Mixed Methods and Triangulation

Introduction

Business problems rarely fit within a single form of evidence. A customer churn model may identify which variables predict cancellation, but interviews may reveal that customers leave because a policy is perceived as unfair. A survey may show declining employee engagement, while observations explain how workload, supervision, and informal norms produce that decline. Quantitative evidence is often strong for estimating prevalence, association, difference, and prediction. Qualitative evidence is often strong for understanding meaning, process, context, and experience.

Mixed-methods research deliberately integrates quantitative and qualitative approaches within one study or programme of inquiry. It is not simply the use of two methods side by side. Its defining feature is integration: the researcher connects the datasets, analyses, interpretations, and conclusions so that the combined inquiry produces insights that neither strand could provide independently.

Triangulation is a related strategy for increasing the credibility and explanatory power of findings by examining a phenomenon from multiple perspectives. The perspectives may involve different methods, data sources, researchers, theories, or settings. Triangulation does not mean that all sources must produce identical results. Agreement can strengthen confidence, while well-understood disagreement can reveal complexity, boundary conditions, measurement problems, or competing explanations.

For business analysts, mixed methods are especially valuable when decisions involve both measurable performance and human behaviour. This chapter explains the logic of mixed methods, major designs, integration procedures, triangulation, conflict resolution, meta-inferences, quality criteria, ethics, and practical implementation.

Learning Objectives

By the end of this chapter, you should be able to:

1. Explain the rationale, philosophical foundations, and terminology of mixed-methods research.
2. Distinguish convergent, explanatory sequential, and exploratory sequential designs.
3. Select an appropriate design for a business analytics problem.
4. Plan integration through connecting, building, merging, and embedding.
5. Apply triangulation without assuming that agreement is always required.
6. Analyse and explain conflicting quantitative and qualitative findings.
7. Develop defensible meta-inferences from integrated evidence.
8. Evaluate quality, ethics, feasibility, and reporting requirements in mixed-methods studies.

1. The Logic of Mixed-Methods Research

1.1 What mixed methods means

A mixed-methods study collects, analyses, and integrates both quantitative and qualitative data in response to a research question or set of linked questions. Quantitative data may include transaction records, survey responses, experiments, operational metrics, or structured observations. Qualitative data may include interviews, focus groups, field notes, documents, open-ended responses, or digital traces interpreted in context.

Three elements distinguish mixed methods from merely conducting several analyses:

• Methodological plurality: both quantitative and qualitative methods are intentionally used.
• Methodological priority and timing: the researcher specifies whether one strand is dominant, whether both have equal status, and whether data are collected concurrently or sequentially.
• Integration: findings are brought into relationship to produce a combined interpretation.

A report containing a statistical analysis in one chapter and interviews in another is not necessarily a mixed-methods report. Integration might occur when interview results explain an unexpected statistical pattern, when qualitative themes are used to construct a survey instrument, or when both results are compared in a joint display.

1.2 Why integrate evidence?

Mixed methods can serve several purposes:

• Complementarity: different methods illuminate different aspects of a phenomenon.
• Development: findings from one strand inform the design of another.
• Expansion: the study addresses distinct but related questions, such as both performance outcomes and implementation experiences.
• Initiation: contrasting findings stimulate a new interpretation or research question.
• Confirmation: convergent findings increase confidence in a conclusion.

The aim is not to create an artificial average of different kinds of data. A percentage and a narrative account are not interchangeable measurements. Instead, integration asks how the forms of evidence relate and what conclusion is justified when they are considered together.

1.3 Philosophical foundations

Mixed-methods researchers commonly use pragmatism, which focuses on the research problem and the consequences of using particular methods. A pragmatic researcher may accept that numerical patterns and interpreted meanings answer different parts of a question, provided that the design is coherent and the conclusions are warranted.

Other philosophical positions are also possible. A critical realist may treat observed data as imperfect indicators of underlying mechanisms and use both statistical patterns and contextual accounts to investigate those mechanisms. A transformative approach may combine methods to examine inequality and support the interests of marginalized groups. A constructivist approach may emphasize that qualitative meanings are co-produced by participants and researchers, while still accepting a limited quantitative component for practical purposes.

The important issue is philosophical coherence. Researchers should explain what they believe counts as evidence, how knowledge is produced, and why the selected methods are appropriate. Mixed methods is not a licence to combine incompatible assumptions without reflection.

2. Core Designs

Design notation helps communicate timing and emphasis. In the notation below, QUAN represents a quantitative strand and QUAL a qualitative strand. Capital letters indicate greater priority; a plus sign indicates concurrent collection; an arrow indicates sequence.

2.1 Convergent design

In a convergent design, quantitative and qualitative data are collected during approximately the same phase, analysed separately, and then merged for comparison and interpretation.

A typical structure is:

QUAN + QUAL → integration → meta-inference

For example, a retailer may analyse customer satisfaction scores and conduct interviews during the same quarter. The survey estimates the prevalence of dissatisfaction, while interviews identify the experiences behind it. The strands may converge, complement one another, or diverge.

Convergent designs are suitable when:

• both forms of evidence are important at the same time;
• the researcher wants a broad and contextual account of a current issue;
• neither strand necessarily depends on the other for instrument development or sampling;
• time permits parallel data collection.

A major risk is that the researcher treats the two strands as independent mini-studies and performs integration only in the conclusion. Integration should be planned before data collection, including the constructs to compare, the populations represented, and the standards for interpreting agreement and disagreement.

2.2 Explanatory sequential design

An explanatory sequential design begins with quantitative research and follows with qualitative research to explain, elaborate, or investigate the quantitative results.

The structure is:

QUAN → QUAL → integration

Suppose a logistics company finds that delivery delays are statistically associated with a particular distribution centre. Interviews with drivers, supervisors, and planners may explain whether the association reflects scheduling practices, equipment failures, local geography, or differences in reporting.

The qualitative sample is often selected from the quantitative sample. Researchers may purposefully select typical cases, extreme cases, outliers, or groups with contrasting scores. This is called connecting the strands because results from the first phase influence participant selection or questions in the second phase.

This design is useful when numerical results identify an important pattern but do not explain its meaning or mechanism. It is less suitable when the quantitative instrument fails to capture the relevant concepts in the first place.

2.3 Exploratory sequential design

An exploratory sequential design begins with qualitative inquiry and uses its results to develop a quantitative instrument, intervention, classification, or test.

The structure is:

QUAL → QUAN → integration

For example, a bank may interview small-business customers to understand how they define “digital service quality.” Researchers can use the themes and language from those interviews to construct survey items, pilot the instrument, and then test the prevalence or structure of the resulting dimensions in a larger sample.

This design is useful when:

• a construct is poorly understood;
• existing scales were developed in another culture, industry, or population;
• the researcher needs participant-generated categories;
• a qualitative intervention or programme model must be tested quantitatively.

The central danger is premature quantification. Themes should be operationalized carefully, with attention to conceptual coverage, wording, response scales, reliability, validity, and whether frequency of mention should be interpreted as importance.

2.4 Embedded and multiphase designs

An embedded design places one type of data within a larger design dominated by the other. For example, a randomized training evaluation may be primarily quantitative but embed interviews to understand implementation fidelity and participant experience. The embedded strand answers a subsidiary question and should not be presented as if it had the same sampling logic as the main strand.

A multiphase design links several studies over time. A programme may begin with exploratory interviews, proceed to survey development, evaluate an intervention quantitatively, and then conduct follow-up interviews. Multiphase work is common in organizational change and product development but requires careful documentation of how decisions in one phase affect later phases.

2.5 Choosing a design

Design selection should follow the research problem rather than habit. The following questions are useful:

Researchers must also assess access, budget, skills, timeline, participant burden, and the feasibility of integrating the evidence. A theoretically attractive design that cannot be implemented is not a defensible design.

3. Planning Integration

Integration is the methodological centre of a mixed-methods study. It can occur at four connected points.

3.1 Integration through connecting

Connecting occurs when results from one strand guide sampling, instrument development, case selection, or questions in another strand. In explanatory research, survey scores may identify interview participants. In exploratory research, qualitative categories may become survey variables.

Connecting requires traceability. The researcher should record why particular cases were selected and whether the second sample represents the full first sample or only selected segments.

3.2 Integration through building

Building occurs when one strand constructs a tool or procedure for another. Interview findings may build a questionnaire, coding framework, experimental stimulus, or implementation checklist. Conversely, quantitative results may build an interview protocol focused on significant predictors or unusual cases.

A built instrument should be tested. Qualitative statements should not automatically become survey items; they require review for ambiguity, redundancy, double-barrelled wording, and conceptual fit.

3.3 Integration through merging

Merging involves bringing results together during analysis or interpretation. A useful device is a joint display, such as a matrix that places statistical results beside qualitative themes for the same groups or cases.

A joint display makes the logic visible. It prevents the researcher from selecting only quotations that support the statistical findings and encourages comparison across cases.

3.4 Integration through embedding

Embedding occurs when one strand is nested within the procedures of another. For instance, observational field notes may be collected during a controlled implementation study. The researcher must specify the role of the embedded evidence: explaining delivery, assessing acceptability, identifying contamination, or documenting context.

3.5 Integration at the interpretation stage

The final interpretation should distinguish:

1. what each strand found independently;
2. how the findings relate;
3. what conclusion follows from their relationship;
4. what remains uncertain.

A useful reporting structure is a meta-inference: an overall conclusion derived from the integrated evidence. A meta-inference is defensible only when the underlying data are appropriate, the analyses are credible, the strands address related questions, and the reasoning linking them is explicit.

4. Triangulation

4.1 Forms of triangulation

Triangulation can be classified into several forms:

• Data triangulation: using different people, times, locations, or data sources.
• Method triangulation: using different methods, such as a survey and interviews.
• Investigator triangulation: involving multiple analysts or researchers.
• Theory triangulation: interpreting findings through more than one theoretical lens.
• Environmental or setting triangulation: examining the issue across organizational or operational contexts.

Mixed methods most visibly uses method triangulation, but triangulation does not require both quantitative and qualitative data. Two survey instruments, several administrative databases, or multiple qualitative methods may also be triangulated.

4.2 Agreement, complementarity, and contradiction

Triangulation outcomes can be classified as follows:

• Convergence: different sources support a similar conclusion.
• Complementarity: sources address different dimensions of the same issue.
• Expansion: one source extends the scope of another.
• Discordance: sources appear inconsistent.

Convergence can increase confidence, but agreement is not proof. Two sources may share the same bias, use the same flawed operational definition, or be dependent on one another. Conversely, discordance is not automatically a failure. It may show that an outcome varies by subgroup, time, context, or level of analysis.

4.3 Quality of triangulation

Good triangulation requires more than collecting multiple sources. The sources should be relevant to the claim, sufficiently independent where confirmation is intended, and analysed with appropriate procedures. Researchers should state whether triangulation is being used for confirmation, explanation, discovery, or critique.

5. Resolving Conflicting Findings

Conflicting findings are common. A survey may indicate high satisfaction while interviews reveal strong complaints. An experiment may show no average effect while some participants describe substantial benefit. Resolving such conflict requires investigation, not forced reconciliation.

5.1 Check whether the findings address the same construct

Terms such as satisfaction, trust, engagement, and performance may be operationalized differently. A single survey item may measure an overall evaluation, whereas interviews may elicit specific incidents. The apparent conflict may reflect construct mismatch.

5.2 Check population and sampling differences

The quantitative sample may represent the full customer base, while interviewees may have been recruited from complainants. Neither is necessarily wrong. The findings may describe different populations. Researchers should compare inclusion criteria, response rates, missing data, and nonresponse patterns.

5.3 Check timing and change

Data collected at different times can legitimately disagree. A policy change, service disruption, or economic event may alter experiences between phases. Time should be treated as an explanatory variable rather than ignored.

5.4 Examine measurement and response processes

Survey respondents may avoid extreme answers, misunderstand a question, or answer according to a general impression. Interview participants may emphasize memorable events, seek recognition, or interpret questions through a particular relationship with the organization. Administrative data may contain coding errors or incomplete records.

5.5 Analyse subgroups and cases

An average effect can conceal heterogeneous effects. Researchers may stratify quantitative analysis by relevant groups and compare qualitative themes across those groups. Outlier cases deserve special attention: they may be errors, rare exceptions, or theoretically important cases.

5.6 Use a conflict-resolution matrix

The outcome may be a qualified conclusion: for example, “Overall satisfaction remained high, but interviews and subgroup analysis indicate that new customers experienced serious difficulties during onboarding.” This is more informative than declaring one strand right and the other wrong.

6. From Findings to Defensible Meta-Inferences

A meta-inference is a conclusion about the phenomenon that depends on the integration of quantitative and qualitative evidence. It should not merely repeat both results.

A defensible meta-inference has four characteristics:

1. Traceability: the reader can follow the conclusion back to specific data and analyses.
2. Coherence: the interpretation fits the research questions and design.
3. Qualification: scope, uncertainty, exceptions, and alternative explanations are stated.
4. Practical usefulness: the conclusion supports an action, theory, measurement decision, or future inquiry without exceeding the evidence.

Researchers can develop meta-inferences through a structured sequence:

1. State the independent quantitative finding.
2. State the independent qualitative finding.
3. Compare points of convergence and divergence.
4. Examine context, timing, measurement, and subgroup explanations.
5. Formulate an integrated proposition.
6. Identify evidence that would weaken the proposition.
7. Translate the proposition into a cautious implication.

For example, a churn model may identify price as a predictor of cancellation. Interviews may show that customers rarely mention price directly; instead, they describe loss of value after features were removed. The meta-inference might be that price sensitivity is partly a value-perception mechanism, suggesting that retention efforts should address feature utility rather than offer discounts alone.

7. Quality and Rigour

Mixed-methods quality includes the quality of each strand and the quality of their integration.

7.1 Quantitative quality

Researchers should consider construct validity, reliability, sampling error, statistical conclusion validity, missing data, model assumptions, and the distinction between association and causation.

7.2 Qualitative quality

Important criteria include credibility, dependability, confirmability, and transferability. Practical techniques include purposeful sampling, reflexive documentation, transparent coding, negative-case analysis, participant engagement where appropriate, and comparison among analysts.

7.3 Integration quality

Integration quality concerns whether the design logically connects the strands, whether the samples and constructs are compatible, whether contradictory evidence is addressed, and whether the meta-inference is supported. A study can have excellent survey statistics and rich interviews but weak integration.

A mixed-methods protocol should therefore specify the integration point before data collection. The protocol should include a design diagram or written sequence, priority of strands, sampling connections, analytic procedures, joint displays, conflict-resolution procedures, and criteria for drawing conclusions.

8. Ethics, Governance, and Data Protection

Mixed-methods studies may increase participant burden because people may complete surveys, participate in interviews, or permit access to records. Consent materials should explain the stages of the research and whether data will be linked across stages.

Linkage creates confidentiality risks. A participant who appears anonymous in a survey may become identifiable when a rare score is connected to an interview quotation. Researchers should use pseudonyms, role-based access, data minimization, secure storage, and a documented retention schedule. Sensitive organizational findings may affect employment, reputation, or commercial relationships.

Researchers must also avoid treating qualitative participants as merely illustrative quotations for a predetermined statistical story. Participants’ accounts should be analysed fairly, including evidence that challenges the dominant interpretation. If AI tools assist with transcription, coding, summarization, or pattern detection, the researcher remains responsible for confidentiality, accuracy, bias assessment, audit trails, and human verification. Sensitive data should not be uploaded to systems that lack appropriate governance or contractual protection.

9. Applied Case: Improving Digital Banking Retention

The following is an illustrative case, constructed for teaching purposes.

Background

A regional bank observes that digital-only customers have a higher annual churn rate than branch-supported customers. Management wants to know whether the problem is price, usability, trust, service recovery, or customer expectations.

Problem and design

The research team chooses an explanatory sequential design. First, it analyses twelve months of account and usage data and administers a satisfaction survey. Second, it conducts interviews with carefully selected customers to explain the quantitative patterns.

The quantitative phase identifies three patterns: churn is associated with low frequency of successful log-ins, repeated failed payment attempts, and low perceived value. However, the model does not establish why these factors occur or whether the associations apply equally to all customers.

For the qualitative phase, the team selects customers from four groups: high-use customers who stayed, high-use customers who left, low-use customers who stayed, and low-use customers who left. This purposive maximum-variation sampling allows the team to examine contrasting experiences rather than interviewing only recent complainants.

Analysis

Interviews reveal that many customers who report “low value” are not primarily objecting to fees. They believe that important features were difficult to find and that payment failures were not explained clearly. Customers who stayed despite low usage often had a branch adviser who helped them during onboarding. Customers who left described failed payments as evidence that the bank was unreliable, even when the technical failure was temporary.

The team creates a joint display linking customer segment, behavioural data, survey results, and interview themes. It discovers that the quantitative variable “low usage” combines two different situations: customers who rarely need banking services and customers who attempted to use the application but abandoned it after failed experiences.

Integrated solution

The bank does not respond with a general discount. It redesigns onboarding, improves payment-failure messages, introduces a guided feature tour, and creates an escalation route for repeated failures. A follow-up survey measures usability and trust, while operational data track successful payment completion and support contacts. Interviews are repeated with a smaller sample to assess whether the redesign changed customer experience.

Outcome and learning points

The integrated evidence produces a more precise conclusion: digital churn is not explained by usage volume alone. For some customers, low use reflects low need; for others, it reflects unsuccessful encounters that reduce trust. The qualitative evidence explains the mechanism behind a statistical association and prevents management from implementing an ineffective price-based intervention.

The case demonstrates five lessons:

1. Quantitative analysis can identify where explanation is needed.
2. Purposeful selection can reveal contrasting mechanisms.
3. A joint display can expose problems in variable interpretation.
4. Conflicting or incomplete measures may indicate construct heterogeneity.
5. Meta-inferences can guide more targeted operational action.

10. Contemporary Developments and Best Practices

Digital platforms create new opportunities for mixed methods. Clickstream data can be combined with usability interviews; customer sentiment can be compared with service records; and experiments can be followed by implementation observation. Nevertheless, large datasets do not automatically provide representative or meaningful evidence. Digital traces reflect platform design, access, and user behaviour under particular conditions.

Best practice includes:

• begin with an integrated research question rather than two unrelated questions;
• justify the design, priority, timing, and philosophical assumptions;
• define constructs consistently across strands;
• plan sampling connections before collection;
• analyse each strand competently before integration;
• use joint displays to make comparisons auditable;
• investigate disagreement rather than suppressing it;
• distinguish evidence, interpretation, and recommendation;
• report limitations, including the limits of generalization;
• preserve an audit trail of decisions, transformations, and coding;
• involve domain stakeholders without allowing organizational preferences to determine findings.

Chapter Summary

Mixed-methods research integrates quantitative and qualitative evidence to answer complex research questions. Its value lies not in simply using two methods but in connecting, building, merging, or embedding the strands so that an integrated conclusion is possible. Convergent designs collect both strands concurrently; explanatory sequential designs use qualitative evidence to explain quantitative results; exploratory sequential designs use qualitative findings to develop quantitative instruments or tests. Embedded and multiphase designs support more specialized or extended inquiries.

Triangulation compares evidence across methods, sources, researchers, theories, or settings. Convergence can strengthen confidence, while complementarity and discordance can reveal context, heterogeneity, measurement differences, or mechanisms. Conflicting findings should be examined through construct comparison, sample assessment, timing, measurement analysis, subgroup analysis, and rival explanations.

A defensible meta-inference is traceable, coherent, qualified, and useful. Rigour requires quality in each strand and in their integration. Ethical practice requires careful consent, confidentiality, linkage governance, fair representation of participant accounts, and human oversight when technology or AI supports analysis. For business analysts, mixed methods provides a disciplined way to connect performance patterns with the experiences and processes that produce them.

Key Terms

Mixed-methods research: Research that intentionally collects, analyses, and integrates quantitative and qualitative evidence.

Quantitative strand: The numerical component of a mixed-methods study.

Qualitative strand: The interpretive component using non-numerical, contextual, or textual evidence.

Integration: The process of connecting findings, datasets, analyses, or interpretations across strands.

Convergent design: A design in which quantitative and qualitative data are collected concurrently and merged during interpretation.

Explanatory sequential design: A quantitative-first design followed by qualitative inquiry to explain numerical findings.

Exploratory sequential design: A qualitative-first design followed by quantitative development or testing.

Embedded design: A design in which one strand is nested within a larger strand or study.

Multiphase design: A series of linked mixed-methods phases conducted over time.

Connecting: Using results from one strand to guide sampling or procedures in another.

Building: Using one strand to develop an instrument, intervention, or analytic procedure for another.

Merging: Bringing quantitative and qualitative results together for direct comparison.

Triangulation: Examining a claim through multiple methods, sources, researchers, theories, or settings.

Convergence: Agreement among relevant sources of evidence.

Discordance: A meaningful difference or inconsistency among findings.

Joint display: A table or visual structure that presents linked quantitative and qualitative findings.

Meta-inference: An integrated conclusion derived from the relationship between strands.

Maximum-variation sampling: Purposeful selection of cases that differ on relevant characteristics.

Reflexivity: Examination of how the researcher’s position and decisions influence the inquiry.

Construct heterogeneity: The condition in which one measured variable combines substantively different experiences or meanings.

Review Questions

Short Answer Questions

1. What distinguishes mixed-methods research from conducting separate quantitative and qualitative studies?
2. Define a convergent design and identify one situation in which it is appropriate.
3. What is the purpose of a joint display?
4. Name three forms of triangulation.
5. What is a meta-inference?

Long Answer Questions

1. Compare explanatory sequential and exploratory sequential designs, including their purposes, timing, sampling implications, and limitations.
2. Explain how a researcher should investigate apparently conflicting findings from a survey and interviews.
3. Discuss the role of pragmatism in mixed-methods research and explain why philosophical coherence remains important.
4. Describe how integration can occur through connecting, building, merging, and embedding.
5. Evaluate the ethical and governance issues involved in linking survey responses, interviews, and administrative records.

Analytical / Application Questions

1. A retailer’s survey shows high satisfaction, but interviews with customers contain repeated complaints about delivery failures. Design a mixed-methods investigation to determine whether the findings truly conflict.
2. A university wants to understand why an analytics training programme improves performance for some employees but not others. Select a mixed-methods design and justify the sampling, integration, and meta-inference strategy.
3. A bank’s churn model identifies low application usage as a predictor of cancellation. Explain how qualitative follow-up could distinguish low need from failed user experiences and recommend an integrated decision process.

Multiple Choice Questions with Answers

Question 1

Which feature most clearly distinguishes mixed-methods research from using two methods independently?

A. The study includes a large sample.

B. The study uses statistical software.

C. The quantitative and qualitative strands are integrated.

D. The study collects data at more than one time.

Correct Answer: C. The quantitative and qualitative strands are integrated.

Question 2

Which design begins with quantitative data collection and follows with qualitative inquiry to explain the results?

A. Convergent design

B. Explanatory sequential design

C. Exploratory sequential design

D. Embedded qualitative design

Correct Answer: B. Explanatory sequential design

Question 3

A researcher interviews users first and then develops a survey instrument from the resulting themes. Which design is being used?

A. Exploratory sequential design

B. Explanatory sequential design

C. Convergent design

D. Cross-sectional design

Correct Answer: A. Exploratory sequential design

Question 4

What is the main purpose of a joint display?

A. To replace qualitative coding

B. To display only statistically significant results

C. To present linked quantitative and qualitative findings for comparison

D. To calculate the response rate

Correct Answer: C. To present linked quantitative and qualitative findings for comparison

Question 5

Which example illustrates connecting the quantitative and qualitative strands?

A. Using survey results to select interview participants

B. Collecting two unrelated datasets

C. Deleting qualitative cases that disagree with a model

D. Reporting methods in separate publications

Correct Answer: A. Using survey results to select interview participants

Question 6

Which statement about triangulation is most accurate?

A. All sources must produce identical findings.

B. It can involve methods, data sources, investigators, or theories.

C. It is used only with experiments.

D. It eliminates all research bias.

Correct Answer: B. It can involve methods, data sources, investigators, or theories.

Question 7

A survey measures overall satisfaction, while interviews focus on specific service failures. An apparent disagreement may result from differences in:

A. Construct operationalization

B. Statistical significance thresholds only

C. Random assignment

D. Citation style

Correct Answer: A. Construct operationalization

Question 8

Which characteristic is essential to a defensible meta-inference?

A. It ignores exceptions to produce a simple conclusion.

B. It is traceable to the evidence and appropriately qualified.

C. It gives priority to numerical findings in every study.

D. It combines all results into one average score.

Correct Answer: B. It is traceable to the evidence and appropriately qualified.

Question 9

In an embedded design, the embedded strand usually:

A. Has no relationship to the main study

B. Is nested within and addresses a subsidiary purpose of the main design

C. Must always be qualitative

D. Must always be collected before the main strand

Correct Answer: B. Is nested within and addresses a subsidiary purpose of the main design

Question 10

Why might a researcher purposefully select both extreme and typical cases for interviews after a survey?

A. To avoid analysing the survey data

B. To explore contrasting experiences and possible mechanisms

C. To guarantee statistical representativeness

D. To eliminate the need for consent

Correct Answer: B. To explore contrasting experiences and possible mechanisms