Hello Rihab,

I'll assume that your questionnaire has been properly translated (and back translated, to assure accuracy of the conversion), any required item reversals applied before scoring, and correctly entered in your data worksheet. As you suggest, one reason that estimates of internal consistency reliability such as Cronbach's alpha are low is that the measure is multidimensional, not unidimensional.

For that reason, an EFA sounds like a good idea at this stage. I would use all of the items together initially, to get a better understanding of how well or how poorly item scores appear to affiliate. Here's a couple of instructional videos in YouTube on using EFA in R:

https://www.youtube.com/watch?v=y-rz1gQwD9U

https://www.youtube.com/watch?v=Ilf1XR-K3ps

There are plenty of others, but this should get you started.

Good luck with your work!

Hi David,

Thanks so much for your response. It is the first time I do this type of work and I am very happy to have feedback from someone from the field! Well I am not sure what you mean by "properly translated (and back translated, to assure accuracy of the conversion)"? What does back translation mean and how do we do it? I did reverse score negatively worded items and verified that the data is correctly entered but I did not do any back translation. Please enlighten me on ways and techniques we can ensure data accuracy when we work across different languages.

Hello Rihab,

Once a measure is translated into another language, back translation means having someone else translate it from the target language version back to an original language version. You then compare the initial, untranslated (in original language) version to its back translated counterpart to see whether any changes in meaning have accrued as a result of the translation process.

For example, in American English, "boot" means sturdy footwear, typically for outdoor use. In UK English, "boot" might be interpreted as the trunk of an automobile. Hence, a "translated" version might not have equivalent meaning.

Good luck with your work!

See a reference for translation in this article:

https://www.researchgate.net/publication/51881861_The_courage_to_change_Patient_perceptions_of_12-Step_fellowships?_sg=l62bAx3BDj9NG7U4Xr_mSv7DEnFekCOBfaKfQp7kxrpdk89t4o79iPj2oJ7Cmb6nTAx0bifiIiphdU9s7qvVtjLE7HdN1tcpzPHb1xWL.p8qgegg_gw8m4TUe1vqHha_7ZpuZ6cRX59HwEYKYZF7aYgtpeVIzWMxlz6eUKQTURRYVNey89FEOfGYP_JTvvw

Best

John kåre

Thank you all!

I have done PCA to see the number of components and I got 10 components in all (view attached file). I am not sure now how many to choose for EFA and on what basis? I also did a corelation plot (view 2nd file) and there seem to be no strong correlations. Does this mean that my data fails to measure the components I intended it to measure?