BU,
In the Experiment Reichenbach 1962-1964 (Report Page 17) were presented some proportions between
standard deviations, compared with proportions between model dimensions, Corresponding proportions
have been calculated in this Experiment. Table 6.13.
Experiment No. of ss, Sa P/s, ^ t/s,
observ,
Pecnÿ 1966-1968 all scales 8117 0.68 0.64 0.95 1.00 1.39
Reichenbach 1962-1964 13320 0.46 0.58 0,77 1.00 1.27
Monti de Revóira 2845 0,51 0.57 0.76 1,00 1526
Weighted mean 24282 0.54 0.60 0.83 1.00 1.33
Theoretical proportions from 0.48 0.63 0.79 1.00 1.28
model dimensions
We found rather large deviations in the Experiment Pecny from the theoretical proportions for the ne-
gative scales 1:3 500 and 1:6 000 but not for 1:12 000. The weighted means must however be repre -
sentative,
The standard error in space is presented in figure 6.11 for every participant and model in relation
to the general means for every negative scale. The figure indicates a rather homogeneous result
of this Experiment.
6.2 Statistical tests of comparison points
About 20 % of the comparison points in the group of models had constant errors or lacking symmetry
around zero table 6.21. In the Experiment Reichenbach 1962-1964 there were 75 % such points and
in the Experiment Monti di Revóira 1958-1960 no constant errors. From a practical point of view the
comparison points could be considered free from constant errors,
About 60 % of the coordinates were not normally distributed. Table 6.22. The same result was found
in the Experiment Reichenbach 1962-1964, This result is aJso confirmed in section 8.
Homogeneous variance was found at the 1 % level. Heterogeneous variance was found in the Experi -
2 8
ment Reichenbach 1962-1964, Homogeneous variance for x and y but not for z was found in the
Experiment Monti de Revóira 1958-1960. Remark. Bartletts test could indicate deparature from nor-
mality rather than heterogeneity of variance, So in practice homogeneous variance exists,