus that it generates data to get a first impression of the variability and effect sizes, what can later on be used to plan the actual study (i.e. judge benefits and losses based on expected effect sizes, and, based on this, define an adequate alpha and beta, and, eventually, determine the required sample size).
But as you said, a pilot study can also be used just to see if a further inverstigation is feasable at all, or simply to train people before the "hot data" is generated (a "burn-in", so to say). The *available* sample size would be the sample size of the actual study anyway, independent of the variance and effects determined in the pilot study, and without controlling error rates in the actual study. The size of such a pilot study may be quite low, just enough to see if there might be problems and to get a little confident with the method/protocol (depending on the skills of the people), so n=1 or 2 might be enough.
Hi, Jochen -
Determining effect size is not the main purpose of conducting pilot studies. Due to the fact that pilot studies allow for the opportunities to address so many areas of a study's design (and peripheral issues also), I do not think that a sample size of 1 or two would be efficient. There is a whole range of other issues that they address.
Since these are smaller versions of an intended study, they help to refine the methodologies for larger studies; therefore, the setting, participants, data collection and analysis methods mirror those of a larger study. These studies are a must for large scale studies as they not only provide an opportunity to identify potential problems with the design of the study, they also justify the feasibility of the study. Another important purpose of pilot studies is that they speak to the competence of the researcher within that area of study. It is recommended that all large-scale studies have either pilot work or other preliminary work as evidence of feasibility of the project and to demonstrate the competence of the investigator with the area of study.
Feasibility issues that might be addressed in a pilot study: the availability of subjects, estimating the recruitment time of subjects, how the investigation is conducted, and the cost of the study. Pilot studies are of crucial importance when planning studies with sensitive or not easily accessed populations; they are opportunities to develop or refine sampling methods and to evaluate the representativeness of a sample, and are also designed to test new instrument.
Musil (2011) captures the range and extent of the importance of these types of studies:
"Pilot work can be used to develop, to test, or to refine a study protocol, including the treatment or intervention to be used in an experimental or quasi-experimental study. Sufficient pilot work is necessary to support the efficacy of an intervention prior to proposal submission for a large-scale intervention study. During a pilot study, extraneous variables that had not been considered in the design may become apparent, and methods to control for them can be introduced when the larger study is designed.
Pilot work also allows the development or refinement of data collection instruments, including questionnaires and equipment. The performance of instruments with a particular sample under specific conditions also can be evaluated in the pilot project. When collecting quantitative data, the reliability and validity of instruments and the ease of operation and administration can be evaluated prior to data collection in a large-scale study. This is an important step whether the data collection instruments are interview schedules, questionnaires, computers databases, or equipment to gather biophysical data. For example, during pilot work, questionnaires can be evaluated for clarity of instructions, wording of questions, reading level, and time required for completion. For qualitative studies, pilot work may be important for gaining experience in interacting with the sample and with aspects of data collection, coding, and analysis.
The results of a pilot study are likely to be significant for the larger proposed study. If the pilot study is of sufficient size, estimates about the relationships between variables and of effect sizes can be made. This is essential not only for statistical power analysis but for a better understanding of the phenomena under study. Pilot studies often provide important insights into the problem being investigated and may lead to reconceptualization of the problem or refinement of the research questions." (para. 3 – 5)
I think that following the size recommendations in the extant literature (which I posted earlier) will enhance a researcher’s credibility as all research design decisions should be grounded in the literature.
Ref:
Musil, C.(2011). Pilot study. In Encyclopedia of nursing research.
Ruth, thank you for this comprehensive post. I agree. But still: to my opinion, any "attempt", "trial" or "experiment", as small as it may be, can be considered a pilot study. Maybe it is a wording problem. Maybe I should not call such early-attempt-trials with n close to 1 "(pilot-)studies". Maybe I should better call it simply "experiments" or "trials", something like "pre-pilot-studies".
For instance, if I aim to measure the concentration of a cellular protein in cell-culture system, I start with extracting some of the traget cells and to take them in culture. If I see that this is not working (I don't get the right cells or they don't grow under the selected conditions), I consider this already as pilot study (with possibly n=1 or 2), efficient enough to recoginze that my isolation of cell culture protocol does not work as I thought. That was what I meant. Surely, nothing to be published. For the reasons you cited, larger pilot studies may well be worth being published.
Oh - I see what you mean!
Yes, the language of research can be very exacting, hence the dependence on conceptualizations based on the literature. Yes - I can see that in some circumstances your observations would be applicable. Qualitative studies tend to use small samples - e.g. in case studies a sample might consist of 1. So that would make sense, especially when testing an open-ended questionnaire for comprehension of the wording, for example. Now you have piqued my curiosity further, so I will have to investigate this aspect a little more :)
Hello every one ... Thanks for your elaboration on the doubt. I would like to place an article on pilot studies which i hope would be helpful ....kindly check attachment ...
Hi Manu,
there are some algorithms to calulate the appropriate sample size depending on the alpha- und beta-error and the estimated effect size. The computation of the "right" sample size can be done easily with the program gpower3. More information about the program and a download opportunity are available at
http://www.gpower.hhu.de/
In my opionon one of the most helpful computer programs.
KR from cologne, Christian
Thank you all for your insightful contributions. It's an issue I am currently dealing with; to pilot or not to pilot.
I dont think there is a set rule in how to fix a pilot study. I have seen suggestions for conducting the pilot survey and actual survey in different locations so that the pilot survey respondents dont influence the actual survey responses. Statistically speaking, in social sciences, 10% of the household size is considered as a good sample size for actual survey.
Sample size is mainly determined by the estimated varience in the population. So, the thumbrule for me is to chose a size that can give a good estimate of the varience for my actual survey. Sample size shouldnt be a matter for the objectives of checking the design flow of the questionnaire, potential for mis-interpretation etc. Just my opinion.
It depends on your research method. If you are using econometric model, (although there is no clear-cut rule), pilot study using at least 100 sample is recommended.
You may use this as reference
https://www.researchgate.net/publication/260082130_R_D_Management_in_State_Universities_and_Colleges_in_the_Philippines_Sampling_in_Business_and_Management_Research?ev=prf_pub
Data R& D Management in State Universities and Colleges in the Ph...
Or you may use these quick solutions
https://www.researchgate.net/publication/262564897_Easy-to-use_Sampling_Formulae?ev=prf_pub
Data Easy-to-use Sampling Formulae
My clients typically have an upper bound on the sample size that they want to consider. I usually compute the sample size required to have a specified power for a testing problem, such as 80%. Sometimes we find that the required sample size is smaller than their upper bound and we use the required sample size; sometimes we find that the required sample size is not much larger than their "upper limit" and they will use the required sample size. Most often, the required sample size exceeds their upper bound and I will show them the consequences of their choice, such as the power of the test or the width of the confidence interval for an estimation problem,
Hi All,
And given folks are providing links to power calculators - here's another that can be very helpful at times, especially for those that don't deal with this stuff on a daily basis:
http://www.sealedenvelope.com/power/continuous-superiority/
You select the appropriate power calculator based on the purpose of your trial (e.g., superiority, equivalence) and the type of outcome (e.g., binary, continuous). Just this act alone will be educational. Also note the nifty validation of their work.
Sach
It is not just the matter of mechanical calculation of sample size because variety of calculators are available. One should understand the philosophy behind it. The philosophy makes ideas clear and hence why the sample size is to be calculated is necessary to understand.
Hi
Classically a pilot study enrols 30 patients per group. The results from this pilot study allow to get data (i.e. mean and SD of the parameter, crude estimate of the expected difference....) to build a second study using these parameters to calculate sample size.
Luca: "the answer depends on what the purpose of your pilot is. You should further specify the research question (of the pilot) to have an appropriate answer. For some question the "minimum sample size/power" kind of question is not relevant at all." I am signing under this statement with both hands.
You can consider using Epi-info or Open Epi to calculate such sample size
You might consider powering the pilot trial based on the postulated size of the main trial. See this study of ours.
Article Sample size calculations for pilot randomized trials: A conf...
I provided an answer to a similar question about the appropriate sample size for a pilot study in an earlier thread. Responses were also provided by other contributors. See the responses at https://www.researchgate.net/post/How_do_you_determine_the_appropriate_number_of_participants_for_a_pilot_study
the pilot study usually dont consider the sample size as a major issue because the purpose of the pilot is figure out the troubles that the researcher may encounter in the real study. such as the calrity of the tool, the communication with the participants, accessability...etc.
however, it is recommended to have the pilot to be 5% to 15% of the large study. depends on the sample size and availability of the participants.
the pilot study usually dont consider the sample size as a major issue because the purpose of the pilot is figure out the troubles that the researcher may encounter in the real study. such as the calrity of the tool, the communication with the participants, accessability...etc.
however, it is recommended to have the pilot to be 5% to 15% of the large study. depends on the sample size and availability of the participants.
I agree with Muayyad Ahmad. As the authors of the article linked below put it, "A pilot sample size is instead based on the pragmatics of recruitment and the necessities for examining feasibility." HTH.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3081994/
With cancer chenotherapy old tradition spoke of phase I studies where you take a drug from what is tolerated in animals to a frist level of drug with an escalaition scheme for analyzing effect. Positive and negative. If the phase I studies showed sign of effect and the drug was tolerated it was taken into a phase two trail so see how often a certain dose would have effect. So for phase I studies you wont know from start how many steps would be needed. For phase II studies you have no idea of the sample size. For a study with no response the sample size will be smaller than that for a drug showing effect. So from old past Ibe believe you need to use another category than pilot study. We have a long tradition for calculating sample size for phase III studies but these trials are not pilot studies.
Ruth is spot on - and supported by some literature. Some of those studies I perused - seems like good advice. I'd say your 30 is a fair number. G*Power is helpful in determining the balance between sample size and power. I can't condone its use in pilot studies as it will likely call for sample sizes more similar to those that you would use in the actual study. Good luck with your research.
I am also conducting a pilot study. I followed the steps based on literature guidance
a) For pilot study, I have identified subgroups in my target segment. As idea about number of sub groups helps in identifying suitable sample size.
b) Now, some researchers said that 12 participants per group (Julious, 2005) would be good for pilot study; whereas, some other literature suggest 10-50 sample per group (Hertzog, 2008).
c) I have identified 4 sub groups representing my target population. Hence, I have to take 4*12= 48 (about 50) sample for pilot test.
I have also found many extant literature suggested sample size for pilot study could be 10% of total sample size of main study. I have planned to conduct main study based on 400+ samples. In that case, sample size could be 40+.
But I feel sample size based on points (a) and (b) could be used for pilot study.
Regards,
Gobinda
if the sample size is 350, is 19 participants appropriate for a pilot??
kindly justify with any reference if you have that 20 is better or 30 .....
In our paper (Cocks and Torgerson - Sample size calculations for pilot randomized trials: a confidence interval approach J Clin Epi 2012) we reviewed several previous papers on the topic and the consensus seemed to be 20 as the previous person said. But 30 would produce more stable estimates of the variance - so if possible I would go for 30 but have 20 as the minimum
Literature search produced the minimum figure of 10 and maximum of 30.Its better to take 30 as this figure is enough to check that your questions are appropriate and easily understandable and does the question measures what it intend to measure.
It depends on what do you want to know in your pilot study. I agree with dr Torgerson reply that you can use 30 if you base what you want to do in normal law (statistically), but this could be a very low number if you want to compare two drugs or too high if you want to test a new surgical procedure for untreatable pain. Furthermore, if you are interested in comparing results using subgroups (males vs females; old vs Young patients) probably 30 patients would be a very low number. A further question refers to the available knowledge. If there are many studies on the same topic your simple size have to be at least greater than the average if you want to publish your results
Open epi tool is quit good, handy, robust and easy to use with different study designs.
http://www.openepi.com/SampleSize/SSCohort.htm
I think it is much better to make a pilot study before embarking on the proper study. The pilot study is a small-scale study carried out on a small number of individuals under conditions similar to those of the final study. The size of the pilot study is a matter of convenience, time and money. The purposes of the pilot study are to have an idea about:
a. The time required in the study as a whole.
b. The cost of the final study
c. The skills required by the investigators and whether the instructions given to them are efficient and adequate.
d. The adequacy of the questionnaire, the ease of handling, the efficiency of its layout, the clarity of the definitions and the adequacy of the questions themselves. Any problems with the questionnaire contents
e. The extent of non-response
The sample size in pilot study is not a major issue except in two situations. The first if it involves using expensive technique and the second if certain parameters are required to be estimated for the purpose of calculating the final sample of the proper study.
Manu -
Pilot studies may have multiple purposes, but when used to determine sample size needs, the important thing is that methods for sample size requirement estimation are all concerned with standard errors, which are determined by population standard deviations (which are constant for a given population or stratum/category within one), and the sample size being sought. Sample designs (generally randomized) vary, and complicate this at times (see a book such as Cochran (1977), Sampling Techniques), and in the case of "prediction" (regression modeling, not necessarily forecasting), the standard deviation you seek is for estimated residuals or the estimated random factors of estimated residuals in weighted least squares regression). But the point is that your pilot study is used to estimate the population standard deviation (or for estimated residuals), or standard deviations by subpopulation, and ... assuming no bias, which is an important assumption which needs attention, but is unfortunately likely often ignored ... that standard deviation (or those standard deviations) estimated from a pilot study, or "guessed" from other info, are used in estimating sample size needs for the final study, even if you are looking at type II error probabilities for an hypothesis test.
So, how big does the pilot study sample need to be to get a good idea of a population standard deviation? That depends upon the population(s) itself.
Suggestion: When estimating a population or subpopulation standard deviation, if you can start with just a few observations, estimate standard deviation, and add a few more observations and re-estimate, etc. You could see when your estimate of such a fixed standard deviation becomes stable. This is not always going to work extremely well, especially if the population is oddly shaped, but it will probably give you better results than are commonly used.
So standard deviation estimates may be obtained from a pilot study, by incrementally increasing sample size until you get an estimate for standard deviation which seems stable enough. You then use the estimated population standard deviation (or subpopulation standard deviation estimates), to estimate what sample size is needed to likely obtain appropriate standard error(s) in the final study.
The sample size needed for the pilot study may be found quickly enough from the incremental approach above, considering this is all what should be considered "rough" estimation, as there are various places for things to change in the course of your study. Your design may not even lend itself well to incremental sampling, but you need to do something to "guesstimate" standard deviations for the "formulas" for your final sample size requirement(s).
Best wishes - Jim
Everybody is posting some theoretical facts and talking about statistics. Every day studies are made with 20-30 subjects and are accepted for publication meaning how practise is something different entirely.
I want to examine the specific effects of aerobic exercise on asthma patients.I have only 6 subjects and will mark the study as Pilot study.
Most of the scientific studies are conducted on a subset (sample) instead of whole population, but the inference is drawn about the population. To draw the inference from a sample about the population, theory of probability is applied, but it requires a random sample from the population. A small sample in size reduces the cost of investigation but did not yield the valid inference. Similarly a large sample size though, provides valid inference but increases the cost of investigation. Therefore, a sample adequate in size is to be determined so as a valid inference may be drawn with minimum cost of investigation. In view of the above arguments, a sample should follow the following criteria;
1. Sample should be adequate in size
2.Sample should be a true representative of the population
How to get a representative sample is the content of sampling technique. Here our concern is only of sample size determination.
Sample size determination is based on the following three components:
a. Probability of type I error, denoted by “a” known as level significance
b. Probability of type II error, denoted by beta, where 1-beta is known as power of the test.
c. Minimum allowable difference, denoted by ∆
Now what should be the sample size of your study, totally depends on your objective, your study design, non response rate and your sampling techniques used in the study.
I hope it helps.
Thanks
Pilot study is a small sized study which may be used before larger scale study of any type. it aims at testing procedures, steps, methodology ,,,, before begining the larger study to detect if its feasibility so you can detect obstacles remove difficulties and modify steps or protocol of study in order to save time, efforts and funding beside conducting a well designed study. it may be used in a future studyto calculate sample size although this has been argued
Pilot study usualy precede new intervention to detect its effect. may also be used before conducting exprimental studies
Sample size of pilot study is determined by patient flow, budget, time frame,,,,etc and doesn't need to be based on previous data so any size may be used based on these items
Data provided by pilot study is excluded from the following larger study as it is used to test methodology and steps of the study to modify protocol as needed so any difference in protocol even if minor may change the effect of the tested question.
Indeed Results of pilot studies can be published separetly and sure you will refer that this is a pilot study. some journals does not welcome pilot studies.
When publishing the final study you may refer that it is based on the pilot study wether it has been published or not
Check this article for more information
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3081994/
Good answers but I think that coefficient of variation must be considered too
cheers
C
There is no hard and fast rule for determining the sample size for pilot testing. The sample size varies for type of study (qualitative vs quantitative), type of data collection method (survey, interview, focus group) adopted and minimum requirement to apply statistical test on pilot data must be considered to determine the sample size for pilot testing. The level of confidence the researcher has about which the pilot test option is adopted will also key to determine the sample size.
Sample size to pilot study can be 5% of total sample size of the actual study in quantitative study.
And what is your goal? Why you are conducting pilot study? To test the measurement instrument? Conduct an exploratory study prior to main study? Is it qualitative or quantitative (the second one I suppose?). In quant 5 -10% should be enough but it is not only the percentage, you need to place it in the context of your study.
My opinion
sample size to pilot study can be 5% of the total sample size of the actual study in the quantitative study. sometimes the level of confidence the researchers has about which the pilot test option is adopted will also key to determine the sample size of the population.
I have a personal experience as to pilot study. In the 1970ties I got more than 500 blood samplesto study LDH isoenzymes as tumor markers for testicular germ cell tumors. So I asked the laboratory in my university hospital whether they would analyse the samples. The administratice chief said we start to measure in a pilot study five samples. Then we decde whether to go on.
Well I had the obligatin to analyse the 500 samples so I never persued the trial plot study.
n fact anaysing the __LDH isoenzyme pattern of 500 patients later gave an amazing positive result. the p value for detection ofraised LDH-1 for a patient with disease versu patient with no disease was impressive. p < 10-24.
That as reflection of valueof pilot study versus full size study.
Pilot study, i think 30% of projected sample size is sufficient but it should not be less than 30 in number. However for large projects it is better to carry out primillary study before starting large scale projects.
30 minimum that you can go for.
reference :
Amy L Whitehead,1 Steven A Julious,1 Cindy L Cooper2 and Michael J Campbell1. Estimating the sample size for a pilot randomised trial to minimise the overall trial sample size for the external pilot and main trial for a continuous outcome variable. Statistical Methods in Medical Research 2016, Vol. 25(3) 1057 –1073.
There's a lot of factors one needs to consider to determine the optimal sample size. Some things to consider are the size of the population, margin of error, alpha and beta error tolerance, effect size, etc.
There is a good explanation here: http://www.columbia.edu/~mvp19/RMC/M6/M6.doc
Article A simple formula for the calculation of sample size in pilot studies
Article Estimating the sample size for a pilot randomised trial to m...
If you, at least partly, want to assess the factor structure of a scale that you'll be using, the size of the pilot study sample should be related to the number of participants needed to conduct valid factor analyses.
One rule of thumb has been that researchers need 10 times the number of participants as they have items on a scale, but there are other criteria, some less demanding, that might need to be met, depending on the nature of the data. However, unfortunately the nature of the data can usually not be determined before the data have been collected and analysed - so rules of thumb might be the best thing to work with.
There are many calculations for the sample size of pilot study, yet according to Hill, it is suggested from 10-30 participants. Pilot study is a mini experimental study to develop the related research.
What is the purpose of conducting pilot test? Usually, for checking the reliability of the instrument for data collection (e.g. questionnaire). Thus, the sample size for pilot test need not to be that large. 10-15 respondents is recommended.
Whitehead et al (2015) - for a main trial designed with 90% power and two-sided 5% significance, we recommend pilot trial sample sizes per treatment arm of 75, 25, 15 and 10 for standardised effect sizes that are extra small (≤0.1), small (0.2), medium (0.5) or large (0.8), respectively.
This editorial from the Journal of Neuroscience can provide some important input for this discussion:
*Consideration of Sample Size in Neuroscience Studies*
Journal of Neuroscience 20 May 2020, 40 (21) 4076-4077; DOI: https://doi.org/10.1523/JNEUROSCI.0866-20.2020
It suggests the differentiation of 2 phases in research, and «how sample size planning depends» on the goal of each phase: one aimed to «determine the presence of an effect» and the other «to obtain accurate estimates of the effect.»
In my opinion, it should be a subsample of the global sample, since the context where the problem occurs or develops is the same for both the subsample and the sample where the study is to be carried out. In addition, the method of sample selection should be the same for the subsample and for the sample itself.
In many areas yolu will come along by analyzing previous literature to define the characteristics you want to estimate in a subsample.
I have been wondering what to do when researchers ask questions already discussed here. Would a whisperer suggesting there’s a similar topic already help? Would some research phase related caregories?
A minimum of 30 samples that converts to 30 data points are required to plot a decent data distribution curve.