Statistics is an instrument of truth.

You can make truer statements with statistics than without. Statements that apply to some fraction of people for some fraction of the time can be stated as such, instead of as absolutes.

No one ‘always’ does anything. If you said to someone “You always do this!” the statement can only be false.

‘All’ people of a certain demographic don’t do any one thing. “All women (or all men) behave a certain way”, is less true than a statement that specifies what fraction of men or women you mean, and how you estimated this fraction.

Degree of knowledge and applicability can only be expressed with completeness using statistical tools like sample size, probabilities, and likelihoods. Without them, you are stuck with either the wishy-washy ‘sometimes’ or the overconfident ‘always’.

The phrase ‘lies, damn lies, and statistics’ was created in response to bad practitioners of statistics — people waving their hands and deflecting from the truth with jargon they know their audience doesn’t understand, trying to give an impression they know is incorrect.

Horrible as this is, statistics is not inherently prone to promoting falsehoods. With greater literacy, statistics can be an instrument of truth with nuance and accuracy.

Originally appeared on Medium.

One thought on “Statistics is an instrument of truth.”

  1. I agree. The problem is the manipulation of data/statistics to paint a picture not inherently implied. I worked for an animal pharmaceuticals/biologicals company for years and once behind the veil…it was startling. We see this a lot in our mainstream media and social media outlets. I watched as the CDC changed the scale of their graphs to paint a certain kind of picture over the last 18 months. Nevermind the data disasters that happened.

    I like what you mention here about bad practitioners of statistics. I am a service provider and much of the time those who have had a bad experience with my industry will say “I tried that, it didn’t work”. Yes, but you tried a poorly executed version…not sure that counts. Same here. You can’t say statistics are bad. They can be misrepresented or misused…but the numbers are such as they are at the time.

    This was very impactful to me “expressed with completeness using statistical tools like sample size, probabilities, and likelihoods.” Using and understanding these tools and their interplay is valuable in forming an opinion.

    Thank you!

Leave a Reply

Your email address will not be published.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