Second fieldwork data collection and analysis practice questions.

Returning to a launderette for a second round of data collection, a student is directed by the owner to speak to him rather than to the customers, and he answers on their behalf about why people come at particular hours. What is the most serious problem with the data that results?

• The interview was not recorded, so the exact wording of the answers cannot be quoted anywhere in the written analysis.
• The owner's account has been substituted for the accounts of the people whose practice is in question.
• A sample of one person is too small to support a statistical claim about the customers.
• The owner is likely to have been mistaken about the opening hours he described to the student.

Answer: The owner's account has been substituted for the accounts of the people whose practice is in question.

Whoever controls access often also offers an authoritative version of the setting, and it is easy to accept because it is coherent and freely given. The customers' own reasons were the data this round was designed to collect.

A student's second round produced four interviews. The written analysis retells each conversation in turn, in the order in which it happened, and closes with a paragraph saying that the interviews were interesting. Which criticism identifies the main weakness?

• Four interviews are too few, and the round should be repeated with a larger and more representative group.
• The transcripts should have been reproduced in full so that a reader can check every claim independently.
• The interviews should have been replaced by a questionnaire, which would have produced comparable answers.
• The material has been narrated rather than analysed, since nothing organises it across the four interviews.

Answer: The material has been narrated rather than analysed, since nothing organises it across the four interviews.

Analysis means bringing material under themes that cut across the cases, so that similarity and difference can be shown and explained. Retelling preserves the order of the fieldwork rather than the structure of an argument.

In her second round a student is told by every regular at a community kitchen that places are taken first come, first served; her observation notes from the first round show three people consistently served ahead of the queue. How should the analysis treat the discrepancy?

• Report the observation only, since what people do is more reliable evidence than what they say.
• Report the interviews only, since participants are the authority on their own arrangements.
• Set both records aside as inconclusive, and build the analysis only on the material on which the two sources happen to agree.
• Treat the discrepancy as the finding, and ask what the stated rule accomplishes for those who state it.

Answer: Treat the discrepancy as the finding, and ask what the stated rule accomplishes for those who state it.

A rule that is stated but not followed is doing something: asserting fairness, protecting a claim, or managing the presence of an outsider. Discarding either record removes the only evidence that this work is being done.

In her second round a student interviews four regulars of a community centre in the main hall, within earshot of the manager, and afterwards talks to two of the same people outside on the way to the bus. The second set of answers is markedly less approving. What should the analysis do with the difference?

• Treat where each account was given, and in front of whom, as part of the evidence.
• Use the second set only, since answers given away from the manager must be the honest ones.
• Use the first set only, since those four interviews were at least conducted under identical conditions.
• Set both sets aside, since participants who say two different things cannot be relied upon.

Answer: Treat where each account was given, and in front of whom, as part of the evidence.

What can be said differs with who is listening, so the situation of an interview is part of what it produced rather than a nuisance to be controlled away. Scott's fieldwork in a Malaysian rice-farming village between 1978 and 1980 turned on exactly this: deference to landholders in public, mockery of them among themselves, and neither version the real one.

A student is drafting the questions for her second round at a launderette. Which wording is least likely to shape the answer she receives?

• 'You find waiting for a machine frustrating, don't you?'
• 'Do you agree that the machines in this launderette are not maintained as they should be?'
• 'Tell me what you did the last time you came and all the machines were in use.'
• 'Why do people put up with the queue instead of going somewhere else?'

Answer: 'Tell me what you did the last time you came and all the machines were in use.'

A question that carries its own answer produces evidence about the question rather than about the setting, and the agreement it collects is hard to spot afterwards in a transcript. Asking what happened on a particular occasion lets the evaluation, if there is one, arrive unprompted; the last wording still assumes that people are putting up with something.

A student has three hours of recorded interviews and cannot transcribe them all in the time available. The soundest way to transcribe selectively is to

• transcribe the first twenty minutes of each recording and leave the remainder unlisted
• summarise every recording in her own words and transcribe nothing verbatim at all
• index each recording by time and topic, then transcribe in full the passages she intends to quote or analyse closely
• transcribe only the passages in which the speakers agree with one another, since these carry the pattern, and summarise the passages in which they differ from one another

Answer: index each recording by time and topic, then transcribe in full the passages she intends to quote or analyse closely

The index preserves access to everything and the full transcription protects the wording of what will actually be used. Transcribing openings loses whatever emerged once the speaker relaxed, and selecting only agreement builds the conclusion into the record.