Home' HR Monthly : June 2015 Contents 22
FEATURE: BIG DATA
FOUR STEPS TO
WRESTLE THE DATA
Dr David Schmidtchen
group manager of the human
capital research and evaluation
group of the Australian Public
WRESTLE THE DATA
1. SET GOALS.
Ask the questions:
Why do you need the data?
What are you tr ying
Why are you engaged in this?
2. DEVELOP A THEORY.
By thinking about how the
data might fit together, you
have a framework in which
you can then go and collect
information, giving you
a hypothesis that can be
3. WHAT DATA IS
You may be using the
organisation’s data or,
depending on your goals,
it could be your company’s
Twitter feed. What do you
need to collect to inform or
test your hypothesis?
4. IS THIS A PARTIAL VIEW OR
A COMPLETE VIEW?
Consider comparing your data
with other large data sets
available from organisations
such as the Australian Public
Service’s employee census.
Advance your career by enrolling in the AHRI Practising
Certification Program - a new industry recognised,
2 year, part-time postgraduate program.
This work integrated program will equip you with
globally benchmarked HR capabilities which are
essential for today’s practising HR professionals.
EnrolmEnts closE 10 July 2015
Enrol toDAy: AHRI.COM.Au/EduCATION
140109AU - Half page adv HRM.pdf 1 31/07/2014 8:45:32 AM
tapping into, such as exit interviews and assessment centre
results [see below]. It uses some ex ternal tools, but it’s internal
information about the company.
Recruitment and staffing is one area of HR where people
analytics could lead to major improvements. Polaris Consulting
director, Geoff De Lacy says that, despite the existence of
recruitment models that rely on data from the government
and major corporations, they remain “mind-bogglingly
“Recruiters still rely on line models that translate into
face-to-face interviews against some sort of structured or general
questioning. At the end of the day, they are proven to be limited
in terms of reliability,” he says.
Assessment centres, which some organisations use extensively
for executive and graduate recruitment, are a slightly better
method, he says. The centres combine inter views, ex tensive
testing, experiential exercises and the views of two or more
observers. However, both methods still fall short of the ideal
that people analytics makes possible, says De Lacy.
“It’s said that individual companies will develop ‘secret
recipes’ for sourcing, analysing and evaluating hires based
on their own data and statistical analysis of the makeup of
their ideal employee,” De Lacy says. “With big data not only
increasing in volume, but also in velocity and variety, it’s time
to consider the f ull utilisation of available research in addition
to the traditional panel or one-on-one interviews and
If we consider this type of internal information as ‘structured
data’, or data that is pu rposefully collected, then we begin to see
how the traditional big data – the numbers in a spreadsheet –
make sense as a point of comparison.
At the Australian Public Ser vice Commission, Dr David
Schmidtchen, group manager of the human capital research and
evaluation group, says it runs four big data collections each year,
trawled from personnel data in public sector HR systems.
“We collect attitude and opinion data in the Australian public
ser vice [APS] employee census, with two-thirds of employees –
99,000 people – responding to the survey,” he says. While much
of this data is purposefully collected, some of it is captured
‘unstructured data’, information generated as a result of an
organisation or company’s primary function.
“Out of the employee census, I get about 45,000 usable
comments which are essentially unstructured data. From the
APS employment database, we end up with 3.8 million data
points every quarter,” says Schmidtchen.
On that scale, human behaviours that are usually quite
difficult to measure start to yield meaning. For example, a small
sample of diversity data on Indigenous visibility is difficult to put
into context. Is this small amount of data indicative of a greater
trend? But with some pointed questions and intense filtering, the
APS is able to draw conclusions from particular groups because
of the sheer volume of unstructured data the census captures.
The real skill, says Schmidtchen, is bringing the discovered
and deliberately collected data together, “ensuring that we
understand what it means, and how that is going to fit into »
25/05/15 4:37 PM
Links Archive May 2015 July 2015 Navigation Previous Page Next Page