A research method is not merely a technical preference. It determines what can be examined, what can reasonably be claimed, and how readers can challenge the resulting study. The central distinction is therefore not “numbers versus words.” It is a division between different kinds of evidence, analysis and explanatory claims.
The Office for Health Improvement and Disparities’ GOV.UK guidance on mixed methods provides a useful starting point: quantitative data is generally better at answering questions such as “What is the effect of your digital product?”, while qualitative data can show how and why those results occurred. That example is not the boundary of either approach, but it identifies the logic beneath them.
What Quantitative Research Actually Does
Data: Quantitative research begins with observations structured through defined measures. The relevant decisions concern which variables, outcomes and comparisons are needed, as well as how observations will be recorded so they can support the intended claim.
Analysis: Quantitative analysis organises those observations for comparison. Depending on the design, this may involve describing measured patterns, examining relationships between variables or assessing an intervention against a comparison. The analysis must be appropriate to the question rather than selected because a particular technique is available or familiar.
Claim: Quantitative work is well suited to an effect claim: what changed, in which direction and under what conditions. The GOV.UK mixed-methods guidance specifically identifies questions about the effect of a digital product as a case in which quantitative data is generally stronger. It does not follow that numerical results automatically establish an effect. The design must make that claim credible, and the analysis must be reported clearly enough to inspect.
The boundary matters. A precise-looking result does not automatically explain the process that produced it. Nor does collecting a large amount of quantitative material remove uncertainty about confounding, missing information or inappropriate study design.
What Qualitative Research Actually Does
Data: Qualitative research uses observations, accounts or other material capable of revealing experiences, practices and processes. Its value lies not in replacing numerical records with personal narratives, but in preserving evidence about how something happens and what it means in context.
Analysis: Qualitative analysis examines relationships within that material to develop an account of patterns, mechanisms, experiences or contextual conditions. The GOV.UK guidance characterises this as the approach that can show how and why a result was obtained.
Claim: Qualitative research is well suited to process and interpretation claims. It can connect an observed outcome to the activities, decisions or conditions surrounding it. It can also reveal experiences that a predetermined measure did not capture.
The boundary is equally important. An interpretation is not strengthened merely because it appears in quotations or seems plausible. Readers still need to assess whether the analysis supports the proposed explanation and whether the proposed explanation fits the study context. Qualitative evidence may contribute to an effect claim, but it does not acquire the same evidential role merely by being combined with a quantitative result.
When Mixed Methods Becomes the Relevant Choice
Mixed methods is not a compromise between opposing approaches. The GOV.UK guidance defines it as a study that combines quantitative and qualitative data collection and analysis. Its rationale is that the two approaches answer different questions, allowing the combined study to provide more in-depth findings.
Use mixed methods when an effect and its explanation are both necessary to the research problem. A quantitative component may establish the observed result; a qualitative component may examine how and why it occurred. The study must still explain what each component contributes. Combining unlike materials does not remove the need to justify either design.
How the Method Choice Shapes the Study
Method choice is a claim-strength decision.
The Cochrane Handbook, chapter 3, states that a review’s scope is defined by its populations, interventions and comparisons, and outcomes. It also says that included study designs must be justified in relation to appropriateness to the review question and potential for bias. The implication extends beyond evidence synthesis: the research object, comparison, outcome and design should form a coherent set.
Changing the method changes the study in several visible ways.
- It changes the available claim. A question centred on an effect calls for evidence capable of supporting an effect claim. A question centred on how or why calls for evidence that can preserve process and context.
- It changes the evidence a reader will demand. Quantitative claims require attention to measurement, comparison, analysis and sources of bias. Interpretive claims require attention to how the material was obtained and how the interpretation was developed.
- It changes the basis for criticism. A reader cannot assess a design using a checklist designed for a different design. The Cochrane Handbook, chapter 25, gives non-randomised studies their own risk-of-bias domains, illustrating that each design must be appraised through criteria relevant to its own weaknesses.
For intervention comparisons, the Cochrane Handbook, chapter 3, states that randomisation is the only way to prevent systematic differences between baseline characteristics across intervention groups in terms of both known and unknown factors. That is a specific strength within a particular design; it is not a guarantee that every later stage of a study is free from bias.
Why Reporting and Bias Are Part of the Choice
A defensible method is not enough if the study reporting it is too unclear to appraise. The Cochrane Handbook, chapter 7, notes that essential information for assessing risk of bias is frequently missing and describes domain-based assessment rather than reliance on a single overall quality scale.
The chapter reports that, across 20,920 randomised trials included in 2001 Cochrane reviews, the percentages judged to be at unclear risk of bias were 49% for random sequence generation, 57% for allocation sequence concealment, 31% for blinding and 25% for incomplete outcome data. The Handbook, citing Dechartres and colleagues from 2017, also reports that more recent trials were less likely to be judged at unclear risk of bias, suggesting that reporting has improved over time.
These figures show why “the study used quantitative data” is an incomplete description. The relevant questions concern the particular design, the domain being assessed, the information reported and the judgement that can properly be made from it. Percentage figures are not substitutes for examining a study’s methods.
The same principle applies to qualitative and non-randomised work. A named approach identifies the starting point for appraisal, not completion of it.
A Decision Path You Can Apply Tonight
Start by writing the claim you want the study to support. Then follow the question rather than the method label.
- If the question is “What is the effect of X on Y for population P?”, begin with a quantitative approach. Specify the population, intervention, comparison and outcome. Then ask which design can support that claim and what sources of bias remain within it.
- If the question is “How is X experienced in context C, and why does that process occur?”, begin with a qualitative approach. Identify the material needed to examine experience and process, then check whether the proposed analysis can support the interpretation rather than merely illustrate it.
- If the question needs both an effect and an explanation, consider mixed methods. State what the quantitative component must establish and what the qualitative component must explain. Combining data collection and analysis is justified by the combined question.
- If some intervention or outcome is difficult to address in randomised trials, do not select a design by default. The Cochrane Handbook, chapter 3, says authors should justify whether to restrict a review to randomised trials or include non-randomised studies. State the reason and assess the bias relevant to the chosen design.
- If the proposed claim does not fit the data, revise the claim before searching for evidence. A measure may not support the intended outcome; an experience may not answer the effect question; a broad explanation may exceed the accessible context.
For each candidate research direction, ask: What data would count? What analysis would change or explain those data? What claim would that analysis warrant? What would an informed reader challenge? If any answer remains unclear, the direction is not yet checkable.
From Method Choice to Critical Reading
Once the design is chosen, the next problem is reading the studies that report it. Begin with fit: population, intervention, comparison and outcome. Then inspect the design, the relevant bias domains and any missing information. Finally, compare the authors’ conclusion with what the design and reported analysis can support.
On Any.ac, the useful result is not a black-box verdict but a research direction whose object, method and scale can be inspected. That makes method choice an opening move rather than an endpoint—and places critical study appraisal at the centre of the next research decision.