Quantitative evidence shows the scale, frequency or pattern of a people issue; qualitative evidence helps explain the meaning, experience and mechanism behind that pattern. Neither is inherently stronger. The right evidence depends on the decision, the uncertainty involved and the consequences of being wrong.
A people dashboard might show that early turnover has increased. That is valuable quantitative evidence, but it does not tell leaders whether employees are leaving because the role was misrepresented, the workload is unsustainable, managers lack capability, induction is weak or external pay has shifted. Interviews, listening groups and exit narratives can add the missing explanation. Conversely, a few powerful stories may identify a concern but cannot show whether it is widespread. Good people practice uses both forms of evidence deliberately.
What each type of evidence contributes
| Evidence type | Typical forms | Best contribution | Main limitation |
| Quantitative | Headcount, turnover, absence, survey scores, pay data, time to competence | Shows scale, variation, change over time and potential relationships | May hide experience, context or cause |
| Qualitative | Interviews, focus groups, observations, open comments, case notes | Explains why and how people experience a process or decision | May not represent the wider workforce |
| Mixed methods | Survey patterns followed by focus groups; operational data paired with manager interviews | Combines breadth and explanation | Requires careful sequencing and integration |
CIPD notes that qualitative data can be useful for understanding the “what”, “why” and “how” of an issue, while quantitative data supports analysis of numerical patterns (CIPD, n.d.). The distinction is practical rather than academic: each type answers different questions.
Start with a decision question
The question should determine the method. If a people director needs to know whether turnover is concentrated by role, tenure or location, quantitative data is necessary. If the organisation needs to understand why recently promoted managers report reduced confidence, qualitative interviews or structured conversations are likely to be needed. If both scale and explanation matter, use a mixed approach.
| Decision question | Evidence approach | Reason |
| Which teams have the highest early turnover? | Quantitative analysis of leavers, tenure, role and location | Identifies pattern and priority population |
| Why are employees leaving those teams? | Exit themes, interviews, manager and employee listening | Explores experience and plausible mechanisms |
| Did a new induction process improve outcomes? | Cohort data plus new-starter feedback | Tests outcome and experience together |
| Is flexible working experienced fairly? | Survey segmentation, policy data and focus groups | Reveals both distribution and lived impact |
| Which development intervention should be scaled? | Capability measures, work-application evidence and participant insight | Avoids treating attendance as evidence of impact |
A mixed-methods approach should not simply place two datasets side by side. The analysis should show how one form of evidence informs the other. For example, survey data may identify a lower belonging score among remote staff; conversations can then test whether access to informal information, manager contact, career visibility or home-working conditions explain the pattern.
Quality in quantitative evidence
Quantitative evidence needs defined populations, consistent measures and sufficient context. A change in absence rate could be caused by seasonality, policy changes, workforce restructuring, data-capture differences or a genuine shift in health and workload. Before interpreting a result, check the denominator, period, exclusions and data completeness.
| Quality question | Why it matters |
| Who is included and excluded? | A rate may differ materially if agency workers, leavers or part-time staff are treated differently |
| Is the definition stable? | A new coding rule can create a false trend |
| Is the sample large enough? | Small groups can produce unstable percentages and privacy risks |
| What is the comparator? | A number needs a prior period, relevant group or benchmark to be meaningful |
| What else changed? | Organisational events may explain apparent relationships |
The correct response to a pattern is often “investigate further”, not “act immediately”. See correlation, causation and bias in people analytics for how to avoid treating associations as proof.
Quality in qualitative evidence
Qualitative work requires equal discipline. Questions should be open enough to allow genuine insight but focused enough to support the decision. Researchers should record how participants were recruited, what prompts were used, how themes were identified and what perspectives may be missing. A leader should not select only quotations that support an existing view.
| Risk | Better practice |
| Senior voices dominate discussion | Use facilitated sessions, separate groups where appropriate and accessible channels |
| Comments are treated as proof | Compare themes across participants and alongside other evidence |
| Employees fear being identified | Explain confidentiality, remove identifying detail and report themes carefully |
| Analysts over-interpret a single phrase | Check context and seek disconfirming as well as confirming evidence |
| Focus groups replace representative data | Use them to explain patterns, not claim organisation-wide prevalence |
Qualitative evidence is particularly important when a people issue involves trust, inclusion, fairness, leadership or change. These experiences cannot be fully understood through a single numerical score.
✅ Get Your Assignment Done by Experts
Workplace application: Rivermark Housing
Fictional Rivermark Housing sees a fall in employee engagement after introducing a new case-management system. Its executive team initially decides that more training is needed. A mixed-methods investigation produces a different picture.
| Evidence | Finding | Interpretation |
| Survey results | Confidence with the system is lower among mobile officers and newer staff | Experience varies by role and tenure |
| System data | Mobile officers have more failed log-ins and slower case completion | Access and workflow may be contributing factors |
| Listening groups | Officers report that training examples do not reflect field conditions and that devices lose connection | The issue is design and infrastructure, not only individual capability |
| Manager interviews | Managers lack a clear route to escalate recurring technical barriers | Governance and response speed need improvement |
Rivermark redesigns field-based learning, improves device support and creates a technical escalation route. It then monitors system access, case completion, employee confidence and service quality. The decision is stronger because it combined numerical trends with staff experience and operational evidence.
Sampling, inclusion and ethics
Both forms of evidence can exclude people. A survey sent only through a desktop portal may under-represent front-line workers. A focus group held during standard office hours may exclude shift workers or carers. A dataset with incomplete demographic information may make inequity invisible. Plan access and representation from the outset.
Ethical evidence use also requires proportionality. Gather only the information needed for a legitimate purpose; protect confidentiality; explain how information will be used; and avoid publishing results that identify small groups. The people analytics ethics guide examines these obligations in more detail.
Triangulation, disagreement and implementation
When different evidence sources disagree, that is often the most useful finding. Rivermark might find that survey scores indicate confidence is recovering while field-based interviews still describe barriers to system use. Rather than choosing the more convenient result, leaders should ask whether the evidence refers to different groups, time periods or aspects of the experience. Divergence can reveal implementation gaps that an average score hides.
Triangulation means using several sources to test a conclusion, not forcing them to say the same thing. Before implementing an intervention, record the working explanation, the evidence supporting it, the main uncertainty and the indicators that will be reviewed. This creates an accountable learning loop: if the expected changes do not occur, the organisation can revisit its assumptions rather than blame employees or dismiss the evidence. It is a practical way to turn mixed evidence into better people decisions.
Integrating evidence into a decision
A simple integration table can prevent one evidence source from overpowering the others.
| Evidence source | What it suggests | Confidence and limitation | Decision implication |
| Workforce data | Early turnover is highest at months three to six | Strong pattern, but no causal explanation | Investigate onboarding and role experience |
| New-starter interviews | Role expectations differ from actual workload | Rich insight, but limited participant group | Test themes with a wider survey or manager data |
| Research evidence | Realistic job previews can improve fit and retention | Transferability must be checked locally | Pilot clearer job information and review results |
| Manager expertise | Workloads vary sharply between sites | May be affected by local perception | Compare workload and staffing data by site |
This is evidence-based practice in action: use research, organisational data, stakeholder concerns and professional judgement together, as explained in the four sources of evidence guide.
Designing proportionate evidence collection
The evidence design should match the scale and risk of the decision. A local change to a team meeting routine may need short feedback conversations and basic operational measures. A major redesign of work, reward or workforce structure may justify a wider survey, segmented data analysis, focus groups and an external research review. Rivermark should be explicit about this proportionality: collecting more data than needed can delay action and create privacy burdens, while collecting too little can make a high-impact decision depend on anecdote. A written plan specifying the question, population, method, access arrangements, limitations and review date creates discipline without making everyday people practice unnecessarily complex.
Frequently asked questions
Is qualitative or quantitative evidence better in HR?
Neither is automatically better. Quantitative evidence is useful for scale and pattern; qualitative evidence is useful for meaning and explanation. Many important people decisions require both.
Can employee survey comments be treated as qualitative evidence?
Yes, but analyse them systematically, protect confidentiality and avoid presenting a few comments as representative of the entire workforce.
What is mixed-methods research in people practice?
It combines quantitative and qualitative approaches to understand both the scale of an issue and the reasons or experiences behind it.
How do I choose the right method?
Start with the decision question, identify what is already known, consider who may be affected and select the simplest combination of evidence that can reduce the important uncertainty.
References
CIPD (n.d.) People analytics. Available at: https://www.cipd.org/en/knowledge/factsheets/analytics-factsheet/ (Accessed: 24 August 2026).
Creswell, J.W. and Creswell, J.D. (2018) Research design: Qualitative, quantitative, and mixed methods approaches. 5th edn. Thousand Oaks, CA: Sage.
Bryman, A. (2016) Social research methods. 5th edn. Oxford: Oxford University Press.