Three kinds of falsehoods: lies, damned lies and STATISTICS? – Elizabeth Merrall, S-cubed statistician, fights back!

The COVID-19 times have been a nice time for reconnecting with old friends. A chat with some university friends brought up that line that usually comes up about my profession, when talking in spheres outside it: “lies, damned lies and statistics”! Having spent most of my working life as a statistician, I thought it was time to tackle this common misnomer head-on and here are my initial thoughts.

It’s not us! Honest!

Interestingly enough, the earliest reference to the quote has been traced back to an accountancy journal in 1886, and so appears to be a misattribution to statistics (phew!), actually alluding to what we have come to know as creative accounting:

“Whereupon counsel on the other side was heard to explain to his client that there were three sorts of liars, the common or garden liar … the damnable liar who is fortunately rather a rara avis in decent society, and lastly the expert, …” (For more info:

Creative Statistics

Creative Statistics

Statistical Whispers? Communication Problems?

Communication Problems

Communication Problems

So, we could stop there – but there is something to be said about the delicate relationship between statisticians and other specialists, and the potential for misunderstandings. As the COVID-19 pandemic has unfolded, statisticians have had an important role to play in understanding the limitations of the available data and interpreting associated analyses (a nice example here:; and this communication has not always been effective (as illustrated here….: and consolidated here: As is the case for all technical disciplines, I would agree that communication with other disciplines is a constant challenge that we are always working on.

Problem of too many cooks analyzing the broth

Statistical Ponies

Statistical Ponies

Can statisticians manipulate data and conjure up statistics to fit a given strategy or message? Assuming that the statistician is behaving professionally, then no. It is possible, that if a researcher presents a problem to three different statisticians, he or she may be presented with three different solutions. This is understandably frustrating for the researcher but alas, another famous statistics-related quote applies here: “all models are wrong, but some are useful”. Different study designs, how the data are collected, levels of missing data and different analyses of the data will make different assumptions about the problem at hand and these assumptions need discussion with the subject matter experts or testing for their robustness, should they not apply in reality. And as with all realms of science, peer review by a second statistician is also important for catching any blind spots in the proposed solution. It has to be said that we cannot do our job well without good collaboration with both statisticians and other specialists.

Statistical Defence

In summary, I hope that my lines of defence are convincing – although sparring is always welcome! Statistics is a misunderstood, and sometimes misrepresented, discipline but not a dishonest one! And whilst we statisticians are typically leaning towards the more introverted end of the spectrum* – possibly due to our delving-into-the-details natures – we love to have a chat about a meaty data or analysis issue and are always here to help.

*An alternative hypothesis I am now curious to test!

Need a meaty statistical discussion?

Contact our Biostatistics team for a good discussion or for their help, or reach out via the contact page here.