How to be a data analyst

In which I try to lift he fog of war on numbers and reveal it all to be a feint.

Windows taskbar with Excel window

As a data analyst I have often believed that it is more important to look right rather than be right. Accuracy will always lose to style; it seems to be the way of the world. I have been told by previous managers that “these numbers don’t feel right” and “could I go and check them”. Later after changing nothing but the colour scheme and some borders the figures all of a sudden feel better.

Double Decker buses are also a standard unit
Double Decker buses are also a standard unit

It’s like evidence depends purely on the appearance of the information, this is why we live in a world where infographics seem to be the fad du jour. Losing a percentage of the rainforest each day can only be measured in Wales (a standard unit of measurement, much like Double Decker buses for length or Eiffel towers for height) rather than hectares.

I once ‘wowed’ a manager by placing a button on a spreadsheet that allowed him to print it off. The fact I had placed it directly below the print button in the toolbar (and it simply was a macro that replayed a press of that button) was beside the point. It was a button that printed.

I often try to follow good chart principles when I present my work, yet I know most people are amazed by a 3D pie chart (despite this being the refuge of an analyst who obviously has no clue what they are doing) rather than a clean chart with no visual noise.

Pic-Bad-Pie-Chart

 

People fall for confidence tricksters; they are easily swayed by shiny pennies and can be manipulated. The good data analyst should play by the rules and be neutral in their data presentation. The bottom line is that it takes a lot of intelligence to figure out what is real in data and what, if anything, is even remotely meaningful

But the successful data analyst practises ‘Follow the lady’ with tables and charts.

Among some of the key rules for successful data analysis are:

  • If any discrepancies stop you from telling a good story then ignore them.
  • Should the results provide good news then they are accurate. If they are bad the data is flawed or insignificant.
  • Round up income, round down costs.
  • Scale is optional, a 50% increase on £2 is better than a 25% increase on £2m
  • People do not understand numbers bigger than ten, if they cannot count it on their fingers do not use it or find a way to bring it down
  • The majority of people do not understand why the average of a set of average is bad
  • Typeface is of more importance than actual numbers
  • Bury bad news in tables, put good news in charts
  • Never underestimate the power of a double border around a table
  • A drop in anything is due to external factors, an increase is because of something we did (the reverse for costs).
  • Analysis should be there to conform not inform.
  • A dashboard should look like a car dashboard. Sales performance is analogous to how much petrol is in the tank.
  • Never overestimate the power of a complicated sounding methodology.

Obviously as a good-hearted analyst I would never practice any of these underhand methods of manipulation. To prove the point here is a chart showing how many times I have used such nefarious tactics.

CHART-Rubbish Chart

Write a new post in response to today’s one-word prompt. Not sure how to participate? Here are the steps to get started.

Source: Fog

Author: Daddysaurus

Ah, so you worked out the riddle. You just needed to use dwarfish and the doors to Geek Ergo Sum opened. Or perhaps you just used Google. Either way you are here, on my little corner of the Internet.

4 thoughts on “How to be a data analyst”

  1. “Round up income, round down costs.” Can confirm. The amount of times I’ve been forced to round exact results into approximates for people to look better blows my mind.

    Like

Think inside the box, feel free to leave a comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s