Thanks. Yes, we are looking at sub-constructs of PCK by synthesizing two models: Magnusson et al.'s model for science education (Orientation, knowledge of curriculum, knowledge of students, knowledge of assessment, and knowledge of instructional strategy) and Ball et al.' model for math education (knowledge of curriculum, KCS, and KCT). So far, we have four sub-constructs: 1) awareness/preference of using questions to respond to students; 2) knowledge of the curriculum used by students; 3) knowledge of students' understanding/struggles/difficulties about a specific concept; 4) knowledge of appropriate questions that could potentially be effective in guiding students. 

Just as a note, the link you provided seems to be broken or does not lead to a functional website.

Thanks for the note. I will fix it soon. 

Those are GREAT suggestions. Thanks a lot. I totally agree that emotional scaffolding is critical for students in a self-paced lab-oriented learning environment. This is the advantage of SAs because they are more knowledgeable peers who have experienced the same process of physics learning and are more likely to reach "resonance" with students. On the other hand, SAs are less prepared or skilled in dealing with emotionally hard time from students, so they are more likely to be an answer fairy. It's challenging for SAs to find a balanced point between shielding students from overwhelmingly negative feelings and encouraging them to struggle through difficulties prior to the uh-huh moment. I think this is also related to the "flexibility of knowledge" that you mentioned. It's definitely worth of attention, but sometimes it's hard to capture students' or SAs' emotions. More accurately speaking, it's hard to infer their emotions from video cues like their actions or facial expressions, and transfer that information into quantitative data. Any further suggestions would be appreciated. 

Thanks. To your first question, we have several candidate strategies suggested by research, such as instructor modeling appropriate questions, SAs role play, and SAs reflecting on their own videos. Actually, we have been trying some of them. The difficulties are: 1) The time of SA-training sessions is limited. SAs, especially undergraduate learning assistants (LA), may not have sophisticated content knowledge. Thus, the priority is to walk SAs through the curriculum, which leaves limited time for pedagogical training. 2) We are developing the instrument to measure SAs' PCK-Q, without which, it's hard to reflect on the efficacy of certain strategies. To your second question, we will develop two instruments to measure SAs' PCK-Q, i.e. coding schema to analyze SAs'  performed knowledge of questioning and written tests to assess SAs' narrated knowledge of questioning, which will hopefully validate each other. Both are content-dependent because that is the nature of PCK, and both contain elements that are applicable in other STEM disciplines. For the former, it is the types of questions and patterns of SA-student interaction (e.g. students' lab result does not match a theory, what should a SA do?) that are transferable. For the latter, what transferable is the process of developing quizzes from authentic classroom scenarios and the context or background information provided in question stems.

This is a good question. Validity and reliability of instruments are our top concerns. We will have two instruments, coding schema for classroom video analysis and written tests. For coding schema, we will check theoretical validity (whether codes are supported by educational theories), predictive validity (practice coded at different time points are consistent with each other or match a reasonable pattern of practice development), convergent validity (refer to narrated knowledge measured by written tests, correlation and multi-linear regression will be used. Correlation is for the assumption that knowledge and practice are parallel to each other, and regression is for that knowledge determines practice.), inter-rater reliability (ICC probably will be used). For written tests, we will check face validity (feedback from the external reviewer and test respondents),  construct validity (Rasch model for item difficulty and SA ability, and 2-parameter or 3-parameter IRT after we convert open-ended test questions to multiple-choice ones), theoretical validity, predictive validity, convergent validity (refer to coding schema), inter-rater reliability (ICC again for open-ended test questions), and test-retest reliability (the same SAs take tests several times? but we are not sure as tests may be edited). Concurrent validity will not be considered because to our knowledge, there has been no such a tool to quantitatively measure SAs' PCK-Q. Please correct us if we are wrong, and inform us of the existing tools.

Thanks for the encouragement.