LESSON: Uses and Misuses of Statistics

Misleading Statistics in the News

Numbers are being used every day. In the news, in politics, in our jobs, and even in our social interactions. They are used to convince, explain, but also to deflect from what is truly happening. Too often, numbers appear as abstract, objective and difficult.

We've learned quite a bit about surveys, observational studies and experiments. We know what each of these data collection methods require in order to produce accurate results. But these results can be presented in ways to say whatever we want them to say! That's why it's so important that when we hear numbers thrown at us, we interpret and act on them carefully. 

In this video, Dr. Sanne Blauw will explain 5 ways to look out for that statistics can be misrepresented to us.

The Good Looking Graph

Dr. Blauw cautioned us to be careful when presented with graphs, and more specifically, "The Good Looking Graph". Graphs can be very dramatic. It's easy for people to look at them quickly and make judgements without getting bogged down in the context of where the numbers on the graphs come from. But the context is precisely what is most important. 

Graphs can be manipulated to dramatize results. Oftentimes, by just modifying the y-axis of a graph, one can drastically change the story that is being told. Take a look at this graph:

Changing the starting point of the y-axis from 0 to 3650 makes the increases and decreases in the line look much more dramatic. 

While it's not important that all graphs start at 0, it's important to know the context of the information being presented. Once we know the context of the graph and what it is representing, then it's easier for us to make rational interpretations on how it's presented to us. 

Watch this video to learn more:



The Polluted Poll

We have learned that there are many ways in which polls and surveys can be biased:

• NonResponse Bias
• Undercoverage
• Leading Questions
• Social Desirability
• Convenience and Voluntary Samples

But just because there are rules to follow to create unbiased surveys doesn't mean people always follow them. This can be intentional or unintentional. But we will keep being presented with the results of these surveys regardless. 

The most important thing with surveys is for people to be educated about how to interpret the results that are presented to them and how to identify sources of bias. 

We've covered this topic quite a bit back in Unit 1.

Correlation vs. Causation

Dr. Blauw mentioned correlation vs. causation in her speech, which is one of the most common ways that statistics can be misused and misrepresented to us. Maybe you've heard the phrase, "Correlation does not imply causation." Let's talk briefly about what that means.

Correlation means that two things are related to each other in some way. 

Causation means the one thing causes another thing to happen. 

Consider a simple example:

An ice cream truck drives around a neighborhood on a sunny day. The ice cream truck and the sunny day are related to each other because both things bring a lot of kids outside, so they are correlated. However, the sunny day did not CAUSE the ice cream truck to come out. Neither did the ice cream truck CAUSE the sun to come out.

The driver of the ice cream truck put the keys in the ignition and pressed on the gas. Therefore the driver driving the truck caused it to drive around the neighborhood. This would be causation.

Many times statistics are presented to us to imply causation when in fact it's only correlation. 

We'll talk about this more in upcoming lessons.