Professional Documents
Culture Documents
TABLE OF CONTENTS:
2010s2
2010s2
(1.a.i)
Assumptions:
1. There is a fixed number of trials, n
2. Each trial can be categorised as either 'failure' or 'success'
3. The probability of a particular trial being a 'success' is a constant, p, between 0 and 1
4. The trials are independent - one trial's outcome is not affected by that of another
(1.a.ii)
Mean = E(x) = np = 1*p = p
Variance = Var(X) = np(1-p) = p-p^2
(1.a.iii)
The drawing of the 5 items cannot be modelled as a binomial experiment since the items are drawn without
replacement and therefore these 'trials' would not be independent. The alternative is to treat each instance
of 5 items being drawn as a 'trial', but because only one such 'trial' is conducted, the only possible values
of 'p' that we could get would be increments of 1/n, i.e., 0.2, 0.4, 0.6, 0.8, and 1.0.
(1.b.i)
Since the 10 items are being drawn from a 'large' shipment of items, it can be assumed that each act of
drawing without replacement has an insignificant effect on every other draw. The draws are therefore
practically independent and can be treated as 'trials' in a binomial experiment.
(1.b.ii)
Let X be the number of defective items in the sample. As per (1.b.i), this can be modelled as a binomial
experiment. n=10, p=0.1 (given). For exact probability, must not use tables.
P(1<=X<=3) = P(X=1) + P(X=2) + P(X=3)
= (10C1)(0.1^1)(0.9^9) + (10C2)(0.1^2)(0.9^8) + (10C3)(0.1^3)(0.9^7)
= 0.6385263615...
(1.b.iii)
Normal approximation to the binomial: X~N(np, np(1-p)) i.e. X~N(1, 0.9)
now P(1<=X<=3)
~= P(0.5 < X < 3.5), applying continuity correction factor
= P((0.5-1)/0.9 < Z < (3.5-1)/0.9)
(1) Hypotheses:
H0: F=15 (females work same number of hours)
H1: F<15 (females work less)
(2) Standardise Observation to Test Statistic:
We observe f=13.5, so Z=(13.5-15)/[(10/2.326)/sqrt(36)] = -2.0934
P(Z<-2.0934) = 0.5-P(0<Z<-2.0934)
~= 0.5-P(0<Z<-2.09)
= 0.5-0.4817 = 0.0183
(3) Decision rule:
From (2), we reject H0 for all alpha > 0.0183 i.e. 1.83%
This is a reasonable significance level for a one-tailed hypothesis test. Therefore, for all significance
levels greater than 1.83%, we reject H0 in favour of H1, i.e. we find that females do in fact work less
hours than males on average.
A Type I error is when the null hypothesis is correct but has been erroneously rejected. In this case, a type
I error would have been to conclude that females work less hours than males if, in fact, they did not. The
probability of committing a type I error is known as the significance level, which in this case is only 1.83%.
A Type II error is when the null hypothesis is incorrect but we have failed to reject it. In this case,
a type II error would have been to conclude that "females work the same number of hours" passes the
hypothesis test if, in fact, females worked less hours.
(2.f)
It is not necessary to assume that the distribution of hours worked by females is normal in part (e) since the
sample size (n=36) large enough to invoke the Central Limit Theorem, which states that for large (n>30)
sample sizes, the sampling distribution of the mean is approximately normal regardless of the underlying
population distribution.
(3.a)
A point estimator provides a single possible value for the parameter. An interval estimator provides a range
of possible values for the parameter.
(3.b)
n=114, which is large enough (n>30) to invoke Central Limit Theorem and thus assume that X-bar is
normally distributed. Hence X-bar~N(10.88, (3.20^2)/n), i.e. X-bar~N(10.88, 10.24/114)
For 99% CI, alpha = 0.01, alpha/2 = 0.005
CI half-width = Z_alpha/2 * sqrt(variance) = Z_0.005*sqrt(10.24/114)
= 2.576*sqrt(10.24/114)
= 0.7720464162...
Hence CI = [10.11, 11.65].
The sampling method used by HRD is a questionnaire that recipients could choose not to respond to. If
managers' choice whether or not to complete the questionnaire is correlated with how many years the
manager has been in the firm, then this sampling method is biased.
(3.c)
The confidence interval could be changed by altering the confidence level (a greater confidence level
increases the CI width) or by altering the sample size (a larger sample size decreases the CI width).
Now for 95% CI, alpha = 0.05, alpha/2 = 0.025
CI half-width = Z_alpha/2 * sqrt(10.24/n) =1.96*sqrt(variance) = 0.5*0.7720464162...