You’re only as good as your data collection!!

Richard Marshall Data Collection 0 Comments

One of the things that has fascinated me in the aftermath of the 2016 US Presidential election is the discussion around polling data and how pre-election predictions of a Clinton victory were so far off the mark.

It appears that, just as in the recent Brexit vote or even in Australia’s last federal election where we saw an unexpected re-emergence of One Nation, there’s a U.S. demographic whose views were not captured in the polling data.  Why this occurred is not yet clear, however I have read a variety of potential reasons which include:

  1. The expense of traditional polling methods.
  2. Access to the internet enabling survey participation.
  3. The reticence of people to participate in polling.

There’s no denying that data is expensive to gather, process and interpret.  Traditional methods rely on people power, particularly to gather data, and there’s no wonder we have seen a shift to online surveys as a way to reduce costs.  However, if a significant proportion of a population doesn’t / can’t use or is afraid of / mistrusts the internet, how can online polling be representative?

The apparent failure of representative data collection around elections provides us some timely reminders for data collection around other aspects of our lives.

As an engineer, the collection of representative data has formed a significant part of my career.  It’s harder than it looks and takes planning and diligence.

We collect data to provide us with certainty about what has been occurring within the systems we operate on a day to day basis.  This information needs to be representative of what has happened otherwise our decisions will be flawed.

While it is important to be cost efficient, there’s no point spending money on data collection if you take a least cost approach and measure the wrong things, or not enough of the right things.

It also means that we have to pay attention to the quality of the data we measure and utilise.  For example, sensors need to be maintained and routinely calibrated to deliver the quality information we need.  Laboratories need to be audited and staff trained to ensure the analyses they provide meet the standards against which they are required to deliver.

As we have just seen – Rubbish in, Rubbish out!!!

Let's Talk

Explore how this could be applied in your organisation.

binary-1327492_1920

Leave a Reply

Your email address will not be published.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