01 According to one of the seminal books on participant observation from 1980, there’s a continuum in the level of researcher involvement from non-participatory to passive, moderate, active, and complete participation. See (new edition) Spradley, J. P. (2016). Participant Observation. Waveland Press.
02 See #TiSDD 5.1.3, Data collection, for more information on potential biases.
With this approach, the people who are being observed know that researchers are present and that they are currently being observed in situations that are relevant to the research question. This is the difference compared to non-participant observation, where research subjects do not know that they are being observed. Since researchers are visible, it is important to manage the “observer effect” – the influence researchers have on their environment and on their research participants’ behavior simply by being present. There’s a fluid transition between participant observation and contextual interviews, and often these go hand in hand. Try to balance out biases like the observer effect by cross-checking with other (non-participant) research methods. 
Researchers can observe situations that include digital and physical actions with or without other people and/or machines. In this context, participant observation is particularly useful to understand cross-channel experiences, as the method focuses on people and not on one particular channel. Depending on the research question and context, observations might be at the participants’ workplace, in their homes, or even following them throughout a process like a holiday trip.
During participant observations it is important to observe not only what people are doing, by interpreting their body language and gestures, but also what people are not doing (e.g., do they ignore instructions or refrain from asking for help or assistance?).