Reading from CRT versus Paper: The CRT-Disadvantage Hypothesis Re-Examined

Abstract

This paper examines the possibility that work done on a CRT is more difficult to perform than similar work in a traditional paper format. Several recent papers have reported data bearing on this issue. Thus far, two tasks have been considered: proofreading for errors and reading text for comprehension. Some consistent trends have been noted: (1) There is no difference in either number of proofreading errors made or reading comprehension scores when proofing or reading material presented in either mode; (2) Proofreading from CRT appears to take longer than proofreading from paper. The data on text reading speed are less consistent. Given the proofing data, a similar difference in speed with the two modes is predictable, i.e., slower reading from CRT than from paper. However, that data on this point are mixed. In some instances, a CRT-disadvantage is reported (1,2), but in at least one instance (3) it is not. Interpretation of these data is complicated by the fact that the experimenters who do report the predicted CRT-disadvantage have failed to control for a number of critical variables which normally exist between CRT and paper presentations, in particular, difference in presentation format (e.g., number of characters per line, number of lines per page, etc.). Cumulatively, these lack of controls bias the data in favor of the CRT-disadvantage. In addition, the use of between-groups designs has failed to control for individual differences in reading ability which can exist between subjects in different groups and which also may account for the obtained CRT disadvantage. The present study was undertaken in order to determine whether or not the CRT handicap for speed of reading text actually exists. It was designed to incorporate the following: (1) proper controls for presentation format and reading ability; (2) varying difficulty levels of the material to be read (based on the observation that the material read by Ss in prior experiments has been unduly difficult; simple, clear and interesting material has been neglected); (3) two readings of the same material, in order to evaluate the potential effect of experience on CRT reading performance. Adult subjects read to completion, then answered multiple choice questions about, two short articles which differed in degree of difficulty. The difficult article was taken from a college level standardized reading test; the easy passage was taken from a popular women's magazine. The latter was edited to match the former in presentation format. Subjects served as their own controls; their participation was split over a two day test session. On Day 1, Ss read both articles, one from each presentation mode. On Day 2, they reread both articles from alternate modes. Order of presentation of both articles (easy, difficult) and separately, presentation mode (CRT, paper) was counterbalanced. The primary dependent measure of interest was reading time to completion; percent correct comprehension was also recorded. Separate analyses of Days 1 and 2 revealed the following similarities: (1) Reading time was longer for the difficult than for the easy passage; (2) There was no difference in reading time for the two presentation modes. In short, this test, with its adequate and proper experimental controls, did not support the CRT-disadvantage hypothesis. Two differences in the data between Days 1 and 2 were observed. Not unexpectedly, overall reading times were faster on the second day. More importantly, on Day 1, there was a significant mode x difficulty level interaction, i.e., the hard article was harder to read on the CRT than paper, but the easy article was equally easy in both conditions. (A floor effect is discounted, since reading times were reduced the second day.) This interaction was absent in Day 2, presumably, the second reading of the hard article renders it “easy.” Discussion of these data will include: (1) the comparison of these data (notably, the difficulty level x mode interaction) with existing literature supportive of the CRT-disadvantage hypothesis; (2) their implication for software designers, particularly with reference to the cognitive complexity of the human interface; (3) their theoretical implications when considered in light of the proofreading data; and (4) proposed research which further evaluates the “experiential” factor, i.e., repetitions of the same reading material versus continuous reading of connected (but non-repeating) prose.