Updated: Jun 10
At the start of this week I noticed this bar graph (below). It had been reposted by a LinkedIn 'Strategic Advisor' and subsequently 'Liked' by several LinkedIn members.
Suitably inspired, I posted an observation:
"So what? Data without meaning."
then sat back, like an angler waiting for a bite.
It didn't take long. The next post appeared from C, who described himself a student intern at a large corporation.
"In my view, the data shows that the world's most performance orientated and innovative technology companies intentionally focus on a young workforce. It seems that older employees with many years of working experience are often not the most optimal in a rapidly changing technological environment."
(Ironically, C's current home is one of the world's largest, most successful technology companies that doesn't appear on the list... we'll revisit that absence shortly).
Soon after, an Associate Professor called ‘T’ wrote a reply, generously advising C against 'jumping to simple and discriminating conclusions'.
"If the data had shown that 80% of the employees were male in all the above companies,’ T continued, ‘would you conclude that females are not the most optimal?"
W wrote next, bringing logistics and IT experience to the party. He pointed out the probable correlation between company duration and employee age. Functions such as Customer Service usually include a large proportion of young employees - a function then outsourced from more mature organizations. That will impact the age data.
W’s final suggestion was to track employees’ age at start-up failures as well as the successes. The lion’s share of early stage employees would be fresh from college.
"How would you think about the conclusion that young people shouldn't be in startup companies?"
T and W both highlighted – with succinct thought – how poorly presented data can tap into an individual's biases, consciously or otherwise.
EVERY NUMBER HAS A STORY
The graph shows an average age, which the small print calls the 'Median'. At least 'Business Insight' (the source being reposted by LinkedIn) acknowledged that choice of metric. In the accompanying article it doesn't explain why the original research source - PayScale - had chose that option.
Median is the middle measurement in a series. In this simple data set - 22,22,24,26,29,33,41,41,41 - the median is 29 (the fifth point in a series of nine).
By comparison, the Mean (the sum of ages divided by number of measures) is 31. And the Mode - the most common measurement in the series - is 41.
Hence, the average is 29, 31 and 41.
Which average would you use to illustrate your story? The ability to attract a young workforce (29) or the implied experience of a mature workforce (41)?
Or consider this: how many data points had been taken for each company on the Business Insight / LinkedIn bar graph? Plotting the quantity of each age in the sample will tell much more than just a single number.
Skew sometimes speaks.
Datum doesn't exist in isolation. It always has a context:
The ‘Top Tech Companies’ bar graph includes Apple (the world’s #1 tech company), but omits Samsung (the #2)?
If Oracle (#8) is on the list, why not SAP (#12)?
If Intel , where's Taiwan Semiconductor, or Jabil Circuit?
If Dell, where’s Lenovo or Fujitsu or Sony?
There’s Cisco but no Qualcomm.
Does HP include HP Enterprise? If so, how about TATA?
Is that not US-centric enough – then how about AT&T, Verizon and Comcast?
And where’s Danaher, Thermo Fisher Scientific, Western Digital, even a warhorse like Texas Instruments? Too boring and hardware-centric? Try Computer Sciences, Symantec, Expedia and NetApp.
All these names come from the 2016 tech subset of the Fortune 500 – as C might describe them: "performance oriented and innovative corporations". Delivery of growth is the only metric that gets you included.
Telsa Motors is private, so it isn’t on the Fortune list. Same applies to LinkedIn - prompting the question 'will it appear at all as a when it's gobbled up by Microsoft by the end of the year?'
The ‘How Old at Top Tech’ chart was reposted without context by a 'Strategic Adviser' tapping into a series of biases. Bias about age, bias about geography (California), bias about the definition of ‘tech’.
In the same way that a symphony, a film, or a dance can be 'read' for meaning, so can any histogram, table, bar graph or pie chart.
However simple the message and 'easy' the supporting evidence, there'll always be an agenda. Accept nothing at face value.
And finally, let's give ourselves a collective pat on the back for reaching the end of this post without referring to ‘lies, damned lies, and statistics’.
For Rutherposts direct to you inbox, subscribe *here*