226 billion dollars is lost annually in the United States due to absenteeism¹. It’s a hidden bottom-line killer, so companies would be wise to minimize it.
Many companies use a points system to try to combat absenteeism, where managers don’t judge valid absences, but “points” are given according to invalid absences, tardiness, and when an employee doesn’t call in at all.
For example, a tardy employee may receive 0.5 points, an absent employee may get 1 point, and not calling in at all gets 3 points. These add up to different consequences, such as a verbal warning at 6 points, a written warning at 7 points, a final warning at 8 points, and termination at 9 points.
Other tactics include dis-qualifying promotions, pay upgrades, or other benefits. However, all these responses are reactive, instead of proactive. Further, they don’t reduce valid absentee hours, which still cost the company money. What if we could solve the problem at its root, instead of addressing the symptoms? What if we could understand the reasons behind absenteeism, and address them accordingly?
Startup101: How Do I Divide Equity Among My Cofounders And Employees? | Data Driven Investor
So you have just spent six months building a prototype with an early team, punting the question on how to divide the…
Analyzing Absenteeism Data
To answer these questions, I analyzed an absenteeism dataset from a courier company with the no-code AutoML tool Apteo. I selected
Absenteeism in hours as the KPI column, and all the other columns as attributes.
Right off the bat, we can see how each attribute impacts absenteeism time. At the bottom of the list are things like whether or not the employee is a social drinker or smoker, which means that these attributes are basically meaningless in predicting absenteeism.
The most important attribute is
reason for absence. The UCI source describes this as “absences attested by the International Code of Diseases stratified into categories,” which means that this refers to health-related absences.
Out of 740 records of absenteeism, only 43 records don’t have a medical reason for absence, and 33 records have the value
26, or “unjustified absence.” Thus, almost 90% of absences are for valid medical reasons.
We can see that category
9 has a particularly high number of hours, with a median of 16 hours lost to absenteeism, and a third quartile of 32 absent hours. The original dataset tells us that this category refers to “diseases of the circulatory system,” which means that serious diseases are responsible for the highest average absent hours.
This is not where the most hours are lost overall, however. It’s just where the average absence has the highest hours lost. To find where the highest number of hours are lost, overall, we can plot the sum of absenteeism hours against the reason for absence.
19 are responsible for the highest number of hours lost, referring to musculoskeletal diseases and general injuries, respectively.
Musculoskeletal diseases include tendinitis, carpal tunnel, bone fractures, and so on. The dataset is sourced from a courier company, which means that this data reflects relatively hard labor, compared to a typical desk job. Research shows that musculoskeletal diseases are indeed quite common in the courier industry, as repetitive, heavy lifting takes its toll on the body.
Reducing Absentee Hours
Since most absentee hours are lost due to these injuries, courier companies could effectively reduce absentee hours by implementing a more ergonomic workplace, training proper lifting and posture, and reducing overtime — all things that would reduce injuries.
Looking at the original list of key factors, we can see that the second-most important attribute is
work load average/day, providing evidence to the hypothesis that overworking leads to greater injury, and thus absenteeism hours.
One research paper found that “high mental stress at work” was an important risk factor for work-related injury in the courier industry, which means that companies could also reduce absentee hours by implementing workplace wellness schemes and better supporting stressed employees.
How Any Company Can Reduce Absentee Hours
Perhaps you’re in a completely different industry, in which case, your employees’ absentee hours may have completely different causes.
Fortunately, the power of predictive insights is that it’s widely applicable. You can find predictive insights in any dataset that has a KPI column and attribute columns, so you can run the same analysis in any industry. If you don’t have a dataset that looks like the one we’ve analyzed, you can simply make one.