Calibration of LEA SYMBOLS® 1993
This calibration followed the same routine that was used in previous calibrations of
these symbols (Hyvärinen et al. 1980):
Figure 1. Test charts used in calibration of the LEA SYMBOLS®
1. Eight experienced subjects were tested using eight charts with the lay-out depicted in Figure 1A. These charts consisted of rows of five symbols in random order. In three of the charts the symbols were the 0.5 (20/40, 6/12 equivalent) line symbols of the previous LH chart, two were 96% and two were 92% of that size. The subjects were aware of the fact that different optotype sizes were used and thus the symbols would be recognisable at different distances.
Since all subjects had visual acuity better than 1.6 (20/12, 6/4), test distances were close to 10 meters and thus easy to measure. The subjects were asked to read five symbols on one line at a time from right to left or from left to right while moving closer to the test half a step at a time until the threshold.
The threshold for each person was individual, i.e. the level at which the person functioned most repeatably. Therefore the threshold varied from person to person between 60 to 80% correct. The person was decided to be at threshold if the same threshold was repeatable at least three times for 100 responses each time. Results of these measurements confirmed that the optotype 'square' was more easily recognised than the other symbols. The results also showed that there were unexpectedly large differences between the different subjects in the way they saw the optotypes.
During testing it was apparent that this type of testing contains a source of error. If a subject recognises a symbol as the one which is the next on that row, also the next symbol is recognised wrongly. This can lead to a situation that a whole row is given wrong responses at a genereal level of 80% correct. Therefore testing was repeated.
2. The same eight subjects were tested a second time in the following way: Now only one symbol was shown at a time through a circular opening of 2cm in diameter. The next symbol was chosen randomly by the tester and the subjects were informed that the same symbol might be shown repeatedly. The distances where each subject was at threshold were recorded using again the criteria of three times 100 responses. During this test all responses were recorded to observe which confusions were most common (Figure 2.).
The test results were identical with the first testing: different persons saw different
symbols in a very variable way. The precent of wrong responses (Table 1.) was
lowest for the symbol 'square' varying between 11.8% and 30.2% with a mean of 19%,
the 'apple' between 13.6.% and 33.0% with a mean 23%, 'ball' between 19.0 and 37%
with a mean of 28.5% and 'house' between 17.4% and 37.5% with a mean 28.7%.
The threshold for seeing 5M Landolt C symbols (Figure 1B) was measured in an identical
way as that for the symbols. It was surprising to see that one of the subjects
regularly saw the Landolt C with an opening to the left better than the Landolt C
with an opening to the right. The thresholds were again based on three times 100
responses. The Landolt C chart had five lines of five optotypes and could thus be
turned in four orientations. Therefore only one chart was needed.
Originally this testing was planned to be used for calibration of the optotypes, i.e. the correct size of the optotypes would be defined based on the relationship betweeen the distances at which the symbols on the one hand and the Landolt C on the other hand were recognised. Since the optotypes were found to have different visibility from each other, the measured distances could not be used for calibration. Therefore a third test situation was designed. It is described in the main text.