Appendix 1 1(a), 1(c), 1(d) - Interviews, coding and ethical representation

Choosing an interview approach

An interview can reveal why people use or avoid a place, including meanings that a count of visitors cannot show.

Suppose the geographical enquiry is: How does access to a town square differ between its users? A pedestrian count can measure how many people pass at stated times; an interview can reveal whether a route feels safe, whether seating is usable or why someone stays away. Using both methods connects observed use with explanations. Neither method automatically validates the other: an empty square at midday may reflect work schedules rather than dissatisfaction.

In a structured interview, the same questions are asked in the same order. Comparisons are easier, but an unexpected issue may never emerge. A semi-structured interview uses common opening questions and neutral follow-up prompts. This preserves some comparability while allowing a participant to explain, for example, a difficulty using the entrance. An unstructured interview develops around the participant's account. It may reveal important meanings, but topics and detail can vary greatly between participants. Choose a mixture for a reason, rather than assuming one method is always superior.

Compare two prompts:

A follow-up such as “Can you describe an occasion when that happened?” asks for evidence without supplying a cause. “Was it because the new café is expensive?” inserts the researcher's explanation. Pilot the wording, check that one question does not ask two things at once, and record whether responses are verbatim audio, checked notes or summaries. A summary can be useful, but should not be presented as an exact quotation.

Whose accounts enter the sample

The people a researcher can reach are often different from the people whose experiences the enquiry needs to understand.

Interviewing whoever is beside the café at 11 am on a weekday is convenient. It omits many workers, evening users and people who avoid the square. A larger café sample would still inherit this coverage problem. For the access question, purposive recruitment might seek contrasting experiences: frequent users, nearby residents, traders and people who stopped visiting. The labels must fit the question; do not assume everyone within a category agrees.

An invitation through a residents' group can reach knowledgeable people, but its organiser is a gatekeeper who may filter access. Recruiting through several appropriate routes can widen the range. Snowball sampling, in which a participant refers another, can reach less visible networks; it can also concentrate the sample within one friendship group. State how participants were reached, who declined and which perspectives remain absent. A refusal is not evidence of a negative opinion.

Time, language, interview location and the researcher's role affect whose account is recorded. An interview beside a manager may inhibit a trader's criticism; an unfamiliar technical question may silence someone who has valuable lived experience. Positionality means the researcher's social position and relationship to participants. Reflecting on it helps explain how evidence was produced. Being an insider can create trust but also unspoken assumptions; being an outsider can elicit explanation but also misunderstanding. Neither status guarantees objectivity.

A defensible report might state: “These accounts describe four contrasting experiences of access, recruited purposively; they do not estimate the proportion of all residents with each view.” Avoid “Residents believe...” when only café customers were interviewed. A repeated theme may merit investigation even when its population frequency is unknown.

Coding accounts without losing meaning

Coding groups relevant meanings so that a researcher can compare accounts while preserving the context of each speaker.

The table contains four original illustrative accounts of a fictional square. The letter identifies a speaker, not a population group.