Beruflich Dokumente
Kultur Dokumente
X μ
Z
σ
n
Confidence interval formulae Determining sample size
XZ
σ Z 2 σ2
n n
Finite correction factor = square root { ( N - n ) / (N -1) }
e2
P-value approach to testing
X μ
t n -1
S
n
d.f. = n – 1
S
X t n -1
n
(c) Population proportion
pπ
Z
π (1 π )
n
Confidence interval Estimating sample size
p(1 p)
pZ
n Z2 π (1 π )
n
e2
Hypothesis testing and CI about differences in two means using Z statistic ( population variance unknown )
2 2 X X μ μ2
σ1 σ 2 Z
1 2 1
σ X1 X 2 2
σ1 σ 2
2
n1 n 2 n1 n 2
Confidence Interval
X X t
1 2 n1 n 2 - 2
1 1
S2p
n1 n 2
Independent samples and population variances unknown and assumed unequal
t
X X μ
1 2 1 μ2
2 2
S S
1 2
n1 n 2
Statistical Inference for two related populations
D μD
n
Z D i
σD D i 1
n
n
μD = hypothesized mean difference
σD = population standard dev. of differences
n = the sample size (number of pairs)
σD
DZ
n
Standard deviation unknown σD Known
D μD
n
(D i D) 2 t D t n 1
SD
SD i 1 SD
n 1 n
n
Two population proportions
Z
p1 p 2 π1 π 2 X1 X 2 X X
p , p1 1 , p 2 2
1 1 n1 n 2 n1 n2
p (1 p)
n1 n 2
p1 (1 p1 ) p 2 (1 p 2 )
p1 p2 Z
n1 n2
S12
F 2
S2
Non Parametric
Runs Test
2 n1 n2 (2 n1 n2 n1 n2)
Z
R
2 n1 n2
1
(n1 n2)
R R 2
R
(n1 n2 1) R
n n
1 2
Mann-Whitney U Test
U n n 1
W 1
nn 1 2
1 1
2 n n 1 2
U 2
where :
n n n n 1
n 1
number in group 1
U
1 2 1
12
2
number in group 2
n 2
If n > 15, T is approximately normally distributed, and a Z test is used. If n <= 15, a special “small sample” procedure is followed. The paired data are randomly
selected. The underlying distributions are symmetrical.
n n 1
T
4
nn 12n 1
T
24
Z
T T
T
Kruskal-Wallis Test
12 C T j
2
K 3n 1
nn 1 j 1 n j
where : C = number of groups
n = total number of items
Friedman Test
T j
total of ranks in a group
n j = number of items in a group
K χ 2 , with df = C - 1
Friedman Test
Spearman’s Rank Correlation
6 d
C 2
12
2
3b(C 1)
2
bC (C 1) j 1 R j
r
r s
1
n n 1
2
where : C number of treatment levels (columns) where : n = number of pairs being correlated
b = number of blocks (rows) d = the difference in the ranks of each pair
R j = total ranks for a particular treatment level
j = particular treatment level
2 2
, with df = C - 1
r
( X i X )(Yi Y )
n n
r
cov (X , Y) (Xi X) 2 (Y Y)
i
2
cov ( X , Y ) i 1
SX SY SX i 1
SY i 1
n 1 n 1 n 1
Regression Model
0 r2 1
n
SSE (Y Yˆ )
i i
2
SYX i 1
n2 n2
SSR
S SYX MSR MSR
Sb1 YX b1 β1 d.f. n 2 k F
t
SSX (X i X )2 Sb1 MSE
SSE MSE
b1 tn 2Sb1 n k 1
MSA is mean squares among groups. MSW is mean squares within groups. Degrees of freedom df1 = c – 1 (c = number of groups). df2 = n – c (n = sum of
sample sizes from all populations)
MSA
F
MSW
Tukey-Kramer Procedure
MSW 1 1
Critical Range Q U
2 n j n j'
QU = Value from Studentized Range Distribution with c and n - c degrees of freedom for the desired level of alpha MSW = Mean Square Within nj and nj’ =
Sample sizes from groups j and j’