http://telem-pub.openu.ac.il/users/chais/2008/after_noon/3_1.pdf

By Rakefet Ackerman and Morris Goldsmith.

Most people, including our survey participants, believe that they learn less
efficiently when reading from a computer screen than when reading from
paper. We investigated the validity of this belief within a metacognitive
framework for self-regulated learning. In three experiments, participants
studied expository texts presented on paper or on screen, either in a selfregulated
manner, or according to a fixed study-time schedule. They
predicted their performance on memory and comprehension questions before
being tested. Objective performance, subjective metacognitive monitoring,
and control of study time were measured. The interrelationships between
these variables and the learning media were analyzed within an adapted
discrepancy-reduction framework (Dunlosky & Hertzog, 1998). On-screen
learners performed worse than on-paper learners under self-regulated study,
but not under a fixed study-time schedule. They also exhibited less calibrated
(more overconfident) monitoring and a weaker relationship between
monitoring and control of study time than did on-paper learners. Hence, an
on-screen learning deficit was in fact observed, which appears to derive from
less efficient metacognitive regulatory processes rather than from ergonomic
factors per se. This study highlights the potential of the metacognitive
approach for clarifying the source of differences in learning processes, and
for pointing toward appropriate means of improving these processes.