HR Management & Compliance

More Big Data Dangers, from Distribution to Interpretation

Yesterday’s Advisor outlined five big dangers of big data; today, more potential pitfalls of this information resource, including problems with distribution, stratification, and more.

(Go here for dangers 1 to 5)

6. Distribution Dangers

There are two primary distribution dangers you’ll face—outliers and stratification.

Outliers

Here is the situation. You have surveyed 22 employers in your industry that are similar to your employer in size and sales, and you find that their average cost per hire (CPH) is $3,433. You are feeling pretty smug, because your CPH is $2,800. But dig into the data.

Here are the CPH figures for the 22 companies surveyed:

$2,000

$2,600

$2,300

$2,400

$2,200

$2,300

$2,300

$2,500

$2,100

$2,300

$2,700

$1,900

$2,300

$2,150

$2,450

$2,300

$2,400

$2,200

$2,100

$14,700

$2,400

$15,800

In fact, 20 of the companies have CPHs well below your $2,800. However, two of the companies’ CPHs are much higher, one with $14,700 and one with $15,800. Those two outliers are skewing the data. Research reveals that in the company with $14,700, the only new hire reported was a new president. And the company with $15,800 had to hire from abroad and relocation expenses were included in their figures.

Eliminating the outliers, the average CPH for the remaining 20 companies was $2,300. And your CPH suddenly doesn’t look so good.


It’s the brave new world of HR. Start your strategic thinking with BLR’s new practical guide: HR Playbook: HR’s Game Plan for the Future.


7. Stratification

Another type of danger comes from how you array your data. When collecting data from a large number of subjects, you typically stratify the data. For example, if looking at a number of employees, you typically divide the responses into groups, like 1 to 49 employees, 50 to 99 employees, 100 to 499 employees, etc.

Here’s an example of some weekly readings on some factor:

Info 1

3

Info 2

7

Info 3

22

Info 4

45

Info 5

81

Info 6

23

Info 7

23

Info 8

81

Info 9

45

Info 10

22

Info 11

7

Info 12

3

For ease of charting, you choose to band together the data points in groups of 2. So, Info 1 (3) and Info 2 (7) combine for an average of 5, and so on.

Info 1 to 2

5.0

Info 3 to 4

33.5

Info 5 to 6

52.0

Info 7 to 8

52.0

Info 9 to 10

33.5

Info 11 to 12

5.0

And here’s the chart that produces:

That’s an expected normal curve, so you assume normal distribution. But what if you instead chart all the original data points? Now, the chart looks like this.

Obviously, something is going on with this bipolar distribution. Again, the data are the same in both charts; it’s just the approach to charting that changes.

8. History

You’ve established a program to improve engagement, and your survey shows improvement. But is it due to the program? Has something else happened at the same time that might explain the improvement? For example, is there new leadership with a new outlook? Is there a sharp rise in sales and profits and a great bonus?

9. Interpretation

Similarly, data are subject to different interpretations. For example, say you have been tracking turnover, have installed programs to improve retention, and can show the board or the C-suite that you have improved retention by 22 percent year over year.

One board member says, I disagree with your conclusion; I think your retention is improving because you have hired such weak people that none of them could find a job anywhere else.

Making sure you’re reading your metrics correctly is just one of the many challenges in the brave new world of HR. Are you prepared for changes that are unparalleled in scope and impact?

  • Employees all over the world, many of whom you’ve never met in person
  • Technological advances and big data
  • Talent management challenges like Millennials managing Baby Boomers you once thought would have retired years ago
  • Big data on everything from hiring strategies to retention predictions
  • Sweeping regulatory changes in the areas of health care, immigration, and privacy that have necessitated massive changes in the way you do business
  • And the new normal—doing more … with less

HR 2015? Time to start planning with BLR’s new HR Playbook. Find out more or order here—HR Playbook: HR’s Game Plan for the Future.


It’s a lot to keep track of—and it’s not going to get easier. To help you get your head around big-picture strategies for 2015 and beyond, HR’s Game Plan for the Future provides a detailed rundown of trends, case studies, and best practices in the following areas:

  • Recruiting and Hiring
  • Onboarding
  • Social Media and Technology
  • Human Resources Information Systems (HRIS)
  • Flexibility and Work/Life Balance
  • Outsourcing
  • Diversity
  • Talent Management
  • Employee Engagement and Retention
  • Succession Planning
  • Telecommuting

Find out more or order here—HR’s Game Plan for the Future.

1 thought on “More Big Data Dangers, from Distribution to Interpretation”

Leave a Reply

Your email address will not be published. Required fields are marked *