It depends on your hypotheses. Do you expect the whole scale to be a predictor of academic performance or simply these two dimensions, or are you following an exploratory approach?

Even if you are only assuming a potential influence of just the two dimensions, it might be interesting to report the effects of the other dimensions (or the entire scale) in order to show that it is only the two dimensions that elicited an effect on the IV

Ujwala Cherian ,

In addition to the insights shared by Sven Greving , both your specified research questions and hypotheses provide as the best guide to what you should be doing. I also note that academic performance is a continuous variable - the dependent variable in your case. However, it is also crucial to know the measurement level of each of your independent variables TEI and CS (Precision and Expressiveness styles) in order to determine a suitable method for analysis. For instance, the indicator Expressiveness cannot be a continuous variable hence multiple linear regression cannot be suitable to apply in assessing the influence of Expressiveness style on academic performance since the assumption of linearity in parameters would not be satisfied. In brief, data should satisfy certain key assumptions in order to it to be analysed by specific statistical method(s).

Thanks for the helpful insights @ Sven and Elvis!

Could be a silly follow up question but I have to confess I am still getting around Stats.

If I decide to look at all the subsets can I then input all the values of these subsets along with my other IV - TEI in the IV box in SPSS?