Key questions facing researchers and policymakers are whether teacher education programs are effective in changing teachers' knowledge and practices and whether such changes, if they occur, increase student learning. Answers to these questions will help determine the characteristics of effective programs for reading teachers.
Because there are multiple layers of causal relationships, encompassing teacher educators to students and including materials and environment, researchers typically focus on a few processes of teaching and learning at a time, usually using a specific method to answer the research question. Research on instructional variables, for example, generally examines the interactions between teachers and students in particular contexts of learning.
Experimental vs. Non-experimental Studies
The number of non-experimental studies far exceeded those of experimental and quasi-experimental studies. There were also far more studies of pre-service teachers than of in-service teachers. Experimental studies provide causal evidence of teacher improvement and sometimes of concurrent student achievement, and non-experimental studies use a variety of approaches and methodologies, providing multiple perspectives and rich contextual descriptions of teacher learning. Correlational data suggest that certain aspects of teacher quality characteristics such as certification status and degree in the field to be taught are positively correlated with student results. However, this does not tell us if these characteristics ultimately lead to better student achievement.
Findings of Experimental Research
Some pre-service experimental studies revealed improvements in the knowledge of prospective teachers, but it is unknown whether their new learning impacts classroom practice and student learning. A longitudinal study would have to follow pre-service teachers into their first year of teaching and beyond. Given the differences between sites where teachers from the same programs teach, the power of such a study would be relatively slight, so few of these studies have been done.
The problems are not as severe for studying in-service education because these sites are identifiable and accessible. However, only some in-service experimental studies reported both teacher and student outcomes. The majority of studies that measured either teacher or student results showed significant or modest improvements in either teacher knowledge or student achievement. Those that measured both provide clear evidence that in-service teachers do learn from professional development programs focusing on specific types of reading instruction and that students of those teachers benefited from improved teaching.
Findings of Non-experimental Research
Non-experimental designs predominate in pre-service studies because of researchers 'interest in relating teachers' learning processes, both individually and collectively, to prescribed coursework, field experience, or combinations of these. In general, these non-experimental studies affirm the importance of providing field-based experiences in conjunction with coursework in order to help teachers connect theory and practice. The majority also report favorably on pre-service teacher change, but it is uncertain if this change leads to application, though some research suggests that the use of pre-service training becomes increasingly evident in the first two years of teaching. Still uncertain is the effect on student learning. Few of these studies measure or report student outcomes.
In concert with trends in pre-professional preparation of teachers, substantial numbers of non-experimental studies have focused on the variously conceived practice of reflection to examine the process of change in prospective teachers' beliefs and attitudes in relation to a host of instructional issues. Similarly, the importance of technology has stimulated numerous non-experimental studies of the impact of new technologies on literacy teacher education largely ignored by experimental research-multimedia, hypermedia, and computer-mediated communication. Non-experimental studies have also been instrumental in foregrounding the under-researched issue of teaching reading to culturally diverse learners.
Non-experimental studies of in-service professional development, as with experimental studies, focused on more specific instructional methods and issues compared with pre-service studies. Conceptual tools supported with practical strategies prove to be the most influential, and conferencing with mentors and supervisors is also important.
Experimental research provides evidence of teacher change and its effect on student achievement. To guide change more effectively, we must also understand more deeply teachers' attitudes, beliefs, and conceptualizations of literacy and the changes they undergo while studying practices and outcomes; knowing about the beliefs and attitudes of teachers is important because it indexes a source of teacher behaviors. In one study, for example, correlational analyzes indicated that teachers' philosophical acceptance predicted their use of instructional methods. Improving and non-improving teachers were different in their self-efficacy and willingness to experiment. Because non-experimental studies ask questions different from those asked by experimental studies-focusing on the processes of change and reflection-both kinds of studies are needed. The findings of the non-experimental studies of teacher change do not contradict those of the experimental research, but they need to be designed and reported better to facilitate parallel or follow-up studies. Furthermore, more longitudinal studies that track teachers through their initial years of teaching and studies investigating diversity need to be rigorously pursued.
One of the key assumptions held about teacher education and professional development is that, if it is effective, it should produce "better" instruction (changes in teacher behaviors) and "better" reading by students (higher achievement). However, this assumption does not drive much of the research. Only some experimental studies compared groups as well as outcome measures, but both are needed for establishing links between interventions and performance. Improvements in research conceptualization and design would allow such analyzes to be conducted-analyzes that are key for policy work, for example, in establishing the relative costs of raising reading achievement through different professional development programs.
Improvements in methodology and reporting could also lead to a more integrated and holistic understanding of teaching reading. Some of the imbalance between the numbers of experimental and non-experimental studies and between pre-service and in-service studies can be accounted for by costs, by the questions being asked, or, regrettably, by assuming that researchers chose a methodology and then Found a problem to study. This latter tactic may become less attractive because of the current national policy, which has adopted as its exemplary standard the experimental research design. It must be acknowledged that non-experimental methodologies may be preferable for certain problems. Integration of knowledge would be facilitated, at least, by authors making their assumptions explicit when reporting studies and by journal editors requiring explicit statements of the relationship between questions, methodology, and data. Finally, we observe that researchers rarely cite relevant research from paradigms other than their own. But much can be gained by having authors broaden their view to include research from different methodologies. Blending data from research conducted using different methodologies has the potential to enrich the knowledge base.