64 
SYMPOSIUM PHOTO INTERPRETATION, DELFT 1962 
present study provided an opportunity to look at confidence for rights and 
wrongs as a function of photo mode and of photo quality. For each object 
identified, the interpreter assigned a percentage that represents what he con 
siders to be the likelihood that the object is correctly identified. Table 7 
shows the results obtained for high photo quality. Note that the right responses 
are invariably associated with greater confidence than the wrong responses 
and significantly so in three out of the four modes. This significant difference 
for confidence in right and wrong responses carries over to the total. Similar 
findings are obtained for the low photo quality as can be seen in table 8. 
Table 7. Percent confidence for rights and wrongs by mode for high photo 
quality 
Right 
Wrong 
Positive Transparency * 
57 
46 
Negative Transparency 
60 
56 
Stereo Print * 
73 
61 
Non-Stereo Print 
75 
62 
Total * 
65 
56 
* Right-Wrong differences significant at 
P < .05. 
Table 8. Percent confidence for 
rights and wrongs by mode for low photo 
quality 
Right 
Wrong 
Positive Transparency 
67 
64 
Negative Transparency * 
61 
44 
Stereo Print * 
73 
59 
Non-Stereo Print 
66 
56 
Total * 
68 
56 
* Right-Wrong differences significant at 
P < .05. 
When we compare total mean confidence for the right responses for the 
high and low photo qualities we find that they are substantially the same 65 
and 68 percent, respectively. This is also true for the mean confidence for the 
wrong responses. This implies that across the four modes, the probability of 
correctness of identification based on confidence is the same for both high 
and low photo qualities. However these findings do not rule out the possibility 
that for a given mode this probability could be greater than for another. 
Length of viewing time 
We have already seen some evidence that accuracy goes down and com 
pleteness up as a function of time. We have also noted interpreter performance 
with the exception of the tenuous finding for 30-minute completeness at low 
photo quality is not affected by photo mode. On the other hand, there are 
significant performance differences attributable to photo quality. A perform 
ance over time comparison was therefore only made for photo quality.