
domain. The statistical method allows one to obtain more information on the
reliability of the results.
Because of rando m reading errors and uncontrolled perturbations, the test
values will follow a given distribution. We can model such distributions by
treating x as a stochastic variable.
The probability density function, f ðXÞ , of the variable x is the probability to
find x between X and X þ dx.
Its integral F(X) is the probability of hav ing x < X:
FðXÞ¼probðx < XÞ¼
ð
X
1
fðxÞdx ð7:48Þ
The lower significance limit is the value X
i
for which FðX
i
Þ¼p, where p is a
given probability. The upper significance limit is the value X
s
for which
FðX
s
Þ¼1 p.
The confidence interval [X
i
; X
s
] is the range between the lower and the upper
significance limit (see Figure 7.2). The probability to find x in this interval is
P ¼ 1 2p .
Average
If the same importance is given to all the results, an estimate of the average, ,
of the variable, x, based on N measurements is calculated by:
hxi¼
P
i
x
i
N
ffi ð7:49Þ
where the sum runs over these N measurements (i ¼ 1; ...; N).
If we give more importance to some measurements than to the others, a
weight, w
i
, can be attributed to each value, x
i
, and the weighted average is
calculated by:
hxi¼
P
i
w
i
x
i
P
i
w
i
ð7:50Þ
0
0.1
0.2
0.3
0.4
0.5
-3 -2 -1 0 1 2 3
x-<x>
s
Confidence
interval
Figure 7.2 Significance limits and confidence interval
Common Methods and Techniques 155
Kommentare zu diesen Handbüchern