(Page: 1-8) Version: January 2018

Statistical Formulas

Vrije Universiteit
School of Business and Economics

Formula Sheets

Please do not write on the sheets and return

them afterwards
Basic formulas for Statistics January 2018
position P th-percentile
P
(n + 1)
100
Mean
P P Pc
xi xi j=1fj mj
= x= or x=
N n n
1
xG = (x1 x2 xn ) n
Variance
P
2 (xi )2
=
P N P 2 P 2 P P
2 (xi x)2 xi nx2 xi ( xi )2 =n fj (mj x)2
s = = = or
n 1 n 1 n 1 n 1
1
Chebyshev: 1 100%
k2
s
CV = 100% CV = 100%
x
P
X;Y (xi x) (yi y)
= r = qP qP
2
X Y
(xi x) (yi y)2
Odds:
P (A) 1 P (A)
1 P (A) P (A)
Bayes:
P (B1 \ A) P (AjB1 ) P (B1 ) P (AjB1 ) P (B1 )
P (B1 jA) = = P =P
P (A) P (A \ Bi ) P (AjBi ) P (Bi )

X
= X = E (X) = xi P (X = xi )
X X
Eh (X) = h (xi ) P (X = xi ) e.g. E (X 3)2 = (xi 3)2 P (X = xi )
E (aX) = aE (X)
X
2
= 2
X = var X = (x )2 P (X = x) = E (X )2 =
= E (X E X)2 = E X 2 (E X)2
var (a X ) = a2 var X aX = jaj X

2
 X
X;Y = cov (X; Y ) = (xi X ) (yj Y ) P (X = xi ; Y = y j )
i;j

1 X
N
= (xi X ) (yj Y)
N i=1

1 X
n
sX;Y = (xi x) (yj y)
n 1 i=1
sX;Y
rX;Y =
sX sY
cov (aX; bY ) = ab cov (X; Y )
2 2 2
X+Y = X + Y X+Y = var (X + Y ) = X + Y +2 X;Y

Variables X and Y are independent if and only if for all x and for all y:

P (X = x; Y = y) = P (X = x) P (Y = y)

If X and Y independent:
2 2 2
X+Y = var (X + Y ) = X + Y cov (X; Y ) = 0

Special Distributions:
n n! n!
n Cr = = n Pr =
r r! (n r)! (n r)!
Binomial:
n
P ( X = x) = x
(1 )n x
; EX=n ; var X = n (1 )
x
(normal approximation if n 5 and n (1 ) 5)
(Poisson approximation if n 20 and 0:05)
Hypergeometric ( = S=N ):
S N S
x n x N n
P ( X = x) = N
; EX = n ; var X = n (1 )
n
N 1

(normal approximation if expected frequencies in all cells are at least 5)

(binomial approximation if n=N < 0:05)
Poisson:
x
e
P ( X = x) = ; EX= ; var X =
x!
(normal approximation if too large for table)

3
Geometric:
1
P ( X = x) = (1 )x 1
; for x = 1; 2; : : : EX = ; var X = (1 )= 2

Uniform discrete:
1
P (X = x) = for x = a; a + 1; : : : ; b
b a+1
1 1
EX = (a + b) ; var X = (b a + 1)2 1
2 12
Uniform continuous:
1 1 1
f (x) = for a x b EX= (a + b) ; var X = (b a)2
b a 2 12
Exponential:

f (x) = e x for x 0; > 0

x
P (X x) = 1 e for x 0
1
EX = X =

Normal:
1 1
f (x) = p exp 2
(x )2 ; EX= ; var X = 2
2 2

Estimators.
Estimator for :X
2 2
N n
EX = ; var X = or
n n N 1

X
p N (0; 1) (normal popul.; or symmetric popul. and n 15; or n 30)
= n
X
p tn 1 (normal popul.; or symmetric popul. and n 15; or n 30)
S= n
2
Estimator for : S2
(n 1) S 2 2
2 n 1 (normal populations)

Estimator for : p formerly also b

p or p(n)

(1 ) (1 )N n
Ep = ; var p = or
n n N 1

4
Con…dence intervals:
r
N n
x z =2 p or x z =2 p
n n N 1
r
s s N n
x tn 1; =2 p or x tn 1; =2 p
n n N 1
r r r
p (1 p) p (1 p) N n
p z =2 or p z =2
n n N 1
p
x z =2 x
(n 1) s2 2 (n 1) s2
2 2
n 1; =2 n 1;1 =2

Sample size:
r
(1 )
E=z =2 p or E=z =2
n n

Testing hypotheses
2
x 0 x 0 2
z= or z= q n = (Z + Z )
p N n
n p
n N 1

x 0 x 0
t= or t= q
ps
n ps N n
n N 1
r
n 2
t=r df = n 2
1 r2
p
z = rS n 1
1 1
x n 0 2
x n 0 2
z=p or z=p q
n 0 (1 0) n 0 (1 0)
N n
N 1

2 (n 1) s2
= 2
(normal populations)
0

x1 x2 ( 1 2)
z= p 2 2
1 =n1 + 2 =n2
2
s21 s22
x1 x2 ( 1 +
2) n1 n2
t= p ; df = 2 2
2 2 s2 s2
s1 =n1 + s2 =n2 1
n1
2
n2

n1 1
+ n2 1

5
 x1 x2 ( 1 2) (n1 1) s21 + (n2 1) s22
t= q ; df = n1 + n2 2; s2p =
2 2
sp =n1 + sp =n2 (n1 1) + (n2 1)

p1 p2 ( 1 2)
z=p
p1 (1 p1 ) =n1 + p2 (1 p2 ) =n2
p1 p2 0 x 1 + x2
z=p ; p=
p (1 p) =n1 + p (1 p) =n2 n1 + n2

s21
F = (df1 ; df2 ) = (n1 1; n2 1) (normal populations)
s22

0
X
n
n0 (n0 + 1) n0 (n0 + 1) (2n0 + 1)
W = Ri+ ; EW = ; var W =
i=1
4 24
n1
X n1 (n1 + n2 + 1) n1 n2 (n1 + n2 + 1)
T1 = Ri ; E T1 = ; var T1 =
i=1
2 12
X (fjk ejk )2
2
= ; df = (r 1) (c 1) ejk 5
ejk

Simple Regression:
P P
SSxy (xi x) (yi y) xi yi nxy
b1 = = P 2 = P 2
SSxx (xi x) xi nx2

b0 = y b1 x

X X
e2i = SSE = (yi yb)2

P
2 2 e2i
b = s = M SE =
n k 1

X
n
(et et 1 )2
t=2
DW =
X
n
e2t
t=1

b1 1 b
t= where sB1 = qP ; df = n k 1
sB1 2
(xi x)

6
 r
n 2
t=r ; df = n k 1
1 r2

p p 1 (xi x)2
ybi tn 2b hi resp. ybi tn 2b 1 + hi where hi = +P
n (xi x)2

Multiple Regression
Sum of Mean
Squares df Squares
P
Regression (b
y y)2 k
P i
Residual (y ybi )2 n k 1 b2error
P i
Total (yi y)2 n 1
P P
(b
yi
2 y)2 e2i
R =P =1 P
(yi y)2 (yi y)2

B Std.Error Beta t
b0
(Constant) b0 s (B0 ) s(B0 )
b1
X1 b1 s (B1 ) ::: s(B1 )
b2
X2 b2 s (B2 ) ::: s(B2 )
::: ::: ::: ::: :::
1
P
2 n 1 n k 1
e2i
Radj =1 1 R2 =1 P
n k 1 1
n 1
(yi y)2

partial F

(SSERestricted SSEF ull ) =m

F = Fm;n k 1
M SEF ull
s
1 M SE
V IFj = 2
sb j =
1 Rj SSxj 1 Rj2
leverage

2 (k + 1)
hi
n

7
ANOVA (c columns, n1 ; n2 ; : : : ; nc observations per column)

Source of variation Sum of Squares df

P 2
Between SSA = nj y j y c 1
PP 2
Error (‘Within’) SSE = yij y j n c
P 2
Total SST = yij y n 1

Tukey
s
yj yk 1 1
Tcalc = r Tc;n c ; Crit.Range = Tc;n c M SE +
1 1
nj nk
M SE nj
+ nk

ANOVA (r rows, c columns, m observations per cell)

Source of variation Sum of Squares df
Pr 2
ROWS (A) SSA = c m 1 y i y r 1
Pc 2
COLUMNS (B) SSB = r m 1 y j y c 1
Pr Pc 2
INTERACTIONS SSAB = m 1 1 y ij y i yj +y (r 1) (c 1)
Pr Pc Pm 2
Error SSE = 1 1 1 yijk y ij: rc (m 1)
Pr Pc Pm 2
Total SST = 1 1 1 yijk y n 1

Kruskal-Wallis
!
12 X Tj2
c
H= 3 (n + 1) df = c 1; nj 5
n (n + 1) 1 nj