Economics 250 Formula Sheet

Descriptive Statistics
Population
1
N

N

Mean

Variance

2 =

CV

Covariance

xy =

1
N

Correlation

xy =

xy
x y

i=1

1
N

Sample
xi

N

i=1 (xi

x=
)2

s2 =

100

s
x

N

i=1 (xi

1
n

x )(yi y )

n
i=1

1
n1

xi

n

i=1 (xi

x)2

100

sxy =

1
n1

rxy =

sxy
sx sy

n

i=1 (xi

x)(yi y)

Grouped Data
With K classes, with midpoints mi and counts ci , the sample mean is x =
K
K
1
2
variance is s2 = n1
i=1 ci (mi x) , where n =
i=1 ci .

1
n

K

i=1 ci mi ,

and the sample

689599.7 Rule
For a normal distribution 68% of the observations are in 1, 95% are in 2, and almost all (99.7%)
are in 3.
Normal Distribution
For < x <

 (x )2 
f (x) = (2 2 )1/2 exp
2 2
with mean x and standard deviation x , then
x N (x , x )
x x
z=
N (0, 1)
x
Warning: Some people record the normal distribution as N (x , x2 ) i.e. the second number in brackets is
the variance rather than the standard deviation.
Uniform Distribution
For a x b
f (x) =

1
ba

E(x) =

a+b
2

V ar(x) =

Random Variables
Let x be a discrete random variable, then:
x = E(x) =


x

xP (x)

(b a)2
12

x2 =

(x )2 P (x)

The covariance of x and y is:

cov(x, y) = xy = E(x x )(y y ) =


x

The correlation between x and y is:

xy =

(x x )(y y )P (x, y)

xy
x y

For a continuous rv replace the sums by integrals.

Functions of Random Variables
If y = a + bx then:
E(y) = y = a + bx ,
and

y2 = b2 x2 .

If w = cx + dy then:
E(w) = w = cx + dy ,
and

2
w
= c2 x2 + d2 y2 + 2cdxy .

Sampling Distribution of the Sample Mean

For large samples,



x N ,
n

Probability Theory
P (A) = 1 P (A) (complement rule)
P (A B) = P (A) + P (B) P (A B) (addition rule)
P (A B) = P (A|B)P (B) (multiplication rule)
A and B are independent if

P (A B) = P (A)P (B)

Marginal probabilities add entries in a joint probability table. If Bi are mutually exclusive and exhaustive
events then:
n

P (A Bi )
P (A) =
i=1

Bayess Rule:
P (B|A) =

P (A|B)P (B)
P (A)

Binomial Distribution
For a success probability of p and sample size n:
P (x successes) =

n!
px (1 p)nx ,
x!(n x)!

and the discrete random variable x has mean

x = np
and variance

x2 = np(1 p)

Sampling Distribution of the Sample Proportion

For large samples,


p(1 p)
p N p,
n

Inference for Means

1. For large samples, a 100(1 )% CI for is:
(a)

x z/2
n

when the population variance is known to be 2 ;

(b)

s
x tn1,/2
n

when the population variance is unknown but estimated by s2 .

2. Dierences in Means: Dependent Samples
di = x1i y2i ,
n

then

i=1

d=


and
sd =

di

n

d)2
n1

i=1 (di

can be used with tn1,/2 to form a condence interval: d tn1,/2 sd / n.

3. Dierences in Means: Independent Samples

(variances unknown and not assumed equal):

sx1 x2 =

s21
s2
+ 2
n1
n2

can be used with t,/2 and

=

[(s21 /n1 ) + (s22 /n2 )]2

,
(s21 /n1 )2 /(n1 1) + (s22 /n2 )2 /(n2 1)

or often degrees of freedom approximated by the smaller of n1 1 and n2 1.

Inference for Proportions
1. For large samples, a 100(1 )% CI for p is:

p z/2

p(1 p)
.
n

(Or replace p = x/n by p = (x + 2)/(n + 4) when = 1%, 5%, or 10%.)

2. Dierences in Proportions
p1 p2
has standard deviation:

p1 (1 p1 ) p2 (1 p2 )
+
n1
n2

which can be used with z/2 to form a condence interval. (Or add 1 success and 1 failure to each sample
when = 1%, 5%, or 10%.)
3. Testing the Hypothesis of Equal Proportions
For this test, use the pooled estimate of the common value of p1 and p2 :
ppool =
to form
SEDp

X1 + X2
n1 + n2

1
1
= ppool (1 ppool )
+
n1
n2