[NightBear]

(Original post by Ree69)
Oh..ye. That was embarrassing.

Too much maths and you forget how to read numbers.

It happens,

[CJAW]

(Original post by Ree69)
No, I'm about 90% confident () that I've done tihs correctly.

CI =

=

=

= (4dp).

. Someone tell me where I've gone wrong .

Mine was using 0.0342, you had 0.342

[Ree69]

(Original post by CJAW)
Mine was using 0.0342, you had 0.342

Yeah sorry I used 0.0342 as well, just typed it up incorrectly. The point still stands though - the mark scheme is wrong.

[Ree69]

I want to say here...

[CJAW]

(Original post by Ree69)
I want to say here...

I agree, I think I got 0.9394 (or something similar)

[CJAW]

Q1) i) Can't remember
ii) Mean = 507.6

Q2) Differences test: t=1.444
Q3) Population proportions: z=2.103
Q4) -0.961
Q5) i) Mean 12 Variance 36
ii) Normal distribution,np = nq = 16 (>5). Large poission mean(>15). Sample size sufficiently large for a normal approximation.
iii) 0.9394
Q6) i) a=2/7
ii) 1/7 tsquared (0<x<1)
2/7 t - 1/7 (1<x<4)
iii) G(Y) = F(ysquared)
1/7 (y)^4 (0<y<1)
2/7 ysquared - 1/7 (1<y<2)

g(y) =
4/7 y^3 (0<y<1)
4/7 y (1<y<2)

Q7i) Xsquared = 1.4
ii) I got this wrong, people I know say it was 6.67 or something in that area.
iii) A

Please agree with me and critique. Some might be the wrong way around, particuarly Q3 and Q4. Also, I'm not sure I've remembered 6ii correctly. Obviously all the z and t values needed to be compared with a critical value

[Ree69]

(Original post by CJAW)
I agree, I think I got 0.9394 (or something similar)

yeah i got the same. i was a bit reluctant doing it this way (i'm not sure why...)

how did you find the paper generally? how many assumptions did you write for the paired t-test? (i think question 2...)

[CJAW]

(Original post by Ree69)
yeah i got the same. i was a bit reluctant doing it this way (i'm not sure why...)

how did you find the paper generally? how many assumptions did you write for the paired t-test? (i think question 2...)

I actually let Q=Y-X, just to avoid getting mixed up with minus signs.

Quite good generally, I messed up the last question because I originally had the rows in the contingency table as 'Number of trials' and 'Improvements' but then changed the rows to 'improvements' and 'non-improvements' which was apparently correct but I ran out of time so I got the Oi-Ei part a bit muddled up
It said assumption (I checked) so just the one being, -the differences are distributed normally.

[Ree69]

(Original post by CJAW)
Q1) i) Can't remember
ii) Mean = 507.6 Yes. No actually. I got 507.3 and wrote 507 (3sf).

Q2) Differences test: t=1.444 yes
Q3) Population proportions: z=2.103 yes
Q4) -0.961 yes (i was a bit unsure about this, was the denominator root6.5 or root(6.5/24)?) anyway i got wrote the same answer)
Q5) i) Mean 12 Variance 36 No. mean = -12.
ii) Normal distribution, I think you had to say something about how the binomial and poisson approximate to normal np = nq = 16 (>5). Large poission mean. Sample size sufficiently large for a normal approximation.
iii) 0.9394 yes
Q6) i) a=2/7 yes
ii) 1/7 tsquared + 2/7 yes, although i had another with another interval
iii) can't remember, I think I had G(Y) = F(ysquared) and went on from there. i don't think i had this...
Q7i) Xsquared = 1.4 yes
ii) I got this wrong, people I know say it was 6.67 or something in that area. I got 1.609. Accept H₀, there is insufficient evidence at the 5% sig. level that the outcome is dependent on the treatment...
iii) A yes

Please agree with me and critique. Some might be the wrong way around, particuarly Q3 and Q4. Also, I'm not sure I've remembered 6ii correctly. Obviously all the z and t values needed to be compared with a critical value

Yeah, I don't remember 6ii or 6iii either....

[Ree69]

(Original post by CJAW)
I actually let Q=Y-X, just to avoid getting mixed up with minus signs.

Quite good generally, I messed up the last question because I originally had the rows in the contingency table as 'Number of trials' and 'Improvements' but then changed the rows to 'improvements' and 'non-improvements' which was apparently correct but I ran out of time so I got the Oi-Ei part a bit muddled up
It said assumption (I checked) so just the one being, -the differences are distributed normally.

Yeah, I think I've messed up the last question then...

[CJAW]

(Original post by Ree69)
Yeah, I don't remember 6ii or 6iii either....

Q1) Yes I wrote it to 3.s.f too.
Q4) Surely Root(6.5/24), otherwise we would have got different answers.
Q5) i) Are you sure? I don't remember it being negative and if it was, we would have had different answers to iii)
Q6) ii)Adding in the second interval
iii) Wasn't it Y=X^1/2 so that would become x=y^2
Q7) ii) I originally had that but changed it... There was some disagreement afterwards about the answer in my school

[falcon pluse]

(Original post by Ree69)
Yeah, I don't remember 6ii or 6iii either....

for the difference test was h1: mean d not equal to 0 ?

[CJAW]

(Original post by falcon pluse)
for the difference test was h1: mean d not equal to 0 ?

Yes

[falcon pluse]

(Original post by CJAW)
Yes

good because i got the minus of wht u wrote , so that shud b fine

[CJAW]

(Original post by Ree69)
Yeah, I think I've messed up the last question then...

You should still get 4/8 anyway. One for the hypotheses, one for the critical value, and two more for the right idea eg.0i-Ei

[Ree69]

(Original post by CJAW)
Q1) Yes I wrote it to 3.s.f too.
Q4) Surely Root(6.5/24), otherwise we would have got different answers.
Q5) i) Are you sure? I don't remember it being negative and if it was, we would have had different answers to iii)
Q6) ii)Adding in the second interval
iii) Wasn't it Y=X^1/2 so that would become x=y^2
Q7) ii) I originally had that but changed it... There was some disagreement afterwards about the answer in my school

1) but then surely you would've had 508?
4) yeah i did root(6.5/24) as well, but somewhere in the back of my head i thought i may have lost a few marks here in case it's just root6.5
5i) I think so. E(X-Y) = E(X)-E(Y) = 16 - 28. Oh wait, was it E(Y-X)? ****. No we wouldn't have got different answers to part iii) because of the modulus sign
6iii) i reallu can't remember
7) yeah it would make sense to do what you did. even as i was doing the calculation i knew something didn't quite make sense because of the wording of the question

[CJAW]

(Original post by Ree69)
1) but then surely you would've had 508?
4) yeah i did root(6.5/24) as well, but somewhere in the back of my head i thought i may have lost a few marks here in case it's just root6.5
5i) I think so. E(X-Y) = E(X)-E(Y) = 16 - 28. Oh wait, was it E(Y-X)? ****. No we wouldn't have got different answers to part iii) because of the modulus sign
6iii) i reallu can't remember
7) yeah it would make sense to do what you did. even as i was doing the calculation i knew something didn't quite make sense because of the wording of the question

Q1) Yes, I should have rounded it up in that case.. Not sure why I did that! But I definitely had 507.64
Q5i) Definitely E(Y-X)- I know because I did it the other way in the first place then realised. Could you have done E(Y-X) for one part and E(X-Y) for the other?
Q7) Grrr, I spent 20 minutes on that thinking it couldn't be right, and I still didn't get the right number

[jedhemsley]

i recognise very few of these answers, and i came out thinking it had gone well. For the population proportions about the colourful letters, i remember getting the z statistic as being the same as the 2.5% significance level value? i double checked it and everything... i think it was 1.96 :/ because it was the hypothesis that Psx - Psy > 0.5, so i did 39/60 - 14/40 - 0.5 all over root pq(1/60+1/40)

[Ree69]

(Original post by CJAW)
Q5i) Definitely E(Y-X)- I know because I did it the other way in the first place then realised. Could you have done E(Y-X) for one part and E(X-Y) for the other?

yeah, i think so. I did E(X-Y) for part i) and then wrote E(Y-X) for part iii) - but used the value from part i). i'm such an idiot. i think i'll lose a mark for part i) but then could i still get full marks for part iii?

It just so happens that Var(Y-X) = Var(X-Y) and E(Y-X) = -E(X-Y).

[CJAW]

(Original post by jedhemsley)
i recognise very few of these answers, and i came out thinking it had gone well. For the population proportions about the colourful letters, i remember getting the z statistic as being the same as the 2.5% significance level value? i double checked it and everything... i think it was 1.96 :/ because it was the hypothesis that Psx - Psy > 0.5, so i did 39/60 - 14/40 - 0.5 all over root pq(1/60+1/40)

It was 0.15 anyway - maybe you meant that.
Secondly, root pq(1/60+1/40) is only for psx-psy= 0
Otherwise it's Root psxqsx/nx + psyqsy/ny