Tuesday, June 21, 2016

Don't Believe Statistics

Back in college, I took a statistics class and had to write a paper about the skewing of statistics particularly as done in the media. I think I did pretty well on it. And I did learn not to believe statistics as presented anywhere without thinking about them.

Here's an example: A recent study said that being married or having a college education made you more likely to get brain cancer. I'll bet that result was a surprise to the researchers. But you really need to dig a little deeper into the research to understand that isn't really the truth.

"Using a huge registry of health and other data on all Swedes, European researchers found that people with at least three years of college had about a 20 percent higher risk of glioma (the most common brain tumor) as those with only an elementary-school education. And married men had a 23 percent higher risk of glioma than never-married men." 

This is how they did the research. But it is important to realize that there is more to this than that. Let me clarify on two points: College educated people tend to be wealthier than those with lesser education. Well off people usually go to the doctor more frequently than others, simply because they can afford it. They also might be more aware of differences in their health (slurring words) that would cause them to go to the doctor themselves. Married men also have been shown to go to the doctor more often (at their wives' behest) than unmarried men.

These cancers are also very rare anyway and may not even kill you if you have one. They are slow growing and are often found at autopsy with other causes of death.

So the end verdict on this study:

Being married or educated seems linked to being diagnosed with some kinds of brain tumor, likely because of socioeconomic rather than biological differences — a useful reminder that when studies count diagnoses they’re not necessarily counting every case.

The lesson learned is do not believe statistics. This is why often additional studies are required to verify the results.

No comments: