03 · What You Need to Know
How AI Can Help Apply the Information Power Framework
What Is Information Power in Qualitative Research?
Information power is a framework proposed by Malterud, Siersma, and Guassora (2016) for considering sample size in qualitative interview studies.
Its central proposition is that the more information relevant to the study aim a sample provides, the fewer participants may be needed, depending on the study's analytical purpose.
This does not mean that a small sample is automatically adequate when participants provide lengthy interviews.
Information power depends on the relationship between the study's purpose, the specificity of participants, theoretical guidance, dialogue quality, and analytical strategy.
The framework encourages researchers to justify sample adequacy through methodological reasoning rather than a fixed numerical rule.
What Are the Five Dimensions of Information Power?
Malterud and colleagues identify five considerations that influence information power.
| Dimension |
Central consideration |
Possible implication |
| Study aim |
How broad or narrow is the research question? |
A narrowly defined aim may require fewer participants than a broad exploratory aim. |
| Sample specificity |
How closely do participants match the experience or phenomenon being investigated? |
Highly relevant participants may contribute more focused information. |
| Use of established theory |
Does relevant theoretical knowledge guide the inquiry? |
Appropriate theory may help focus data collection and interpretation. |
| Quality of dialogue |
How rich, relevant, and informative are the interviews? |
Detailed dialogue may provide greater analytical material than superficial accounts. |
| Analysis strategy |
What type and scope of analysis will be conducted? |
Case-oriented and cross-case analyses may require different amounts of information. |
These dimensions should be considered together. None provides a universal numerical threshold that automatically determines the required sample size.
How Does the Study Aim Influence Information Power?
The breadth of the research aim affects how much information may be required.
For example, a study examining how instructors in one university interpret a specific generative AI assessment policy has a narrower scope than a study exploring all challenges associated with AI integration across higher education.
The broader study may involve more institutional contexts, professional roles, and experiences that require investigation.
A narrowly focused aim can increase information power when participants provide material directly relevant to that aim.
However, narrowing a research question merely to justify a smaller sample may weaken the intended scholarly contribution.
The aim should be determined by the research problem, not adjusted opportunistically to make recruitment easier.
What Does Sample Specificity Mean?
Sample specificity concerns how closely the selected participants match the phenomenon or experience the study seeks to understand.
For example, instructors who have personally implemented an institutional AI assessment policy may provide more directly relevant accounts than instructors who have only heard about the policy.
However, specificity does not mean selecting only participants who share the same perspective.
A sample may need to include instructors with contrasting experiences, such as those who successfully implemented the policy and those who encountered substantial difficulties.
Researchers should therefore distinguish relevance from homogeneity.
AI can help identify the characteristics needed for purposive selection, but the researcher must justify which forms of variation are important.
Does a More Homogeneous Sample Always Have Greater Information Power?
No. A relatively homogeneous sample may support a narrowly focused inquiry, but homogeneity can also conceal important differences.
For example, interviewing only instructors who enthusiastically adopted generative AI may produce detailed accounts of successful implementation.
However, those accounts may not adequately explain resistance, uncertainty, or institutional barriers if these are central to the research question.
Information power depends on relevance to the study aim, not simply similarity among participants.
AI should therefore examine whether the sampling strategy captures the variation necessary for the intended analysis.
How Does Established Theory Influence Information Power?
Relevant theoretical perspectives may help researchers focus the research question, interview guide, and analytical interpretation.
For example, a study examining instructors' professional autonomy during AI policy implementation may draw on an appropriate theory of professional agency.
The theory may help identify relevant processes and guide attention toward particular aspects of participants' experiences.
However, theory should not be used merely to force participants' accounts into predetermined categories.
A theoretical framework that is poorly aligned with the phenomenon may restrict rather than strengthen understanding.
AI can help compare possible theoretical perspectives, but researchers must verify their relevance and remain attentive to material that challenges the initial framework.
Does Using a Theory Automatically Increase Information Power?
No. The presence of a theoretical framework does not independently establish that the sample provides adequate information.
The theory must be relevant to the study aim and used coherently in the inquiry and analysis.
For example, selecting a technology acceptance theory simply because the study mentions generative AI may not adequately support an investigation of instructors' professional identities or institutional power relationships.
AI should not treat any theoretical framework as sufficient merely because it is established or frequently cited.
The value of theory depends on its explanatory fit and how it informs the analytical process.
What Does Quality of Dialogue Mean?
Quality of dialogue concerns how effectively interviews generate relevant, detailed, and meaningful information for the research aim.
It may be influenced by the interviewer's preparation, the clarity of questions, participant engagement, the interview relationship, and opportunities for elaboration.
For example, an interview in which an instructor describes specific decisions, institutional negotiations, and classroom experiences may provide richer material than one consisting mainly of brief general opinions.
However, interview length alone does not establish dialogue quality.
A long interview may contain substantial repetition or information unrelated to the research question.
AI should therefore evaluate relevance and depth rather than simply count words or minutes.
Can AI Measure Dialogue Quality From Interview Transcripts?
AI may help identify features such as brief responses, missing follow-up questions, unclear descriptions, or limited elaboration.
For example, it may highlight an interview in which a participant repeatedly refers to institutional support without explaining what that support involved.
However, transcripts do not capture every aspect of the interview interaction.
Nonverbal communication, rapport, contextual circumstances, and the researcher's understanding of the interview setting may also matter.
AI-generated judgments about dialogue quality should therefore be treated as provisional analytical observations rather than validated scores.
Researchers should review the original material and relevant interview context.
How Does the Analysis Strategy Influence Information Power?
The intended analytical strategy affects how much information is required from the sample.
For example, a detailed case-oriented analysis may focus on developing an in-depth understanding of a relatively small number of participants.
A cross-case analysis seeking to compare patterns across several institutions or professional groups may require broader information.
However, the relationship is not simply that case studies need fewer participants and comparative studies always need more.
The adequacy of the material depends on the analytical questions, case complexity, and level of detail required.
AI should examine the actual analytical purpose rather than rely on generic assumptions about qualitative methods.
Can AI Calculate Information Power Using a Formula?
No universal numerical formula is established in the original information power framework.
Malterud and colleagues describe dimensions that inform qualitative judgments about sample adequacy.
They do not provide a validated equation that converts the five dimensions into an exact participant count.
AI may nevertheless generate a scoring system that assigns points to each dimension.
Such a score would be an additional, researcher-created instrument requiring its own justification, not an established calculation from the original framework.
Researchers should avoid presenting invented scores as scientifically validated measures of information power.
Can Information Power Be Expressed as High, Moderate, or Low?
Researchers may use descriptive judgments to discuss how strongly a sample appears to support the study aim.
However, labels such as high or low should be accompanied by a transparent explanation rather than treated as standardized categories.
For example, a researcher might explain that participants have highly specific experience, interviews provide detailed accounts, and the analysis addresses a narrowly defined question.
These observations may support an argument for substantial information power.
However, the conclusion remains a methodological judgment, not a validated classification produced by an algorithm.
How Is Information Power Different From Data Saturation?
Information power and saturation are both discussed in relation to qualitative sample adequacy, but they emphasize different considerations.
Information power
Examines how the study aim, participant specificity, theoretical guidance, dialogue quality, and analytical strategy influence the information required.
Data saturation
Examines whether further data collection contributes sufficiently new information or analytical development under a specified saturation approach.
Information power provides a broader framework for considering sample adequacy throughout the research process.
Saturation may involve monitoring the contribution of successive interviews or analytical batches.
Neither should be applied mechanically, and their appropriateness depends on the methodology.
Researchers should examine whether saturation is an appropriate criterion before using it alongside or instead of information power.
Can Information Power Be Evaluated Before Data Collection?
Yes, provisionally. Researchers can examine the proposed study aim, sampling criteria, theoretical orientation, and intended analysis before recruitment begins.
For example, a narrowly focused inquiry involving participants with directly relevant experience may provide a rationale for planning a relatively focused sample.
However, the actual quality of dialogue and richness of empirical material cannot be established in advance.
Researchers should therefore distinguish an initial sampling justification from the final assessment of achieved information power.
AI can assist with planning but should not guarantee that a predetermined number of interviews will be sufficient.
Can Information Power Be Reassessed During Data Collection?
Yes. The framework can inform ongoing reflection about whether the sample and collected material support the study aim.
For example, early interviews may reveal that participants have less direct experience with AI assessment policies than anticipated.
The researcher may need to refine recruitment criteria or seek additional participants with more relevant experience.
Alternatively, interviews may provide detailed accounts that directly address a narrowly defined research question.
AI can help organize these observations, but the researcher must determine their implications for sampling.
Can Information Power Be Evaluated After Data Collection?
Yes. Researchers may use the framework to explain how the achieved sample and empirical material supported the final analysis.
However, the justification should accurately reflect the study as conducted.
For example, if the interviews were brief and participants had limited experience with the phenomenon, the researcher should not claim strong information power merely because the sample size matched the original plan.
A retrospective assessment should acknowledge limitations and explain how they affect the interpretation of findings.
AI should not be used to manufacture an overly favorable adequacy argument after recruitment has ended.
Can Information Power Justify a Very Small Sample?
Potentially, when the study aim is narrowly defined, participants are highly relevant, the material is rich, and the analytical strategy supports the intended interpretation.
However, a small sample is not automatically adequate simply because the researcher invokes information power.
For example, a study seeking to understand one highly specialized professional experience may have different information requirements from a study comparing AI policy implementation across several institutions.
The researcher must demonstrate how the actual material supports the analytical claims.
Information power should not become a rhetorical justification for recruiting fewer participants than the study requires.
Can Information Power Replace Purposive Sampling?
No. Information power is an approach to reasoning about sample adequacy, not a participant selection procedure.
Purposive sampling concerns how participants or cases are selected according to their relevance to the inquiry.
Information power helps evaluate whether the selected sample and resulting material provide sufficient information for the study aim.
The two concepts may complement one another, but they answer different methodological questions.
Researchers should therefore distinguish the sampling strategy from the rationale used to assess adequacy.
Can AI Evaluate Information Power Without the Actual Interviews?
Only provisionally. AI can assess the coherence of the proposed sampling plan and identify which information power dimensions require attention.
However, it cannot establish dialogue quality or the analytical contribution of material it has not examined.
Even when transcripts are provided, AI-generated evaluations require researcher verification and attention to the original context.
Researchers should also ensure that submitting participant material to an AI service is consistent with consent, confidentiality, and institutional data governance requirements.
How Should Information Power Be Reported in a Research Paper?
A defensible justification should explain how the sample and empirical material relate to the five dimensions of the framework.
For example, the researcher may describe the specificity of the study aim, the relevance of participant experiences, the role of theoretical guidance, the richness of interviews, and the analytical strategy.
The discussion should connect these considerations to the achieved sample rather than simply state that information power was high.
Researchers should also acknowledge limitations such as missing perspectives, uneven interview depth, or restricted institutional coverage.
The justification should be transparent enough for readers to evaluate its methodological plausibility.
How Can You Conduct an AI-Assisted Information Power Assessment?
1. Define the qualitative research aim. Specify the phenomenon and level of analytical breadth.
2. Examine sample specificity. Determine how closely participants' experiences match the research question.
3. Review theoretical guidance. Evaluate whether established theory appropriately focuses the inquiry and analysis.
4. Examine dialogue quality. Review the relevance, depth, and richness of the actual interviews.
5. Clarify the analysis strategy. Identify whether the study emphasizes case-oriented interpretation, cross-case comparison, or another analytical purpose.
6. Identify remaining information gaps. Examine missing perspectives, insufficiently developed accounts, or unresolved analytical questions.
7. Make a researcher-led adequacy judgment. Determine whether additional data collection is necessary for the intended analysis.
8. Document the rationale. Explain how the five dimensions support the achieved sample and acknowledge remaining limitations.
This process uses the information power framework as a guide for methodological reasoning rather than a numerical scoring system.
Watch Out
Do not present an AI-generated information power percentage or numerical score as though it were part of Malterud and colleagues' original framework. The five dimensions require contextual interpretation, and no universal scoring equation establishes an exact participant requirement.
