The Student Room Group

what does this mean - s1?

The questions is:
Some values, (x, y), of a bivariate distribution are plotted on a scatter diagram and a regression
line is to be drawn. Explain how to decide whether the regression line of y on x or the regression
line of x on y is appropriate.

The mark scheme says:
If neither variable contr’d (or indep)
AND want est y from x: use y on x
What does it mean by this
Original post by BULL14
The questions is:
Some values, (x, y), of a bivariate distribution are plotted on a scatter diagram and a regression
line is to be drawn. Explain how to decide whether the regression line of y on x or the regression
line of x on y is appropriate.

The mark scheme says:
If neither variable contr’d (or indep)
AND want est y from x: use y on x
What does it mean by this


I hate these kinds of questions, the answer is always tricky to get to.

Whenever you are stuck with a maths problem, try to picture it with a simple example.

Okay, for the first line about neither being independent variables / control, think of this:

Usually when we have something and call it x, and have something else that is related to x in some way, y, then we investigate it by changing x and measuring y.

Eg x is number of hours of sunlight a plant gets, y is its height.

X would be our control variable there, which is just like saying what the regression y on x is, I.e. What happens to y when x changes.

So when the MS says when neither are a control, it means that it's not immediately clear what you want the relationship to be either way, I.e. We don't know whether the question is 'what happens to y when you change x' (y on x) or 'what happens to x when you change y' (x on y)

An example of this is a plot of height vs weight, which one is the control variable? It could be either!

So that's the 1st line on the MS.

Then the 2nd line is a follow up, e.g. 'If we want how height changes with weight, then regress height on weight'.

This is merely an explanation of what the markscheme is saying, not WHY it is saying it because I would struggle to justify it myself..
Reply 2
Original post by Kevin De Bruyne
I hate these kinds of questions, the answer is always tricky to get to.

Whenever you are stuck with a maths problem, try to picture it with a simple example.

Okay, for the first line about neither being independent variables / control, think of this:

Usually when we have something and call it x, and have something else that is related to x in some way, y, then we investigate it by changing x and measuring y.

Eg x is number of hours of sunlight a plant gets, y is its height.

X would be our control variable there, which is just like saying what the regression y on x is, I.e. What happens to y when x changes.

So when the MS says when neither are a control, it means that it's not immediately clear what you want the relationship to be either way, I.e. We don't know whether the question is 'what happens to y when you change x' (y on x) or 'what happens to x when you change y' (x on y)

An example of this is a plot of height vs weight, which one is the control variable? It could be either!

So that's the 1st line on the MS.

Then the 2nd line is a follow up, e.g. 'If we want how height changes with weight, then regress height on weight'.

This is merely an explanation of what the markscheme is saying, not WHY it is saying it because I would struggle to justify it myself..



Thank you so much:smile:
However if both variable are independent would I draw y on x or x on y?
Original post by BULL14
Thank you so much:smile:
However if both variable are independent would I draw y on x or x on y?


Good question, but I don't think you would ever have this - because if two variables are independent then you won't be interested in a relationship between them (eg shoe size vs pocket money)

Good question though because it helps your understanding to ask 'what if' questions :smile:

Quick Reply

Latest