68 
PSYCHOLOGY: J. A. HARRIS 
and of 8.7% in the case of Observer D, who has a decidedly greater 
scatter in her estimates — that is a far less steady judgment — than either 
of the other observers. Indeed, in every individual experiment her 
standard deviation is higher than that of either of the two other 
experimenters. 
Thus while there is no certain differentiation among the experi- 
menters in personal equation, they differ distinctly in steadiness of 
judgment. 
The influence of previous experience upon personal equation or steadi- 
ness of judgment may be most succinctly expressed in terms of the 
correlation between some quantitative measure of the amount of pre- 
vious experience and the measures of personal equation and steadiness 
of judgment. 
In these experiments the errors of observation were recorded in se- 
quence. A group of fifty consecutive estimates with the accompanying 
determinations of the errors constituted a 'period.' In determining 
correlations one must deal with a number of subgroups for each period. 
It is most convenient to divide each half daily period of 50 estimates 
into five consecutive 'trials,' each of 10 estimates. For each of these 
'trials' the mean personal equation and the standard deviation of the 
errors must be computed. Thus in obtaining the constants discussed 
here it was first necessary to compute 1520 means and 1520 standard 
deviations, which were then treated as units in computing the correlations. 
The main problems involved in the question of experience are two: 
Is there a change in personal bias as a result of constant effort to im- 
prove and opportunity for improvement? Does the judgment become 
steadier, i.e., does the observer make less erratic estimates, as a result 
of experience? 
Both of these questions are really twofold. Is there an improvement 
from period to period? Is there an improvement within the period? 
In short, does the worker improve both from estimate to estimate in 
the same half daily period and also from period to period? 
Personal equation seems to be remarkably little influenced by experi- 
ence. In some experiments it increases, in others it decreases. The 
correlations may be either positive or negative in sign. Numerically 
they are generally low, and are in great part insignificant in comparison 
with their probable errors. Taken as a whole the results indicate a 
slight reduction in personal equation as a result of experience from 
period to period. Within the period there is no demonstrable influence 
of experience upon personal equation. 
Steadiness of judgment is in rather conspicuous contrast with per- 
