Hello,
I am a masters civil engineering student, my study is about comparing rehabilitation techniques of Water Distribution Networks (WDN) between two countries.
My data collection method is an online questionnaire survey (literate questionnaire survey). I used LinkedIn to target my respondents.for the first country I added around 1500 person and after the filtration and visual examination I chose around (350- 400) I only got around 65 responses.
For the second country, the same process was made (i selected around 150) + contacting specific people from the WDN industry( as it's my home country). but I got around 50 responds
I was asked how to defend that number of respondents and make sure that the number of respondents is enough
Your sample size completely depends on the population size you are considering. You can check with different confidence intervals and confidence levels which value suits you and you can define the whole population.
In addition to the population size, you may have think in term of precision as the variation of sample and your targeted confience interveal impact a number of samples impact.
If your responses, for example, have small variation, you require less sample to test hypothesis at the same level of confience.
For your easy reference:
https://www.qualtrics.com/au/experience-management/research/determine-sample-size/
Abdelrahman -
This is about far more than sample size. Even "big data" can be misleading. For quantitative methods, there are two ways to infer to a population from a sample: either a probability of selection (randomized design-based) method, or "prediction" (which just means regression, not forecasting) where you need regressor/auxiliary data on the entire population. You don't appear to have either. This is not my subject matter area, but I suspect the missing information here might be substantial. Work has been done on pseudo-randomization (not my area), but I think you need a lot more information. People like Richard Valliant and Mike Brick have papers on ResearchGate, but I doubt you have the information you need to apply them. Consider the following, for example:
https://www.researchgate.net/publication/331481770_Comparing_Alternatives_for_Estimation_from_Nonprobability_Samples
by Richard Valliant, and the following by Mike Brick
http://www.statcan.gc.ca/sites/default/files/media/14252-eng.pdf
https://www.researchgate.net/publication/261967521_The_Future_of_Survey_Sampling
At any rate, you don't know if your sample is something from which you can infer to the population.
For a qualitative sample, which is not my area, I think the concern is similar: How does your sample relate to the population? A simple random sample for each category of interest might work there. You could sample on each case until you got no more surprises and variability leveled, but that does not look like what you have here.
Whatever quantitative or qualitative information you are researching here, you are not just interested in variance/variability, you have to make a case that you are not ignoring key parts of your population. Or if you are, what might be the expected impact?
Most techniques you might see might be design-based, I think, where randomization is the key. I don't think you have that. I think the best you can do is to speculate from a subject matter perspective. Consider the variability in your sample, and speculate about the impacts of factors you could not control. Lots of caveats may be needed.
Best wishes - Jim
PS - The following may be of interest to some - it is entertaining:
Brewer(2014), "Three controversies in the history of survey sampling," Survey Methodology, Dec 2013 - article Ken Brewer wrote for receiving the Waksberg Award:
https://www150.statcan.gc.ca/n1/pub/12-001-x/2013002/article/11883-eng.htm
305.95 or 306 people, is your required sample size according to formula given in attachment; 115 responses are not enough, try to have more people to respond to your questionnaire
(Formula for sample size calculation given by @Abdullah Noori on RG with this attachment for sample size calculation)
The above formula assumes simple random sampling. We already know that is rather far from true. It is also only for a proportion (i.e., a yes/no question). It does not apply to continuous data questions.
By the way, please note that the standard deviation for a proportion is estimated as SQRT((n/(n-1))(p(1-p))), where 1-p = q, so for n that is not too small, s^2 is approximately pq.
Also, remember, this isn't just about population size either. To find a suitable sample size for each question (the largest sample size needed for any key question then being the survey/questionnaire sample size needed), you need to conduct your data collection in a manner consistent with being able to infer to the population. You need a proper plan and good execution. There are problems which can be daunting. Note that if you have a large number of nonresponses, that can be very problematic.
Please see my first response above.
There is a dedicated formula for it and I can see someone given it to you
The table shown can be used in a two-sample t-test to estimate the sample sizes of an experimental group and a control group that are of equal size, that is, the total number of individuals in the trial is twice that of the number given, and the desired significance level is 0.05. The parameters used are:
The desired statistical power of the trial, shown in column to the left.
Cohen's d (= effect size), which is the expected difference between the means of the target values between the experimental group and the control group, divided by the expected standard deviation.
https://en.m.wikipedia.org/wiki/Sample_size_determination
How to Find a Sample Size Given a Confidence Interval and Width (unknown population standard deviation)
za/2: Divide the confidence interval by two, and look that area up in the z-table: .95 / 2 = 0.475. ...
E (margin of error): Divide the given width by 2. 6% / 2. ...
: use the given percentage. 41% = 0.41. ...
: subtract. from 1.
https://www.statisticshowto.com/probability-and-statistics/find-sample-size/
In continuation with my previous answer
I have calculated sample size according to formula given in attachment reference below" 305.95 or 306 people, is your required sample size according to formula given in attachment; 115 responses are not enough, try to have more people to respond to your questionnaire"(Nazia Asad, September 1st 2020)
(Determining sample size for research activities
Robert V. krejcie university of Minnesota Duluth; Daryle W. Morgan Texas A. & M. University)
(Both attachments for sample size calculation given by @Abdullah Noori on RG )
- Badges
- Science topic
- Similar topics
- Econ
- Manufacturing
- Industry