Session 1, 2011

MATH2089

Page 6

Formula sheet
1
1.1

Probability distributions
The Binomial distribution

Assume X Bin(n, )
domain of variation :
SX = {0, 1, . . . , n}
probability mass function (pmf ) :
 
n x
p(x) =
(1 )nx ,
x

for x SX

cumulative distribution function (cdf ) :

x  
X
n k
(1 )nk
F (x) =
k
k=0

(where x denotes the integer part of x).

expectation :
E(X) = n
variance :

1.2

Var(X) = n(1 )

The Poisson distribution

Assume X P()
domain of variation :
SX = {0, 1, 2, . . .}
probability mass function (pmf ) :

x
,
for x SX
x!
cumulative distribution function (cdf ) :
p(x) = e

F (x) = e

x k
X

k=0

k!

(where x denotes the integer part of x).

expectation :
E(X) =
variance :

Var(X) =

Please see over . . .

Session 1, 2011

1.3

MATH2089

Page 7

The Uniform distribution

Assume X U[,]
domain of variation :

SX = [, ]

probability density function (pdf ) :

f (x) =

1
,

for x SX

cumulative distribution function (cdf ) :

F (x) =
expectation :

E(X) =

variance :

1.4

x
,

Var(X) =

for x SX
+
2
( )2
12

The Exponential distribution

Assume X Exp()
domain of variation :

SX = [0, +)

probability density function (pdf ) :

f (x) = ex ,

for x SX

cumulative distribution function (cdf ) :

F (x) = 1 ex ,
expectation :
variance :

E(X) =
Var(X) =

for x SX
1

1
2

Please see over . . .

Session 1, 2011

MATH2089

1.5

The Normal distribution

Page 8

Assume X N (, )
domain of variation :
SX = (, +)
probability density function (pdf ) :
f (x) =

2
1 (x)
1
e 2 2 ,
2

for x SX

cumulative distribution function (cdf ) :

Z x
2
1 (y)
1

F (x) =
e 2 2 dy,
2

(no closed form)

expectation :

E(X) =

variance :

1.6

for x SX

Var(X) = 2

The Standard Normal distribution

Assume X N (0, 1)
domain of variation :
SX = (, +)
probability density function (pdf ) :
1 2
1
(x) = e 2 x ,
2

for x SX

cumulative distribution function (cdf ) :

Z x
1 2
1
e 2 y dy,
for x SX
(x) =
2

(no closed form, see also the attached Standard Normal table)
expectation :
E(X) = 0
variance :

Var(X) = 1

standardisation : If X N (, ), then
Z=

X
N (0, 1)

Please see over . . .

Session 1, 2011

1.7

MATH2089

Page 9

The Students t distribution

Assume X t
domain of variation :
SX = (, +)
probability density function (pdf ) :


 +1
2
+1
x2
2

1+
f (x) =
,

for x SX

cumulative distribution function (cdf ) :

F (x) =


 +1
2
+1
y2
2

1+
dy,

for x SX

(no closed form, see also the attached t-distribution critical values table)
expectation :
E(X) = 0
variance :

1.8

Var(X) =

for > 2

The 2 -distribution

Assume X 2
domain of variation :

SX = [0, +)

probability density function (pdf ) :

f (x) =

1
2/2 (/2)

x/21 ex/2 ,

cumulative distribution function (cdf ) :

Z x
1
F (x) =
y /21 ey/2 dy,
/2 (/2)
2

for x SX

for x SX

(no closed form, see also the attached 2 -distribution critical values table)
expectation :
E(X) =
variance :

Var(X) = 2

Please see over . . .

Session 1, 2011

1.9

MATH2089

Page 10

The Fisher F-distribution

Assume X Fd1 ,d2

domain of variation :

SX = [0, +)

probability density function (pdf ) :

f (x) =

((d1 + d2 )/2)(d1 /d2 )d1 /2 xd1 /21

,
(d1 /2)(d2 /2)((d1 /d2 )x + 1)(d1 +d2 )/2

for x SX

cumulative distribution function (cdf ) :

Z x
((d1 + d2 )/2)(d1 /d2 )d1 /2 y d1 /21
dy,
F (x) =
(d1 +d2 )/2
(d1 /2)(d2 /2)((d1 /d2 )y + 1)

for x SX

(no closed form, see also the attached F -distribution critical values table)
expectation :
d2
for d2 > 2
E(X) =
d2 2
variance :
Var(X) =

2d22 (d1 + d2 2)
d1 (d2 2)2 (d2 4)

for d2 > 4

Sampling distributions

2.1

Sample mean

2.1.1

known variance

be the sample average from a random sample of size n from a population

Let X
with mean and standard deviation . Under appropriate conditions,

X
Z= n
N (0, 1)

(exact result if the population distribution is normal, approximate result if the

population distribution is not normal but n > 30)

2.1.2

unknown variance

and S be the sample average and standard deviation from a random

Let X
sample of size n from a normal population with mean . Under appropriate
conditions,

X
T = n
tn1
S
If the population is not normal but n is large enough (n > 40), we can also
write

X
N (0, 1)
T = n
S
approximately

Please see over . . .

Session 1, 2011

2.2

MATH2089

Page 11

Sample proportion

Let p be the sample proportion of successes where the number of trials is n

and the true probability of a success is . Under appropriate conditions,

p
np
N (0, 1)
(1 )

approximately when n(1 ) > 5

2.3

Sample variance

Let S 2 be the sample variance from a random sample of size n from a normal
population with variance . Under appropriate conditions,
(n 1)S 2
2n1
2

2.4
2.4.1

Difference in sample means

variances 12 and 22 known

For two independent samples of size n1 and n2 from two populations with
i be the
means 1 and 2 and standard deviations 1 and 2 respectively, let X
sample average of sample i for i = 1 and 2. Under appropriate conditions,
1 X
) (1 2 )
(X
q2 2
N (0, 1)
22
1
n1 + n2

(exact result if both population distributions are normal, approximate result

if they are not but n1 , n2 > 30)

2.4.2

variances 12 and 22 unknown; 12 = 22

For two independent samples of size n1 and n2 from two normal populations
i
with means 1 and 2 respectively and common standard deviation , let X
and Si be the sample average and sample standard deviation of sample i for
i = 1 and 2. Under appropriate conditions,
1 X
2 ) (1 2 )
(X
q
tn1 +n2 2 ,
Sp n11 + n12

where Sp is the pooled sample standard deviation,

s
(n1 1)S12 + (n2 1)S22
.
Sp =
n1 + n2 2

Please see over . . .

Session 1, 2011
2.4.3

MATH2089

Page 12

variances 12 and 22 unknown; 12 6= 22

For two independent samples of size n1 and n2 from two normal populations
i
with means 1 and 2 and standard deviations 1 and 2 respectively, let X
and Si be the sample average and sample standard deviation of sample i for
i = 1 and 2. Under appropriate conditions,
1 X
) (1 2 )
(X
q2 2
t ,
S1
S22
+
n1
n2

where

(s21 /n1 + s22 /n2 )2

(s21 /n1 )2
n1 1

(s22 /n2 )2
n2 1

(rounded down to the nearest integer)

2.5

Ratio of sample variances

Let S12 and S22 be the sample variances from two independent random samples
of size n1 and n2 from normal populations with variances 12 and 22 respectively. Under appropriate conditions,
S12 /12
Fn1 1,n2 1
S22 /22

Simple linear regression

Consider the simple linear regression model

Y = 0 + 1 X +
where N (0, )
The least squares estimators 0 and 1 of 0 and 1 are
SXY
1 =
SXX

0 = Y 1 X

where
SXY =

X
i

i Y )
(Xi X)(Y

SXX =

X
i

2.
(Xi X)

An estimator of is
S=

i (Yi

0 1 Xi )2
.
n2

Under the simple linear regression model :

1 1

tn2
S/ SXX

Please see over . . .

Session 1, 2011

MATH2089

Page 13

0
q0
tn2
2
S n1 + SXXX

Let x0 denote the predictor value for a response yet to be observed :

i) a 100(1 )% confidence interval for the mean response at x0 is
s
)2
1 (x0 x
+
y(x0 ) s tn2;1/2
n
sxx
where y(x0 ) = b0 + b1 x0 ;
ii) and a 100(1 )% prediction interval for the response at x0 is
s
1 (x0 x
)2
y(x0 ) s tn2;1/2 1 + +
n
sxx

ANOVA

Total sum of squares :

ni
k X
X

SSTot =

i=1 j=1

2
(Xij X)

Treatment sum of squares :

SSTr =

k
X
i=1

2
i X)
ni (X

Error sum of squares :

SSEr =

ni
k X
X
i=1 j=1

i )2
(Xij X

Under the assumption of equality of means in each of the k groups of a

one-way Analysis of Variance,
F =

MSTr
Fk1,nk ,
MSEr

where n is the total number of observations.