The main focus for this paper is an overview and critique of an extensive s
eries of empirical evaluations of Integrated Learning Systems (ILS) in UK c
lassrooms which took place between 1994 and 1998, The main purpose of the i
nvestigations was to determine if the use of ILS was effective in producing
worthwhile gains in basic numeracy and literacy. Findings from three phase
s of research offered evidence relevant to this question. The safest conclu
sion is that the investigations failed to find any convincing evidence for
gains, Though 'safe', this conclusion masks some highly significant differe
nces between the findings of the three phases, One produced evidence of sta
tistically significant, positive and educationally worthwhile effects. A se
cond generated relatively neutral findings; a mixture of small positive and
equally small negative effects, while a third led to the conclusion that i
s was "difficult to find anything but an apparently negative effect on achi
evement". A consideration of the nature and implications of these different
; findings forms the main content of this paper. The findings are assessed
from three perspectives. The first explores issues concerning research meth
odology with a view to identifying minimal rules of "good practice" in empi
rical evaluations of the impact of computers on educational achievement. Th
e second considers the findings from a more theoretical perspective; explor
ing the case that conceptual analysis can also provide a source of convergi
ng evidence in evaluations of the educational impact of technology. Finally
, the paper cites evidence from observational data to explore the thesis th
at any impact of ICT on pupils' learning will not be understood without att
ention to the way in which the technology is assimilated into teaching and
learning practices. We argue that, even in relation to systems designed to
support individualised, automated teaching, there are sound theoretical rea
sons to expect that any long-term and cumulative effects on performance wil
l be mediated and influenced by classroom practice, Finally, implications o
f this argument for developing principles of design based on the "system-le
arner-teacher" triangle are explored. (C) 2000 Published by Elsevier Scienc
e Ltd. All rights reserved.