Saturday, August 6, 2016

Statistics Fun!

Yea! Let's have some statistics fun!

fun theme park skipping happy amusement park

What? There's no such thing? Well, that's probably true, but let's give it a whirl anyway.

Let me start out by saying I'm not a statistician. That takes more school and math than I am willing to do, but research and statistics cover most my background and I get paid to make meaning of globs of data, so let's pretend like I'm qualified.

Meaningful statistics are a combination of three components. If any of said components is lacking, your statistics are bullshit. Which I would say 90% of statistics people tout on the Internet are bullshit. See that made up number? It's not real, but it feels real so you want to believe it. Stop doing that.

First, we need a good data sources. Trust nothing that could have an alternative agenda or is a biased organization for raw data sets. If you don't know what I'm talking about, please check out any guns and ammo magazine or pro-life website. If they're gathering data themselves, it's probably wrong. Data collected from the Department of Justice or National Health Organization, those are real data sources; Department of Police Services or Organization for Healthy Living may not. Beware of real-sounding organizations that are just propaganda machines because they are very good at creating names that sound legitimate.


Second, is what's being tested the real data. Wait, I'm asking you to check the data? Yes, because pulling up shit results in shit statistics. If we're looking at the percent of Chipotle burritos with salmonella, it doesn’t make any sense to pull the results of the taco shells. Yes, someone is shitting somewhere, but the taco data doesn’t tell us if the burrito was to blame.

And finally, the data needs to be interpreted properly. I mean, you can’t run a chi-squared test if you’re comparing two separate populations. That would just be ridiculous.

What? Too technical?

How about you can’t say more puppies were adopted than kittens this month because 60% of the puppies were adopted. I mean 60% is obviously more than half, but that isn’t what the statistic is telling us. This goes back to the data set. If we have 100 kittens and only 50 puppies, then 60% of puppies being adopted is 30 puppies adopted. And 40% of kittens being adopted is 40 kittens. More kittens were adopted than puppies even though the statistic made it sound the other way.

fun roller coaster roller coaster

What am I trying to get at here? Don’t trust all the statistics you read. Do your research before you re-quote something, even if it proves your point. Especially if a politician is using it. And stop flinging around bad statistics to back up your bad ideals. To misquote the immortal John Oliver, “Don’t bring feelings to a fact fight.”*

Thank you to for the use of their gifs.

No comments:

Post a Comment