Turn on thread page Beta
    Offline

    2
    ReputationRep:
    (Original post by anndz3007)
    Chi square for the doctor question ?
    Do you remember what the degrees of freedom were etc?
    Offline

    14
    ReputationRep:
    What did people put for q1?

    For 1a I put if the sample required is large, and if the population divides naturally into mutually exclusive strata

    For 1b, I put if sampling errors need to be estimated (can't be done for quota as its non-random), and if the characteristics being sampled are hard to judge
    Offline

    3
    ReputationRep:
    Messed up a few questions I should have got right which is annoying. Somehow didn't get the last q
    Offline

    2
    ReputationRep:
    (Original post by Armpits)
    Do you remember what the degrees of freedom were etc?
    2 i think, and s.l is 10%
    Offline

    6
    ReputationRep:
    Answer for part c lung que?

    Posted from TSR Mobile
    Offline

    1
    ReputationRep:
    I'm pretty confident about this test I thought it went quite well. My answers are:

    1a) More representative, less chance of bias
    1b) it's random, unlikely to have interview bias

    2) 5.614>4.605 so reject H0

    3a) Want to check correlation not linear ability
    3b) 0.75
    3c) 0.75>0.7143 so reject H0
    3D) 0.65<0.6694 so accept H0
    3e) positively correlation but non-linear

    4a) 0.7794
    4b) X-N(760, 302.25)
    4c) 0.0107

    5a) 1.954>1.6449 so reject H0
    5b) CLT, means approx normal, bar sample = var population
    5c) 12.7

    6a) 4.1<5.991 so binomial suitable
    6b) r= 26.78, s=16.07
    6c) 14.65>11.345 so poison not suitable

    7a) [19.15,19.85]
    7b) reduce 20% value as outside interval
    7c) n=44
    Offline

    3
    ReputationRep:
    (Original post by arks_007)
    Answer for part c lung que?

    Posted from TSR Mobile
    Remind me what it was roughly?
    Offline

    0
    ReputationRep:
    For the Po question a lot of people seem to have assumed that lambda is 1.8 by taking the mean of the sample. I personally took the first expected value and equated it to e^-lambda, getting lambda as being 1.78. Not a major difference but it means I didn't subtract an additional degree of freedom in my Chi squared test. Any opinions?
    Offline

    10
    ReputationRep:
    (Original post by lai812matthew)
    got question 1 wrong :/
    i only list the advantages of stratified to simple random sampling at a and quota at b will i get any marks......
    Offline

    2
    ReputationRep:
    (Original post by ZeusOfScience)
    For the Po question a lot of people seem to have assumed that lambda is 1.8 by taking the mean of the sample. I personally took the first expected value and equated it to e^-lambda, getting lambda as being 1.78. Not a major difference but it means I didn't subtract an additional degree of freedom in my Chi squared test. Any opinions?
    Yeah I did the same, got lambda as a natural log and used that. I did subtract a degree of freedom for it though, as it was calculated from the data afterall?
    Offline

    9
    ReputationRep:
    My answers agree with yours - so i am liking that!!!!
    (Original post by SHJBHB)
    I'm pretty confident about this test I thought it went quite well. My answers are:

    1a) More representative, less chance of bias
    1b) it's random, unlikely to have interview bias

    2) 5.614>4.605 so reject H0

    3a) Want to check correlation not linear ability
    3b) 0.75
    3c) 0.75>0.7143 so reject H0
    3D) 0.65<0.6694 so accept H0
    3e) positively correlation but non-linear

    4a) 0.7794
    4b) X-N(760, 302.25)
    4c) 0.0107

    5a) 1.954>1.6449 so reject H0
    5b) CLT, means approx normal, bar sample = var population
    5c) 12.7

    6a) 4.1<5.991 so binomial suitable
    6b) r= 26.78, s=16.07
    6c) 14.65>11.345 so poison not suitable

    7a) [19.15,19.85]
    7b) reduce 20% value as outside interval
    7c) n=44
    Offline

    2
    ReputationRep:
    (Original post by SHJBHB)
    I'm pretty confident about this test I thought it went quite well. My answers are:

    1a) More representative, less chance of bias
    1b) it's random, unlikely to have interview bias

    2) 5.614>4.605 so reject H0

    3a) Want to check correlation not linear ability
    3b) 0.75
    3c) 0.75>0.7143 so reject H0
    3D) 0.65<0.6694 so accept H0
    3e) positively correlation but non-linear

    4a) 0.7794
    4b) X-N(760, 302.25)
    4c) 0.0107

    5a) 1.954>1.6449 so reject H0
    5b) CLT, means approx normal, bar sample = var population
    5c) 12.7

    6a) 4.1<5.991 so binomial suitable
    6b) r= 26.78, s=16.07
    6c) 14.65>11.345 so poison not suitable

    7a) [19.15,19.85]
    7b) reduce 20% value as outside interval
    7c) n=44
    Got same for all. Im safe now
    Offline

    3
    ReputationRep:
    (Original post by SHJBHB)
    I'm pretty confident about this test I thought it went quite well. My answers are:

    1a) More representative, less chance of bias
    1b) it's random, unlikely to have interview bias

    2) 5.614>4.605 so reject H0

    3a) Want to check correlation not linear ability
    3b) 0.75
    3c) 0.75>0.7143 so reject H0
    3D) 0.65<0.6694 so accept H0
    3e) positively correlation but non-linear

    4a) 0.7794
    4b) X-N(760, 302.25)
    4c) 0.0107

    5a) 1.954>1.6449 so reject H0
    5b) CLT, means approx normal, bar sample = var population
    5c) 12.7

    6a) 4.1<5.991 so binomial suitable
    6b) r= 26.78, s=16.07
    6c) 14.65>11.345 so poison not suitable

    7a) [19.15,19.85]
    7b) reduce 20% value as outside interval
    7c) n=44
    1a) Gives fair representation of strata, more suitable for larger samples
    1b)Can estimate sampling error, no interview bias

    2)agree.

    3)a)When data is already ranked or variables are not jointly normally distributed
    b)agree
    c)agree
    d)not sure

    4)agree

    5a)agree
    5b) Not sure you 'assume' CLT but this might be correct, I answered that the people were selected independently.
    5c) can't remember

    6)agree

    7)agree
    Offline

    0
    ReputationRep:
    (Original post by cjlh)
    Yeah I did the same, got lambda as a natural log and used that. I did subtract a degree of freedom for it though, as it was calculated from the data afterall?
    We can't assume they calculated their lambda from the data, they could've completely made up 1.78 - a very poorly phrased question
    Offline

    1
    ReputationRep:
    (Original post by SHJBHB)
    I'm pretty confident about this test I thought it went quite well. My answers are:

    1a) More representative, less chance of bias
    1b) it's random, unlikely to have interview bias

    2) 5.614>4.605 so reject H0

    3a) Want to check correlation not linear ability
    3b) 0.75
    3c) 0.75>0.7143 so reject H0
    3D) 0.65<0.6694 so accept H0
    3e) positively correlation but non-linear

    4a) 0.7794
    4b) X-N(760, 302.25)
    4c) 0.0107

    5a) 1.954>1.6449 so reject H0
    5b) CLT, means approx normal, bar sample = var population
    5c) 12.7

    6a) 4.1<5.991 so binomial suitable
    6b) r= 26.78, s=16.07
    6c) 14.65>11.345 so poison not suitable

    7a) [19.15,19.85]
    7b) reduce 20% value as outside interval
    7c) n=44
    I got virtually the same for the calculations and for the worded questions:

    1a) More representative as reflects population structure, gives estimates for individual strata
    1b) Random process so able to estimate sampling errors, interviewer could introduce bias in quota sampling due to choice of sample

    3a) Use Spearman's when the two variables cannot be assumed to be normally distributed / when the data is already ranked

    5b) Assumptions are sample variance is equal to population variance and that the means (I used Xbar and Ybar) are independent

    7b) 20% is above (I believe they didn't allow 'outside' in one of the past papers I'd seen) confidence interval so he should recommend them to lower their stated fat content
    Offline

    3
    ReputationRep:
    (Original post by cjlh)
    Yeah I did the same, got lambda as a natural log and used that. I did subtract a degree of freedom for it though, as it was calculated from the data afterall?
    I did this initially, however estimated 1.8 correctly afterwards. I believe the 1.8 answer is probably correct (it is nicer). In any case you should have subtracted an extra degree of freedom as the mean has still been calculated from observed data.
    Offline

    6
    ReputationRep:
    Hella pissed I didn't say sample size is large for 1a.
    Can you still get anything if you write along the lines of its results best reflecting the population as proportion of stratum in sample equals proportion of that strata in population. And for point 2 stating population divides into mutually exclusive groups ??
    Offline

    1
    ReputationRep:
    (Original post by Mav1)
    1a) Gives fair representation of strata, more suitable for larger samples
    1b)Can estimate sampling error, no interview bias

    2)agree.

    3)a)When data is already ranked or variables are not jointly normally distributed
    b)agree
    c)agree
    d)not sure

    4)agree

    5a)agree
    5b) Not sure you 'assume' CLT but this might be correct, I answered that the people were selected independently.
    5c) can't remember

    6)agree

    7)agree
    Yeah, my definitions aren't brilliant but I'm pretty sure all my maths is right
    Offline

    2
    ReputationRep:
    (Original post by ZeusOfScience)
    We can't assume they calculated their lambda from the data, they could've completely made up 1.78 - a very poorly phrased question
    Very good point, I think it was poorly worded too. Oh well, suppose we'll have to see haha. Hopefully just one lost mark in this case
    Offline

    9
    ReputationRep:
    My answers - I thought t was fairly standard

    1a) More representative, individual strata estimates availiable
    1b) random process, no interviewer bias

    2) 5.61 and I think you rejected

    3a) When the data isn't bivariate noral
    b) 0.75 and I think you rejected
    You accepted the pmcc question
    e) You HAD TO say about ranks - days where sales of ice cream ranked highly, ranked highly for sunglasses

    4a) 0.7794
    b) 760,302.25
    c) 0.0107

    5a) 1.95 and you rejected
    b) Var of sample = var of population. Samples are independent of each other. CANT MENTION CLT - as its not an assumption, it's a thorem
    c) tricky - 12.7

    6a) Test statistic was 4.1 and you accepted
    b) r= 26.78 s= 16.07
    c) Test statistic was 14.65 you reject

    7a) 19.15%,19.85%
    b) Paul shoud tell them to lover stated weight
    c) n=44

    Who agrees?
 
 
 
Reply
Submit reply
Turn on thread page Beta
Updated: June 29, 2016

University open days

  1. University of Bradford
    University-wide Postgraduate
    Wed, 25 Jul '18
  2. University of Buckingham
    Psychology Taster Tutorial Undergraduate
    Wed, 25 Jul '18
  3. Bournemouth University
    Clearing Campus Visit Undergraduate
    Wed, 1 Aug '18
Poll
How are you feeling in the run-up to Results Day 2018?

The Student Room, Get Revising and Marked by Teachers are trading names of The Student Room Group Ltd.

Register Number: 04666380 (England and Wales), VAT No. 806 8067 22 Registered Office: International House, Queens Road, Brighton, BN1 3XE

Write a reply...
Reply
Hide
Reputation gems: You get these gems as you gain rep from other members for making good contributions and giving helpful advice.