“I only believe in statistics that I doctored myself.” — Winston Churchill
When I recently went to the time-traveling mailbox at my lake house, I found this letter from Winston Churchill inside. Apparently, he really took this question on Quora to heart. Here’s what it said.
Hey, it’s your buddy Winston! Apart from coining this awesome quote about how bogus all this hypothesizing is, I’ve also applied to be trained for cavalry instead of infantry at military school, because the required grade was lower and I didn’t have to learn mathematics, which I HATED.
Also, I failed the entry exam three times, so I figured better make this as easy as possible, LOL!
I think this means I’m the perfect guy for the job of answering this question — with statistics! Let’s do this!!
If you actually applied this statistics nonsense to your question, you’d call this a directional hypothesis: the independent variable level of honesty (for example measured in percentage of statements that are true) has a negative impact on the dependent variable intelligence (for example IQ).
If your hypothesis is correct it indicates the following: The higher the percentage of honest statements one makes, the lower one’s IQ.
Note: I know IQ is not the perfect measure of intelligence, but mine is pretty high so let’s just roll with it ok? JK haha I was the student with the lowest grades in the lowest class!
The trouble with these hypothesis-thingies is you can never prove they’re right, because it’s impossible to check them for every possible observation.
What you can do is flip the hypothesis on its head and then disprove it — because all it takes is one piece of contrary evidence.
Here, we’d say we assume a higher level of honesty has no negative impact on intelligence. Now we can try to find evidence against this.
This is called rejecting the null hypothesis (called H0) — I read that in one of those math books I hated!
Because rejecting a directional H0 hypothesis is only valuable for those few data points you’re looking at, but what you want is generally valid conclusion, you would make H0 non-directional, including both negative and positive effects.
In this case: the level of honesty has no influence on intelligence whatsoever.
If you can now find a correlation, positive or negative, between those two variables for a sufficiently large data set, you can reject the null hypothesis on a high significance level (meaning with great certainty — for us normal people here — geez these math guys).
For example, if we measured the IQ of 10,000 people, along with the average percentage of their honest statements, we could throw the data into a statistics software like SPSS (not that I know about that kind of stuff) and then measure correlation.
I’m pretty sure if we ran the numbers, we would find some correlation, if it’s albeit a significant one, that’s impossible to predict.
So after going through all this trouble, we could maaaaybe reject H0 and are now…
…none the wiser! WTF?!
Now all we’ve shown is that we can’t say with certainty that there’s NO correlation.
So there is likely some effect between the two, but what effect exactly and how strong it is…can’t tell!
DAMMIT! Why is stats so hard? I knew why I went into politics!
Now I need a cigar.
Ahh, much better.
Told you this was all nonsense! We can hypothesize all day and get nowhere.
Let me tell you what the real question here is:
Winston Churchill’s IQ was never tested, but is believed to be in the range of genius (140 to 160) or greater. He was famous for his sometimes even blunt honesty, which would often get him into trouble, but remained one of his highest values all through his life.