Lessons for conducting random assignment in schools
| DOI | https://doi.org/10.1108/17466660200800010 |
| Date | 12 April 2008 |
| Published date | 12 April 2008 |
| Pages | 28-39 |
| Author | Gary Ritter,Marc Holley |
| Subject Matter | Education,Health & social care,Sociology |
28
1Endowed Chair in
Education Policy,
University of
Arkansas
2Doctoral Fellow,
University of
Arkansas
Journal of Children’s Services
Volume 3 Issue 2 September 2008
© Pavilion Journals (Brighton) Ltd
Abstract
The use of random assignment can be effective and appropriate in the evaluation of programmes that
serve children in schools. Because random assignment creates pre-treatment equality between treatment
and control groups, this methodology is particularly effective for understanding the impact of an
intervention. Contemporary research on educational experiments has tended to focus on programme
results rather than on their origin or implementation. While programme results are important, they provide
little guidance to those interested in designing and implementing programme evaluations that use random
assignment. This article shares the practical lessons learned from three educational experiments with
researchers and practitioners interested in pursuing evaluations that use random assignment.
Key words
education; experimental design; random assignment; RCT
Introduction
In 2005, the US Department of Education, one of the
largest funders of education research in the world,
announced that it would favour evaluation that uses
random assignment. The push for random assignment
as a matter of US federal policy, which is also
beginning to be seen in other western developed
countries, indicates that some public policy makers
have begun to understand that the field of
educational research needs to employ the same
rigorous experimental methodologies that are used in
medicine and the applied sciences. The lessons in
this article may help to facilitate the forming of
relationships between researchers and practitioners
that undergird successful social science experiments.
Random assignment studies of education
interventions have had some success in informing
policy debates over the past 20 years in the US. For
example, two prominent randomised controlled
studies, the Perry Preschool Project1(Schweinhart et
al, 1993) and Tennessee’s Project STAR2(Word et al,
1990), have had great influence on decisions
concerning early childhood education and elementary
school class size. Furthermore, results from
randomised evaluations of job training and
employment initiatives have provided much of the
basis for welfare reform debates over the past decade.
Random assignment in programme evaluation has
tremendous import, especially in a time of contested
resources for social programmes, and its importance
for understanding the impact of interventions in the
field of children’s services cannot be overstated.
Perhaps one of the most important primers on
experimental design, Campbell and Stanley’s (1963)
Experimental and Quasi-Experimental Designs for
Research, refers to randomisation as the ‘all-purpose
procedure for achieving pre-treatment equality of
groups, within known statistical limits’ (p6). The
importance of pre-treatment equality of treatment
and control groups is that it allows researchers to
estimate what subjects would have looked like in the
absence of the treatment; in other words, random
assignment creates the best counterfactual to
understand the impact of an intervention.
Boruch (1997) states that, as a result of random
assignment, ‘estimates of the relative differences
among treatments being compared will be unbiased;
Lessons for conducting random
assignment in schools
Gary W Ritter1and Marc J Holley2
To continue reading
Request your trial