Quantitative, Qualitative and Mixed-Methods Research

Overview

Quantitative, qualitative and mixed-methods research are broad approaches to generating health knowledge. They are not ranked levels of evidence, and they should not be selected solely according to researcher preference or whether the planned data are numbers or words.

Quantitative research is particularly useful for estimating quantities, comparing groups, examining measured relationships and evaluating predictions. Qualitative research is particularly useful for understanding experience, meaning, context, process and interaction. Mixed-methods research intentionally combines and integrates quantitative and qualitative strands when neither alone can adequately address the research problem [1-3].

The correct starting point is the question. A study may need to know how many, how much or whether an association exists; it may need to know how, why or what an experience means; or it may need both forms of knowledge in a connected design.

Why this matters

Using the wrong broad approach can produce an answer that is technically competent but irrelevant. A survey may estimate that 38% of patients miss appointments but not explain the practical and relational reasons. Interviews may richly describe barriers but cannot estimate their prevalence in the clinic population. Conducting both does not automatically solve the problem unless their findings are integrated.

Approach selection also affects sampling, instruments, researcher roles, analysis, quality assurance and reporting. A representative probability sample is valuable for population estimation, whereas a purposive qualitative sample may be designed to capture diverse or information-rich experiences. These are different logics, not strong versus weak versions of the same process.

Quantitative research

DESIGN PROFILE

Quantitative research

Core structure: Concepts are operationalised as variables and measured systematically so that quantities, differences, relationships or predictions can be estimated.

Best suited for: Prevalence, incidence, effect estimation, group comparison, diagnostic accuracy, prognosis, prediction and testing pre-specified hypotheses.

Important limitations: Complex meaning may be reduced by measurement; unmeasured context can be missed; bias is not removed by large samples; and statistical association does not automatically establish causation.

Quantitative research commonly uses surveys, experiments, structured observations, clinical measurements, laboratory tests, registries and electronic records. It requires clear definitions of the target population, variables, measurement timing, comparison structure and unit of analysis.

The output is usually numerical: a prevalence, mean difference, risk ratio, diagnostic accuracy measure, association or prediction performance estimate with uncertainty. The analysis belongs in the Biostatistics section, but the methodological design determines whether the estimate is meaningful.

Strengths

• Can estimate the frequency or magnitude of a phenomenon.
• Can compare groups or time periods using explicit metrics.
• Supports reproducible, standardised measurement.
• Can quantify uncertainty and assess generalisability when sampling supports it.
• Can evaluate predictions on new data.

Limitations

• Variables may not capture the full construct or lived experience.
• Standardised response options can impose the researcher’s assumptions.
• Large routine datasets may contain incomplete or poorly defined measurements.
• Statistical adjustment cannot guarantee control of all bias or confounding.

Qualitative research

DESIGN PROFILE

Qualitative research

Core structure: Researchers generate and interpret rich, contextualised data to understand meaning, experience, social processes, interaction or implementation.

Best suited for: Questions beginning with how, why or what it is like; poorly understood phenomena; intervention development; acceptability; implementation; and contextual explanation.

Important limitations: It is not designed to estimate population prevalence; findings depend on context and researcher interpretation; and quality requires reflexivity, analytic transparency and adequate depth.

Qualitative methods include in-depth interviews, focus groups, observation, field notes, documents and digital materials. Sampling is commonly purposive rather than probability based. Participants may be selected because their experiences are relevant, contrasting or information rich.

Analysis may involve thematic analysis, qualitative content analysis, grounded theory, narrative analysis or other approaches. The aim is not to convert every response into a count. It is to develop a defensible interpretation supported by the dataset, context and analytic process, with epistemology, methodology and method kept coherent [4].

Strengths

• Explores meanings and mechanisms not captured by fixed variables.
• Preserves context and participant perspective.
• Supports discovery when concepts are not yet well defined.
• Identifies practical barriers, unintended consequences and implementation processes.
• Can inform instrument, intervention or theory development.

Limitations

• Does not provide a statistically representative prevalence estimate.
• Data generation and interpretation are influenced by researcher–participant interaction.
• Poorly conducted studies may produce superficial themes disconnected from evidence.
• Transferability must be judged through contextual description rather than sample size alone.

Mixed-methods research

DESIGN PROFILE

Mixed-methods research

Core structure: Quantitative and qualitative strands are each conducted rigorously and purposefully integrated to address a shared research problem.

Best suited for: Problems requiring both magnitude and meaning, explanation of quantitative patterns, development and testing of an instrument or intervention, or evaluation of complex implementation.

Important limitations: Requires expertise and resources for both strands; integration can be difficult; and parallel studies without a shared inference are not genuine mixed methods.

Mixed-methods research is more than collecting two data types. The researcher must state why both are needed and how the strands interact. Integration can occur by:

• connecting, where results from one strand guide sampling for another;
• building, where one strand informs an instrument, intervention or data collection in the next;
• merging, where results are compared or combined during analysis or interpretation; or
• embedding, where one strand is nested within a larger design [2].

Common broad designs include convergent designs, explanatory sequential designs and exploratory sequential designs. Detailed mixed-methods designs are addressed in Part 12.

Comparing the approaches

Comparison by methodological purpose

illustrative

Table 1. Quantitative, qualitative and mixed-methods approaches

Feature	Quantitative	Qualitative	Mixed methods
Main purpose	Estimate, compare, associate or predict	Understand meaning, experience, context or process	Address a multifaceted problem through integrated evidence
Typical question	How many? How much? Is there a difference or association?	How? Why? What is the experience or process?	What is the pattern, and how can it be explained or acted upon?
Sampling logic	Often probability, consecutive, census or design-specific sampling	Often purposive, theoretical or information-oriented sampling	Sampling relationship may be parallel, nested or connected
Data	Structured measurements and variables	Text, talk, observation, images or documents	Both, with an explicit integration strategy
Quality emphasis	Validity, precision, bias control, reproducibility	Credibility, reflexivity, depth, coherence, contextual transparency	Quality of each strand plus quality of integration
Main output	Estimates and uncertainty	Interpretive findings, concepts, themes or theory	Meta-inferences connecting both strands

The table describes tendencies, not rules. Some quantitative studies are exploratory; some qualitative studies use structured frameworks; and mixed-methods designs vary substantially.

Choosing the approach

DECISION GUIDE

Select the broad approach from the question

1. Does the study need a numerical estimate, comparison, association or prediction?
2. Does it need contextual understanding of meaning, experience or process?
3. Would one approach leave an important part of the problem unanswered?
4. Can the team conduct each required strand rigorously?
5. Is there a specific plan for integration rather than parallel reporting?

A mixed-methods design should not be chosen merely because it appears comprehensive. It increases workload and requires coherent timing, sampling and integration. A focused single-method study is preferable when it adequately answers the question.

Sampling and quality differ by approach

Quantitative representativeness depends on the target population, sampling frame, recruitment and response. A large convenience sample can still be biased. Qualitative adequacy depends on whether the sample and data generation provide sufficient depth and diversity for the analytic purpose. The number of interviews alone does not establish quality.

In mixed methods, the relationship between samples matters. The same participants may contribute to both strands, a purposive subsample may be selected from quantitative findings, or different samples may represent different system levels. Researchers should explain how the samples support integration.

Interface with statistical analysis

Quantitative and mixed-methods studies require an analysis that respects the design and intended estimand. The statistical method is determined by the outcome type, comparison structure, timing, clustering, repeated observations and missing data.

Qualitative analysis is not a substitute for statistical analysis and vice versa. Counting how often a theme appears does not make it a population estimate. Conversely, a regression coefficient does not explain how participants experience the phenomenon unless the study has gathered evidence capable of addressing that question.

Worked example

WORKED EXAMPLE

Three approaches to retinal-screening uptake

Quantitative study: Use clinic records to estimate the proportion of eligible adults overdue for screening and model associations with age, duration of diabetes, clinic, distance and prior attendance. This identifies the distribution and correlates but may not explain causal mechanisms.

Qualitative study: Conduct interviews with patients who attended, delayed or declined screening and with relevant staff. Explore understanding of screening, referral communication, transport, fear, competing responsibilities and clinic workflow. This provides contextual explanation but not population prevalence.

Mixed-methods study: First analyse clinic records to identify uptake patterns and contrasting patient groups. Then purposively sample patients and staff for interviews. Integrate findings in a joint matrix to determine which measured patterns are explained, contradicted or expanded by the qualitative evidence. The integrated conclusion informs a targeted service intervention.

The best choice depends on the actual objective. If the service only needs a reliable estimate of overdue screening, the quantitative component may be sufficient. If the aim is to redesign the pathway, qualitative or mixed-methods evidence may be necessary.

Suggested Methods wording

Appraising broad methodological fit

APPRAISAL CHECKLIST

Questions for critical appraisal

• Does the selected approach answer the stated research question?
• Is the sampling logic appropriate for the intended claim?
• Are data-generation and analysis procedures rigorous within that approach?
• Does the report use quality criteria appropriate to the methodology?
• In mixed methods, are the strands connected and integrated into a combined inference?

Common mistakes

Knowledge check

A study asks how many patients experience treatment-related fatigue. Which broad approach is the most direct starting point?

Answer and reasoning

A quantitative approach is the direct starting point because the question seeks a numerical frequency or proportion. Qualitative work may be added if the study also seeks to understand how fatigue is experienced or managed.

Can a qualitative study be rigorous without a statistically representative sample?

Answer and reasoning

Yes. Qualitative rigour is judged through alignment, appropriate purposive sampling, depth, reflexivity, analytic transparency and evidence supporting interpretations. It is not intended to produce a population prevalence estimate.

What makes mixed-methods research different from using two methods?

Answer and reasoning

Mixed methods requires a justified quantitative strand, a justified qualitative strand and purposeful integration that produces a combined inference. Two unrelated methods in one project do not meet this requirement.

Why should mixed methods not be selected merely to make a study look comprehensive?

Answer and reasoning

It requires additional expertise, time, sampling decisions and integration. Without a genuine question that needs both forms of evidence, the design may become unnecessarily complex and each strand may be conducted superficially.

Key take-home messages

SUMMARY

Key take-home messages

• Select the broad research approach from the question and intended inference.
• Quantitative research estimates quantities, comparisons, relationships or predictions.
• Qualitative research develops contextual understanding of meaning, experience and process.
• Mixed methods requires rigorous strands, a clear rationale and purposeful integration.
• Sampling and quality should be judged according to the logic of the selected approach.

References

1. Curry LA, Krumholz HM, O’Cathain A, Plano Clark VL, Cherlin E, Bradley EH. Mixed methods in biomedical and health services research. Circ Cardiovasc Qual Outcomes. 2013;6(1):119–123. https://doi.org/10.1161/CIRCOUTCOMES.112.967885
2. Fetters MD, Curry LA, Creswell JW. Achieving integration in mixed methods designs—principles and practices. Health Serv Res. 2013;48(6 Pt 2):2134–2156. https://doi.org/10.1111/1475-6773.12117
3. National Institutes of Health Office of Behavioral and Social Sciences Research. Best practices for mixed methods research in the health sciences. 2nd ed. Bethesda (MD): National Institutes of Health; 2018. Available from: https://obssr.od.nih.gov/sites/g/files/mnhszr296/files/Best_Practices_for_Mixed_Methods_Research.pdf
4. Carter SM, Little M. Justifying knowledge, justifying method, taking action: epistemologies, methodologies, and methods in qualitative research. Qual Health Res. 2007;17(10):1316–1328. https://doi.org/10.1177/1049732307306927