Hi, Good Afternoon

From this below link, you would get the survey methods.

http://www.socialresearchmethods.net/kb/survsel.php

you want to find out the survey method through software, kindly see the below link:

https://www.statpac.com/surveys/sampling.htm

All the best.

for example, you might use the R package sampling

Dear Dr Vishnu

I am not sure about the software. However wish to request you to go potentially, while selecting the sampling methods from a range of methods. Once you have identified the sampling frame and decided the methods, then selection of samples will not be as typical, ya may be time consuming. 

Thanks.

Regards

thank you everybody. but soft-wares give same sample size  for different contexts objectives and aims. I am disappointing about.

Hi,

First you need a sampling frame for your entire population.

Then each member of that population is given a number, say 1 to 1000

Decide on the sample size - this will be determined by how accurate you want your findings to be and whether the findings from your sample can be generalised to the entire population.

Check on the internet - there are numerous sites that create random samples.

http://www.randomizer.org/

http://www.youtube.com/watch?v=q8fU001P2lI

From the sampling frame, choose a random sample of say 100 people.

Basically there are two types of samples; probability also called inferential and non-probability samples. With a probability sample each member of the population has an equal chance of being selected.

Please visit this site

https://www.dssresearch.com/

it can be used for calculating sample size for all types of research

Following are all good sample calculators

http://www.campbellcollaboration.org/escalc/html/EffectSizeCalculator-SMD7.php

http://www.danielsoper.com/statcalc3/calc.aspx?id=47

http://www.surveysystem.com/sscalc.htm

http://www.openepi.com/oe2.3/menu/openepimenu.htm

http://www.cct.cuhk.edu.hk/stat/

Try Gpower: http://www.gpower.hhu.de/

The sample size depends of what you are gonna do. The easiest part of a survey is to calculate the power (if it is not too complex), the problem is to set the correct sample frame, get maps etc, and plan carefully what you are goona do. This is not performed by a calculator but for a person with extremely good knowledge on sampling. ISR from Michigan is the best center in world for that. Not only you have to make the right sample , if necessary clusters, stratas and also you need to calculate weights, take into consideration missing data, calculate the weigths properply including missing so at the end you can have valid inferences, otherwise you will ended up with flawed results. 

Hi Dr Prashant,

As numerated by many researchers, there are many software can help choose sample from the survey you had conducted - some software might be accessed online from Internet, some might be from software you have installed / downloaded.  In fact common software like Microsoft Excel as well as others like R, SPSS etc. can do the job.  May be you might want to consider not only software can select sample, but rather how user friendly the software can help you capture the population in the software, then randomly / purposive select them & straight away proceed to data analysis without any conversion / copy & paste - one of them is  SPSS.  If you want to know  more how to do this in SPSS, you can refer to the following YouTube link:

http://www.youtube.com/results?search_query=how+to+select+random+sample+from+SPSS

Regards,

Fung

You are asking about sampling from a finite population, about which you wish to estimate statistics such as totals, or means, correct?

There are, as has been noted, many sample designs, and your sample size will be partly determined by that, and partly determined by innate variance, how accurate you want your answers to be, and often greatly muddled by nonsampling error, including measurement error. Generally, once you have decided on a design, and obtained some information (enough for good 'guesses') on variance, a calculator or formula for a sample size estimate will give you an estimated sample size that actually assumes no measurement error or other nonsampling error. But if you try to use a sample size that is too large for your resources, you may do much worse than expected, because of the nonsampling error.

Also note that your estimation method has to be consistent with your design. If you use regression (model-based) estimation, then the design is not so critical, though you always need to stratify properly so that the data modeled together belong together.

Note that there are probability-based methodologies (often called "design-based"), model-based methodologies, and model-assisted design-based methodologies. The advantage of modeling (which requires auxiliary/regressor data on the entire population, where perhaps administrative data will do) is that a random (probability) design can, by bad fortune, give you a sample that could be said to not be 'representative' in that estimations of population totals based on that sample alone would not be very good.

A good book, perhaps relatively easy to obtain, to consult for random design-based methods, would be Cochran, W.G.(1977), Sampling Techniques, 3rd ed., John Wiley & Sons. On pages 77 and 78 in that book, Cochran discusses estimation of sample size needs for a simple random sample. A sample size estimator analogous to this, but for a robust model-based estimator, is in the paper at the link below. Cochran has a great deal of good information in a small book, but is mostly on design-based methods. A good book on model-assisted design-based methods would be Särndal, C.-E., Swensson, B. and Wretman, J. (1992), Model Assisted Survey Sampling, Springer-Verlag. My paper, link attached, is the only method of which I am aware which estimates sample size needs from a finite population, based on a strictly model-based approach.

There is also the problem in a survey, of determining sample size needs for more than one question on a survey. This can get rather muddled too. Often for a design-based survey, one is told to concentrate on doing well for just a few key questions. The two books noted above may have something on this, but an Internet search for articles is advisable, in addition. For the paper at the link attached, an iterative approach, looking at one question at a time, is advisable. The sample for one question will impact the sample needs for another.

If you have a very simple problem, which seldom occurs, but perhaps it is sufficient for your purposes, say a simple random sample for one question, then pages 77 and 78 from the third edition of Cochran noted above, may be satisfactory. In that case, if you then need to choose the sample to go with it, any software with a random number generator might be used.

Whatever software you use, you will want some expertise to be certain it is actually doing what you want. Stratification for any sampling method is often very important, but you will need to be able to tell the software how this should be set up.

I think that SAS may have something to help in such simple sample selection, and I suppose other standard software packages do as well.

If you are doing simple random sampling and want to find something online to help select the members of the sample, you will need to look for something with a random number generator.

Note that you should also estimate variances for relative standard error estimates, or confidence intervals, for the aggregate results you report in the end.

Conference Paper Projected Variance for the Model-based Classical Ratio Estim...