Three Kinds of Bad Data

To be effective in influencer or content marketing, you need proof-points. You need to nail your tale with memorable nuggets, compelling case studies and robust research.

Sadly, a lot of material is let down by data abuse that ranges from a lazy neglect of context through to the levels of systematic torture required to extract false confessions.

Here are three abuses of data that are disturbingly common yet never fail to leave, at the very least, a bad taste in the minds of the audience:

Context-less Data

This abuse play’s upon an assumed innumeracy in the audience and is a favourite of tabloid journalists. For example:

“Sunbeds blamed for 100 deaths a year” – how many people die in the UK each year? It’s around 560,000. So sunbeds (or possibly just the misuse of sunbeds) are responsible for 0.018% of all deaths in the UK each year.

Or compare, these two articles on the same topic:

“Universities see 15% slump in UK applicants as school leavers shun huge rise in fees” – from the Daily Mail
“Down but not out” – from the Blighty blog of the Economist

One creates a sensationalist article from a single, large, context-free figure. The other is a measured analysis of the reality behind the number.

Leading Questions

If a helpfully large number tells your story then it’s surely tempting not to dig too deep. But what if you can’t find suitable data in the first place? You create it, of course!

We’ve all encountered survey questions crafted to lead the respondent to the “right” choice:

“Do you believe that the use of biased questions to create desired survey results is a serious problem for content marketing … or are you so wilfully and irredeemably incompetent as not to have noticed?” Vote now!

Wilful Misinterpretation

And if your myopic and distorted use of results from biased surveys still don’t tell the story you need? What then?

There’s at least one research firm that simply misinterprets or misrepresents the actual results. Changing the subject matter to protect the guilty – but not altering the form – I’ve encountered this situation several times in research commissioned by different companies:

Question: Do you believe that government expenditure would be reduced if the Armed Forces were abolished?
Analysis: 95% of people believe the armed forces should be abolished.

Whatever the results of the survey, the question asks about the expected effect of an action. The “analysis” represents the response as support for the action. Very different.

Lies, Damned Lies, Statistics and…

Someone who wasn’t Mark Twain or Disraeli once said,

“There are three kinds of lies: lies, damned lies, and statistics”

I think there’s a fourth, even more pernicious, form: “damned statistics”.

In any situation, but especially in a discipline that relies on an audience going beyond the headline, credibility and integrity are all you have. Anyone tempted to use these distortions in support of an influencer of content marketing approach does so at their peril.