I am doing a research by using convenience sampling, do I need to calculate the sample size of my research with G power calculator?and what is that?

#1
Hi! my friends, I am collecting the samples with convenience sampling method, which means my study using non-probability sampling samples . A friend of mine told me even though I am doing a non- probability sampling research, I still need to calculate the sample size of my study with G- power calculator. is that true? and what is G-power calculator ? how can I know the sample size of a non-probability sampling research with g power?
 

katxt

Active Member
#2
If all your study hopes to achieve is a summary of the data you have collected, then do what you like and explain what you have done.
If you are wanting to use your data to infer something about a larger group, then, (ignoring the all the problems with convenience sampling as opposed to random sampling), the whole idea of a power/sample size analysis is to avoid a "not significant" conclusion. "Not significant" does not mean there is no difference. It is really just a face saving way of saying "After all the work I've done and all resources I've used, I still don't know if there is a difference or not" which is not a good admission in a study report.
If you are sure that your study will be valid with convenience sampling, and you want to make inferences about populations, and you want to be reasonably sure that your work and resources aren't wasted, then you should do a power/sample size analysis to see if your study is feasible.
 
#3
If all your study hopes to achieve is a summary of the data you have collected, then do what you like and explain what you have done.
If you are wanting to use your data to infer something about a larger group, then, (ignoring the all the problems with convenience sampling as opposed to random sampling), the whole idea of a power/sample size analysis is to avoid a "not significant" conclusion. "Not significant" does not mean there is no difference. It is really just a face saving way of saying "After all the work I've done and all resources I've used, I still don't know if there is a difference or not" which is not a good admission in a study report.
If you are sure that your study will be valid with convenience sampling, and you want to make inferences about populations, and you want to be reasonably sure that your work and resources aren't wasted, then you should do a power/sample size analysis to see if your study is feasible.
Many thanks for your explanation. I see, that means G power is just a way to prove if my variables are statistically valid or not , right? it has nothing to do with my population, right ? but how about probability sampling, do I need to predict if my variables are valid or not in a probability sampling research?
 

katxt

Active Member
#4
No. G power is a program which calculates how much data you need to be reasonably sure of getting a "significant" result. It has nothing to do with validity, whatever you mean by that. I imagine that *validity" is to do with conforming to the assumptions underlying your chosen analysis.
 

Dason

Ambassador to the humans
#5
Keep in mind that any sample size calculation needs to be done *before* collecting data. But I honestly don't think it matters at all with convenience sampling because the assumptions underpinning any sample size calculation are almost surely not met.
 

katxt

Active Member
#6
I agree. Dason is very likely right in your situation.
However, it may possibly depend on how you plan your study. Many (most) studies have some form of convenience sampling because it is usually impractical to ensure that every member of your population has an equal chance of being selected and you are limited to what is available. A good design can overcome this problem with randomization, controls, replication, and other statistical techniques.
 
#7
I agree. Dason is very likely right in your situation.
However, it may possibly depend on how you plan your study. Many (most) studies have some form of convenience sampling because it is usually impractical to ensure that every member of your population has an equal chance of being selected and you are limited to what is available. A good design can overcome this problem with randomization, controls, replication, and other statistical techniques.
I know what you meant. Many many thanks !!
 
#8
I agree. Dason is very likely right in your situation.
However, it may possibly depend on how you plan your study. Many (most) studies have some form of convenience sampling because it is usually impractical to ensure that every member of your population has an equal chance of being selected and you are limited to what is available. A good design can overcome this problem with randomization, controls, replication, and other statistical techniques.
many many thanks. if you know G-power analysis, do you have any idea of my second question over here:http://www.talkstats.com/threads/co...ample-size-of-my-study-through-g-power.77445/