The Student Room Group

Edexcel S3 - Wednesday 25th May AM 2016

Scroll to see replies

Original post by L'Evil Wolf
If rankings are tied in spearmans rank say at 5 there are two values which are equal then would we give it a value of 5.5, I tried doing this for a question but the difference, d was not equal to 0.


If tied rankings in spearman don't you take the average (in your case 5.5) and then use pmcc? Im sure I've seen a question where you have had to do that.

Original post by Nijuuninichi
If you work out the expected values for a contingency table in a goodness of fit question and get fractions, do you round it up before working out chi squared??

I was doing a solomon paper just now and in the mark scheme they rounded up their expected values to 2dp so my answer was about 0.002 away from their answer. Do you know if I'd lose the mark if I did that in the real thing? :frown:


It's best to leave it as fractions and then round once you have worked out chi squared. That seems the common theme in edexcel mark schemes, I wouldn't trust slomon as much since its independant and all from edexcel:smile:
Original post by Nijuuninichi
If you work out the expected values for a contingency table in a goodness of fit question and get fractions, do you round it up before working out chi squared??

I was doing a solomon paper just now and in the mark scheme they rounded up their expected values to 2dp so my answer was about 0.002 away from their answer. Do you know if I'd lose the mark if I did that in the real thing? :frown:


In a past paper I've picked at random (June 2011) they accept answers that round correctly to 2 d.p so I'd go with 3 or 4 d.p for all of the values and round it at the end.
Original post by econam

It's best to leave it as fractions and then round once you have worked out chi squared. That seems the common theme in edexcel mark schemes, I wouldn't trust slomon as much since its independant and all from edexcel:smile:



Original post by SeanFM
In a past paper I've picked at random (June 2011) they accept answers that round correctly to 2 d.p so I'd go with 3 or 4 d.p for all of the values and round it at the end.


Thank you so much! :biggrin:
Original post by Nijuuninichi
Thank you so much! :biggrin:


No worries :tongue: a shame this exam isn't on the 22nd, that would have been more fitting.
Original post by SeanFM
No worries :tongue: a shame this exam isn't on the 22nd, that would have been more fitting.


It is a shame because the 22nd is my lucky day of month - you probably guessed that! Guess I have to make up for it with more Solomon papers. :wink:
Original post by Nijuuninichi
It is a shame because the 22nd is my lucky day of month - you probably guessed that! Guess I have to make up for it with more Solomon papers. :wink:


Ah, birthday? :tongue: or won the lottery or something?

Yep, you can never do enough :redface: I don't suppose you do S4 too? :K:
Original post by econam
If tied rankings in spearman don't you take the average (in your case 5.5) and then use pmcc? Im sure I've seen a question where you have had to do that.



It's best to leave it as fractions and then round once you have worked out chi squared. That seems the common theme in edexcel mark schemes, I wouldn't trust slomon as much since its independant and all from edexcel:smile:


Not if the questions states to do Spearman's

It isn't an official question per se,

https://771a1ec81340d97ae9ed29694f73dd633b1c7c70.googledrive.com/host/0B1ZiqBksUHNYV1BTbDkxWXZCVmc/CH5.pdf

It is dereived from the data on page 39, Question 11, where the differene doesn't get to 0
Original post by SeanFM
My knowledge of spearmans is a bit rusty, but does the difference have to be 0? And that, in the case of tied ranks, you can use something else?


You are probably correct, just a general question I had. I think that the differece, d between the ranks has to be 0. It is a check they said to do in the edexcel book.
Original post by L'Evil Wolf
If rankings are tied in spearmans rank say at 5 there are two values which are equal then would we give it a value of 5.5, I tried doing this for a question but the difference, d was not equal to 0.


Which questions was it?
Original post by coolguy123456
Which questions was it?


Edexcel Ex 6D Q10, it gives the spearmansrank but I had this issue.
Original post by L'Evil Wolf
Edexcel Ex 6D Q10, it gives the spearmansrank but I had this issue.


Ex 6D? Do you mean 5D AS there are only 5 chapters..
Original post by coolguy123456
Ex 6D? Do you mean 5D AS there are only 5 chapters..


yes that it is sorry
Original post by L'Evil Wolf
yes that it is sorry


Yeah i didn't see any tied ranks in "age" or "maintenance costs". If you meant that the difference in ranks was 0, yes that is allowed.
june 09 q2 part a)

i have no clue about this qn
Original post by tazza ma razza
june 09 q2 part a)

i have no clue about this qn


With these questions where it asks to find the limits, you have to do the same as you would for CI but remove the root n part. so it's just: 20.1 ± 1.96 × 0.5 = (19.1, 21.1)
Original post by econam
If tied rankings in spearman don't you take the average (in your case 5.5) and then use pmcc?
Im sure I've seen a question where you have had to do that.


hey, what was the question where you had tied ranks - cant remember having come across a question like that!
Original post by AakashG

hey, what was the question where you had tied ranks - cant remember having come across a question like that!


June 2012. You dont have to do it, just explain what you would do.
Please could someone explain exercise I question 14A https://771a1ec81340d97ae9ed29694f73dd633b1c7c70.googledrive.com/host/0B1ZiqBksUHNYV1BTbDkxWXZCVmc/CH3.pdf Thanks :smile:
Original post by tazza ma razza
june 09 q2 part a)

i have no clue about this qn


Just to add a little to what coolguy said. Part a is asking for range of heights for all the individual orchids and part b asking for the range of the mean. The more samples you take to find the mean you would expect to get closer to its true value. This means when finding a confidence interval for the means you divide standard dev by rootn and as a result get a smaller range.

Because in part a you are looking at the range of values for just the orchids you just multiply by standard dev. Getting a larger range, as you would expect when looking at the range of individual orchids.

I think thats right and hope it makes sense:smile:
Original post by econam
June 2012. You dont have to do it, just explain what you would do.


if we were to use the pmcc instead of sr - would we literally rank them as say 1,2,3,4,5,5.5,5.57,8 for example and then literally just use pmcc by looking at sample level in the table or would we have to calculate say sxy,sxx,syy and then do it like that?

Quick Reply

Related discussions

Latest

Trending

Trending

Articles for you