There is an old saying from somewhere: “numbers don’t lie.” That phrase is made up of letters, and it is definitely a lie. Numbers are one of the easiest ways to deceive people. Darrell Huff points this out in his incredible book “How To Lie With Statistics.” When people see numbers they often get uncomfortable. Maybe it’s because of how painful learning mathematics is when we are kids. I can remember nearly coming to tears during lessons of basic algebra and cursing the black-hearted person who came up with binomials.

At that young and tender age, the abstraction of mathematics makes it seem irrelevant to daily life. Do I really need a sine and cosine to figure out that I want ice cream? The only logs I needed were the dry ones to make a bonfire on the dunes in summertime. The frustration of math built and built until I finally got out when I could: the last math class I took was statistics in 12th grade. That was 13 years ago. For my quantitative reasoning requirement in college, I took Oceanography, and I’ve avoided math ever since.

But now I’m starting to understand that it was a huge mistake. Because it might not matter if I personally don’t want to use math, but it matters that other people can use it to get the better of me. And it’s happening all the time.

Statistics is the math that we see the most often since it’s designed to develop understandings which can be passed on to the general public. Statistics are everywhere. Normally statistics are supposed to inspire confidence; few things sound more precise than a percentage with a bunch of numbers after the decimal point. But as Huff attests, that in itself is a reason to put into question what you are seeing.

Huff starts off with putting averages into context. There are three types of averages, and sort of like the music group The Fugees, they are not created equally. Lauryn Hill is the mean, the most popular average but one that can easily sway back and forth and be misleading of the group in general. Wyclef is the median, less popular but more accurate since half of the samples are above it and half below. Pras Michel is the mode, the most obscure way to represent an average. We tend to think of the mean as being the best, but a mean can be meaningless if a few very large or small values are present in the data set. Companies use the mean for things like representing average salaries. It sounds really good for a company to have an average salary of 40K, but if the president makes 1 million per year and 50 other employees make 20K/year that mean doesn’t *mean* anything.

“How To Lie With Statistics” is full of insights like this. From manipulating visual graphs and charts to the difference between percentiles and percentage points, Huff goes through a crash course in basic statistics, revealing how statisticians can make the case for either side of an argument and both be “right.” One of the best parts of the book is that it was written in 1956, back when the average family income in the United States was 3,100 dollars!! (If you just asked yourself which average I used, you’re on the right track!). Huff also uses a bunch of fun words that we should try to bring back, like chicanery (tomfoolery).

Most importantly, “How To Lie With Statistics” gives you an invaluable lesson into the basic ways that we can make numbers say exactly what we want. In order to defend yourself from the chicanery of dubious statisticians, Huff gives us a list of questions to ask whenever we are presented with statistics:

**Who says so?**– Who did the research? Did someone plant a name? If I do a study on Harvard students, that doesn’t mean the conclusions are backed by Harvard.**How does he know?**–*Sorry ladies, this book came from the 1950s, when unmarried women were referred to as “Spinsters.” I’m just going to go ahead and apologize for the entire history of man. I’m sorry. Men were dicks for years and centuries and millennia. Many still are today. I’m sorry. I hope it gets better. I’m trying to do my part, so instead of “How does he know?” let’s change that question to*– What are the statistical conditions of the information? How was it collected? Is it representative? Would the same conclusion be drawn if the test was done again?**“How do they know?”****What’s missing?**– Look out for averages that aren’t explained. Look out for information that is not conditional. Look out for hidden factors. Did the number of cases of a disease rise because the disease is becoming more prevalent, or are there just more people going to more doctors who diagnose more accurately?**Did somebody change the subject?**– Asking people what they do and knowing what they do are two different things.**Does it make sense?**– Use your head. Sometimes a conclusion seems so illogical that it must be skewed. Look out for huge increases, oddly specific decimal points, and anything that claims to generalize across a huge population.

Entrepreneurs spend a lot of time with numbers, from understanding market opportunities to presenting growth plans and generalizing their target personas. Having a strong basis in statistics makes these tasks a heck of a lot easier. Plus, it makes it easier to defend information and plans in front of potential investors, who usually have their skepticism goggles on.

Statistics is arguably the most practical domain of math, and ever since I became an analyst I’ve started to embrace what statistics are capable of. If you cringe at the idea of pulling out a calculator, the good news is that there are some very powerful programs out there that will crunch your numbers for you. But nothing can replace the basic know-how that can draw useful conclusions from a swathe of seeming overwhelming data.