Data analysis is said to reduce workplace bias and increase efficiency. But the truth may not be that simple.
The Translation Bureau is a compilation team that focuses on technology, business, workplace, life and other fields, focusing on foreign new technologies, new ideas, and new trends.
Editor’s note: As technology continues to advance, more and more wearable tracking devices have emerged in our lives. In recent years, there have been special projects that use the data generated by these wearable tracking devices to analyze the specific performance of the research subjects. If the company’s boss lets you wear the device 24 hours a day, so that he can further understand your performance, would you agree? The original title of this article is A New Study Suggests Employers Track Your Every Move to ‘Improve Productivity’. The author, Angela Lashbrook, introduced several similar projects in the article, as well as opinions and opinions of different groups on these projects.
Image source: FitNish Media on Unsplash
If you ask the boss for a raise, the boss tells you in turn that you must wear a special wearable device 24 hours a day. What would you react to?
This approach is not impossible in the future. During Obama’s administration, there have been a large number of workplace health research projects that currently cover more than 50 million workplace employees. In these projects, fitness tracking devices or other devices are often used to assess the health of the employee’s body, and the relevant data affects the rate of personal insurance in many cases.
And a recent new study is more likely to continue to embrace this wonderful new era: a variety of irrelevant data points aggregated through day and night data monitoring, such as how often you look at your phone or how often you go out on weekends. To assess the performance of employees. The study was also partially funded by the Office of the Director of National Intelligence (DNI), and the researchers were mainly from Dartmouth College.
The purpose of this study is to transmit base stations, wearables via geo-based tracking signals.Equipment and mobile applications, “distinguish between efficient and inefficient employees.” Its design and purpose are similar to two other research projects called mPerf and MOSAIC. The theme of both projects is to study how artificial intelligence helps employees (and their employers). But experts warn that there are still many concerns about this type of tracking.
“This study will focus on: how much sleep time the respondents have, how their heart rate behaves, and how much physical exercise they have.” Privacy and Security Nonprofits Center for Democracy And Technology, CDT) policy analyst Natasha Duarte said, “To see if an employee is relatively young and healthy, these are the main indicators…what if the respondents are disabled? If you refer to them If you use sports data to assess your performance, it is too discriminating.”
In this study, there were 554 subjects. Among them, 320 males and 234 females. These subjects come from all walks of life, but mainly in the technology and information industries. They regularly fill out the Workplace Assessment Questionnaire, which includes yes or no questions, such as “Do you show a loyal work attitude to the organization today?” or a sub-topic, such as “How far have you completed your work?” “To what extent do you ensure that tasks are properly completed?” Then, based on their responses, the researchers will determine if they are working efficiently or inefficiently.
At the same time, the researchers also equipped a variety of different tracking devices for these subjects. Everyone wears a waterproof Garmin watch, and their smartphones also have a tracking app called PhoneAgent, which also assigns four signal transmitting base stations to them: one placed in the wallet and one placed in the wallet. On the key chain, one is placed at home and one is placed in the office.
These devices can record the sleeping habits of the subjects, the frequency of leaving the work surface, the frequency of going out from home at night or on weekends, the frequency of unlocking the phone, how much physical activity is completed, and the quality of sleep.
The researchers then collected data from various devices for comparative studies. The results show that the research objects in different industries, and whether the research subjects are in management positions, have a great impact on the results. For example, highly-employed employees in non-management positions spend more time on weekends (unexpectedly), and they rarely travel to other places during the usual weekdays. Efficient employees in the consulting industry rarely participate in physical exercise on weekends, while highly skilled employees in the technology industry rarely participate in exercise on weekdays.
Pino Audia, a professor of management at Dartmouth College, said that based on these data,It can be inferred that efficient employees who rarely travel to other places on weekdays often have fixed daily schedules, and even in difficult work environments, these behaviors can keep them active and resourceful. “If you are always disturbed, your performance may be affected,” Odia said.
In addition, the researchers hope that the data will further eliminate discrimination and unfairness in employee surveys. The current assessment of work “is not only old but also biased,” said Andrew Campbell, a professor of computer science at Dartmouth College. He said they want to understand how to use these motion-sensing data to predict patterns that reflect high performance. At the same time, they also hope that within ten years, employees can review and improve their performance by combining feedback from these data.
In terms of employee performance evaluation, the traditional approach does have certain flaws. According to researchers at the Stanford University’s VMware Women’s Leadership Innovation Lab, when answering open-ended questions, such as “How do employees achieve expectations?” “What is the most prominent skill of the employee?”, managers often rely on stereotypes And bias, rather than combining data to evaluate.
Men usually receive specific feedback about their technical skills, while women usually receive general comments and feedback, such as “You are a great communicator!” Specific feedback can provide direction for employees to progress At the same time, indicating their superior characteristics, and not receiving similar feedback, it is more like a group that is ignored.
Odia said the study could “in the short term allow companies to no longer rely on or reduce their reliance on these surveys. Instead, we can rely more on objective indicators of employee behavior. How should companies treat employees? , compensation and promotion are not subject to gender, race and nationality and show greater fairness?… Some of the technologies used in our research will continue to struggle towards these goals.”
Image source: Unsplash @rstone_design
Of course, there are also some concerns about bias in this study. In addition to the prejudice problems mentioned by Duarte, the main subjects in the study are also males. If the data of these white-collar men are used to predict or evaluate the performance of people who do not belong to such groups, then they may be Factors that are unrelated to their performanceTreat unequally.
“If you apply research data for men in their 20s to women, or to groups over 30, or people with disabilities, it may not hold you.” Stanford Jen King, head of consumer privacy research at the University’s Internet and Social Center, said, “These data are inherently biased.”
For example, if someone causes anxiety due to a mental illness, it increases heart rate or affects sleep quality. In addition, according to the research, efficient employees rarely go to other places after work. However, the mother of a middle school student may need to take the child to the remedial class after work, and may leave the home frequently. It is also unfair to use these data to evaluate them.
Of course, if you use these devices to track and record employee activity and biometric information, especially after work, there will be privacy violations.
In the United States, there are currently few legal provisions for employee privacy protection. Therefore, “required” employees to be monitored by these devices for a long time may not break the law, and the employee health project mentioned in this article has already landed. In response to these projects, not only did people point out that they may be illegal, but they were also strongly encouraged by the Affordable Care Act, which was introduced during Obama’s administration.
“This kind of thing should not happen, but unfortunately, the legal boundaries of these studies are not clear,” Duarte said. She also said that even if employees agree to participate in these projects, these tracking projects still have moral hazard problems, which will make people feel that they are forced to participate in the same way as “good employees”.
King also puts it another way, and employers are likely to use these tracking data to track employees. “Imagine a wretched supervisor who may have a loving heart for a young female subordinate. If you can understand the real-time position of the subordinates at any time and follow the subordinates based on these data, it would be terrible!” she said.
But the founder of the study said that its research is still in its early stages, and if these devices really need to be applied to the workplace, at least a few years later. Campbell said that he also very clearly realized that there is a general prejudice problem with these data.
“But as far as your question is concerned, I will treat it as a general criticism for any machine learning or data-based algorithm. For example, if the data does not contain a group, then the algorithm will completely ignore They,” Campbell said. “As for how to solve this problem, I have not yet thought of a good solution.”
However, it is not just this big project that studies this macro theme. For example, the mPerf project mentioned above, also funded by government agencies, like this project, they are also studying workplace performance and motion sensing.The relationship between the data.
“After many years of use, high school students do not need to take the SAT or ACT exams if they want to enter the university. This is not an incredible thing.” Deniz Ones, principal of the mPerf project As mentioned in a press release, “Students need to download certain applications on their mobile phones, connect them to wearables, and allow universities to collect data from them within a certain number of months.” /p>
In addition, there is the MOSAIC project mentioned above. In these projects, the data of the members of the intelligence department will also be tracked to assess their performance, although it has been pointed out that the intelligence department’s employees accept this data monitoring, which is part of their work.
So, if you really want to improve your employee assessment, is there any other way that does not involve privacy violation tracking?
According to researchers at Stanford VMware Labs, managers should agree on certain employee performance assessment requirements in advance by months. Subsequently, the specific performance of the employee can be compared with these requirements to eliminate bias in the assessment environment.
In addition, managers should further control general or ambiguous praise, such as “She is a great communicator.” In contrast, they should refer to established employee performance indicators to assess their specific communication performance fairly and fairly.
Of course, further research is necessary to educate employers on how to effectively guide their employees to work effectively. However, in addition to the Dartmouth College project and the mPerf project, which may involve the nature of privacy violations, new and sometimes neglected prejudice issues related to technology will still emerge. People may have different opinions on the opinions of others, but in general people still believe in the objective and fairness of these systems, but we can do better.
Translator: Ishii Junichi