Acquiring research skills is considered to be a highly challenging aspect of developing expertise in the social sciences. Because instruction and mentoring in these skills are typically grounded in the self-report of researchers, difficulties in learning the material may be due to the content and accuracy of these explanations. Using a mixed-method, microgenetic design, this study examines the explanations of problem-solving processes by researchers along a continuum of expertise during simulated experiment design and subsequent data analysis. Findings indicate that participants' self-explanations are largely inaccurate. Further, frequency of inaccurate statements is positively associated with the frequency of abstract cognitive processes, such as mental modeling and situation assessment. Implications of these findings for instruction and future research directions are discussed.