477
Table 3. Investigation 1. Correlation Coefficient Between Blood Alcohol Con
centration and Standard Deviation.
Operator
no.
Date
s
X
s
y
s
p
i
11.2
0.535
0.672
0.441
13.2
0. 199
0.324
0.590
2
12.2
0.457
0.070
0.923
18. 2
-0.526
-0.328
0.018
3
12.2
0.410
0.094
0. 167
18.2
-0.492
-0.591
-0.081
4
11.2
0. 144
11.2
0.295
13.2
0.099
13.2
0.331
deviations may be regarded as an estimate of a theoretical standard
deviation only if the corresponding chi 2 -value from Bartlett’s test is
not significant. In investigation 2 every operator measured with the
same precision all day.
The root mean square values of the standard deviations of the hori
zontal parallax measurements for every day are plotted in diagram 1.
The values of operator 3 were much lower during the last days due to
training; unfortunately this experiment was the first time he had
measured with an instrument of this type. Other operators also showed
a similar but less evident effect of training. This dominated operator 3.
This can be explained in the following way: after taking the alcohol
the operator was more careful when setting the mark, knowing that he
was under the influence of alcohol and that his results were being
noted. This can also be applied to other operators as well.
Correlation Between the Blood, Alcohol Concentration and
Standard Deviation
To investigate if there exists a linear relationship between the blood
alcohol concentration and standard deviation of a setting in a stereo-