on pupils’learning outcomes included a number of variables that served as indi-
cators of teacher characteristics such as gender, years of schooling, and degrees
obtained, but for the most part ignored the process involved in learning to teach.
Many large-scale international empirical studies used less-than-ideal proxies for
what teachers knew, such as the number of years of teachers’formal education, the
quantity of courses they have taken, or whether they attended formal teacher
education or development programs (see Fuller and Clarke 1994 ). Studies specif-
ically directed at exploring the impact of teacher education and development had
failed to include measurements that could actually reflect what teachers knew (such
as observations and tests of knowledge and skills).
In a review of the literature, Kennedy ( 1999 ) found that most studies included
indirect measures—such as situated descriptions of teaching by teachers, including
teachers’daily logs or vignettes, non-situated testimony about practice (e.g., teacher
questionnaires and interviews that ask about teaching practices), or testimony about
effects of policies or programs in which teachers are asked to judge how a policy
affects teaching practice. She argued that as one is further removed from direct tests
of knowledge and observations the larger is the risk of self-serving bias, lack of face
validity, estimation errors, and reliability. More recent work explored relationships
between time and learning, pupil cognition, and teacher cognition, and decision
making. Many of these studies have, for the most part, ignored indicators essential
to teachers’learning and teaching quality, such as the knowledge that teachers
acquire as a result of teacher education and how it relates to curriculum and
instruction, and to the values orientation of teachers and their pupils (Good and
Brophy 2000 ).
Although still limited in scope, there were exceptions in the international
mathematics research literature found in the works by Fuson and Kwon ( 1992 ),
Fuson et al. ( 1997 ), LeTendre and Rohlen ( 1999 ), and Linn et al. ( 2000 ). Though
less dominant, the comparative literature also provides examples of the application
of ethnographic and ecological approaches to understanding teacher learning and
quality (Ma 1999 ). A modest number of studies that linked teacher education with
teacher quality and pupil learning used statistical analysis methods, such as HLM or
ML3, to allow for analysis across the nested contexts of schooling; such methods
later came into common use.^10 By the early 2000s, however, no large-scale studies
had linked teacher education with teachers’acquired knowledge for teaching and
with teacher quality and pupil learning.
In sum, the research on teacher education—while making important contribu-
tions to thefield—was lacking in several aspects, ranging from the positing of valid
and relevant research questions to how and whether the research was disclosed for
examination by the users of the knowledge produced. The quality of educational
research as an issue was widely acknowledged by the early 2000s, and the edu-
cational community, with the support of the National Academy of Sciences, agreed
(^10) See Creemers and Reezigt ( 1996 ), Murnane et al. ( 1996 ), Raudenbush and Bryk ( 1986 ), and
Riddell ( 1989 ).
626 M.T. Tatto