Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
B S-T.R i o T E pr inter-correlations) higher than .80, although, to be sure, there are a few such scattered ough the data summarized by Dr. Berry. Still another generalization which may be reached, lough it is not justifiable on the basis of any systematically collected data is that any of the ts, no matter how administered (within wide limits) will suffice to identify the very few cases : linical night blindness which may be found in the service population. More interesting than what was done on various night vision programs is, perhaps, what was done, or was done but rarely, and inadequately. The most conspicuously missing class of ivities has already been suggestedâthat is, the investigation of a wide variety of tests and in- uments under comparable conditions. Only the Weddell experiment (abstract no. 73) incor- â¢ated a very great variety of tests. The results showed that none of the tests were very satis- tory, but continuing efforts at further development were restricted to the more conventional truments, those involving acuity figures. The second missing class of investigations is those which might yield comprehensive sets of idation data. To be sure, a few sets of such data appear, but the criterion performances em- yed too often approximated the performance in the laboratory. The most "naturalistic",âor ce-valid" criterion performance, that of the Weddell experiment, yielded data which indicated, the validity of one or more of the tests studied, but the unreliability of the criterion perform- :e (and hence the less valid as a criterion) the higher the predictive value of the test. Most :h validation studies check performances on only one occasion, or over a very short period; y further seemed to assume that men were performing at their asymptotes. In only one lab- itory was an effort made to measure the reliability, over a period of time, of a criterion per- mance. Evelyn, in Canada, collected such data over several months on a group of men. But criterion performance duplicated closely, in the field, the performances typical of the tests to validated. His reliabilities, incidentally, were not very high.1 Third among the missing is an extensive analysis of the statistical logic of the tests, e.g. of role of chance in determining the scores obtained on tests employing forced guessing. The sumption that a "correction for guessing", such as the "R-l/3 W" employed in the four-choice ts, is adequate to assure that a valid measure of frequency of seeing is obtained was seldom istioned, despite the fact that such a correction is justifiable only when the number of guesses very large. In very few cases indeed was this variable taken into account as significant for determination of cutting scores, or of brightness levels at which testing was to proceed. Finally, there is almost a complete absence of data on the actual utility of the various tests practice. This is perhaps crucial to the usefulness of any test, however reliable, and however 11 validated. If a test lends itself to abuse in practice, and is such that uninterested, not to ntion unskilled, personnel can produce unreproducible results, then that test is not a useful ;, and may even do more harm than good. Hearsay evidence indicates that even those tests signed for administration by the most simpleminded were subject to abuse once distributed to field and put into operation in a large-scale testing program. There is doubt, then, that the work on night vision testing during the war has led to very use- results with respect to the selection of personnel for night visual duties, and it may follow t, if it is still considered desirable to develop such tests, a fresh start will prove more fruit- than the further investigation of tests now existent. Present test methods seem to assure the ruination of the clinically night blind, and any of them may be used for that purpose, but a con- ning need for positive selection procedures can only be served by further, and novel research. The basic fact is that we do not know very much about seeing at night. Seeing at night is >bably also dependent upon a variety of non-visual psychological and physiological factors, such fatigue, motivation, nutrition, situational variables and possibly others which we have not con- ered involved. Visual acuity, contrast sensitivity, absolute sensitivity, and movement sensi- Lty of all parts of the retina, which have not yet been convincingly demonstrated as co-varying,2 isumably contribute to night seeing, as do ocular tremor and habitual eye-movement patterns. a more complex level, so do training in the detection and recognition of objects under various iditions. Visual performance seems to show itself, both within the individual, and among 1 A result also obtained in the field tests on binoculars, to be reported soon. 2 Indeed, there is some evidence that such visual functions are only slightly correlated.
