I am translating a standard questionnaire into a local language. What size of sample should I take for checking the reliability and validity? Is it required 10 times of the questions?
be careful with the size of the sample, not just a matter of numbers, but also, since you get the sample groups.
Theoretically you should consider at least 7 individuals for each item, but you look like you will select the sample according to the type of population. If you intergrar a larger number of participants is better. But beware intending sample.
Simple Random sampling, stratified random sampling, cluster random sampling, for example.
Thanks,... Here i am applying the tool into a specific deceased group.so i can't use this type of sampling technics.theortically means any cut of rules for the sample size.?...
Im agree with Carlos Jimenez-Gallardo. You should consider at least 300 individuals for more than 10 items, but you look like you will select the sample according to the type of population. If you intergrate a larger number of participants is better. But beware intending sample.
Article Rapid assessment of tinnitus-related psychological distress ...
when we have problems of sample size, what I do is take networks, digitize the test and sent it to colleagues, so they can make measurements that contribute to the validity and reliability of the questionnaire. They send me the information and I perform factor analyzes.
You should consider spatiotemporal history of the study unit.
Tinnitus-Fragebogen (TF): Standardinstrument zur Graduierung des Tinnitus-Schweregrades; Ergebnisse einer Multicenterstudie mit dem Tinnitus-Fragebogen (TF)
What do you mean by "validating" ? You cannot validate any measure in our single study. There are many forms of validity (internal, external, convergent, concurrent, predictive, etc.). If you're talking about structure (and plan to perform an exploratory factor analysis or a confirmatory factor analysis), the number of observations is a function of the number of factors you want to extract, and the number of items in the test. The rule of thumb of 7 or 10 observations per item is not grounded on any scientific basis.
I am a little bit confused about the question you are asking. Are you asking about how to create a valid sample for your study or how many people you need to ensure that your translation is valid. The way the question has been written is a little bit confusing.
There's just no answer to your question except 'how many as you can' !
Look at the formula of Cronbach's alpha; it includes the number of items in the scale. Now the covariance matrix needs to be accurate enough so you need a lot of people to get small enough a standard error!
Once again, validating doesn't mean at all reporting any reliability coefficient. The first step is to develop a reliable translation (forth and back), test the scale on a small sample of individuals for understanding (n=30 would be fine). Then, collect data on a larger independent sample along with concurrent, convergent or divergent measures with regard to our hypotheses. Results may depend on the sampling: age range, sex ratio, occupation, education, etc.
Michael, that advice should be considered as minimum (although they mentioned about 5 per item with good factor structure). If you are validating a new measure, you want to go a lot higher than that.