If a response variable has three levels, could these variants be collapsed differently to obtain two analyses? For example, if a response variable has the following three levels: - deletion of a segment; - addition of a segment; and - application of no rules. Would it be appropriate to develop two separate studies by collapsing different levels into the application value? 1 - One study focusing on: the application (i.e., deletion and addition of segments) vs. non-application of rules; and 2 - Another study analyzing: target-like forms (deletion and application of no rules) vs. non-target-like forms (addition of a segment)? I would appreciate your suggestions and comments...
Theoretically, this should be possible, but the collapsing rule has to be complex enough to be able to model it. Consider a use-case example. Deep Learning can be used to model RGB images (with 3 levels) into 2 levels, first containing the Luminance value, and the second containing some information (like an annotation). As I said, it depends on the rule used to model the method.
Thank you very much for your answer! I really appreciate it! The amalgamation of two variants in each pair of the two dependent variables is based on theoretical assumptions whereby: (1) deletion and epenthesis may be conceived of as repair strategies; and (2) the maintenance of the structure and deletion are understood as the L2 target forms... Both models are statistically significant (p < .001) , and different factors were selected as the main predictors in each of them.
📷
- Badges
- Science topic
- Similar topics
- Comment