The Student Room Group

Official Edexcel S3 thread Wednesday 20 May 2015 Morning [6691]

Scroll to see replies

Reply 160
Original post by Navo D.
Does anyone have a list of all the definitions we need to know?

The main ones

You need to be able to define:

central limit theorem



And you'll need to describe

Simple random, systematic, stratified and quota sampling, and compare the merits of each

Whether a product moment correlation coefficient is more significant than the spearman's rank coefficient in a given context

Basic interpretation of goodness of fit tests relating to the context



There might be more but these are the most common
(edited 8 years ago)
Original post by STATER
I think you're trying to say that the sample has to be large enough, since a confidence interval is calculated for individual samples. The sample size must be large enough so that it may be assumed by the central limit theorem that the sample mean is normally distributed. And the x% confidence interval tells you the x% probability that the population mean lies within that interval.

Nope I meant what I said.

The confidence intervals you do have no probability attached to them. Saying the 95% confidence interval tells you with 95% probability the population mean lies in that interval is wrong.

Original post by Karoel
If this were the definition of a confidence interval, then why would the confidence interval use the sample mean of only one sample? Repeating an experiment lots of times kinda goes against this.


This is the definition.
This is what statisticians use.

Ok if yous don't believe me, look on here: http://mathforum.org/library/drmath/view/69143.html
Reply 162
Original post by rayquaza17
Nope I meant what I said.

The confidence intervals you do have no probability attached to them. Saying the 95% confidence interval tells you with 95% probability the population mean lies in that interval is wrong.

This is the definition.
This is what statisticians use.

Ok if yous don't believe me, look on here: http://mathforum.org/library/drmath/view/69143.html


This directly contradicts the S3 textbook. "What a 95% confidence interval tells you is that the probability that the interval contains mu is 0.95."

Edit: This seems to be a disputed topic all over the internet. I'd suggest just using the textbook phrasing, whether it is technically correct or incorrect. You can't possibly lose marks for using their own phrasing word for word.
(edited 8 years ago)
Hey, quick question - in regression and correlation, for PMCC the variables X and Y are jointly normally distributed. Can someone please explain what that means. Thanks.
So after completing all past papers apart form the 2014 standard, r and ial paper, i have scored an average of 65.13 out of 75... with 70+ scores on the last few papers...

Few questions though

Midway through doing goodness of fit, do we use fractions or rounded answers? i used exact fractions and would always be out by 0.01 of the awrt figure in the mark scheme.

Any idea on what A* grade boundary will be, according to last year, 69 was A* and 65 was A - seems a little high???

Also, what are people doing for last minute revision checks etc ???
Original post by tazza ma razza
So after completing all past papers apart form the 2014 standard, r and ial paper, i have scored an average of 65.13 out of 75... with 70+ scores on the last few papers...

Few questions though

Midway through doing goodness of fit, do we use fractions or rounded answers? i used exact fractions and would always be out by 0.01 of the awrt figure in the mark scheme.

Any idea on what A* grade boundary will be, according to last year, 69 was A* and 65 was A - seems a little high???

Also, what are people doing for last minute revision checks etc ???

I round it when writing it down but store the value in my calculator and then use that when calculating.
Yeah, the boundaries I thought are a bit on the high side but then again I suppose this is a Further Maths paper, so the cohort on average is stronger.
I'm just going to go through my notes I think.
Original post by Pri2506
Hey, quick question - in regression and correlation, for PMCC the variables X and Y are jointly normally distributed. Can someone please explain what that means. Thanks.


Meaning that it both events can be modelled by a normal distribution and so you can use bell curve theory to find critical regions etc I think (just read my textbook)
Original post by tazza ma razza
Meaning that it both events can be modelled by a normal distribution and so you can use bell curve theory to find critical regions etc I think (just read my textbook)


Yeah i think that's it.
- If the distribution is symmetrical about the mean
and
- approximately two thirds of the values fall within one standard deviation of the mean
Thank you!
The 13R paper had some really interesting questions :tongue: What was the grade boundary for A/A*/full?
65 for an A
69 for an A*
75 for 100UMS
Original post by Navo D.
The 13R paper had some really interesting questions :tongue: What was the grade boundary for A/A*/full?


65 for an A
69 for an A*
75 for 100UMS
Original post by tazza ma razza
65 for an A
69 for an A*
75 for 100UMS


Thanks, a little steep though huh... :s-smilie:

EDIT: Shouldn't it be 73 for full then?
Original post by Navo D.
Thanks, a little steep though huh... :s-smilie:

EDIT: Shouldn't it be 73 for full then?


A little!?!?! this is insanely high! espicially for the A - that was the A* mark 2 or 3 years ago i believe... Nah, pretty certain it said 75
Original post by tazza ma razza
A little!?!?! this is insanely high! espicially for the A - that was the A* mark 2 or 3 years ago i believe... Nah, pretty certain it said 75


Lol true, I don't mind if the paper is hard as long as there are low grade boundaries.

What are your other modules for FM?
Original post by Navo D.
Lol true, I don't mind if the paper is hard as long as there are low grade boundaries.

What are your other modules for FM?


FP2 and S4, you?

yeah, the fact that boundaries are high puts additional pressure on you i find, espicially if you aim to get 90+ on S3 or are banking on S3 to offset other modules!!!!
Original post by tazza ma razza
FP2 and S4, you?

yeah, the fact that boundaries are high puts additional pressure on you i find, espicially if you aim to get 90+ on S3 or are banking on S3 to offset other modules!!!!


M2/3 S2/3 FP2/3 but I only just got A* in S2 so I still need to work to get that 270

Yeah and I don't understand how the boundaries are so high sometimes, June 2014 R was 67 for A and 71 for A*?!

Also it seems that in recent papers they've removed the old style contingency table questions and instead give you the expected values :O
Original post by Navo D.
M2/3 S2/3 FP2/3 but I only just got A* in S2 so I still need to work to get that 270

Yeah and I don't understand how the boundaries are so high sometimes, June 2014 R was 67 for A and 71 for A*?!

Also it seems that in recent papers they've removed the old style contingency table questions and instead give you the expected values :O


Woah - got your work cut out for you then :wink: I am currently doing the r paper now (q4) so far so good haha :smile:

I noticed that as well! They were easy marks :frown: instead you now have to work out the expected values in binomial or poisson distributions. Still, S3 isn't that hard, it is the Chi squared degrees of freedom that screws you 4 out of 7 marks if you get it wrong :smile:
Original post by STATER
The main ones

You need to be able to define:

central limit theorem


And you'll need to describe

Simple random, systematic, stratified and quota sampling, and compare the merits of each

Whether a product moment correlation coefficient is more significant than the spearman's rank coefficient in a given context

Basic interpretation of goodness of fit tests relating to the context


There might be more but these are the most common


When would one use product moment correlation coefficient over rank correlation and vice versa?
What if I give two answers to one question or three assumptions where they ask for two? Will I still get marks if one of them is wrong?
How about multiple attempts at a question?
Original post by simonli2575
What if I give two answers to one question or three assumptions where they ask for two? Will I still get marks if one of them is wrong?
How about multiple attempts at a question?


Multiple attempts at a question are fine, not sure about the other query. I think in the markscheme it say ignore non relevant answers, so it should be fine

Quick Reply