Abstract
Purpose: :
To determine how measures of reading capability estimated from reading performance are related to measures of reading ability estimated from patient reports.
Methods: :
Reading speed in words per minute was measured in 109 low vision patients at 14 print sizes ranging from 0.4M to 8M in 0.1 log unit steps using MNRead. The same patients were asked to rate the difficulty of a subset of 120 reading tasks in the Activity Inventory. Maximum reading rate, log reading threshold, and the log of the ratio of critical print size to reading threshold were estimated from a least-squares fit of a template function to each patient’s reading speed vs. print size data. Log reading speeds were binned into 8 ordinal categories and Rasch analysis was performed on the binned data, treating each print size as a separate item, to obtain estimates of reading capability on an interval scale for each patient. Rasch analysis also was performed on the difficulty ratings of the reading tasks to obtain an estimate of self-perceived reading ability on an interval scale for each patient.
Results: :
Item measures estimated by Rasch analysis of the reading speed data were linear with log print size. Mean square fit statistics indicated that both estimated item measures and estimated person measures of reading capability are unidimensional and valid. Estimated reading capability measures were strongly correlated with the log reading threshold (r = -0.91), moderately correlated with maximum log reading rate (r = 0.38), and weakly correlated with the log ratio of the critical print size to the reading threshold (r = -0.21). The mean square fit statistics for both item and person measures of self-perceived reading ability were shown to be unidimensional and valid. Estimated measures of self-perceived reading ability were moderately correlated with estimated measures of reading capability (r = 0.55), with a strong linear trend in the relationship between the two variables. Self-perceived reading ability was moderately correlated with log reading threshold (r = -0.42) and maximum reading rate (r = 0.30), and weakly correlated with the log ratio of critical print size to reading threshold (r = -0.21).
Conclusions: :
A single summary measure of reading capability estimated from reading speed vs print size is influenced primarily by the patient’s reading acuity and secondarily by the patient’s maximum reading rate. Self-perceived reading ability depends on reading capability, as estimated from performance measures, but also is influenced by other variables independent of capability.
Keywords: low vision • reading • clinical (human) or epidemiologic studies: systems/equipment/techniques