Is it advisable for a researcher to use non-probability sampling for example in an interview or observation and a probability sampling for a questionnaire in the same study?
You would need to be clear on what your reasoning was for each portion of your study, as well as your linkage between the studies. For example, in mixed methods research, your design might be either Exploratory Sequential (qual --> QUANT) or Explanatory Sequential (QUANT --> qual), and either of these cases would allow you to combine different sampling techniques, as long as your choice was well justified.