There are all sorts of problems with simple observation. There are places where observers are prohibited - in government, in much of business, and often in the bedroom, but even where a person is allowed in the existence of a passive observer sat in a corner perhaps taking notes disturbs the participants and makes them change their behaviour. We see this in the docusoaps. One solution is to observe a situation where the presence of observers is considered more normal, but these are few and far between. Another strategy is to be in a place so often and for so long that it becomes normal to be there.
The best solution is for the observer to get involved with the activities of the group being studied. The observer thus becomes a participant observer. This method is associated with social anthropology and symbolic interactionism. It can be used with people who still want to study the large-scale issues that are seen to determine social reality and by those phenomenologists who are simply interested in meaning.
The first question is how do you get in. You can become a covert researcher. You get in by pretending you're a new member of the group and joining in. You never tell anyone why you are really there. When done you go away and they are left puzzled as to why you have gone. You may invent a story for that or just become out of reach. Or you may tell everyone a partial truth - tell them you're looking for ideas for a book, for example. Or maybe you can tell some people more than others - like the leaders know who you are but no one else does. Or in time you may find a friendly privileged individual within the group who you tell more about yourself because you can use that person to get more information. Well it's an ethical nightmare and even if you do choose to be covert you might get found out. They might not tell you they've found out who you are but alter their behaviour, which ruins the validity of the data, or indeed they bring you to a sticky end one dark rainy night.
If you are successfully covert then at least no one is altering their behaviour for you - however, you must be seen to have become a full participant. Unless you have the mental constitution of an expert in espionage, and there are few of those, this can be psychologically dangerous to the researcher, and could change the character of the researcher. The researcher can lose the ability to stand back - losing the ability to be an effective observer. However, it is possible even with openness to become involved in the subculture of any group and take on characteristics of members of the group. Empathy and sympathy as a research technique to build trust becomes your own empathy an sympathy, and this can challenge the validity of data. To some extent, but perhaps in a way unacceptable to my subjects, this happened to me.
However overt or covert, you must build trust. This takes time. It also takes time to wait for what you are interested in to arise in terms of activities and their talking. Again you might find a privileged person to talk to. This happened to me - I used someone who spilled the beans and gave lots of background information which would have taken ages to acquire by turning up week after week. This one person gave me biographies and his own assessment of the situation, and it was important too that he was rather semi-detached from the group, not affecting the central events of the group. In return I told him more about what I was doing than I told the others. Later on there was a privileged female when I wanted to find out what had happened, and she wanted to know what I had written, and there was the perosnal interest with her too.
So when do you record the information? You might have a concealed small tape recorder but it would need to be quiet and need opportunities to change tapes. You might also carry a small camera for covert filming. Of course this could be illegal and quite unethical. If you sat there taking notes it would not be participant observation - unless taking notes was common within the group (say in a workplace). But you remind people that you take notes, and people always want to know what you have said about them. I told my privileged person what I was writing about him, I did not (with one exception) tell others what I was writing about them. You need to avoid the situation where they ask you what you are writing. What you do is write up the notes that night in as detailed a way as possible. I wrote down actual conversations. Someone said this and then they said that and this was said back and it happened there where so and so was present and so and so was out of the way. Someone did this and then they did that and then the other did that and it happened there where so and so was present and so and so was out of the way. You could do it like writing a play with actors and scenes.
The big advantage of participant observation is that you see the world you are studying, hopefully with a minimum of disturbance. You find out what they do, not what they say they do. I did interview my subjects when I told them what I was doing near to the start, and they gave very rosy accounts of their activities, and then I went on to observe all the rich variations of what they actually did. You also find yourself taking up the sympathy and empathy of the participants and you becomes more able to tell the story of the subculture that was previously foreign to you. You also get over the problem of people innocently losing their memory or filling in the gaps as they do during interviews. You know what they did because you were there and you recorded it. They will tell you (as happened to me) that certain things never happened but you were there and you know that they did happen. The other thing about participant observation is that it is good fun and you can make friends.
Participant observation is very often inductive in actual method so you find certain things happening that put a whole new impetus into research, and you then follow up what is in effect a new hypothesis perhaps using other research methods.
But there are disadvantages. One is the time required, another is the demands of participation which may be illegal or dangerous. There is the potential of psychological change to who you are. Less dramatically, the researcher can be accused of being subjective, especially if seen to be sympathetic, but even if only because of the necessary interpetation and retelling of the data. Furthermore you can never get away from the researcher effect, unless it is successfully covert. I knoew this and partly by accident and partly by design I put my foot into several situations - I then recorded the responses and reactions. I'm not sure if that was bad method or an opportunity gained, but there we are. Well it can be accused of being unscientific evena participant observer can classify responses to situations and find points where variables are isolated. This might be helped by comparing different groups, perhaps at the same time and perhaps at a different time and in some cases you study, stop studying and then go back to the same group (longitudinal study). In my case I studied two groups which overlapped in time and one person had attendance at both groups.
Haralambos, M., Holborn, M., Sociology: Themes and Perspectives, London: Collins Educational, 740-747.