The story behind statistics

On our Literature Review page, we described how chance can influence the outcome of research studies. We used this example: if 50 experimenters flipped a coin 100 times, most of them would get around 50/50 heads and tails. But some would get 70/30 or an even more dramatic ratio.

That's one way a study can come out with dramatic and unexpected results - pure chance. But other times, there are confounding variables. These are unexpected factors that influence the outcome of what you are studying.

--

For example, say your hypothesis is that eating Smarties cures a cold. You get a cold on Day One. You eat a lot of Smarties. On Day Five you are feeling better. Did Smarties cure your cold?

Probably not. What are the confounding variables here?
- Time (colds usually go away after five days on their own).
- The cold medicine you took on Day Two.

Another way this is expressed is "correlation is not causation". 

The Smarties were eaten, and you started to feel better. They happened at the same time - they were correlated.

But no reasonable person would accept that based on your experience, the Smarties caused your cold to get away. There was not causation. 

What if you firmly believed the Smarties cured your cold? How could find out for sure?

You'd have to do a lot more research. One strategy would be to study two groups of people who were equally sick. One would eat Smarties, one wouldn't, and we'd see if they recover at the same rate.
--

When it comes to breed and dog bite research, there are many possible confounding variables and many concerns with how data is collected and analyzed. For example:

- who is assessing dog breed?
- are they qualified to do so?
- is breed being recorded the same way every time?
- have steps been taken to see if these IDs are accurate?
- how are mixed or unknown breeds accounted for?
- could there be another factor affecting bite rates? (For example, area, socioeconomic factors, ownership, known risk factors, etc).

Never accept statistics at face value! Track down their source and use your critical thinking skills to determine if they are showing you the full picture. 

 ---

Source and further reading:

https://www.psychologyinaction.org/psychology-in-action-1/2011/10/30/what-is-a-confounding-variable

https://www.avma.org/KB/Resources/LiteratureReviews/Pages/The-Role-of-Breed-in-Dog-Bite-Risk-and-Prevention.aspx