Comparison of reading performance tests concerning difficulty of sentences and paragraphs and their reliability

Abstract

In research and practice, sentences or paragraphs of reading tests may be randomly chosen to assess reading performance. This means that in addition to test reliability, all sentences or paragraphs should be reliable and equally difficult to read. The sentences and paragraphs of five (un-) standardised Dutch reading tests were investigated in this regard. Methods: Tests were performed with 71 normally sighted persons (mean age 55 [18-86] years). All sentences and paragraphs had equal print size. The relative difficulty of sentences and paragraphs from the five Dutch reading tests was tested with linear mixed models (reading speed) and generalised linear models (mistakes). Results: Reading speed in standard words per min ranged from 179 (Radner) to 142 (De Nederlanders). Reading mistakes per 100 characters ranged from 0.25 (Radner) to 0.40 (Colenbrander). On the Colenbrander charts 7/24 sentences were read significantly faster vs 5/24 read slower (sentence reliability 0.56-0.87); International Reading Speed Texts 3/10 vs 3/10 [0.94-0.97]; Laboratory of Experimental Ophthalmology 14/55 vs 15/55 [0.64-0.92]; De Nederlanders 2/6 vs 3/6 [0.83-0.94]; Radner 4/24 vs 3/24 [0.73-0.87]. Agreement between tests differed from 1 to 36 standard words per minute and 0.01 to 0.14 mistakes per 100 characters. Conclusion: The Radner, with the highest number of equally difficult sentences, is appropriate to measure reading acuity as well as reading speed in a heterogeneous population; the International Reading Speed Texts, with the highest paragraph reliability, provides long paragraphs to measure reading speed. The Colenbrander and Laboratory of Experimental Ophthalmology are suitable for daily practice; however, for research or inspection purposes, reliable sentences must be chosen. Although the clinical relevance of the differences between the tests is debatable, use of the De Nederlanders as a reading test remains questionable.