When we say something is statistically significant what does that mean?
In: Mathematics
Simply put: something is statistically significant when you can be more certain than not that the result actually represents reality, and wasn’t just unlucky coincidence.
Imagine I have a ball pit. Exactly one of every three balls is red, the rest blue.
If I pull out balls at random I might very well be unlucky and pull out 6 red balls in a row. This might lead me to the false conclusion that *all* the balls in the ball pit are red. If I pull out enough balls however, the ratio starts to balance, and I’ll end up with a *roughly* 1:3 ratio. That’s when I’ve a statistically significant result, because the odds I have an outlier becomes very small.
It means it’s more likely that there is an effect than there no effect.
Let’s take an example: you roll a dice repeatedly. Usually you’d expect to get roughly equal amounts of each number after rolling for a while. You roll ten times and get the 6 four times. You suspect the dice might be unfair because you expexted only 1.66 times.
But 10 is a small sample, and so this result is not statistically significant. But if you did it 10,000 times and got 4,000 times the 6 it would be pretty likely that the dice is manipulated.
I will spare you the math behind, but in many cases you define a niveau (often 5%) about how unlikely a result would need to be to assume it didnt occur naturally. It of course could always have happened. Unlike many people assume there is no “statistically impossible” result. Even getting a 6 each time is no real proof, but it’s statistically significant that this dice is not normal
Latest Answers