In this part of the research, I compared 10 high- and 10 low-rated theses with the intention to reveal typical learner errors in terms of reference. Though RCA (as described at Stages 2 and 3) was not directly in the center of our attention for the discussion of error types here, it was of course an inevitable part of the procedure of collecting reference errors. Unfortunately, this RCA revealed two major issues that lead to the modification of the thesis corpus used in this study in subsequent parts of this book. An interesting outcome of the method of RCA introduced here is that it shows a markedly different structure for empirical and theoretical research papers. On the other hand, out of the empirical research papers, case studies also had a structure that both resembled theoretical papers and had extensive samples of interview data, which would have distorted the comparative data concerning empirical papers. Therefore, 5 papers were replaced (3 theory-oriented papers and 2 cases studies) in the MA thesis corpus.