Describing correlation (1 Viewer)

[Hehehe22]

So I know a few of you, @qweeosh @gossipgirllll @bigupsanky are discussing how to describe correlation using r. I need someone to confirm once and for all what the criteria for strong/moderate/weak is, because it really shouldn't be this hard

This is what Cambridge says:

And this is from the New Senior textbook:

 

[gossipgirllll]

i do standard, so for us they taught us what the new senior textbook says and ive been doing hsc questions and been getting it right soo it should be the second way which is correct i guess. not sure what cambridge is saying

 

[qweeosh]

So I know a few of you, @qweeosh @gossipgirllll @bigupsanky are discussing how to describe correlation using r. I need someone to confirm once and for all what the criteria for strong/moderate/weak is, because it really shouldn't be this hard

This is what Cambridge says:
View attachment 49309
And this is from the New Senior textbook:
View attachment 49310
View attachment 49311

I was taught with the new senior textbook so I would naturally lean towards the latter. However, the Cambridge isn't exactly incorrect either? I think it depends on the context the question is asked e.g. whether a graph is also given with the question and putting a judgement based on the appearance.

 

[qweeosh]

I was taught with the new senior textbook so I would naturally lean towards the latter. However, the Cambridge isn't exactly incorrect either? I think it depends on the context the question is asked e.g. whether a graph is also given with the question and putting a judgement based on the appearance.

although I personally would categorise -+ 0.5 as moderate.

 

[gossipgirllll]

this is what my standard textbook sayss

 

[gossipgirllll]

although I personally would categorise -+ 0.5 as moderate.

yeah i feel like in an exam id write down the correlation coefficient and then state its strength with justification so the marker can understand ifykwim

 

[reniiiblaseee]

View attachment 49313
this is what my standard textbook sayss

yep i remember doing a question once where it was 0.65 and i said moderate but the exam paper said it was strong i was so confused

 

[qweeosh]

yep i remember doing a question once where it was 0.65 and i said moderate but the exam paper said it was strong i was so confused

Moderately strong?

 

[gossipgirllll]

whatttt im so confuseddd

 

[qweeosh]

whatttt im so confuseddd

no it was a bad take don't take it serious please

 

[reniiiblaseee]

Moderately strong?

nahhhh it was strong

 

[coolcat6778]

heck you for not using maths in focus. i got a band 6 in 2024 paper with maths in focus only

 

[coolcat6778]

also this is borederline memorisation and bullshit. if something as subjective as this was required to be memorised, they would have put it in the reference sheet, similar to how NESA defines an outlier as Q1 - 1.5IQR or Q3 + 1.5IQR

 

[cheesynooby]

in the hsc it would generally be pretty obvious (since all the rules above are just rules of thumb after all)
and also a question would likely be a short answer that asks you to describe the data (like i think they did last year) and they're more looking for whether you remember to talk about the strength of the correlation

 

[Hehehe22]

Aight thanks guys. I didn't realise that this was std/adv overlap, but I did suspect that it was on the subjective side; I just didn't want to take any risks going into the exam. Looks like I made someone angry

 

[Eagle Mum]

The historical context is probably relevant. “r” Is the correlation coefficient developed by Karl Pearson. He didn’t provide specific criteria for interpretation when he used it to study the relationship between various biological/physiological variables, so there’s no single authority on this. It’s been used a lot in medicine and psychology by researchers who applied different numbers of categories and different cut offs.

Make sure you can recognise whether a correlation is direct (r is positive) or inverse (r is negative) and very strong/high or weak etc and, if necessary, be able to compare different graphs, but there’s no point trying to memorise specific cut offs because even the Wikipedia page on Pearson correlation coefficient states that they are arbitrary.