Predictive Analytics in HR: Do you have the data needed to back that up?
Some organizations have begun to use predictive analytics as part of their hiring process. I filled out questionnaires used for analytics while applying for HR Manager positions. Supposedly, there are no right or wrong answers. The employer just wants to know “how we are wired” to move on to the next phase.
An excerpt from one of these questionnaires appears below. Section one lists one hundred ten adjectives from which the applicant chooses to describe him/herself. Section two presents the same list and asks that the applicant use them to describe how one should behave in their current work environment.
Do you really know how your predictive analytics are supposed to work? It has a growing presence in talent acquisition but has never been proven effective in this context. We do know that these algorithms must draw from a wealth of data to reach any meaningful conclusion. Does your company have enough data to determine who will make a great hire?
I applied for two positions for which the hiring manager applied this methodology. A “thumbs up” from the predictive assessment was required to move on to the next step in the process. Despite claims of “It’s not pass or fail,” I never received a follow-up call after either test. When I asked for feedback, none was provided.
The test is pass or fail. Don’t believe otherwise.
You passed if you are called to the next phase. You failed if you aren’t.
The soul-sucking part of this assembly line process lies in the fact that there is no way to know where you went wrong. You are selecting adjectives with no situational context. Does it really make sense to exclude a candidate because of vacuous word selection? If someone did this manually, they would be considered a crackpot, but because the actual logic is hidden within an automatically executed algorithm, the result is treated as gospel. This is not logical, or rational, or in keeping with any expert understanding of computer science. It is nothing but snake oil software.
Depending on our age, our culture, upbringings, education, and experience, the way we answer these surveys will vary. When implementing this type of strategy do you really know what you’re looking for as a recruiter or as an organization? Do you know what kind of mental-wiring you’re looking for?
I am at a point in my life and my career that I picked the same words for Section 1 & 2. If I was younger and new to the workforce my answers will be different. If this was supposed to be no right or wrong answers, nor is it pass or fail, what is the point of this process? Why spend the money on this assessment if you’re only trying to detect how a person is wired? Wouldn’t this be at some point a discriminatory/prejudicial process that may impact people based on age, gender, ethnicity, and disability since how we answer may predict these things? Not knowing which section is given more weight and how we are scored, how do we improve or meet the implicit expectations?
I met a couple of data scientists who use predictive analytics in a variety of business domains. One works for a health insurance company, and I can see how datasets available to insurers could drive useful analytical outcomes. Unfortunately, they also took on the task of extending these methods for use in talent acquisition. Neither has ever actually worked in the recruiting field! We need to stop letting non-HR people define the essential decision-making processes we use to do our jobs! Would you let someone with no HR expertise walk up to your desk, rifle through your files, and start ordering around? Let’s wake up to the fact that this is exactly what’s being done, being blinded by those who package the clueless interference in a piece of software and label it “analytics.”
I’m not a fan of any personality assessments. I don’t think they are useful indicators of who we are as professionals or indicative of what we bring to the table. If anyone out there has rigorous studies they can quote to contrary, I’ll gladly listen to them, but I’m not holding my breath. The business world has a long and infamous history of rabidly seizing upon trendy practices without a shred of evidence that attests to their effectiveness.
I’m also not a fan of hiring for cultural fit, because what constitutes cultural fit should and will evolve over time. A static, inflexible notion of cultural fit is a surefire way to achieve organizational sclerosis. If we want to assess our candidates, let’s assess the skills they bring to the table After all, isn’t our current concern about skills gap, and not some nebulous black-box notion of “how we are wired?”
Predictive analytics is the practice of extracting information from existing data sets in order to determine patterns and predict future outcomes and trends. Predictive analytics does not tell you what will happen in the future (wabopedia.com).
The Data: “Lack of good data is the most common barrier to organizations seeking to employ predictive analytics” (hbr.org).
To make predictions about what kind of employees we should employ, we need these kinds of data points:
- Characteristics and accomplishments of current employees
- Characteristics and accomplishment of past employees.
- Reasons employees stayed
- Reasons employees left
If you lack this kind of data, along with some means to feed it into the analytical model, then the analytical model is useless. If you’re a small-medium company or a start-up, it’s particularly likely that this is the case.
The reason predictive analytics work in domains such as sales is the wealth of data available to the model. The same goes for healthcare.
The same just can’t be said of talent acquisition. Period. Even worse, applying these algorithms may lead to bias and perhaps discrimination, if the outcome includes disproportionate impact on protective groups. If you can’t open the black box, how do we know there’s not a bigot inside? The software vendors pushing this snake oil provide nothing but vacuous hand-waving when they assure us this isn’t the case. This is why many HR pros who are also practicing attorneys don’t like assessments that fail to actually measure person’s ability to do the job.
Your Predictive Analytic Model Sucks!
As I mentioned earlier on, I have taken the predictive analytics tests twice. This was in the Summer of 2017. Both of those employers have since re-posted those jobs on multiple occasions. I never reapplied because they haven’t changed their requirements. I got a call from an external recruiter yesterday and she thought that I was a perfect candidate for their client based on my resume. When she told me who the client was, I laughed and told her that I’d already been disqualified without an interview. I let her know that I appreciate her thinking of me and wished her luck!
Clearly, the employer is failing to realize that their predictive analytic model sucks! My husband who is a software enterprise architect thinks they got scammed into buying a technology that was full of promise but failed to deliver. If they admit to drinking snake oil it will reflect badly on those involved in the decision, so backing down isn’t an option.
I’m not claiming their model sucks because I didn’t get the job. I’m saying this because they’ve been trying to fill this position since the summer of 2017. They have hired multiple staffing agency to help them fill this position every time it gets reposted. There are a plenty of great HR managers in the Midlands. This is not about a skills gap.
When will the employers utilizing these products wake up to the fact that they’re cheating themselves out of great hires? By leaving the HR Manager post vacant, the company I applied to is leaving themselves vulnerable to compliance issues.
Meet your candidates.
Have the conversation.
Stop hiding behind junk software that reduces your workload by doing your work badly.
Photo Credit: media.licdn.com
#HR #TalentAcquisition #HRTech #PredictiveAnalytics