Canadian HR Reporter

Human Resource (HR) analytics and related topics -- big data, people ana- lytics, data visualization, algorithmic selection, and the like – have featured prominently in every edition of the Society for Industrial and Organizational Psychology's Top 10 Workplace Trends since that annual list was introduced in 2014. HR analytics refers to a practice "enabled by information technol- ogy that uses descriptive, visual, and statistical analyses of data related to HR processes, human capital, organizational performance, and external eco- nomic benchmarks to establish business impact and enable data-driven de- cision-making" (Marler & Boudreau, 2017, p. 15). In a 2018 survey of British Columbia businesses, most respondents saw potential value in HR analytics but did not have an individual – let alone a team – focused on HR analytics (Neale, 2018). HR analytics is at an early adoption stage in Canada and elsewhere (Coo- per & Jackson, 2018; Marler & Boudreau, 2017). Almost three-quarters of respondents in the BC survey indicated that a "lack of analytics talent among HR" was at least somewhat of an obstacle (Neale, 2018). Lack of analytical skill is the most common impediment to implementation. Moreover, "there is a concern that should HR Analytics be adopted it will not be controlled by HR professionals but by others who may misinterpret or misspecify the analyses" (Marler & Boudreau, 2017, p. 19). A tangible example of this problem was identified in the retail sector – a poorly-conceived staffing algorithm mini- mized labour costs at the expense of performance and profitability (Angrave et al., 2016). Rasmussen and Ulrich (2015) agreed that HR analytics is being taken over by other functions, but argued that it is positive for HR to be inte- grated into end-to-end business analytics. HR professionals should prepare to participate in analytics even if they don't run the statistical analyses. As I proposed at the Conference Board of Canada's Evidence-Based HR 2017 event, there seems to be value for HR professionals and others to position analytics within a broader framework. Evidence-Based Management (EBMgt) and HR analytics are distinct yet complementary perspectives. EVIDENCE-BASED MANAGEMENT The basic idea underlying EBMgt is to make use of the best available infor- mation to support high-quality decisions and actions. The most well-known EBMgt framework consists of four elements – evidence from the local con- text, formal research evidence, ethics and stakeholder concerns, and practi- tioner judgement and expertise (Briner, Denyer, & Rousseau, 2009). Local Evidence HR analytics has a close connection with EBMgt's local evidence element (e.g., data within an organization). HR analytics extends the information tech- nology and data analysis tools implied in EBMgt's local evidence element. However, focusing on data analysis is misguided. Design, measurement, and analysis are all critical aspects of the research trinity (e.g., Kline, 2009). Fan- cy statistics cannot compensate for weak measures or research designs that cannot answer research questions of interest. EBMgt proponents consider a more extensive array of research designs (e.g., quasi-experiments) than an analytics strategy focused on mining datasets. Other tips: Beware of statis- tical fishing expeditions. Also, use existing knowledge to identify measures, research tactics, and conceptual frameworks that can improve local analytics work. Formal Research Insights from decades of published research can complement local evidence and stakeholder perspectives in the decision-making process. The formal re- search element extends the analytics perspective. You won't find direct an- swers to your organization's challenges in any book or article – that is neither promised nor implied. Nevertheless, armies of nerds are likely to have in- vestigated issues relevant to your circumstances. Finding and adapting their insights can help you add value to the work of data scientists who may not be familiar with relevant literature. Paywalls are a problem for database access. Literature search and appraisal skills are challenging to refine. But you can meet those challenges. Tips: Some professional associations (e.g., Center for Evidence-Based Management) provide database access you might not have through your employer. Also, try searching Google Scholar. Ethics and Stakeholder Concerns The EBMgt framework reminds us to consider ethics and stakeholder con- cerns. This element should be interpreted broadly and include, but not be limited to, research ethics. Analytics professionals may attend to data priva- cy concerns but, unfortunately, some of that interest appears to be reactive (as evidenced by various personal data scandals). I'm not sure how analytics students are learning about the norms, standards, and legal requirements inherent in behavioural research or human resource management. A year ago, I glanced through the curricula of about a dozen master's programs in analytics. I only found one that included a course title related to research ethics or data privacy. In addition to applying relevant legal requirements, HR professionals can complement their own profession's ethical duties and standards (e.g., respect, confidentiality, privacy safeguards) with principles and practices supporting the Ethical Conduct for Research Involving Humans (Interagency Secretariat on Research Ethics, 2010). Tip: Complete, and have analytics staff complete, the free online Course on Research Ethics (http:// tcps2core.ca/welcome). Even in the absence of mandated compliance, you can use those federal guidelines to promote good practice (and help stay out of trouble). Practitioner Judgement and Expertise The EBMgt framework encourages explicit, critical consideration of hu- man experience and judgement as part of the decision-making process. A well-developed knowledge network is a useful source to consider. Expertise and insights should be respected. However, we must navigate a tricky bal- ance between having confidence in our judgements and appropriate humil- ity. There is ample evidence of biases in human decision-making, and gen- uine expertise is elusive in business. Critical reflection on human judgment and biases may, in some instances, encourage the use of well-developed algorithms to increase fairness and validity in decision-making (Kahneman & Kline, 2009) -- e.g., mechanical versus subjective combination of predictors in employee selection. However, algorithms designed to capture and repli- cate human decision-making processes or make predictions based on pro- hibited grounds could perpetuate biases and legal problems. Adopt tools thoughtfully to aid human judgements and support effective decisions. CONCLUSION HR analytics continues to attract attention. Adopting a broader EBMgt frame- work may help HR professionals contribute to analytics work, whether or not they identify as data scientists. Asking the right question is at the core of the EBMgt approach. Likewise, failure to "start with a business problem" puts HR analytics at risk of becoming a management fad (Rasmussen & Ulrich, 2015, p. 238). HR analytics shouldn't be only about the numbers. By R. Blake Jelley, Ph.D., CPHR University of Prince Edward Island *References are available upon request from the author (bjelley@upei.ca) CPHRPEI.CA Chartered Professionals in Human Resources of PEI (CPHR PEI) • 101 Kent Street, PO Box 2151, Charlottetown, PE C1A 8B9 Evidence-Based Management and HR Analytics

• Cover