Featured Recruitment Advertising

The pros and pitfalls of data-driven recruiting

Marketing department and analytics team are discussing campaign results

Most of us demand data to support our beliefs and strategies. But this has resulted in data being muddled with the truth. This has some not-so-positive implications for our ability to understand, explain, and improve the way things work.

After a long history of relying on subjective tools as evidence of our opinions, most of us today demand some form of data to support our beliefs and strategies. But according to New York University’s Professor of Data Science Andrea Jones-Rooy, this has resulted in data being muddled with the truth, which can have not-so-positive implications for our ability to understand, explain, and improve the way things work.

How reliable is data?

According to Jones-Rooy, “Data is an imperfect approximation of some aspect of the world at a certain time and place. It’s what results when humans want to know something about something, try to measure it, and then combine those measurements in particular ways”.

We know there is considerable value in data, but why use the word ‘imperfect’? In her opinion, data is a “fundamentally human construct” and therefore subject to limitations, biases and other imperfections.

“Data is a necessary ingredient in discovery, but you need a human to select it, shape it, and then turn it into an insight. Data is therefore only as useful as its quality and the skills of the person wielding it”.

According to Jones-Rooy, there are four main ways that imperfections occur in data – random errors, systematic errors, errors of choosing what to measure, and errors of exclusion. Let’s look at them in a bit more detail.

  1. Random errors in data

These occur when we measure something and then due to error, dysfunctional tools or equipment, or carelessness in its collection, the recorded data is wrong. An example Jones-Rooy gives is in the area of medical screenings, in particular for cancer. A positive result may suggest we have cancer, however we won’t if a ‘false positive’ is determined by further testing.

“There are lots of reasons this might happen, most of which boil down to a misstep in the process of turning a fact about the world (whether or not we have cancer) into data (through mammograms and humans)”.

On the down side, false positives in a cancer screening can lead to mental health consequences, however they can also lead to more vigilant screening, and a belief that future errors are statistically random and will cancel out over time.

  1. Systematic errors in data

These refer to data that consistently appears in data collections at the expense of other data, which can potentially lead to faulty conclusions. According to Jones-Rooy, it can occur for different reasons including who you sample and when. ‘Selection bias’ is a common type of systematic error.

“Medical studies are sadly riddled with systematic biases. They are often based on people who are already sick and who have the means to get to a doctor or enrol in a clinical trial. If everyone who has an Apple Watch, for example, could just send their heart rates and steps per day to the cloud, then we would have tons more data with less bias. But this may introduce a whole new bias: The data will likely now be skewed to wealthy members of the Western world”.

  1. Measurement errors in data

These occur when we think we are measuring one thing, but we’re actually measuring something else. An example Jones-Rooy gives in the recruitment sector is when employees turn to data and metrics to make more objective hiring decisions, including attracting top talent.

According to her, it is important to ask yourself if your data is measuring what you think it’s measuring. And even further, why are you measuring data in this way in the first place? Is there another way you could more thoroughly understand candidates? And given your current data, can you adjust your filters to somehow reduce any bias?

For example, when choosing candidates for particular roles, is your preference for those who have a university degree?  Because according to Jones-Rooy, “rather than that being a measure of talent, it might just be a measure of membership in a social network that gave someone the ‘right’ sequence of opportunities to get them into a good college in the first place”.

  1. Errors of exclusion in data

These happen when whole populations are ignored in data sets, which can then set a precedent for further exclusion. An example Jones-Rooy cites is the fact that more women in the US die from heart attacks than men, which is thought to be partly due to most cardiovascular data being based on men. Men also experience different systems to women, which can lead to incorrect diagnoses.

Choosing to study something can also incentivise further research on the topic, which Jones-Rooy believes is a bias in itself. This is because it’s easier to build on existing databases rather than creating your own, so researchers often study certain topics at the expense of others. If this behaviour is repeated enough times, then certain groups become ‘defaults’, for example, men are the ‘default’ in US heart disease studies as opposed to women.

How is data used in recruitment?

In its simplest form, recruitment data is gathered from your current and prospective employees and used to garner key insights. These insights can lead to ‘datafication’ (which is the transformation of data into new forms of value), and you can leverage it in a variety of ways. These include designing better processes and making more effective decisions, to better understanding your job candidates, employees and the organisations you’re competing with.

Using advanced analytics including using predictive methods can also give you insights into why things happened, and what will or should happen in the future. So for example, analysing a candidate’s data and comparing their attributes to what’s needed in the current workforce, can help you identify the best ways to attract top talent who stay with their employers for longer.

So … is data still worth leveraging?

In terms of the value of data in recruitment, the leveraging of ‘structured data’ alone (like employees’ personal information and employment history) is not sufficient. You should use it in combination with ‘unstructured data’, which includes things like emails, survey data, social media posts and photos, videos and audio recordings, because over time, the ability to analyse it will become more and more critical to the recruitment industry.

It’s also worth remembering that the value of data management doesn’t come from ‘completeness’ or ‘effectiveness’, but from you being able to understand its limitations.

As Jones-Rooy summarises, “Just as we want to analyse things carefully with statistics and algorithms, we also need to collect (data) carefully, too. We are only as strong as our humility and awareness of our limitations. We should not just ask what does it say, but ask who collected it, how did they do it, and how did those decisions affect the results?”

What are your experiences with using data for your recruitment strategies? Share a comment below and continue the conversation on LinkedIn

Sources:

Quartz – July 2019

Digital List Magazine – March 2019

With a career history of over 20 years in the marketing sector working for some of Australia’s top ad agencies, Kaye launched her own copywriting business, KK Productions, in 2014. 

Her client list includes Australia’s largest retail travel outlet, the world’s largest insurance company and one of the country’s top retail stores; she has also written content for a variety of other sectors from recruitment, real estate and educational organisations to health and service-oriented industries. Kaye’s writing experience spans the full gamut of advertising and marketing copy. When she is not immersed in the world of writing, she loves reading, travel and bushwalking with her husband and dog, Barney.

Related posts

Kinnect attracts people who do amazing things all day, every day

Susanne Mather

The art of saying “no”: how to let candidates know they haven’t got the job

Ricky Fritsch

Buyer’s remorse

Susanne Mather

Leave a Comment