"Evidence-based HR" doesn't mean only using numbers — it means systematically weighing multiple types of evidence, including data that's inconvenient for a popular initiative, before making a decision that will affect real people's jobs.
The four types of evidence in evidence-based HR
HRM5080 uses Rousseau and Barends' framework of four evidence types: scientific evidence (peer-reviewed research on what actually works, e.g., in selection or training), organizational evidence (the organization's own internal data, like turnover or engagement metrics), stakeholder evidence (values, concerns, and professional judgment of the people affected), and practitioner evidence (the HR professional's own accumulated experience and expertise). Students learn that genuine evidence-based decision-making weighs all four rather than substituting scientific evidence alone for organizational context, or practitioner intuition alone for what the data actually shows.
HR analytics and avoiding common data traps
The course covers practical HR analytics — turnover analysis, engagement survey interpretation, predictive attrition modeling — while explicitly teaching common analytical traps: correlation mistaken for causation, survivorship bias (only studying employees who stayed, missing why others left), and small-sample overconfidence. Students practice critically evaluating a vendor's or consultant's data-driven pitch for a new HR program before recommending it be adopted.
Key topics in HRM5080
- The four types of evidence: scientific, organizational, stakeholder, and practitioner
- HR analytics fundamentals: turnover analysis, engagement metrics, predictive attrition models
- Common analytical traps: correlation vs. causation, survivorship bias, small-sample overconfidence
- Critically evaluating HR vendor and consultant data claims
- Building an evidence-based business case for an HR initiative
- Balancing data-driven decisions with ethical and stakeholder considerations
Working on an HR analytics project or an evidence-based business case?
Our business experts build HRM5080-level coursework with genuine data-analysis and evidence-weighing rigor.
Worked example: weighing four types of evidence on a 4-day workweek proposal
- Scientific evidence: Published pilot studies show mixed but generally positive productivity effects in some industries
- Organizational evidence: Internal engagement survey shows high employee interest, but current turnaround times suggest scheduling risk
- Stakeholder evidence: Client-facing teams express concern about coverage; internal teams are enthusiastic
- Practitioner evidence: HR's own experience suggests a phased pilot in one department first, rather than a company-wide rollout
- Decision: A limited pilot in a non-client-facing department, evaluated against clear metrics before wider rollout
Get Help With HRM5080
HR analytics projects and evidence-based decision-making assignments.
Place Your OrderView All ServicesRelated courses
Frequently asked questions
Rousseau and Barends' framework identifies scientific evidence (findings from rigorous, peer-reviewed research on HR practices, like what selection methods actually predict job performance), organizational evidence (the specific organization's own internal data — turnover rates, engagement scores, exit interview themes), stakeholder evidence (the values, concerns, and professional judgment of the people who will be affected by or must implement the decision, such as line managers or employees), and practitioner evidence (the HR professional's own accumulated experience and pattern recognition from prior similar situations). HRM5080 teaches that evidence-based HR does not mean relying on scientific evidence alone and ignoring the others — a practice with strong scientific support might still fail in a specific organization if it conflicts with stakeholder values or organizational context, so genuine evidence-based decision-making means deliberately weighing all four sources rather than treating any single one as sufficient justification on its own.
Survivorship bias occurs when an analysis only examines the population that "survived" some selection process, systematically missing information from those who didn't — in HR analytics, a classic example is analyzing what makes top performers successful by only studying current high performers, without also examining people who had similar starting characteristics but failed or left, which might reveal that the same traits look different in outcome depending on factors the analysis never captured. Another common HR example is analyzing engagement survey results only from current employees, entirely missing the perspective of employees who already left — potentially the group with the most valuable information about what's actually driving attrition. HRM5080 teaches students to explicitly ask "whose data is missing from this analysis, and could that missing group change the conclusion" as a standard check before drawing conclusions from any HR dataset, since survivorship bias is one of the most common and least visible ways HR analytics can mislead decision-makers.