Subscription pricing
Can you take the inverse of a positive correlation? For example:
The more cholesterol we have in our blood, the higher the risk of heart attack (increase one, increase the other)
Can I infer a decrease in cholesterol correlates to a decrease risk of heart attack?
0
9 comments
@sorooshianh185 said:
Positive correlation = grows in same directions
negative = grows in opposite direction
Yep you're right.
seems like for a correlation the terms direct/positive are interchangeable, and negative/inverse
Also from what u said (x = y is -x = -y) it seems like "inverse" is the correct term???
but the Inverse correlation is a fundamentally different relationship.
What you are referring to is simply reducing one parameter (let's say cholesterol) and the point slides to the left on graph.
Irrelevant to my Q but now I'm curious
I think the word inverse starts to get confusing.
There is a logical inverse E.g. x=y to -x = -y. Mathematical inverse, when you make the x and the y switch sides. Eg. y=2x and the inverse x=2y.
And the inverse correlation is just a negative one(downward slope).
Thanks for your responses!
@jacquestoupin470 said:
@hbochjk116 said:
I appreciate the explanation, but what I wanted to know was whether "inverse of a positive correlation" is an appropriate terminology. @jacquestoupin470
Oh, sorry I was tired last night
You were correct. The inverse of x=y is -x = -y, so it would imply a different relationship. Positive/negative just refers to the slope (up or down), so it's fixed to the correlation.
Positive correlation = grows in same directions
negative = grows in opposite direction
So the negative slop of positive correlation could also be a decrease,decrease relationship right... which is why I called it inverse bc I don't know what else it's called.... except for a negative - positive correlation hahahah
Also from what u said (x = y is -x = -y) it seems like "inverse" is the correct term???
Irrelevant to my Q but now I'm curious
@hbochjk116 said:
I appreciate the explanation, but what I wanted to know was whether "inverse of a positive correlation" is an appropriate terminology. @jacquestoupin470
Oh, sorry I was tired last night
You were correct. The logical inverse of x then is -x then -y, a mathematical inverse is switching the function's x and y. and Positive/negative direct/inverse correlation refers to direction of the relationship (up or down), so it's fixed to the correlation.
I appreciate the explanation, but what I wanted to know was whether "inverse of a positive correlation" is an appropriate terminology. @jacquestoupin470
@hbochjk116 said:
I was a lowly humanities major though, so someone who majored in science or math might be able to enlighten me. Anyone...?
lol.
Science major here.
These graphs from Wikipedia might help
Here are some different forms of correlation.
For the graph on the top left, it's probably safe to assume the formula is that of a straight line, Eg. x=y
as x or y increase the other follows.
If for the cholesterol/heart attack example, the correlation is like x=y, then yes, if one decreased the other would as well.
but context is key, without more information the correlation could be anything or only within certain parameters. I would proceed with caution.
@hbochjk116 said:
I think it's a reasonable inference to a certain point. I'm not sure the term for that is "inverse of a positive correlation" though. I know there's inverse (i.e. negative) correlation, but inverse of a positive correlation...
I was a lowly humanities major though, so someone who majored in science or math might be able to enlighten me. Anyone...?
Hey, I'm a humanities major! :D
I think it's a reasonable inference to a certain point. I'm not sure the term for that is "inverse of a positive correlation" though. I know there's inverse (i.e. negative) correlation, but inverse of a positive correlation...
I was a lowly humanities major though, so someone who majored in science or math might be able to enlighten me. Anyone...?
Not absolutely, because too much of a decrease in cholesterol could potentially increase your risk of heart attack beyond a certain level. I suppose you could infer a reverse correlation assuming you know the starting point and there aren't any wild variables.
This also isn't what an "inverse" correlation is. An inverse correlation would be where A increases and B decreases, or where A decreases and B increases. So, as cholesterol level increases, heart attack risk decreases.
inverse correlation = negative correlation
direct correlation = positive correlation