Problems with participant observation

You are here

Problems with participant observation

Shipman suggests that there are three key questions that should be asked about any piece of social research:

If the investigation had been carried out by someone other than the author, but using the author's methods, would the same results have been obtained? (reliability)

The answer for P.O. is very unlikely. The characteristics of the author are essential to the account produced. We don't need to get complicated here, crucial variables will be age, sex, ethnicity, personality.

Could Willis replicate Pryce's Endless Pressure, could an old, fat, lazy woman replicate Parker's View from the Boys?


Do the results really reflect the influence of the factors under examination, or did extraneous influences interfere? (validity). Unknowable, but at least you might hope that a researcher has a better chance with this method than with most others.

What relevance do the results have beyond the actual research, for instance, to what extent can the results be applicable to situations similar to that studied? (generalisation).

A number of problems here: first, it might not be the intention to generalise, it might be a unique case, second, have other similar studies been done?

Most often the conclusion must be that generalisation is fraught with difficulty because of the limited size of the group studied.

How many boys did Willis hang out with? The West Midlands, typical of what?

P.O. produces interesting and thought provoking work, it is messy, inconclusive and of dubious parentage. It could be good but is most likely to be used by the least experienced, and knowledgeable researchers and the groups that are 'open' to such study are limited-it tends to be the poor, the weak, the disadvantaged. P.O. is frequently associated, in my mind, with studies of the weird and the powerless.


The perception of any observer is not passive. An impression is first selected and then interpreted within the mind of the observer. The observers attitudes, values and beliefs intervene, as do any theoretical models held. This has been called the pollution of interpretation and is clearly a part of the problem of imposition.


Control is used in research in order to detect and reduce errors in observation. In a 'natural' setting replication is impossible. Consequently, comparison with other similar settings is often used as an alternative method. Additionally, checklists, film and tape recordings could be used.


This is nonsense as far as P.O. is concerned because control introduces factors not present in the actual research if the research is any good. Control introduces interference, to control a situation is to change it. To identify significant controls is to specify, in advance, what are significant variables. In brief, control imposes pre-selected boundaries of significance - OK for cabbages but will it do for people? This is not really a problem for the method but a problem of evaluation. For 'scientific' sociology there is the deep question of how they think they can control human interaction.


Reliability requires detachment. But detachment requires disengagement with the subjects of research; we stand back and ask 'what do I think?'. And we assume that what we think is detached. There is no way to objectively consider evidence. P.O. provides stories just like statistics provide stories, you decide what resonates with you.

It is also likely to be the case that in the process of P.O. research that the initial period and the final period are those where detachment is most sought. During the actual joining in period subjective understanding is what is sought.


All researchers have some conception of what they consider ethical, so do we all. The problem is that we don't always do what we would ideally like to do. Ethics is simply a question of how much guilt we can handle, and how many valid excuses we can construct. Sociologists, in the UK at least, are part of an academic community in which prestige and advancement are based (at least in part) upon prowess in research.

Prowess is associated with finding out new and interesting things. Think on, think particularly about the source of our ethical standards - they are not philosophical absolutes but class based relative accords of reasonable behaviour. And do they stick to them? Simply, ethics is a minefield, so don't let the opposing camp tell you what they are!

Clearly there is a need for some standard, but, at present one does not exist. The problem in research , at present, is the pretence of neutrality while pursuing political agendas.

There is no reason to suspect P.O., rather than other research techniques, of ethical bias. Certainly we should consider the ethical implications of the use made of social research by people who are seeking personal advancement, and they are always there.


In any research into people there is the problem of imposition. At heart, this means getting the answers you get or the conclusion you reach, because of the way that you, as a researcher, handle the data.

Interactionist research has often believed that it bypassed the problem of imposition because it refrained from asking direct (tick in the box questions).

Certainly P.O. generally refrains from asking tick in the box questions. Thus it gets round the imposition problem of: asking specific questions. And seeking specific answers, but:

Imposition is still there:

  • The researcher has to choose what to observe.
  • The researcher chooses what to record.
  • The researcher chooses what to use.
  • The researcher decides significance.

It all amounts to the pollution of interpretation.

Problems with participant observation

A perfect design for research would be:

  • Fully controlled to ensure reliability.
  • Fully participant to ensure maximum validity.
  • Carried out with scientific rigour.
  • Done with the full knowledge of the observed.

P.O. is a valuable research tool, it is not able to deliver, objective, testable, replicable research, but it might provide insight, the starting point for understanding.