DESCRIPTIVES

Notation
The following notation is used throughout this chapter unless otherwise stated:

Xi Value of the variable for case i

wi Weight for case i

N Number of cases
Wi Sum of the weights for the first i cases

Xi Mean for the first i cases

Moments
Moments about the mean are calculated recursively using a provisional means
algorithm (Spicer, 1972):

j
Wj = ∑w i
i =1

vj =
wj
Wj
3X j − X j −1 8
W
+
2
− 3w j Wj −1 v W
M j4 = M j4−1 − 4v j M 3j −1 + 6v 2j M j2−1

j
w 3j

4
j j −1Wj

M 3j = M 3j −1 − 3v j M 2j −1 +
Wj Wj −1
w 2j
3W − 2w 8v
j j
3
j

Wj Wj −1
M j2 = M 2j −1 + v 2j
wj
X j = X j −1 + v j

W0 = X 0 = M 02 = M 03 = M 04 = 0

1
2 DESCRIPTIVES

After the last observation has been processed,

WN = sum of weights for all cases

X N = mean

∑w 3X 8 r
r
MN = i i −X
i =1

Basic Statistics
Mean

XN

Variance

S2 = M N
2
1
WN − 1 6

Standard Deviation

S = S2

Standard Error

S
SX =
WN
 DESCRIPTIVES 3

Minimum

min X j
j

Maximum

max X j
j

Sum

X N WN

Skewness and Standard Error of Skewness

3
WN M N
1 6 1W 1
6WN WN − 1 6
g1 =
1W − 161W
N N 6
−2 S 3
se g1 =
N 61 61
− 2 WN + 1 WN + 3 6
If WN ≤ 2 or S 2 < 10 −20 , g1 and its standard error are not calculated.

Kurtosis (Bliss, 1967, p. 144) and Standard Error of Kurtosis

1 6
WN WN + 1 M N
4
− 3M N
2 2
MN WN − 1 1 6 1 6 4
4 WN2 − 1 SE g192 1 672

g2 =
1W − 161W
N N − 261W − 36S
N
4
se g2 =
1W N − 361W + 56
N

If WN ≤ 3 or S 2 < 10 −20 , g2 and its standard error are not calculated.

4 DESCRIPTIVES

Z-Scores

Xi − X N
Zi =
S

If X i is missing or S ≤ 0 , Zi is set to the system missing value.

References
Bliss, C. I. 1967. Statistics in biology, Volume 1. New York: McGraw-Hill.

Spicer, C. C. 1972. Algorithm AS 52: Calculation of power sums of deviations

about the mean. Applied Statistics, 21: 226–227.