Is Variability Important?
Yes it is, and especially when you’re studying the relationship between on variable with another e.g. depression with peer bullying.
What do we mean?
Lets say you had a school where everyone was a bully, and 100 students suffered from depression. Could you attribute depression to the bullying? Why not?
Here, there is no variability for bullying. Everyone of our 1000 students are bullies. So we can’t tell if its causing depression in our 100 students.
Lets take another extreme example: Lets say we have 500 students who are depressed and rated themselves around 3 – 4. While we have almost no one reporting bullying. Here variability is low for depression, and almost no variability for bullying. Its going to be hard to find any association between depression and bullying.
What do we mean?
Lets say you had a school where everyone was a bully, and 100 students suffered from depression. Could you attribute depression to the bullying? Why not?
Here, there is no variability for bullying. Everyone of our 1000 students are bullies. So we can’t tell if its causing depression in our 100 students.
Lets take another extreme example: Lets say we have 500 students who are depressed and rated themselves around 3 – 4. While we have almost no one reporting bullying. Here variability is low for depression, and almost no variability for bullying. Its going to be hard to find any association between depression and bullying.
So, when we design studies and collect data, we would ideally want as much variability in our sample for the variables we are measuring.
The more diverse your school is, the greater the confidence we will have on results you report on the relationship between depression and bullying.
The more diverse your school is, the greater the confidence we will have on results you report on the relationship between depression and bullying.