I want to translate and validate a standard questionnaire. The original version contains 26 items. How many samples do I have to take for validation?...
Hi Deepak I don't think any standard sample size calculation software is going to answer your question. Sample size software is about power to detect differences, so unless you have a specific question about a difference you want to be able to detect it's not going to tell you how large a sample you need to assess your instrument. As always, one answer is, "the more the better." One very important thing, however, is that you want to be sure to get as large a range of responses as possible. In other words, you want your sample to be heterogeneous is as many ways as possible You won't get very much information if all the answers are at one end of the scale or some answers are never chosen. This is a rare case in quantitative research where random sampling isn't so important, becuse it isn't necessary to generalize back to a population. An aside, it is very important to back translate the instrument by independent translators to see if you obtained a good translation before you even try the instrument. There is no point collecting data if your translation isn't good. As for things you might do, if there is a published factor analysis, for example, you might try to replicated that. For an example, you can see a paper called "A Closer Look at the Measurement of Burnout" on my profile. It's not a translation, but we changed several aspects of the original measure and then showed we could pretty much reproduce a published analysis of the original. Doing a confirmatory factor analysis like that would require more data then a simpler item analysis or correlation analysis or even an exploratory factor analysis. Bob
"There are no general criteria for the required sample size in a validation study. A sample size of at least 50-100 participants is generally recommended. However, certain methods require larger numbers of participants" . See the following website for further information.
Taken from: http://www.emgo.nl/kc/preparation/research%20design/8%20Questionnaires%20selecting,%20translating%20and%20validating.html
You may need to ask someone much more experienced in this area than myself, but what I know, as with so much in clinical research, there are no black and white rules where it comes to assessing the reliability and validity of questionnaires, depending on the need to assess what type and level of validation is sufficient for your purposes.
You also need to make sure the you have enough power to detect a meaningful change on both scales. G-Power is a good free software program you can download to calculate power. There are also a lot of online calculators for determining the right sample size to ensure sufficient power.
Hi Deepak I don't think any standard sample size calculation software is going to answer your question. Sample size software is about power to detect differences, so unless you have a specific question about a difference you want to be able to detect it's not going to tell you how large a sample you need to assess your instrument. As always, one answer is, "the more the better." One very important thing, however, is that you want to be sure to get as large a range of responses as possible. In other words, you want your sample to be heterogeneous is as many ways as possible You won't get very much information if all the answers are at one end of the scale or some answers are never chosen. This is a rare case in quantitative research where random sampling isn't so important, becuse it isn't necessary to generalize back to a population. An aside, it is very important to back translate the instrument by independent translators to see if you obtained a good translation before you even try the instrument. There is no point collecting data if your translation isn't good. As for things you might do, if there is a published factor analysis, for example, you might try to replicated that. For an example, you can see a paper called "A Closer Look at the Measurement of Burnout" on my profile. It's not a translation, but we changed several aspects of the original measure and then showed we could pretty much reproduce a published analysis of the original. Doing a confirmatory factor analysis like that would require more data then a simpler item analysis or correlation analysis or even an exploratory factor analysis. Bob
I think i would be useful to check some papers about the development or validation of a scale of similar characteristics as the one that you want to validate. It will be useful to analize the decisions of the original authors about the sampling process. Maybe I could help if you tell us more information about the questionnaire.
The number of subjects included in a factor analysis is a matter
of debate. Rules-of-thumb vary from four to 10 subjects per variable, with a minimum number of 100 subjects to ensure stability of the variancee covariance matrix
Also, You can find further details in this paper:
Quality criteria were proposed for measurement properties
of health status questionnaires. Journal of Clinical Epidemiology 60 (2007)
There is a review paper on this particular topic is available: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4275948/#CR39. Again there are no clear evidence based approach towards your query. Vincent.
Determining the sample sizes involve resource and statistical issues. Usually, researchers regard 100 participants as the minimum sample size when the population is large. However, In most studies the sample size is determined effectively by two factors: (1) the nature of data analysis proposed and (2) estimated response rate.
For example, if you plan to use a linear regression a sample size of 50+ 8K is required, where K is the number of predictors. Some researchers believes it is desirable to have at least 10 respondents for each item being tested in a factor analysis, Further, up to 300 responses is not unusual for Likert scale development according to other researchers.
Another method of calculating the required sample size is using the Power and Sample size program (www.power-analysis.com).
It may take you some time to figure out how to use this software and run the analysis successfully, but here are the sample size power recommendations for CFA and growth models using MPlus