‘Big
Data’ by Jonathan Kettleborough, Training Journal, April and May 2014
I am
reviewing here the two articles that have appeared in the Training Journal
these last two months and will follow up with a further review when the third
one is published next month. I chose the
topic because, as the first article admits, ‘Big Data’ is one of the latest
corporate buzzwords, usually linked to ‘analytics’, which seems to be a fancy
word for analysis. I was interested to
see how the author would make a connection with Learning & Development.
The
first article defines what is meant by ‘Big Data’; it is a way of describing
the vast increase in information available due to improved technology. Useful examples are retailers’ information on
customers’ buying habits, the vast number of Facebook entries and the US
Government’s spying capability. The
author quotes the ‘three Vs’ that have increased the power of information
available – Volume, Velocity and Variety.
He also makes a good point that there are two other Vs that have to be
in place to make the information useful – Validity and Value.
The link
to Learning & Development is made by the author’s suggestion that we can use such data to
improve decisions; he quotes Michael Lewis’s book ‘Moneyball’ (reviewed in anearlier blog) – as an example of this principle being applied to the discovery
of talent. But, rather less convincingly,
he quotes a similar approach being used by the – maybe soon to be sacked - Sam
Allardyce, manager of West Ham United! His parallel argument in an L&D
context is that we have much more data than we realise – recruitment,
performance reviews, staff turnover, absentees etc - and we should be using
this to highlight trends and issues.
He quotes
examples of companies using such data to discover that their past assumptions
about recruitment effectiveness have been completely misguided. An anonymous example of a financial services
company was not convincing but the experiences of Google, Sprint and Cisco were;
they were apparently able to predict which employees were most likely to leave
from the analysis of all the data from their CV and employment records.
This
brought an end to the first article and my reaction at that stage was one of
scepticism; I am no statistician but I do recall from studies many years ago
that correlation is not causation and it is highly dangerous to assume so. And, to give credit to the author, this is
exactly what he addresses in the second article. The problem however is that he does so in a
rather ponderous and patronising way.
First he labours over the fact that an average in a survey does not mean
that every respondent has that same number; then he goes on at great length
about the dangers of making wrong assumptions from correlations.
He also
makes the other valid point which those of us who are interested in politics
will know – that people will focus on the interpretations of statistical data
that match their own misconceptions. He
calls it ‘data blindness’; I would call it confirmation bias. We seek and select data that matches our
prejudices, sometimes unconsciously, sometimes deliberately.
Again
the author is guilty of repetitive and patronising content as he tells us how
business success, at individual or corporate level, might be correlated to
investment in L&D but this does not necessarily mean that the investment
drove the success. He was right to
stress this obvious point – which is not always understood by those who provide
simplistic solutions to course evaluation – but not to labour it to such an
extent; it was more or less the only message in the second article. He does make a brief reference to the more
advanced analysis techniques which work on multiple correlations but does not
take this further.
In the final article,
he promises to look at the ramifications for L&D and the potentially
‘dark areas’. I will be hoping for more
challenging content to convince me that Big Data is a major issue for L&D
professionals.
Read the articles in full here;
No comments:
Post a Comment