Prehire Assessments Can Speed Up Screening, Reduce Bias

​Employers that use prehire assessments to evaluate job candidates have always faced two significant hurdles: recruiters' perceptions that the tests can scare off top applicants if they are too lengthy or onerous, and doubts whether the tests accurately predict future job performance. But new technologies are slashing the time needed for candidate testing while also accurately scoring unstructured data—like video, audio or text—from candidates.

More employers, just as concerned about quality-of-hire as they are with filling open jobs quickly, are using these tests. A 2018 study from Aptitude Research, a Boston-based advisory and research firm specializing in talent acquisition technologies, found that 71 percent of companies are now using at least one assessment provider and 57 percent are using two or more providers. The study's authors said the move is driven in part by recruiters' needs for more science and less bias in the talent acquisition process.

Candidates also are seeing new opportunities for using skill assessments. A recently added feature on Indeed, for example, allows job seekers to add results of assessments to their profiles to help improve their marketability to potential employers.

[SHRM members-only platform: SHRM Connect]

New Technology-Driven Capabilities

While recruiters remain focused on hiring available candidates as rapidly as possible, the organizations they work for want to ensure that those hires are a good fit and don't quickly leave for greener pastures.

As a result, employers are using assessments not just for leadership positions but also for customer-facing roles and positions requiring high-volume hiring. Some also have started using them to evaluate the ability of existing employees to assume new or expanded roles.

Regardless of who is being tested, innovations from startups as well as established players in the assessment space consolidate steps in the screening process, evaluate candidates in new ways and, in some cases, reduce bias in the hiring process.

For example, "The ability to accurately score unstructured data gives assessment providers many more data points than just the outcome of a multiple-choice test," said Jonathan Kestenbaum, managing director of Talent Tech Labs, a talent acquisition technology research firm in New York City. "Collecting those data points in a video interview or other format allows providers to deliver accurate results in a smaller time period than in the past. What may have been an hourlong assessment can sometimes now take 20 to 25 minutes."

HireVue is among the assessment providers using machine intelligence and deep learning to capture and analyze such data points. The company's technology can analyze candidate word choice, intonation, nonverbal gestures and more in video interviews.

"We extract about 50,000 data points from video interviews and then analyze and score from 15,000 to 25,000 of those points deemed the most important predictors of future performance for specific job roles," said Nathan Mondragon, chief industrial/organizational psychologist for HireVue. "We believe it's a more accurate and user-friendly way of predicting performance than giving candidates a 200-question test, for example."

Shaker International is another assessment provider that uses machine learning and predictive modeling in its testing. The Cleveland-based company uses a multimethod approach in its "virtual job tryouts" to measure candidates for roles in call centers, banks, health care institutions and manufacturing plants.

The assessment includes exercises like simulated customer and co-worker interactions and evaluates behavioral work styles, situational judgement and cultural fit.

Eric Sydell, executive vice president of innovation at Shaker, said use of deep learning techniques allows the company to score what candidates say for meaning and quality, instead of measuring for simple data points like word count or spelling of typed words.

Candidate-Friendly Features

One of the biggest obstacles to prehire assessments is a belief that they take too long and will be a turnoff to candidates who have other good options for employment. But experts say assessment length rarely proves to be a problem—within limits—if candidates believe the tests are directly relevant to the job.

"At the right time in the process for the right candidate, the duration of an assessment tends to be a nonissue," said Elaine Orler, co-founder and CEO of The Talent Function, a talent acquisition consulting firm in San Diego.

Orler said years of data from the CandE awards, the candidate experience competition sponsored by the Talent Board, show that applicants typically are not bothered by the length of assessments as long as they feel the tests are relevant to skills or knowledge they need to succeed on the job.

Shaker's virtual job tryouts bear this out. The simulation takes about an hour to complete and evaluates candidates in work-specific scenarios. "We see very little drop-off during that one-hour assessment," Sydell said.

In a study of 225,000 job applicants, Shaker found that if candidates are going to opt out, they usually do so in the first 10 minutes, regardless of assessment length. In addition, sharing more-challenging aspects of a job doesn't deter candidates from completing the assessment, the study found.

Other providers, like HireVue, have shortened assessment length by consolidating steps in the screening process. The company's video-based assessments, where candidates are typically asked to answer six or seven questions, replace phone screens and take about 25 minutes to complete, Mondragon said.

Hilton Hotels used the HireVue process to help reduce its time-to-hire from 42 to five days. Hilton's high-volume recruiting process had previously relied on assessments with more than 100 questions that took over an hour to complete, according to Amber Weaver, director of volume recruiting and talent research at the hotel chain. The multistep process had low completion rates, too.

"By condensing steps in the process, they not only reduced time-to-hire but also improved quality-of-hire," Mondragon said. "Previously, they had to interview four people to get one new hire. But now more often when they interview four, they report finding two candidates worth hiring."

Pymetrics is another assessment provider focused on creating tests that not only have a high degree of predictive accuracy but also prove to be candidate-friendly. The New York City-based company uses neuroscience games and machine-learning algorithms to measure cognitive and emotional functions as well as other traits.

The assessments last 20 to 25 minutes and have an average completion rate of 97 percent, said Frida Polli, CEO of Pymetrics.

"At every stage of the assessment, candidates receive feedback and they're also given a report at the end," Polli said. "Our system also allows candidates to match to other roles if it's found they're not best-suited for the original role they apply to."

Dave Zielinski is a freelance business writer and editor in Minneapolis.