03 · What You Need to Know
How to Establish Meaningful Alignment Between Research and Interview Questions
Research Questions and Interview Questions Perform Different Functions
A research question identifies what the study seeks to investigate or understand. It may contain theoretical concepts, analytical relationships, or methodological language appropriate for scholarly inquiry.
An interview question is a data-generation prompt. Its purpose is to invite a participant to provide an account that can contribute to the researcher's understanding.
These functions are related but distinct.
For example, a research question might ask, "How do institutional governance arrangements shape teachers' professional autonomy in AI-mediated instruction?"
That question establishes an analytical relationship between governance and professional autonomy.
An interview prompt might instead ask, "Can you describe a recent teaching decision involving an AI tool and what influenced that decision?"
The participant's response may reveal institutional requirements, professional judgment, practical constraints, or other influences without requiring familiarity with the researcher's theoretical terminology.
The interview question is aligned because it can generate relevant evidence, not because it repeats the research question.
Why Directly Repeating Research Questions Can Be Problematic
Research questions are often written for scholarly audiences. They may contain abstract constructs that participants interpret differently or do not ordinarily use.
Terms such as agency, institutional legitimacy, epistemic trust, pedagogical autonomy, or sociotechnical mediation may be useful analytical concepts but difficult interview prompts.
Asking participants to respond directly to these concepts may encourage them to speculate about definitions rather than describe their experiences.
It can also impose the researcher's conceptual categories prematurely.
For example, asking, "How has algorithmic governance reduced your agency?" assumes that algorithmic governance exists in the participant's context and that it has reduced agency.
A more open prompt might ask, "Can you describe how decisions about using AI tools are made in your institution?"
Follow-up questions can then investigate whether participants experience restrictions, discretion, negotiation, or other arrangements.
The objective is to obtain evidence relevant to the concept without requiring participants to adopt the concept itself.
Alignment Should Be Based on the Evidence Required
A useful way to develop interview questions is to begin with the evidence the analysis will require.
For each research question, identify the experiences, decisions, interactions, interpretations, or circumstances that could illuminate it.
Then develop prompts that invite participants to describe those matters.
Consider a study asking how teachers respond to uncertainty about institutional AI policies.
Relevant evidence might include specific occasions when policy guidance was unclear, actions taken to seek clarification, decisions made despite uncertainty, and consequences for teaching practice.
Interview prompts can address these experiences without asking participants to provide an abstract explanation of institutional uncertainty.
This approach helps establish a chain of reasoning from the research question to the interview material and eventually to the analytical interpretation.
One Research Question May Require Several Interview Questions
A complex research question often cannot be adequately explored through one prompt.
For example, a study investigating how teachers develop confidence in using generative AI may need to examine initial experiences, opportunities for professional learning, unsuccessful attempts, institutional support, and changes in practice.
Each area may require a different prompt.
One question might invite a participant to describe their first experience using an AI tool. Another might explore a situation in which they felt uncertain about the output. A later question could examine how their approach changed.
These questions contribute different forms of evidence to the same research objective.
There is no methodological requirement that the number of interview questions equal the number of research questions.
One Interview Question May Contribute to Several Research Questions
Participant experiences do not necessarily fit neatly within the categories used to organize a research proposal.
Suppose a teacher describes deciding not to use an AI-generated assessment because institutional policy was unclear and the output appeared unreliable.
The account may contribute to research questions concerning professional judgment, institutional governance, and perceptions of AI reliability.
Researchers should not force the account into only one category because it emerged in response to a particular interview question.
Alignment is therefore often many-to-many rather than one-to-one.
An alignment matrix can support planning, but it should not become a rigid rule governing how data are interpreted.
How to Develop an Interview Question Alignment Matrix
An alignment matrix is a planning tool that connects research questions with the evidence needed and the prompts intended to generate that evidence.
It can help researchers identify missing areas of inquiry, redundant questions, and prompts that do not contribute meaningfully to the study.
A practical matrix might include the research question, relevant evidentiary domain, proposed interview prompt, possible probes, and intended analytical contribution.
| Research question |
Evidence needed |
Possible interview prompt |
Potential analytical contribution |
| How do teachers make decisions about using generative AI? |
Concrete instructional decisions and influencing circumstances |
Can you describe a recent situation in which you decided whether to use an AI tool? |
Decision processes and contextual influences |
| How do institutional expectations shape these decisions? |
Experiences with policies, guidance, and institutional practices |
What institutional expectations, if any, were relevant to that decision? |
Perceived institutional influences |
| How do teachers respond to uncertainty about AI outputs? |
Experiences of uncertainty and subsequent actions |
Can you recall a time when you were unsure whether an AI-generated response was suitable? |
Interpretations of uncertainty and response practices |
| How do teachers' practices develop through experience? |
Accounts of earlier and current practices |
Has your approach to using AI changed since you first encountered these tools? |
Reported development and perceived change |
The matrix demonstrates conceptual alignment without requiring the interview prompts to reproduce the wording of the research questions.
It should remain a planning resource rather than a predetermined coding framework that excludes unexpected findings.
Interview Questions Should Use Language Participants Can Understand
Questions should be appropriate to participants' experiences, professional backgrounds, and language.
Researchers should avoid unnecessary theoretical terminology, ambiguous expressions, and questions containing several separate tasks.
For example, asking, "How do you evaluate the reliability, transparency, accountability, and pedagogical alignment of generative AI systems?" may require participants to interpret several concepts simultaneously.
A more accessible approach might begin with a concrete experience: "When an AI tool produces material for your teaching, how do you decide whether to use it?"
Follow-up questions can explore particular considerations as they become relevant.
This does not mean that theoretical concepts should never be introduced. Their use should be justified by the participants' familiarity with them and the purpose of the inquiry.
Concrete Experiences Often Provide More Useful Starting Points
Questions asking participants to describe specific events can provide contextual detail that broad evaluative questions may not elicit.
For example, "What challenges do you face when using AI?" may produce a general list of concerns.
Asking, "Can you describe a recent occasion when using an AI tool did not go as expected?" may generate an account of the circumstances, actions, and consequences involved.
Neither question is universally superior. Broad prompts may be useful for identifying participants' initial concerns, while event-based prompts can support more detailed exploration.
The choice depends on the research question and the type of evidence needed.
Researchers should also recognize that accounts of past events are reconstructed through memory and interpretation rather than direct recordings of what occurred.
Use Follow-Up Probes to Connect Experiences with Analytical Concerns
Follow-up probes can help researchers explore the relevance of an experience without introducing leading assumptions.
Suppose a participant describes using AI-generated lesson materials despite uncertainty about their accuracy.
The interviewer might ask what made the material seem uncertain, what the participant did next, and whether institutional expectations influenced the decision.
These probes may contribute evidence about professional judgment and institutional context.
However, the interviewer should not assume that uncertainty necessarily reduced trust or that institutional policy constrained the participant.
Procedures for avoiding leading interview questions are important when researchers use established theories to guide probing.
Alignment Does Not Mean Every Interview Must Be Identical
In semi-structured interviewing, participants may address common topics through different experiences.
One teacher may discuss AI-supported assessment, while another describes instructional planning. Both accounts may contribute to a research question about professional decision-making.
Researchers should identify essential topics while allowing relevant variation in examples, question order, and probing.
The decision about whether all participants should receive the same questions depends on the degree of standardization required by the research design.
Similarly, flexibility within a semi-structured interview guide can support alignment when it helps clarify participants' experiences without changing the study's purpose.
How Much Should a Conceptual Framework Influence Interview Questions?
A conceptual framework can help identify relevant phenomena, relationships, and potential areas of inquiry.
However, using a framework does not require every question to correspond to a predetermined construct.
In some theory-informed studies, researchers deliberately investigate how participants experience concepts identified in existing literature.
In more exploratory approaches, the framework may provide background orientation while leaving greater room for participants' own categories.
The appropriate relationship depends on the study's epistemological position and analytical purpose.
Researchers should distinguish questions designed to investigate a theoretical proposition from questions that presuppose the proposition is correct.
Can Interview Questions Address Emerging Issues Not Included in the Original Research Questions?
Yes, when the emerging issue contributes meaningfully to the study's purpose and the methodology permits flexible inquiry.
For example, participants discussing AI adoption may unexpectedly identify institutional procurement as a major influence on their decisions.
If procurement arrangements are relevant to the research question, follow-up questions may be justified.
However, an interesting topic does not automatically justify expanding the study into an unrelated investigation.
Researchers should evaluate whether the issue clarifies, complicates, or challenges the existing inquiry.
Substantial changes to the research objectives or approved interview procedures may require protocol review.
What Is the Risk of Overly Rigid Alignment?
An excessively rigid alignment matrix can encourage researchers to treat participants' accounts as material that must be placed into predetermined categories.
This may reduce attention to contradictions, unexpected experiences, and interpretations that do not fit the initial framework.
For example, a study organized around perceived usefulness and ease of use might overlook institutional trust, professional identity, or ethical concerns if the interview guide permits discussion only of the two predefined constructs.
The problem is not the existence of an alignment matrix. It is using the matrix to restrict what participants can meaningfully contribute.
Researchers should retain opportunities for participants to introduce issues they consider important and explain how unexpected evidence will be handled analytically.
How Should Interview Question Alignment Be Evaluated?
Before data collection, researchers can review the interview guide to determine whether its prompts are relevant, understandable, and capable of generating evidence for the research questions.
Expert review may identify missing areas or questions that introduce unsupported assumptions.
Pilot interviews can reveal whether participants understand the prompts and whether their responses provide the expected type of information.
Researchers should avoid treating expert agreement as conclusive proof that an interview guide is valid.
Unlike a standardized psychometric instrument, a qualitative interview guide may be evaluated primarily through methodological coherence, question quality, contextual appropriateness, and the usefulness of the material it generates.
Kallio and colleagues (2016) emphasize systematic development and testing of semi-structured interview guides, while Castillo-Montoya (2016) provides a framework for preparing and refining interview protocols through alignment, conversational organization, feedback, and piloting.
How Should Alignment Be Reported in the Methodology?
A methods section should explain how the interview guide was developed from the research questions, relevant literature, conceptual framework where applicable, and intended analytical approach.
Researchers may include an alignment matrix in an appendix to demonstrate how the principal topics contribute to the study.
However, the matrix should not be presented as evidence that every response can be assigned to one research question or that the guide's methodological adequacy has been conclusively established.
Report significant guide revisions, pilot testing, and the degree of flexibility permitted during interviews.
Transparency about these decisions allows readers to evaluate how the interview material was generated and how it supports the findings.
