As part of my research I need to translate the tool which is developed in the USA. Therefore we need to test their reliability and validity in the community that we need to use it for. I need to use a tool with reliability and validity being tested in children with different diagnosis such as learning disability, ADHD, Cerebral Palsy etc. We tested th reliability and validity in Iran using for ADHD children only. I believe as the original validity and reliability was good across the diagnosis therefore what we needed to work on was only testing the tool considering the culture therefore this tool now can be applied to different participants group with different diagnosis because this has been tested in the original version. But some of my colleagues think the tool should be tested for each one of the diagnosis again in this country. Your thoughts are really appreciated. And thank you
I think that it's better to build new tools and not only translated it, because the cultural aspects are very important in a lot of areas en human sciences. For example USA is a very religious country and not GB or France (at 50% or more agnostic or atheist) so the bases and the grounds of relations are not exactly explained by the same link or by the same representations of the world. The roots of moral are very differents... but I can give a lot reason to built specifics tools for a human science research.
You should do both: calculate validity and reliability to the general and to the specific groups. However, other multigroup analysis should tell you about group invariance or different item functioning (DIF).
In SPSS, you can check both reliability and validity. Reliability can be checked using Cronbach Alpha.
By using SPSS, one can check both reliability and validity. The strength of alpha values will be considered. For applying measure or assessment tool in another culture, do a pilot study on a particular culture respondents by using original measure and then can test the reliability and validity, and if found measure reliable, then one can use it comfortably.
I agree partly with Jacques. I think you should consider the tool made in USA but I would firstly check how the construct specified in USA sound in your country. Your cultural and language differences may render the original construct obsolete for your application. If you translate, you should translate in both directions, from english to your used language and the reverse after, to be sure the translation does not introduce any type of biais (mainly due to different way to see the world). You should proceed after like a new instrument, face validity with experts, pilot test, construct validity etc. The validity of the instrument should not be only the statistical analysis with a model of measure but also considers the future use made with this instrument. See Messick and/or Kane for more discussions about validity or Markus & Borsboom (2013). Hope this may be usefull.
Reliability is sample specific, so I would run it in the actual samples you are using per your colleagues suggestion. A measure may be reliable in one sample, and then fail to be reliable in another sample. If the measure itself was reliable as people sometimes describe reliability to be, then it should have the same (or very similar) reliability statistics regardless of the sample. However, this is not always the case, which is why it should be considered sample, and not measure, specific. I think their suggestion is a good suggestion. It will only add credibility to the measure.
- Badges
- Science topic
- Similar topics
- Human-Computer Interaction
- Collaboration