Ashish, the convenience sampling is no excuse for lack of the formula. Sorry. Some researchers suggest that sample size is linked with number of variables used in research. Also you can calculate sample size using free G Power application available online. 

Good luck

Withold Orlick -  Sir this a study which was done among truck drivers in Croatia and the sampling procedure was convenience sampling . This study does not mention the use of any sample size calculation 

In this context i am really confused 

I kindly request you to go through the highlighted portion of the article so that you can provide me useful  add on advises 

Thankyou 

Dear Ashir Pundhir,

                                     Generally the pilot study is being conducted for testing the reliability and validity of the data collection tool. Your sample selection method is convenient. In this context the pilot study sample size may be 10% of the sample size proposed for the main study. If you are selecting 50%, you are wasting your valuable golden time. If the tool fulfills the stability and internal consistency, then go ahead in your task.   

From the word itself, your explanation would be "it is the most convenient sample size that would be able to answer your problem". You have no way to cover a probability sample (due to cost, non-feasibility, threat to your life, and inaccessibility of a probability sample) and convenience sampling is the only way.

Dear Ashish,

Don't be confused about 'sample size' and 'sampling'.  This are completely differnt term in statistics but related to each other. Sampling means how to select your sample it can be SRS, SYS, Cluster Sampling, Sterified Sampling etc for probability Sampling. But when it is not possible to collect sample by strictly following any probability sampling then one should use non probability sampling methoads. convenience sampling is one of non probality sampling. But after collecting sample using non probability one should justify their sample size using power calculation.To justify sample size using power calculation we need some information related to study design, objective and characteristics of outcome variable from previous study. If it is not possible to find previous as similar study then researcher should perform pilot study. The sample size of pilot study is enough 10-20% of total sample size which you are able to collect from your study period.

sample size and sampling address two different issues: a sufficient sample size ensures you can measure whatever you want to measure in your population with sufficent precision (narrow confidence intervals). The sampling method has to do withavoiding bias. A very large convenience sample does not address bias. It is hard to advise on justifying a sample size if we don't know what you are measuring

Are we speaking about a qualitative or quatitative study? With the former we look for patterns and not for figures, in the latter we are interested in quantities and thus numbers and figures.

The sample size in a qualitative study can very well be a handfull of subjects - until you have found enough a variety of patterns.

Dear Ashish Pundhir

The study you attached (On the Road: Croatian Truck Drivers, Commercial Sex and HIV/AIDS) was name a “pilot study” by the authors, probably simply because they used convenience sampling. The main problem with convenience sampling is that you have no control on the representativeness of the results due to possible selection bias, as mention by Michael Edelstein in his post above.

To answer your question you have to let us know what the main purpose of your study is?

Is the purpose to test group difference in HIV/AIDS knowledge?

Two self-administered questionnaire scales to measure knowledge of HIV/AIDS was used in the above mention study. These scales had really bad Cronbach alpha coefficients, and therefore the variances of the scale scores are probably high. If the standard deviation of the scale score e.g. is 4 scale scores and if you would consider a scores difference of e.g. 3 scale scores as a relevant group difference, you can calculate the sample size using standard sample size calculation tools.

If on the other hand your study is of a more descriptive nature (as the above mentioned study), and you are using a convenience sampling, you should worry more about the representativeness of the study results than about the sample size.

If the purpose of your study is to validate the HIV/AIDS knowledge scales you would probably need a much larger sample (+100) to do confirmative factor analysis and item bias and item response theory analysis.

Therefore, please inform us about the purpose of your study.

Regards Kim

I appreciate the efforts of all the  researchers for the explaination - 

The aim of the thesis is to assess sexual behaviour among truck drivers in meerut,India

The objective are

1. To determine sexual partners of truck drivers

2. To determine the Knowledge and Patterns of Condom use

3. To assess awareness and knowledge of transmission and Prevention for HIV/AIDs

I have not used any scale to assess knowledge of anything only close ended questions

Further the design of my study is mixed methodology using both quantitative and qualitative method(in depth interview) and convenience sampling was done

the sample size of pilot study was 50 and main study was 100

Please kindly give your expertise opinion how i should justify why i chose sample size of 50 and for the main study 100 . I did not use any sample size formula

Please kindly give any reference or link or pdf so that i can quote in my thesis your suggestion- i mean which will corroborate your suggestion 

Thankyou

Good morning Ashish Pundhir

Generally the purpose of a pilot study is different from the purpose of a subsequent main study. Usually the purpose of a pilot study is to test some aspect of the study design, e.g. to see if the response types of your closed ended questions are working well without any ceiling of floor effects, or excessive use of “don’t know” answers etc. 

So, if testing the study design was the purpose of your pilot study, a sample size of 50 can be justified simply by saying something like the following:

The purpose of the pilot study was to test the appropriateness of the response types of the closed ended questions, and to test the in-depth interview methodology, etc. etc. A convenient sample of 50 subjects from the target population was used in the pilot study. The sample size of 50 subjects was an arbitrarily choice, but as the pilot study did not involve validation of any questionnaire scales, 50 subjects was thought to be sufficient to roughly capture the variation in the target population.

Now the main study

As your study use mixed methodology you have a build in problem; in your qualitative part you seek to explore all possible variation in your target population, and in your descriptive (epidemiologic) quantitative part your seek representativeness. This is hardly feasible in one and the same study (unless it is very big), so maybe you can use the following strategy.

If you have made, or intend to make, 100 in-depth interviews, that must be considered sufficient in any qualitative study, and fare more informative than a much larger quantitative study. Therefore, regarding the qualitative part of study you can quite rightly argue that a sample size of 100 subjects was considered to be sufficient to assess all aspects of sexual behavior among truck drivers in Meerut, India. This could probably be substantiated by the finding from your study, where you probably have found (will find) that you did not find any new aspects after you had completed the first 50 in-depth interviews or so. With respect to the quantitative part of your study, I think you could argue, that this part is merely an add-on to your qualitative part, to get some idea of the prevalence of whatever your have close ended questions about. As I said my previous post, as you study is of a descriptive nature (as opposed to a hypothesis testing study) you should be more concerned about the representativeness of your study. As you use a convenient sample, I think you need to admit this as a weakness of your study, and clearly state, that the prevalence found in your study may differ substantially from that in the full target population due to possible selection bias.

I can’t give you any reference for my views above; I think it will be hard to find any.

I wish you all the best with your study.

Regards Kim

(PS: “Kim” is a masculine name here in DK :-)

Kim Lingby - Thank you sir 

Sir what does variation imply in the aforementioned answer

Dear Ashish Pundhir

In my previous post I said "... in your qualitative part you seek to explore all possible variation in your target population, and in your descriptive (epidemiologic) quantitative part your seek representativeness."

By variation I did not mean variation in any specific statistical sense.

One typical way to sample data in a qualitative study is to use "maximal variation sampling", trying to maximize the diversity in the sample, by selecting cases that are as different from each other as possible, or keep on including cases intil you do no longer get any new information.

Regards Kim

I learnt so much from following the thread of this question for my own pilot test study planning, thank you all!

Best regards,

Paramjeet