New Technologies to Balance “Speed-Accuracy Trade-Offs” in Hiring

Thomas Mielke | ASSESSMENTS & TRAINING, HR STRATEGY, HR TECHNOLOGY, LEADERSHIP, PERFORMANCE MANAGEMENT, RECRUITMENT

A Growing "Need for Speed"

(First publication rights with HotelExecutive)

We predict 2021 will be “the year of two extremes.” These first quarters are likely to involve continued pressure on operations and financial distress, but the last quarters could well bring a significant upswing in various bookings. A surge in business ― fuelled by the global vaccination programs — should efficiently usher a recovery and rebound period. This view is not unrealistic optimism. People have ruminated on missed experiences over the past 12-months, and several consumer surveys indicate significant pent-up demand for entertainment and escapism due to the extended COVID lockdowns. Also, as social and travel restrictions are lifted and consumers’ fears and anxieties are alleviated, people will be more apt to make needed or desired purchases. It is not surprising therefore that industry commentators have openly talked about a return of the “roaring 20s” and some even touted the notion of “revenge tourism,” which refers to burgeoning travel and leisure activities that will come post-pandemic. This latter parallels the “revenge spending” of Chinese consumers in the 1980s after the country’s economy was reopened. We expect therefore that it is not a question of “if” but rather “when” the rebound reaches critical mass. Of course, a company’s ability to act swiftly and decisively impacts whether it will be “ahead of the curve” relative to its competitors. This applies especially to many organizations that reduced their front-line staff and now face the challenge of ensuring that enough of the right people are in the right positions to meet brand promises. Reassessing and possibly beefing up talent bench-strength is certainly important but also potentially urgent.

“Speed—Accuracy” Trade-Offs are Looming

Faced with a need to recruit and build-up teams relatively quickly and with shorter notice, service-hospitality organizations are confronted with a classic psychological conundrum known as the “speed-accuracy trade-off.” This principle states that decision-speed tends to be inversely correlated with decision-accuracy. In other words, decisions can either be “fast” or “accurate” but typically not both simultaneously. Herein lies the problem: best practice companies know that wrong hires cost significant money, so they value due-diligence during candidate screening and selection. Yet, this process usually takes time to do well and therefore any lags can result in losing out on top talent if competitors are willing or able to make quicker decisions. Tactical operations is another area where productivity is affected by the speed-accuracy trade-off. Here it concerns the time needed for a new hire to ramp-up successfully by learning new systems or SOPs and start adding value to the business. A new hire’s learning curve, efficiency, and effectiveness are certainly influenced by their intelligence and motivation levels. Companies thus need people who are “hardwired” to think and thrive in service-driven cultures — team members with what we call the “Hospitality X-Factor” who consistently deliver on brand promises. These individuals are competitive, self-directed, and intrinsically driven to exceed performance expectations.

Strategic Merger of Two Technologies

The preceding clarifies the dual-tier challenge for businesses poising for their rebound and recovery, namely “the need to efficiently and accurately boost internal bench-strength with “X-Factor” team members, who themselves are hard-wired to efficiently and accurately execute on brand promises.” There are new technologies that significantly facilitate this goal, but most operational and HR leaders have probably not heard of them. These innovations relate to new knowledge of key attributes to be assessed with psychometric testing, as well as new knowledge about how to measure these key attributes inconspicuously and precisely. These two technologies are the product of modern computer processing capacities that bring the “speed” and the machine learning algorithms (i.e., Artificial Intelligence: AI) that deliver the “accuracy.” Research and application of such systems has been a specialty of the Laboratory for Statistics and Computation at ISLA-Vila Nova de Gaia in Portugal, headed by psychometrician and data scientist Dr. Rense Lange. Lange says that, “Recent developments in machine learning allow assessment systems to be implemented online in cloud-based systems in extremely cost effective ways.” This combination of improved hardware and software has proved to quickly and accurately profile people with the Hospitality X-Factor, or what many people might call “service superstars.” To clarify, standardized testing of high- and low-performers in “guest-facing” and “non-guest-facing” roles has been quietly conducted over the last four years in cooperation with Dr. Lange and researchers in the School of Hotel Administration at Cornell University. This hospitality field research used AI modelling to refine a concept in organizational psychology known as “contextual performance.” Although many companies have traditionally hired candidates based on their specific “task performance,” this approach actually misses what matters most in actual practice. Rather than focus on a fixed set of performance skills, it is much more productive to screen and select for what organizational psychology refers to as “contextual performance.” This term denotes a broad set of knowledge, skills, and abilities that are relevant and transferable across a wide array of jobs and settings. For example, the ability to effectively use a point-of-sale system represents “task-specific performance,” as it may be essential for restaurant servers, but unnecessary for cooks. However, teamwork and related behaviours are important to effective job performance for servers and cooks, and many other jobs, regardless of the function-specific tasks.

Answering Two Critical Questions

Recent AI-based modelling of data collected from the field research has helped to answer two practical and critical questions: “How do you define contextual performance?” and “How can it be measured quickly and accurately?” To explain, a few qualitative and quantitative findings must be unpacked. First, the field research noted above correlated several psychometric variables with performance evaluations. Global companies that participated in the studies created four groups using their incumbents (i.e., “Very Poor, Poor, Good, Very Good”), who were categorized by consensus using a standardized set of criteria that addressed contextual performance (or the X-Factor). Analysis revealed that thirteen specific attitudes and behaviours defined a robust psychometric scale of the Hospitality X-Factor (see Figure 1). These characteristics can be further subsumed by four general categories described by the acronym “CHAT.” That is, high-performers in service-driven cultures are collectively “Conscientious, Hospitable, Adaptable and Trainable.” Furthermore, scores on this new model of the Hospitality X-Factor showed a strong, positive, and statistically significant correlation with the overall performance rankings of “Very Poor, Poor, Good, and Very Good.”

Figure I: Psychometric Profile of Contextual Performance (aka, the "Hospitality X-Factor")

Second, while we now have a modern profile of contextual performance, it is not necessarily a straightforward task to measure its underlying traits and tendencies quickly and accurately. Indeed, pre-screening hiring tools are often criticized for being easily manipulated by applicants. Self-report measures based on personality traits or personal characteristics are notoriously susceptible to “social desirability” biases. This means the tendency for individuals to answer questions in a manner that will be viewed favourably by others. It can take the form of over-reporting “good behaviour” or under-reporting “bad,” or undesirable behaviour. This natural behaviour can pose serious problems when conducting research with self-report assessments. However, AI-based modelling helped to create a new way to measure the CHAT Model that mitigated social desirability biases. It is well-known to testing experts that there tend to be “hidden” patterns in people’s responses to tests and questionnaires. In fact, this knowledge is routinely used in psychology and education alike to improve test questions, equate different versions of test questions, and even to catch “cheaters.” These covert patterns are known to specialists as “differential item functioning,” and they can be used in ways that summed or averaged scores cannot. To begin with, test takers are unaware of their existence, but such knowledge would do little to help them anyway. Particularly, it takes sophisticated analyses to identify these patterns as they are not obvious in their “direction” (i.e., positive or negative) or “focus” (e.g., men versus women). Nevertheless, they are a special type of “fingerprint” that can reliably categorize people apart from simple raw scores. It is no surprise that people high in the Hospitality X-Factor indeed showed a covert fingerprint that special analytics can detect. This fingerprint detector was validated in research using a “test” that was not based in “words” or “statements” which are especially vulnerable to social desirability biases. Rather, the detector consisted of a gamified, visual exercise that requited only four minutes to complete. And this process could even be done easily on a mobile device. This Hospitality X-Factor exercise delivered a “quick” evaluation decision, but how “accurate” were these evaluations? Analysis of the field data showed high accuracy rates when test-takers are categorized in terms of three levels of contextual performance: “low, medium, and high” (see Table 1). Overall, the study found an average accuracy rate of “83%” for the simple four-minute task. Thus, there is compelling evidence that new technologies consisting of more precise psychometric models and more precise measurement of these models can mitigate the “speed-accuracy trade-off” that typically plagues rapid decision-making in screening and selection.

Thinking Ahead

Academics know that good science equals good measurement, and business leaders likewise know that good decision-making depends on good measurement. The hospitality industry has been, at times, at the forefront of such data-driven business strategies and decisions. For example, we referenced earlier the 1980s, a period in which “revenge spending” was coined. However, it was also the decade in which the airline sector began introducing dynamic pricing to optimize its financial results. This later evolved into revenue management, which nowadays is best practice amongst all hospitality verticals. Similarly, although not able to claim the prize for having invented it, the hospitality industry is fairly advanced when it comes to its customer loyalty programs —a tool that provides organizations with invaluable data to better, and more efficiently, target and upsell to customer bases. Other examples stem from different technology verticals, from the proliferation of e-distribution in the hospitality sector to data-driven digital marketing campaigns and service automation via AI and the Internet-of-Things (IoT) that make ordinary devices “smart.” On a larger scale, there also are virtual and augmented reality capabilities that have helped the event segment during the pandemic, as well as smart property management systems that collect, measure, and advise on energy consumption, waste reduction and other measures helping to improve sustainability credentials. In other words, the hospitality industry has shown that it can become “smart.” It is now time for this to translate into “smart(er)” hiring decisions. Our case study of screening and selection to prepare organizational bench-strength for the eventual rebound is therefore only one way that new technologies can confront “speed-accuracy trade-offs.” In fact, AI-based models can facilitate tailor-made predictions that are context dependent. So yes, thanks to AI and machine learning we can now simultaneously improve decision speed and accuracy in business - or, as we might put it, hospitality can have its cake and eat it too.