How Statistics Lie

The first part of this website was about pseudoscience, which I found important to the mission of bringing awareness to misinformation. Now, I'm transitioning to another common way we are fed faulty information: statistics. Statistics tend to be numbers, whether percentages or proportions and they are invaluable for summarizing and presenting data. Unfortunately, with great importance, come great potential to be misused. On this page, I will be going over the most common deceptions people should be aware of in the world of statistics.

1. Graphs

Graphs are among the most diverse ways to manipulate and display data. Ranging from things like changing the scales, to completely removing the axes. Here are a few things to look for when you see graphs on the news or any media outlet.

a. Make sure the y-axis starts at 0, or at least an appropriate number.

-Graphs usually have the y-axis starting at zero, unless you're dealing with large numbers, then check how much the intervals are increasing on the y-axis. Otherwise, it can make information look something like this. . .

This looks like the taxes would have an enormous increase right? Well, actually it's just a little less than 5%, but because the y-axis starts at 34, it really messes with the graph's image.

b. Make sure the graph shows the whole picture, and not just what they want you to see. This happens a lot when someone wants you to believe something that just isn't true, so they'll only show you part of the picture.

This graph had an overhead label that said, "Global Warming Out of Control!"

Yes, it shows temperatures increasing, but look at that x-axis. . . it starts in the winter and ends in the summer, so of course, it will only show temperatures increasing.

c. Make sure the graph actually has axes and labels

At first glance, it seems that Lanacane does a much better job than Hydrocortisone, however, due to the lack of any labels, we really can't discern ANYTHING about this comparison, and you truly cannot trust it because they are leaving out the most important piece of information in statistics - numbers.

-Sources of graphs: https://www.statisticshowto.com/misleading-graphs/

2. Context is Everything

This one applies to everything in life, to be honest. In the world of statistics, however, this is also true. With all of the headlines and claims being thrown at us, it's important to do proper research and know your sources are trustworthy and relevant. Otherwise, you might hear an unbelievable statistic, but when put into context you would find out it really isn't that big of a deal.

Example: "People who eat processed meats are at a 20% higher risk of getting bowel cancer."

-Wow! At first, that seems very alarming, however, if you do a little more research, you'll realize the chance of getting bowel cancer specifically is around 4.4%, and eating processed meats increases that by 20% which is just to 5.2%. Not that alarming.

Source of study: https://academic.oup.com/ije/article/49/1/246/5470096#133824902

3. Correlation Does NOT Equal Causation

To keep it simple, a correlation is when one variable is RELATED to another variable, and causation is when one variable CAUSES another variable to happen. This is one of the most important distinctions in statistics. Two graphs can look identical and you might be tempted to say one causes the other, but there truly is a difference between two things being related, and them actually having an effect on each other.

Does this mean eating more cheese will cause more people to die from bedsheet entanglement? Definitely not, it just means that they happen to be very closely correlated.

Source of graph: https://www.tylervigen.com/spurious-correlations

4. Wording is (also) Everything

There are science experiments and surveys happening every single day, and with these, comes the need for scientists to be extremely careful with how they word questions. It may not seem like much, but studies have actually shown that the wording used on surveys to participants can wildly change the data.

Example: 1 in 4 women are sexually assaulted on campus.

-This was a study that came out in 2015 and was headlined by the New York Times. This is by far the most concerning statistic we've seen so far, but the reality is that it just isn't true. Once the work had been fully published and examined, it appeared that the survey questions were very unclear. One example was along the lines of, "have you ever felt uncomfortable at a party because a man looked at you?" These kinds of questions resulted in a larger portion of women saying they had been sexually assaulted when the real statistic is much lower.

Source of article: https://www.nytimes.com/2015/09/22/us/a-third-of-college-women-experience-unwanted-sexual-contact-study-finds.html