Agree with Mohammed - all psychometric properties are potentially context-dependent. I recall a colleague translating an established English language scale into Korean and getting substantially different results than expected. It turned out that the literal wording in English did not accurately reflect the way Koreans think about the topic. As a result the scale underestimated the health condition in Koreans and had to be revised.

Agree with Mohammed - all psychometric properties are potentially context-dependent. I recall a colleague translating an established English language scale into Korean and getting substantially different results than expected. It turned out that the literal wording in English did not accurately reflect the way Koreans think about the topic. As a result the scale underestimated the health condition in Koreans and had to be revised.

Article Insufficient cross-cultural adaptations and psychometric pro...

Shalini Quadros, I notice that you say that the scale you want to use has "well established psychometric properties", but even those properties might not be as valid as many believe them to be.

So, that is an increased reason why it would be good, as indicated by the above contributions, to check the psychometric properties of the scale, not only when translated, but also in the context of your own research.

Did you translate the measure yourself? If the measure was already translated, then whomever used it should have published psychometric properties. With any measure that has been validated psychometrically, if you manipulate the measure in any regard, you are straying from the original properties and it is no longer reliable and valid until you test it.

Non-english versions of measures are extremely useful! You should absolutely test and publish it's properties!

Shalini Quadros, since contributing my post up above, I have realised that I didn't respond sufficiently to your question. You asked whether you needed to test all the psychometric properties of the translated scale - and I didn't respond to the issue of "all".

I'm happy for others to disagree with me, but I think the short answer is no, you don't need to test ALL the psychometric properties of the translated scale. For example, it might not be necessary to check for the full set of indicators of metric invariance because those indicators can be quite "demanding" and are often not satisfied within cultures let alone across them. An interesting, refreshing, and informative piece about that was posted on the internet recently by David Funder:

https://funderstorms.wordpress.com/2020/01/31/misgivings-some-thoughts-about-measurement-invariance/

Depending on your sample size(s), it might not even be possible to check the factor structure of the translated scale, though doing so would be desirable. A lot of researchers trot out coefficient alpha values, presumably believing that they are useful (which they often aren't), but, if you believe that finding them would be helpful, I recommend going ahead and obtaining them.

Obtaining other forms of reliability (e.g., test-test reliability) as well as a number of forms of construct validity might also be beyond your resources - and the resources of many researchers. In essence, obtaining a range of psychometric properties of a scale is a quite large task.

So, if I were you, I'd look carefully at the kinds of psychometric properties that you are capable of obtaining and then go about obtaining them as competently and honestly as you possibly can.

Whatever you're able to do, I certainly recommend that you not make assumptions about the psychometric properties of either the original or the translated scale.

A systematic review of the psychometric properties of the cross-cultural translations and adaptations of the Multidimensional Perceived Social Support Scale (MSPSS)

Jermaine M. Dambi1,2* , Lieselotte Corten1 , Matthew Chiwaridzo1,2 , Helen Jack3,4 , Tecla Mlambo2 and Jennifer Jelsma1

Definitely psychometric properties take on context-specific, operational definitions that may need refinement cross-culturally. I would wonder that the purpose to which you are applying such metrics would further define the relevance and context to which such constructs would apply.

I agree with everyone. Absolutely the translated tool needs revalidation both for being able to convey the intended point of the language, and for the cultural meaning that might be inferred by the question that could differ substantially from the original.

I agree with Nancy and the others. These days, it is strongly recommended that you validate the psychometric properties of any instrument that is used in a new or different context even if it is in the original language that it was developed. In Shalini's case, the fact that you are translating the instrument into a new language makes it more important to validate it due to possible effects of the study context and participants.

It is very important to check the psychometric properties of translated version of a gold standard scale so as to determine its validity and reliability. This will insure the authenticity of the data collected and findings made.

a test score is a function of "true score", "systematic error" and "random error". You can only limit the systematic error to some extent but variation in other two variable can never be even expected. Therefore, anytime you measure a social construct you should check whether psychometric properties meet the threshold

If yiu translate it you will have to check the chronbach of the same before going for any test as reliabilyy is now crucial .

In my view, it might be good to check Cronbach's alpha (more properly referred to as coefficient alpha), but I think researchers should be aware that it can be a very deceptive and uninformative metric. See the following references, for example:

Schmitt, N. (1996). Uses and abuses of coefficient alpha. Psychological Assessment, 8(4), 350–353. https://doi.org/10.1037//1040-3590.8.4.350

Taber, K. S. (2018). The use of Cronbach’s alpha when developing and reporting research instruments in science education. Research in Science Education, 48(6), 1273–1296. https://doi.org/10.1007/s11165-016-9602-2

If it's possible, I would prefer to check the factor structure of the translated scale to see whether that structure conforms to the original scale. That structure could be influenced either by the new scale itself, or by a "new" sample.