Based on theories, the construct (translation strategies) I would like to examine by the scale contains four dimensions (four types of translation strategies). So I develop scale items describing these four types of strategies, with one item on one strategy. Participants (students) choose the response alternative according to how well they can use the strategy.

As students can be familiar with strategies of different types to the same degrees while being familiar with strategies of the same type to different degrees, not all the strategies (item) of the same type (dimension) can be clustered together if I put all the items together in one EFA. Therefore, my questions are: could I do EFA dimension by dimension (i.e., conducting four EFAs)? Could I conclude the scale has acceptable construct validity if only one factor is attained in each of the four dimensions?

Besides, if I could do so, is there any published article using similar approach of EFA when developing scales or sub-scales?


1 Answer 1


Yes, absolutely; you could run 4 EFAs to assess the dimensionality of sub-scales. However, you could also run a confirmatory factor analysis (CFA) with four latent factors - one for each translation strategy. If you are already reasonably certain there are four sub-scales, then why not go ahead and assess whether such a factor structure can adequately reproduce your data?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.