I apologize if this is a simple question. I have created dummy variables for highest degree earned. I have high school grad, AA degree, BA degree, MA, and doctorate. My reference group is high school grad. My results did not make sense so I added high school grad in and I didn't get an error for collinearity with STATA dropping one of the categories. I looked closer at the codebook and there are categories labeled 97, 98, and 99 for refusal, missing, and legitimate skip respectively. Do I need to drop all of these people it wold be over 1,000 observations?