Review questions
Integrated summary: Questions and research design
Begin with a question that the intended evidence can answer, rather than with the name of a collection tool. ‘The library is poor’ is an unbounded judgment: whose experience, which service and what circumstances are involved? Qualitative inquiry can examine meanings, steps and obstacles in participants’ accounts. It therefore helps to separate the observed issue from an assumed explanation. Low workshop attendance may justify an inquiry, but it does not establish why people stayed away. A few interviews cannot estimate the proportion of all library users who experienced a particular obstacle. A useful qualitative question bounds experience and context without presuming a cause. Interviews can illuminate a journey, but a few messages cannot yield a population rate or causal conclusion.
A method follows the question and intended inference; it is not a declaration of the researcher’s preferred tool. If a library team asks ‘How do users understand an error message?’, it needs an account of meaning, sequence and differences in experience. A semi-structured interview lets a participant interpret the wording and the interviewer ask a neutral follow-up. ‘What proportion of booking attempts stopped at the payment step?’ instead requires defined attempts, a defined stopping event, a denominator and a record covering the relevant period. Neither question replaces the other. A rich account does not produce a rate, and a counter does not necessarily explain why a person stopped. One project may need both questions when the decision depends on extent and explanation. A mixed methods design then specifies in advance where the results will be integrated. The question determines evidence: interviews illuminate experience, defined records estimate rates, and mixed design links a pattern with an account when both are needed. Integration is planned explicitly; temporal association is not causal proof.
Approach names are methodological commitments, not decoration. A case study examines a bounded system, such as one programme’s move to online booking during a particular term, using accounts, records and observations to understand how its parts worked in context. If the question concerns the meaning of waiting as lived by people who experienced it, a lived-experience approach may fit; the focus is their experience rather than the performance of one organisation. To describe a group’s practices and shared meanings in its setting, a passing interview is not enough to label the work ethnography. Appropriate presence and observation are needed. To develop an explanation of a changing social process, such as how learners move from hesitation to help seeking, a grounded theory approach may fit if comparison and ongoing analysis guide recruitment of later cases. The differences concern purpose, unit of analysis and construction of evidence. A name alone cannot supply depth. Choose an approach by question, unit of analysis and feasible evidence. A case examines a bounded system; lived experience examines meaning; ethnography studies group practice; grounded theory develops process through iterative comparison. An honest descriptive study is better than an unsupported label.
A report is not mixed methods merely because it contains a numerical table and a quotation. The research question must call for both kinds of evidence, and the investigator must show when and how the strands connect to answer something neither could answer alone. If a usage log shows a drop in completion at one booking step, selected interviews may explore what happened there. If early interviews reveal several meanings of ‘ease of use’, they may inform construction of a later measurement item. In a parallel design, usage records and accounts from the same period can be brought together to examine agreement and divergence. These are different purposes. A strong numerical result cannot compensate automatically for weak interviews, and a detailed story cannot repair an unreliable log. Mixed methods needs a clear two-strand question and specified connection. Numbers may guide explanatory interviews, interviews may build a measure, or strands may run in parallel. Record and investigate discordance rather than conceal it.
A protocol does not promise that every participant will respond. It records decisions that can be reviewed before data collection. State the question, why the approach fits, the unit of analysis, inclusion and exclusion criteria, who will invite people, what they will be told and when they may decline. Then describe each data source, its instrument and storage, and how raw material will support a bounded inference. For mixed work, add strand sequence, integration point and the decision rule when results disagree. Do not defer privacy planning until after the interview. Participants may mention names or distinctive events that identify them even after their direct name is removed. Access rules and separation of identity from transcripts belong in the design, not an incidental technical appendix. A protocol ties a question to obtainable authorised inputs, sampling, analysis, timing, risks and a fallback when evidence is missing. Organisational access, individual consent and ethics review are distinct; invented training data confer no permission for a real study.
Review questions
Test your understanding with these questions. Marking and saved results are available with a free account.
Create a free account to mark your answers and save your progress.