xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam a22 u 4500
controlfield tag 008 c20009999azu 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E11-00145
Educational policy analysis archives.
n Vol. 8, no. 1 (January 01, 2000).
Tempe, Ariz. :
b Arizona State University ;
Tampa, Fla. :
University of South Florida.
c January 01, 2000
Teacher quality and student achievement : a review of state policy evidence / Linda Darling-Hammond.
Arizona State University.
University of South Florida.
t Education Policy Analysis Archives (EPAA)
xml version 1.0 encoding UTF-8 standalone no
mods:mods xmlns:mods http:www.loc.govmodsv3 xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govmodsv3mods-3-1.xsd
mods:relatedItem type host
mods:identifier issn 1068-2341mods:part
mods:detail volume mods:number 8issue 1series Year mods:caption 20002000Month January1Day 11mods:originInfo mods:dateIssued iso8601 2000-01-01
1 of 44 Volume 8 Number 1January 1, 2000ISSN 1068-2341 A peer-reviewed scholarly electronic journal Editor: Gene V Glass, College of Education Arizona State University Copyright 2000, the EDUCATION POLICY ANALYSIS ARCHIVES. Permission is hereby granted to copy any article if EPAA is credited and copies are not sold. Articles appearing in EPAA are abstracted in the Current Index to Journals in Education by the ERIC Clearinghouse on Assessment and Evaluation and are permanently archived in Resources in Education Teacher Quality and Student Achievement: A Review of State Policy Evidence Linda Darling-Hammond Stanford UniversityAbstract Using data from a 50-state survey of policies, stat e case study analyses, the 1993-94 Schools and Staffing Surveys (SASS), an d the National Assessment of Educational Progress (NAEP), this stu dy examines the ways in which teacher qualifications and other scho ol inputs are related to student achievement across states. The findings of both the qualitative and quantitative analyses suggest that policy inves tments in the quality of teachers may be related to improvements in stude nt performance. Quantitative analyses indicate that measures of tea cher preparation and certification are by far the strongest correlates o f student achievement in reading and mathematics, both before and after cont rolling for student poverty and language status. State policy surveys a nd case study data are used to evaluate policies that influence the overal l level of teacher qualifications within and across states. This analy sis suggests that policies adopted by states regarding teacher educat ion, licensing, hiring, and professional development may make an important difference in the qualifications and capacities that teachers bring t o their work. The
2 of 44implications for state efforts to enhance quality a nd equity in public education are discussed. (Note 1) Introduction For many years, educators and researchers have debated which school variables influence student achievement. As policymakers beco me more involved in school reform, this question takes on new importance since their many initiatives rely on presumed relationships between various education-re lated factors and learning outcomes. Some research has suggested that "schools bring little influence to bear upon a child's achievement that is independent of his ba ckground and general social context" (Coleman et al., 1966, p. 325; see also Jencks et a l., 1972). Other evidence suggests that factors like class size (Glass et al., 1982; Mostel ler, 1995), teacher qualifications (Ferguson, 1991), school size (Haller, 1993), and o ther school variables may play an important role in what students learn. As new standards for student learning hav e been introduced across the states, greater attention has been given to the role that t eacher quality plays in student achievement (National Commission on Teaching and Am erica's Future, 1996; National Education Goals Panel, 1998). In the last few years more than 25 states have enacted legislation to improve teacher recruitment, educati on, certification, or professional development (Darling-Hammond, 1997a). While some ev idence suggests that better qualified teachers may make a difference for studen t learning at the classroom, school, and district levels, there has been little inquiry into the effects on achievement that may be associated with large-scale policies and institu tional practices that affect the overall level of teachers' knowledge and skills in a state or region. This paper reports on one such study, which combines state level case studies and quantitative analyses of state-level achievement data to examine whether and how state policies may influence teachers' capabilities and student learning. Using data from a 50-state policy survey conducted by the National Commission on Teaching and America's Future, case studies of s elected states conducted under the auspices of the Center for the Study of Teaching an d Policy, the 1993-94 Schools and Staffing Surveys (SASS), and the National Assessmen t of Educational Progress (NAEP) sponsored by the National Center for Education Stat istics, the study examines the ways in which teacher qualifications and other school in puts, such as class size, are related to student achievement across states, taking student c haracteristics into account. In addition, these data and state case study data are used to evaluate policies that influence the overall level of teacher qualifications within and across states.Previous Research Despite conventional wisdom that school i nputs make little difference in student learning, a growing body of research suggests that schools can make a difference, and a substantial portion of that difference is attributa ble to teachers. Recent studies of teacher effects at the classroom level using the Tennessee Value-Added Assessment System and a similar data base in Dallas, Texas, have found th at differential teacher effectiveness is a strong determinant of differences in student lear ning, far outweighing the effects of differences in class size and heterogenity (Sanders & Rivers, 1996; Wright, Horn, & Sanders, 1997; Jordan, Mendro, & Weerasinghe, 1997) Students who are assigned to several ineffective teachers in a row have signific antly lower achievement and gains in achievement than those who are assigned to several highly effective teachers in sequence (Sanders & Rivers, 1996). Teacher effects appear to be additive and cumulative, and
3 of 44generally not compensatory. These studies also find troubling indicators for educational equity, noting evidence of strong bias in assignmen t of students to teachers of different effectiveness levels (Jordan, Mendro, & Weerasinghe 1997), including indications that African American students are nearly twice as likel y to be assigned to the most ineffective teachers and half as likely to be assig ned to the most effective teachers (Sanders & Rivers, 1996). These studies did not, ho wever, examine the characteristics or practices of more and less effective teachers. These issues have been the topic of much other research over the last 50 years. Variables presumed to be indicative of teachers' co mpetence which have been examined for their relationship to student learning include measures of academic ability, years of education, years of teaching experience, measures o f subject matter and teaching knowledge, certification status, and teaching behav iors in the classroom. The results of these studies have been mixed; however, some trends have emerged in recent years. General Academic Ability and Intelligence While studies as long ago as the 1940s have found positive correlations between teac hing performance and measures of teachers' intelligence (usually measured by IQ) or general academic ability (Hellfritsch, 1945; LaDuke, 1945; Rostker, 1945; Skinner, 1947), most relationships are small and statistically insignificant. Two reviews of such st udies concluded that there is little or no relationship between teachers' measured intelligenc e and their students' achievement (Schalock, 1979; Soar, Medley, & Coker, 1983). Expl anations for the lack of strong relationship between measures of IQ and teacher eff ectiveness have included the lack of variability among teachers in this measure and its tenuous relationship to actual performance (Vernon, 1965; Murnane, 1985). However, other studies have suggested that teachers' verbal ability is related to student achievement (e.g., Bowles & Levin, 1968; Coleman et al., 1966; Hanushek, 1971), and th at this relationship may be differentially strong for teachers of different typ es of students (Summers & Wolfe, 1975). Verbal ability, it is hypothesized, may be a more sensitive measure of teachers' abilities to convey ideas in clear and convincing w ays (Murnane, 1985). Subject Matter Knowledge Subject matter knowledge is another variable that one might think could be related to teacher effecti veness. While there is some support for this assumption, the findings are not as strong and consistent as one might suppose. Studies of teachers' scores on the subject matter t ests of the National Teacher Examinations (NTE) have found no consistent relatio nship between this measure of subject matter knowledge and teacher performance as measured by student outcomes or supervisory ratings. Most studies show small, stati stically insignificant relationships, both positive and negative (Andrews, Blackmon & Mac key, 1980; Ayers & Qualls, 1979; Haney, Madaus, & Kreitzer, 1986; Quirk, Witte n, & Weinberg, 1973; Summers & Wolfe, 1975). Byrne (1983) summarized the results of th irty studies relating teachers' subject matter knowledge to student achievement. The teache r knowledge measures were either a subject knowledge test (standardized or researche r-constructed) or number of college courses taken within the subject area. The results of these studies were mixed, with 17 showing a positive relationship and 14 showing no r elationship. However, many of the "no relationship" studies, Byrne noted, had so litt le variability in the teacher knowledge measure that insignificant findings were almost ine vitable. Ashton and Crocker (1987) found only 5 of 14 studies they reviewed exhibited a positive relationship between measures of subject matter knowledge and teacher pe rformance. It may be that these results are mixed be cause subject matter knowledge is a positive influence up to some level of basic compet ence in the subject but is less important thereafter. For example, a controlled stu dy of middle school mathematics
4 of 44teachers, matched by years of experience and school setting, found that students of fully certified mathematics teachers experienced signific antly larger gains in achievement than those taught by teachers not certified in math ematics. The differences in student gains were greater for algebra classes than general mathematics (Hawk, Coble, & Swanson, 1985). However, Begle and Geeslin (1972) f ound in a review of mathematics teaching that the absolute number of course credits in mathematics was not linearly related to teacher performance. It makes sense that knowledge of the mate rial to be taught is essential to good teaching, but also that returns to subject matter e xpertise would grow smaller beyond some minimal essential level which exceeds the dema nds of the curriculum being taught. This interpretation is supported by Monk's (1994) more recent study of mathematics and science achievement. Using data on 2,829 students from the Longitudinal Study of American Youth, Monk (1994) f ound that teachers' content preparation, as measured by coursework in the subje ct field, is positively related to student achievement in mathematics and science but that the relationship is curvilinear, with diminishing returns to student achievement of teachers' subject matter courses above a threshold level (e.g., five courses in math ematics). In a multilevel analysis of the same data set, Monk and King (1994) found both positive and negative, generally insignificant effe cts of teachers' subject matter preparation on student achievement. They did find s ome evidence of cumulative effects of prior as well as proximate teachers' subject mat ter preparation on student performance in mathematics. Effects differed for highand lowachieving students and for different grade levels. In a review of 65 studies of science teachers' characteristics and behaviors, Druva and Anderson (1983) found students' science a chievement was positively related to the teachers' course taking background in both e ducation and in science. The relationship between teachers' training in science and student achievement was greater in higher level science courses, a result similar to t hat found by Hawk, Coble, and Swanson (1985) in mathematics. It may also be that the measure of subjec t matter knowledge makes a difference in the findings. Measures of course-taking in a subjec t area have more frequently been found to be related to teacher performance than hav e scores on tests of subject matter knowledge. This might be because tests necessarily capture a narrower slice of any domain. Furthermore, in the United States, most tea cher tests have used multiple-choice measures that are not very useful for assessing tea chers' ability to analyze and apply knowledge. More authentic measures may capture more of the influence of subject matter knowledge on student learning. For example, a test of French language teachers' speaking skill was found to have significant correl ation to students' achievement in speaking and listening (Carroll, 1975). Despite concerns that education majors ma y be less well prepared in their subject areas than are academic majors (Galambos, 1985), co mparisons of teachers with degrees in education vs. those with degrees in disciplinary fields have found no relationship between degree type and teacher performance (Murnan e, 1985). This may be because certification requirements reduce the variability i n course backgrounds found for teachers with different degree types. For example, many states require the equivalent of an academic major or minor in the field to be taugh t as part of the education degree for high school teachers, regardless of the department granting the degree (NASDTEC, 1997). Given the standardizing influences of licens ing requirements within states but substantial differences in licensing requirements a cross states, within-state studies are likely to find less variation in teachers' educatio n backgrounds than might be found in cross-state studies.
5 of 44 Knowledge of Teaching and Learning Studies have found a somewhat stronger and more consistently positive influence of educati on coursework on teachers' effectiveness. Ashton and Crocker (1987) found sign ificant positive relationships between education coursework and teacher performanc e in four of seven studies they reviewedÂ—a larger share than those showing subject matter relationships. Evertson, Hawley, and Zlotnik (1985) reported a consistent po sitive effect of teachers' formal education training on supervisory ratings and stude nt learning, with 11 of 13 studies showing greater effectiveness for fully prepared an d certified vs. uncertified or provisionally certified teachers. With respect to s ubject matter coursework, 5 of 8 studies they reviewed found no relationship and the other three found small associations. Reviewing findings of the National Longit udinal Study of Mathematical Abilities, Begle (1979) found that the number of credits a tea cher had in mathematics methods courses was a stronger correlate of student perform ance than was the number of credits in mathematics courses or other indicators of prepa ration. Similarly, Monk's (1994) study of student's mathematics and science achievem ent found that teacher education coursework had a positive effect on student learnin g and was sometimes more influential than additional subject matter preparation. In an a nalysis of science teaching, Perkes (196768) found that teachers' coursework credits in science were not significantly related to student learning, but coursework in scie nce education was significantly related to students' achievement on tasks requiring problem solving and applications of science knowledge. Teachers with greater training in scienc e teaching were more likely to use laboratory techniques and discussions and to emphas ize conceptual applications of ideas, while those with less education training placed mor e emphasis on memorization. In a study of more than 200 graduates of a single teacher education program, Ferguson and Womack (1993) examined the influences on 13 dimensions of teaching performance of education and subject matter coursew ork, NTE subject matter test scores, and GPA in the student's major. They found that the amount of education coursework completed by teachers explained more tha n four times the variance in teacher performance (16.5 percent) than did measure s of content knowledge (NTE scores and GPA in the major), which explained less than 4 percent. In a similar study which compared relative influences of different kin ds of knowledge on 12 dimensions of teacher performance for more than 270 teachers, Guy ton and Farokhi (1987) found consistent strong, positive relationships between t eacher education coursework performance and teacher performance in the classroo m as measured through a standardized observation instrument, while relation ships between classroom performance and subject matter test scores were pos itive but insignificant and relationships between classroom performance and bas ic skill scores were almost nonexistent. Another program-based study by Denton and Lacina (1984) found positive relationships between the extent of teachers' profe ssional education coursework and their teaching performance, including their students' ach ievement. It may be that the positive effects of su bject matter knowledge are augmented or offset by knowledge of how to teach the subject to various kinds of students. That is, the degree of pedagogical skill may interact with subje ct matter knowledge to bolster or reduce teacher performance. As Byrne (1983) suggest ed: It is surely plausible to suggest that insofar as a teacher's knowledge provides the basis for his or her effectiveness, th e most relevant knowledge will be that which concerns the particular topic be ing taught and the relevant pedagogical strategies for teaching it to the particular types of pupils to whom it will be taught. If the teacher is to teach fractions, then it is
6 of 44knowledge of fractions and perhaps of closely assoc iated topics which is of major importance.... Similarly, knowledge of teachi ng strategies relevant to teaching fractions will be important. (p. 14) The kind and quality of in-service profes sional development as well as pre-service education may make a difference in developing this knowledge. Several recent studies have found that higher levels of student achievemen t are associated with mathematics teachers' opportunities to participate in sustained professional development grounded in content-specific pedagogy linked to the new curricu lum they are learning to teach (Cohen & Hill, 1997; Wiley & Yoon, 1995; Brown, Smi th, & Stein, 1995). In these studies, both the kind and extent of professional d evelopment mattered for teaching practice and for student achievement. The National Assessment of Educational Pr ogress has also documented how specific kinds of teacher learning opportunities co rrelate with their students' reading achievement. On average, in the 1992 and 1994 asses sments, 4th grade students of teachers who were fully certified, who had master's degrees, and who had had professional coursework in literature-based instruc tion did better than other students on reading assessments (NCES, 1994; NCES, n.d.). While these relationships were modest, the relationships between specific teaching practic es and student achievement were often quite pronounced, and these practices were in turn related to teacher learning opportunities. NAEP analyses found that teachers wh o had had more professional training were more likely to use teaching practices that are associated with higher reading achievement on the NAEP tests--use of trade books and literature, integration of reading and writing, and frequent visits to the lib rary--and were less likely to engage in extensive of use of reading kits, basal readers, wo rkbooks, and multiple choice tests for assessing reading, practices that the NAEP analyses found to be associated with lower levels of student achievement. Interestingly, stude nts of teachers who had had more training in phonics instruction did noticeably less well than other students in both years. Often, this kind of training, narrowly cast, is foc used heavily on the use of basal readers and workbooks rather than an integrated approach th at teaches decoding skills in the context of other important reading skills and langu age development strategies. Other studies have found that students ac hieve at higher levels and are less likely to drop out when they are taught by teachers with c ertification in their teaching field, by those with master's degrees, and by those enrolled in graduate studies (Council for School Performance, 1997; Knoblock, 1986; Sanders, Skonie-Hardin, & Phelps, 1994). However, like the NAEP analyses described above, th ese are simple correlational analyses that do not take into account other school resources or student characteristics like poverty or language background that may affect student performance. Continuity of teachers' learning may also matter. In earlier work, Hanushek (1971) demonstrated that the recency of voluntary educational experience was related to teacher performance. Penick and Yager (1983) found that tea chers in exemplary science programs had higher levels of education and more re cent educational experiences than others, even though they were older than the averag e science teacher. As Murnane (1985) suggests, these findings may indicate that i t is not only the knowledge acquired with ongoing professional development (which may re present more recent advances in the knowledge base) but also the teacher's enthusia sm for learning that relates to increased student achievement. Teaching Experience Other studies of the effects of teacher experience on student learning have found a relationship between teachers' effectiveness and their years
7 of 44of experience (Murnane & Phillips, 1981; Klitgaard & Hall, 1974), but not always a significant one or an entirely linear one. While ma ny studies have established that inexperienced teachers (those with less than three years of experience) are typically less effective than more senior teachers, the benefits o f experience appear to level off after about five years, especially in non-collegial work settings (Rosenholtz, 1986). A possible cause of this curvilinear trend in experie nce effects is that older teachers do not always continue to grow and learn and may grow tire d in their jobs. Furthermore, the benefits of experience may interact with educationa l opportunities. Veteran teachers in settings that emphasize continual learning and coll aboration continue to improve their performance (Rosenholtz, 1984). Similarly, very wel l-prepared beginning teachers can be highly effective. For example, some recent studi es of 5-year teacher education programs--programs that include a bachelor's degree in the discipline and master's in education as well as a year-long student teaching p lacement--have found graduates to be more confident than graduates of 4-year programs an d as effective as more senior teachers (Andrew & Schwab, 1995; Denton & Peters, 1 988). It is also possible that uneven effects o f experience in cross-sectional studies can be the result of cohort effects (for example, cohor ts of teachers hired in times of shortage may be less well-qualified than those hired when sc hools can be more selective) or of attrition effects (for example, disproportionate ea rly attrition of more able teachers may leave a less capable senior force on average) (Murn ane & Phillips, 198; Vance & Schlechty, 1982). Presumably, the direction of this effect would change if retention policies kept the most able beginning teachers in t he profession. Since experience is also correlated with teacher education and certification status, these variables may be confounded in some analyses. Certification Status Certification or licensing status is a measure of teacher qualifications that combines aspects of knowledge a bout subject matter and about teaching and learning. Its meaning varies across th e states because of differences in licensing requirements, but a standard certificate generally means that a teacher has been prepared in a state-approved teacher education prog ram at the undergraduate or graduate level and has completed either a major or a minor i n the field(s) to be taught plus anywhere from 18 to 40 education credits, depending on the state and the certificate area, including between 8 and 18 weeks of student t eaching. (The norm is about 30 education credits and about 12 to 15 weeks of stude nt teaching.) There are only a few states that have requirements outside these paramet ers; however, individual teacher education programs often require more preparation t han the state demands in education, in clinical practice, and in the content area(s) to be taught. Most states now also require one or more tests of basic skills, subject matter k nowledge, and/or teaching knowledge or skills as the basis for the initial or continuin g license or for admission to teacher education. While most states have been increasing th eir standards since the 1980s, more than 30 states still allow the hiring of teachers who ha ve not met their licensing standards, a practice that has been on the increase in some stat es as demand has grown in recent years. Some allow the hiring of teachers with no li cense. Others issue emergency, temporary, or provisional licenses to candidates wh o, depending on the state, may or may not have met varying requirements (e.g., a bach elors degree, a certificate in another teaching field, a basic skills test). More than 40 states have also initiated alternate route provisions for candidates who enter through postbac calaureate programs. Most of these are master's degree programs which offer an educati on degree that meets all of the normal state requirements but does so in a fashion tailored to individuals, like mid-career entrants, who already have a bachelor's degree. Some states allow candidates
8 of 44to complete a short summer course of study and assu me full teaching responsibilities, with or without completing additional coursework. In times of relatively low demand, like m ost of the 1980s, virtually all teachers were certified and there was too little variability to find effects of this variable in large-scale studies. Most studies of the influence of training and certification on teacher performance are from the high-demand era of the 196 0s and 1970s and from the 1990s when demand increased again. Studies in different s ubject matter fields that compare teachers with and without preparation have typicall y found higher ratings and greater student learning gains for teachers who have more f ormal preparation for teaching. In addition to the studies of science and mathematics teachers cited earlier, these include reading and elementary education (Hice, 1970; LuPon e, 1961; McNeil, 1974), early childhood education (Roupp et al., 1979), gifted ed ucation (Hansen, 1988), and vocational education (Erekson and Barr, 1985). In a review of research, Evertson, Hawley, and Zlotnik (1985) concluded: (T)he available research suggests that among studen ts who become teachers, those enrolled in formal preservice preparation pro grams are more likely to be effective than those who do not have such traini ng. Moreover, almost all well planned and executed efforts within teacher pr eparation programs to teach students specific knowledge or skills seem to succeed, at least in the short run (p.8). Other studies point out the differences i n the perceptions and practices of teachers with differing amounts and kinds of preparation. A number of studies suggest that the typical problems of beginning teachers are lessened for those who have had adequate preparation prior to entry (Adams, Hutchinson, & Ma rtray, 1980; Glassberg, 1980; Taylor & Dale, 1971). Studies of teachers admitted with less than full preparation--with no teacher preparation or through very short altern ate routes--have found that such recruits tend to be less satisfied with their train ing (Darling-Hammond, Hudson, & Kirby, 1987; Jelmberg, 1995), and they tend to have greater difficulties planning curriculum, teaching, managing the classroom, and d iagnosing students' learning needs (Bents & Bents, 1990; Darling-Hammond, 1992; Lenk, 1989; Feiman-Nemser & Parker, 1990; Gomez & Grobe, 1990; Grady, Collins, & Grady, 1991; Grossman, 1989; Mitchell, 1987; National Center for Research on Tea cher Learning, 1992; Rottenberg & Berliner, 1990). Principals, supervisors, and colle agues tend to rate them less highly on their instructional skills (Bents & Bents, 1990; Je lmberg, 1995; Lenk, 1989; Feiman-Nemser & Parker, 1990; Gomez & Grobe, 1990; Mitchell, 1987; Texas Education Agency, 1993), and they tend to leave tea ching at higher-thanaverage rates (Darling-Hammond, 1992; Lutz & Hutton, 1989; Stodda rt, 1992). These findings are reflected in Gomez and Grobe's (1990) study of the performance of alternate route candidates in Dallas who receive a few weeks of summer training from the district before they assume full teaching responsibilities. Although these candidates were rated near the average on som e aspects of teaching, they were rated lower on such factors as their knowledge of i nstructional techniques and instructional models. The performance of alternate route candidates was also much more uneven than that of trained teachers, with a much g reater proportion of them--from 2 to 16 times as many--rated "poor" on each of the teach ing factors evaluated. The strongest effects of this unevenness were seen in students' a chievement in language arts, where the achievement gains of students of alternate route te achers, adjusted for initial student scores, were significantly lower than those of stud ents of traditionally trained teachers.
9 of 44 Two studies of alternative certification in Texas have reportedly failed to find such gaps in the performance of students of alterna tive and traditionally licensed teachers (cited in Goldhaber & Brewer, 1999). A stu dy of Houston's alternative certification program by Goebel, Romacher, and Sanc hez (1989) reported no evidence of differential student outcomes and little evidence o f teacher effects. However, this study did not control for students' initial test scores a nd did not match comparison teachers by years of experience. First year traditionally train ed teachers were compared to two groups of alternative certification recruits, one w ith 1-4 years of experience and the other with 5-7 years of experience. Thus, this study did not include adequate controls to allow measurement of effects. Another study by Barnes, Sa lmon, and Wale (1989) reported second-hand that two districts reported equivalent outcomes for alternative and traditional program teachers but did not present an y empirical data or discussion of methodology. The study's table listing program type s evaluated included 1 to 2-year university-based master's programs (which are calle d "alternative" in Texas because they are not undergraduate models) as well as district a lternative programs that generally offer only a few weeks of summer training. In this case, the "alternative" group included programs providing extensive graduate level trainin g along with those with very little preparation, thus preventing assessment of the effe cts of preparation on teacher effectiveness. With non-comparable groups and no co ntrols, it is impossible to draw inferences from either of these studies. Some recent multivariate studies of stude nt achievement at the school and district level have found a substantial influence of teacher s' qualifications on what students learn, especially when scores on licensing examinat ions are included. In an analysis of nearly 900 Texas school districts that evaluated th e effects of many school input variables and controlled for student background and district characteristics, Ronald Ferguson (1991) found that combined measures of tea chers' expertise--scores on a licensing examination, master's degrees, and experi ence--accounted for more of the inter-district variation in students' reading and m athematics achievement (and achievement gains) in grades 1 through 11 than stud ent socioeconomic status. An additional, smaller contribution to student achieve ment was made by lower pupil-teacher ratios and smaller schools in the elementary grades Of the teacher qualifications variables, the strongest relationship was found for scores on the state licensing examination, a test t hat measures both basic skills and teaching knowledge. The effects were so strong, and the variations in teacher expertise so great, that after controlling for socioeconomic status, the large disparities in achievement between black and white students were a lmost entirely accounted for by differences in the qualifications of their teachers Ferguson also found that every additional dollar spent on more highly qualified te achers netted greater increases in student achievement than did less instructionally f ocused uses of school resources. Another study (Strauss & Sawyer, 1986) fo und that North Carolina's teachers' average scores on the National Teacher Examinations (a licensing test which measures subject matter and teaching knowledge) had a strong influence on average school district test performance. Taking into account per-capita in come, student race, district capital assets, student plans to attend college, and pupil/ teacher ratios, teachers' test scores had a strikingly large effect on students' failure rates on the state competency examinations: a 1% increase in teacher quality (as measured by NTE scores) was associated with a 3 to 5% decline in the percentage of students failing th e exam. The authors' conclusion is similar to Ferguson's:
10 of 44Differences in State Policies Regarding Teaching Despite logical presumptions and research evidence that student learning depends substantially on what teachers know and can do, sta tes differ greatly in the extent to which they invest in teachers' learning as a key po licy lever. At the front end of the career, there is wide variation in the standards to which entering teachers and teacher education institutions are held. Licensing standard s are noticeably different from state to state, as are state commitments to enforcing these standards. Later access to professional development is also widely disparate. In high-standards states like Wisconsin o r Minnesota, for example, a prospective high school teacher must complete a bachelor's degr ee that includes a full major in the subject area to be taught plus coursework covering learning theory, child and adolescent development, subject matter teaching methods, curri culum, effective teaching strategies, uses of technology, classroom management, behavior and motivation, human relations, and the education of students with special needs. I n the course of this work, the teacher must complete at least 18 weeks of student teaching in Wisconsin (at least a college semester in Minnesota) under the supervision of a c ooperating teacher who meets minimum standards. In Minnesota, this experience mu st include work in a multicultural setting and with special needs students. If teacher s are asked to teach outside the field of their major for part of the day, they must already be licensed with at least a minor in that field, and can receive a temporary license in the n ew field only briefly while completing a major. By contrast, in Louisiana, prospective hig h school teachers can be licensed without even a minor in the field they will be teac hing. The state does not require them to have studied curriculum, teaching strategies, cl assroom management, uses of technology, or the needs of special education stude nts, and they can receive a license with only six weeks of student teaching (NASDTEC, 1 997; Darling-Hammond, 1997a). In addition to differences in the standar ds themselves, there are great differences in the extent to which they are enforced. Whereas s ome states do not allow districts to hire unqualified teachers, others routinely allow t he hiring of candidates who have not met their standards, even when qualified teachers a re available. In Wisconsin and eleven other states, for example, no new elementary or sec ondary teachers were hired without a license in their field in 1994. By contrast, in Lou isiana, 31% of new entrants were unlicensed and another 15% were hired on substandar d licenses. At least six other states allowed 20% or more of new public school teachers t o be hired without a license in their field (Darling-Hammond, 1997a, Appendix A). Studies of teacher hiring show that even when there are an adequate number of qualified teac hers in the labor market--which was the case nationally and in most states from the ear ly 1980s through the mid-1990s--some districts hire unlicensed teachers because of cumbe rsome and poorly managed hiring procedures that discourage qualified entrants, pere nnially late hiring (e.g. waiting until late August or September to hire), patronage hiring preferences for hiring lower salaried staff, and inequalities in salary schedules caused by state funding formulas and by local decisions to use budgets for purposes other than te acher salaries (see e.g. Haberman, 1995; Johanson and Gips, 1992; Pflaum and Abramson, 1990; National Commission on Teaching and America's Future; Wise, Darling-Hammon d, and Berry, 1987). More than 30 states allow teachers to be hired on temporary or emergency licenses without having completed preparation or having met other licensing requirements. During the late 1980s and early 1990s, at least 50, 000 emergency or substandard licenses were issued annually by states (NCTAF, 1996). Natio nally, in 1994, 27% of those who were new entrants into public school teaching held no license or a substandard license in their main teaching field (Darling-Hammond, 1997a). Even the rigor of these restricted
11 of 44licenses varies. States such as Minnesota will issu e a restricted license only to a teacher who has already been fully prepared in a teaching f ield but who needs to complete additional coursework in order to enter from out-of -state or switch to a new field or teaching level. Such a license is only good for one year, while the necessary coursework is completed. Others, including Louisiana, will iss ue an emergency license to a person who does not even hold a bachelor's degree and will allow it to be renewed for several years while the candidate makes little progress tow ard becoming licensed. It is certainly true that differences in student enrollment growth, coupled with teacher production rates and attrition, construct d ifferent levels of teacher demand that can affect the ease or difficulty of hiring within states. While incentives to enter and stay in teaching are affected by policies governing sala ries, working conditions, and teacher education funding, student enrollments are less ame nable to policy control. It is reasonable to ask whether these differences in oper ational teaching standards are mostly a function of demographic trends beyond the control of state policymakers. In examining state variations in hiring practices, however, it i s clear that a number of high-growth states have enacted and maintained high standards f or entry to teaching while many low-growth states have not. Policies appear to be a t least as important as demographics in determining the qualifications of teachers hired and retained. Because of these differences in licensing standards and enforcement, in 1994, more than 80% of high school teachers of academic c ourses in Wisconsin and Minnesota had fully met state certification requirements and had at least a college major in the field they teach. Four other states--Connecticut, Iowa, M ontana, and North Dakota--reported similarly well-qualified teaching forces in that ye ar. The comparable proportion of teachers with full state certification and a major in their field in Louisiana was only 64%. (An additional six states had fewer than two-thirds of their teachers similarly prepared.) Interestingly, students in Minnesota and Wisconsin have typically scored at the top of the distribution on national assessments of reading and mathematics, along with the four other states who share similarly well-qual ified teachers. Together these states held six of the top ten spots in the national ranki ngs in reading and mathematics in 1994 and 1996. Students in Louisiana have typically scor ed near the bottom of the NAEP distributions--no higher than 47th of 51 states in any of the assessments reported by 1996. The other six states with similar proportions of teachers holding a license and a major in their field all fall in the bottom quartil e of states in the national rankings of average student achievement scores (Campbell et al. 1996; Darling-Hammond, 1997a, pp. 13, 26; Reese et al., 1997). Some have quipped that state-level student achievement in the U.S. can be best predicted by proximity to C anada--which in turn may be a proxy for variations among states in factors ranging from demographics (e.g., student poverty, parent education, and race) to political culture an d spending on education. The distributions of scores described above could indee d partly support the "Canada hypothesis," which I test below. States also differ greatly in the levels of funding they allocate to preservice and inservice teacher education, in the standards they apply to teacher education institutions and to schools, in the types and extent of professi onal learning opportunities and the incentives for professional study they make availab le to educators, and the extent to which they require or fund induction supports for b eginning teachers. To illustrate these differences, in 1997 only three states required pro fessional accreditation for schools of education and only nine funded induction programs t hat provided a structured program of mentoring for beginning teachers, including trai ned, state-funded mentors. Student teaching requirements ranged from 5 weeks in Massac husetts to 18 weeks in Wisconsin. As of 1994, the proportions of academic high school teachers teaching with both a
12 of 44license and a major in their field ranged from a lo w of 52% to a high of 85% across states. The proportions of mathematics teachers tea ching with less than a minor in the field ranged from a low of 9% to a high of 56% (Dar ling-Hammond, 1997a, Appendices A and B). This means that a student in one state mi ght have only one chance in ten of being taught by an out-of-field teacher, while a st udent in another state might have more than a 50% chance of being taught a subject by a te acher who is not adequately prepared in that subject. In every category of possible investment in teachers' knowledge and in every area in which standards for teaching are set (e.g., lice nsing, accreditation, advanced certification, onthe-job evaluation), there are s ubstantial differences in the policies and practices employed by states. States with some of t he highest, most consistently enforced standards for teachers have tended to cluster in th e upper Midwest (Minnesota, Wisconsin, Iowa, Nebraska, North Dakota, Missouri, Montana, Kansas). States with the lowest and least well-enforced standards have tende d to include many in the southeast (Louisiana, Mississippi, Georgia, South Carolina) a nd in remote locations (Alaska, Hawaii). Some states have developed relatively ambi tious standards for teaching but do not enforce them for large numbers of candidates (C alifornia, New York). Others have made major investments in preservice and inservice teacher development in recent years that have affected a substantial share of the teach ing force (e.g., Connecticut, Kentucky, North Carolina, West Virginia). The possible outcom es of these cross-state differences are discussed below.Trends in Student Achievement: Policy Hypotheses In their book, The Manufactured Crisis Berliner and Biddle (1995) noted that while U.S. secondary school students tend to score below the median in international assessments of mathematics and science, students in some states score as high as those in the top-ranked countries in the world while stud ents in others score among the bottom-ranked. U.S. students also perform relativel y better in some fields than others. For example, U.S. students have compared favorably with students in other countries in reading and at about the median in general science. However, in mathematics and physical science, U.S. students do much more poorly : In the most recent international assessments, 8th graders ranked 18th out of 25 coun tries that met the TIMSS guidelines in mathematics and 17th out of 25 countries in phys ics. Twelfth graders did even more poorly (Darling-Hammond, 1997a, pp. 28-29). Although it may be purely coincidental, t hese differences in rankings are similar to the differences in teacher qualifications across th ese fields. Since the early 1980s, the U.S. has made major investments in teacher preparat ion in the area of reading. Not only are almost all elementary school teachers fully cer tified (more than 95%), an increasing number have been prepared in programs that have a s trong emphasis on training to teach reading; there has also been a large increase in th e number of reading specialists throughout the 1980s. In general science and biolog y, where U.S. middle and high school students scored at about the median on the m ost recent international assessments, there are relatively few uncertified or out-of-fiel d secondary teachers (about 18% of the total). By contrast, in mathematics and physical sc ience, where U.S. students fall well below the international norms, teacher qualificatio ns are much weaker. In addition to the fact that most U.S. elementary teachers have had li ttle background in mathematics, about 30% of U.S. mathematics teachers and 50% of physica l science teachers at the high school level have been teaching with less than a mi nor in the field, many of them uncertified (Darling-Hammond, 1997, p.28 and Append ix Table 3). While these are only
13 of 44casual observations, other evidence point in simila r directions. Long-term Achievement Trends by State Not only do U.S. students appear to perfo rm least well in the fields in which U.S. teachers are least well prepared, the states that r epeatedly lead the nation in student achievement in mathematics and reading have among t he most highly qualified teachers in the country and have made longstanding investmen ts in the quality of teaching (see Figures 1-3). The three long-time leaders--Minnesot a, North Dakota, and Iowa--have all had a long history of professional teacher policy a nd are among the 12 states that have state professional standards boards which have enac ted high standards for persons entering the teaching profession. They are recently joined at the top of the achievement distribution by Wisconsin, Maine, and Montana, stat es that have also enacted rigorous standards for teaching and that are among the few w hich rarely hire unqualified teachers on substandard licenses. Iowa, Minnesota, Montana, North Dakota, and Wisconsin have among the lowest rates of out-of-field teaching in the country and among the highest proportions of teachers holding both certification and a major in the field they teach. (Note 2) Maine joined these states in requiring cer tification plus a disciplinary major when it revised its licensing standards in 1988. These states have also been leaders in re defining teacher education and licensing. Minnesota was the first state to develop performanc e-based standards for licensing teachers and approving schools of education during the mid-1980s and has developed a beginning teacher mentoring program in the years si nce (for details, see Darling-Hammond, Wise, & Klein, 1995). Wisconsin wa s one of the first states to require high school teachers to earn a major in the ir subject area in addition to completing extensive coursework in a teacher prepar ation program. Thus, teacher education in Wisconsin is typically a four-and-ah alf to five year process. Maine, Wisconsin, Iowa, and Minnesota have all incorporate d the rigorous new standards developed by the Interstate New Teacher Assessment and Support Consortium (INTASC) (Note 3) into their licensing standards an d have encouraged universities to pilot performance-based assessments of teaching usi ng these standards.
14 of 44 Figure 1. State Trends in Mathematics Achievement, Grade 4 (NAEP scores, 1992-1996)Source: National Center for Education Statistics, NAEP 1996 Mathematics Report Card for the Nation and the States, Table 2.2, p. 28.
15 of 44 One can still wonder whether policies are the source of these states' strong student outcomes or whether the "Canada effect" (general ed ucation spending combined with low rates of student poverty) is responsible. Among these six states, four spent below the per pupil national average in current expenditures in 1995, and the other two spent just above the average. All, however, spent a larger per centage of their expenditures on instruction than the national average. While these states did have a lower proportion of low-income students than the national average, none fell near the tail of the distribution. There were at least twelve states with lower propor tions of low-income students who scored less well on the NAEP than any of these stat es. However, the relative contribution of student population characteristics and school inputs is an important one to pursue further. That question is raised again be low. State Achievement Gains Another important question is whether inv estments in teaching could raise achievement in states that do not have a long histo ry of this sort. Over the last decade of reform, a few states undertook major initiatives ai med at improving the quality of teaching. From a survey of state policies, we ident ified five states that enacted unusually comprehensive reforms of teaching during the late 1 980s and 1990s: Connecticut and North Carolina enacted the most ambitious teacher l egislation of any states nationally, followed by Arkansas, Kentucky, and West Virginia, which also initiated multi-faceted reforms of teacher preparation, licensing, professi onal development, and compensation, accompanied by substantial investments in teacher l earning. Of the 50 states, North Carolina and Conn ecticut undertook the most substantial and systemic investments in teaching during the mid -1980s. Both of these states, which share relatively large high-poverty student populat ions, coupled major statewide increases in teacher salaries and improvements in t eacher salary equity with intensive recruitment efforts and initiatives to improve pres ervice teacher education, licensing, beginning teacher mentoring, and ongoing profession al development. Since then, North Carolina has posted the largest student achievement gains in mathematics and reading of any state in the nation, now scoring well above the national average in 4th grade reading and mathematics, although it entered the 1990s near the bottom of the state rankings. Connecticut has also posted significant gains, beco ming one of the top scoring states in the nation in mathematics and reading (ranked first at the 4th grade level in mathematics and reading and in the top five at the 8th grade le vel), despite an increase in the proportion of low-income and limited English profic ient students during that time. North Carolina's reforms, launched with o mnibus legislation in 1983, did many things simultaneously: (a) boosted salaries in the mid-1980s and again in the 1990s, (b) created a career development program that rewarded teachers for greater education and for achieving National Board Certification, (c) lau nched an aggressive fellowship program to recruit hundreds of able high school stu dents into teacher preparation each year by entirely subsidizing their college educatio n, (d) required schools of education to become professionally accredited by the National Co uncil for the Accreditation of Teacher Education (NCATE), (e) increased licensing requirements for teachers and principals, (f) invested in improvements in teacher education curriculum, (g) created professional development academies and a North Caro lina Center for the Advancement of Teaching, (h) developed teacher development netw orks like the National Writing Project and an analogous set of professional develo pment initiatives in mathematics, (i) launched a beginning teacher mentoring program, and (j) introduced the most
16 of 44wide-ranging set of incentives in the nation for te achers to pursue National Board certification. North Carolina now boasts more Board -certified teachers than any other state. The state was recognized in the recent Natio nal Education Goals Panel report (NEGP, 1998) for having made among the greatest gai ns in teacher mentoring of beginning teachers as well as the greatest achievem ent gains for students. These extensive investments in teaching o ccurred alongside sizable investments in early childhood education and general K-12 spending increases which lowered pupil/teacher ratios slightly. In the early 1990s, new curriculum standards were introduced and accompanied by an extensive program of professional development for teachers statewide. In 1993, the state enacted an a ssessment system linked to the curriculum standards and substantially aligned to t he NAEP tests. This assessment program, which was implemented in 1994-95, occurred too late to account for most of the gains in achievement. Its effects would require several years to appear, but it may have had some modest influence on the gains after 1 994. A recent analysis of student achievement gains on the National Assessment of Educational Progress (Grissmer & Flanagan, 1998) at tributed much of the NAEP score increase in North Carolina between 1990 and 1996 to the test-based accountability system. However, the new standards and assessments were not on-line until 1995, and the rewards and sanctions component of the accounta bility system was not enacted until 1997, so it was clearly not a factor in these trend s. Grissmer and Flanagan also note the state's large-scale investments during the 1980s in early childhood education, reduced class sizes, teacher salary increases, teacher educ ation upgrades, and extensive professional development. All of these factors coul d have influenced the achievement gains observed during this time period. North Carolina's 1997 Educational Excelle nce Act furthered efforts to upgrade the quality of teacher preparation and teaching quality pouring hundreds of millions of dollars into a new set of reforms. The Act created a professional standards board for teaching and required that all colleges of educatio n create professional development school partnerships to provide the sites for year-l ong student teaching practicums. It also funded a more intensive beginning teacher mentoring program, further upgraded licensing standards, created pay incentives for tea chers who pursue master's degrees and National Board certification, and authorized funds to raise teacher salaries to the national average. It will be useful to watch future trends i n the state. Connecticut's strategies were similar. Th e state's 1986 Educational Enhancement Act spent over $300 million to boost minimum beginn ing teacher salaries in an equalizing fashion that made it possible for low-we alth districts to compete in the market for qualified teachers. At the same time, th e state raised licensing standards by requiring a major in the discipline to be taught pl us extensive knowledge of teaching and learning as part of preparation; instituted perform ance-based examinations in subject matter and knowledge of teaching as a basis for rec eiving a license; created a state-funded mentoring program which supported trai ned mentors for beginning teachers in their first year on the job; and created a sophi sticated assessment program using state-trained assessors to determine which first-ye ar teachers could continue in teaching. An analysis of the outcomes of this initiative foun d that it eliminated teacher shortages and emergency hiring, even in the cities, and creat ed surpluses of teachers within three years of its passage (Connecticut State Department of Education, 1991). Connecticut also required teachers to ear n a master's degree in education for a continuing license and supported new, contentbase d professional development strategies in universities and school districts. In a National Education Goals Panel (1998) report highlighting Connecticut's strong performanc e and large gains in mathematics,
17 of 44state officials pointed to the salary increases and teacher education investments as central to their progress. These investments includ e an intensive professional development program in mathematics, science, and te chnology which, since 1983, has offered 4-week institutes with follow-up support to elementary, middle, and high school teachers. The state has more recently invested in n ew curriculum frameworks and a statewide assessment system for students using exte nded performance tasks and constructed response items intended to measure high er order thinking and performance skills. Launched in 1995, this system, which is tie d to statewide reporting of scores and substantial new professional development, may suppo rt future gains in student achievement. In addition, the state has further ext ended its performance-based teacher licensing system to incorporate the new INTASC stan dards and to develop portfolio assessments modeled on those of the National Board for Professional Teaching Standards (NBPTS). The new teacher assessments, whi ch are tightly linked to the student standards, require beginning teachers to de monstrate that they can implement content-based teaching standards within their subje ct matter field and can analyze student work and learning. Finally, as part of ongo ing teacher education reforms, the state agency is supporting the creation of professi onal development schools linked to local universities as sites for clinical training o f entering teachers. The Connecticut and North Carolina reform s both featured substantial investments in pre-service and in-service education for teacher s linked to standards that incorporate much of the current knowledge base about teaching a nd learning (those of NBPTS, INTASC, and/or NCATE). While the reforms also inclu ded salary increases, the dollars were linked to improved quality via heightened lice nsing standards. Both states sought to increase not only the quality of preparation for teachers, but also the consistency with which they enforced their standards, sharply reduci ng the hiring of unlicensed and under-prepared staff. Kentucky also realized substantial achiev ement gains during the 1990s, after undertaking perhaps the most extensive systemic edu cation reforms of any state in the 1980s. These included major equalization of school funding along with large increases in teacher salaries and overall spending; changes i n school organization, including multi-age primary grade classrooms; investments in early childhood education; the introduction of standards and curriculum frameworks along with portfolios and performance assessments. Changes in teacher educati on and licensing accompanied these reforms, including the adoption of the INTASC licensing standards developed by a consortium of more than 30 states, the introduction of new licensing tests and teacher education requirements, incentives for colleges of education to meet national professional accreditation standards; and massive i nvestments in professional development. All of these efforts undoubtedly combined to produce the steep gains in achievement experienced in Kentucky. By 1994, data from the Schools and Staffing Surveys showed that Kentucky teachers were much bet ter prepared in terms of their content and teaching coursework background than in 1988 and had experienced more extensive professional development than teachers in any other state (Darling-Hammond, 1997a). A recent survey of Kentucky teachers also f ound that more than 80% of beginners who graduated from Kentucky colleges of e ducation felt well-prepared for virtually all aspects of their jobs (Kentucky Insti tute for Educational Research, 1997), in contrast to reports about teacher education from pr evious studies elsewhere. Although somewhat less ambitious in their reforms, Arkansas and West Virginia also raised teacher salaries and licensing requirements and req uired national accreditation of
18 of 44education schools during the late 1980s or early 19 90s, while investing in more professional development for in-service teachers. T hese states also realized steeper gains in student achievement than the national average. In a recent report, Grissmer and Flanagan (1998) focused on Texas and North Carolina for their large gains in average student a chievement. They attributed Texas' gains primarily to the state's accountability syste m, although they also mention its shifts of resources to more disadvantaged students through school finance equalization, class size reductions, and the creation of full day kinde rgarten. The school funding investments that occurred in the 1980s and were con tinued into the following decade may indeed have made some difference in Texas stude nts' achievement in the 1990s. However, the state's new assessment and accountabil ity system was not initiated until 1994 and not fully implemented until 1995-96, so it could not have accounted for gains between 1990 and 1996. Texas was not included in the above analy sis of state test score gains because it was not one of the states that made large comprehen sive investments in teaching during the 1980s. (Texas did make some noteworthy investme nts in teacher salaries and professional development in the 1990s.) In addition however, there are questions about the stability of scores in Texas and the extent to which the posted gains are real. First, Texas included fewer than 45% of its students with disabilities in the testing pool, a much smaller share than most states (NCES, 1997, Ta ble D3). Excessive exclusions of low-scoring students from the testing pool can caus e gain scores to appear much larger than they would otherwise be. In addition, recent s tudies in Texas have raised concerns that much of the ostensible gain registered by Afri can American and Latino students has been a function of grade retentions and dropouts or pushouts, which have increased substantially in recent years. These practices also make average test scores look higher by eliminating lower scoring students from the test ing pool (Haney, 1999; Kurtz, 1999; Mexican American Legal Defense and Education Fund, 1999). Assuming that some of the gains in Texas are not spurious, however, it is worth noting that, in addition to the equalization of funding and investments in kinderga rten and reduced class sizes, Texas was among the few states recognized by the National Education Goals Panel (1998) for large gains since the early 1990s in the proportion of beginning teachers receiving mentoring from expert veterans. Texas has also had a growing number of 5-year teacher education programs in response to an earlier reform eliminating teacher education majors at the undergraduate level. State reform strategies during the 1980s that did not include substantial efforts to improve the nature and quality of classroom work ha ve shown little success in raising student achievement, especially if the reforms reli ed primarily on student testing rather than investments in teaching. For example, the firs t two states to organize their reforms around new student testing systems were Georgia, wi th its Quality Basic Education Act (QBE) of 1985, and South Carolina, with its Educati on Improvement Act of 1984. These states developed extensive testing systems coupled with rewards and sanctions for students, teachers, and schools. Although both stat es also mandated tests for teachers, they did not link these assessments to emerging kno wledge about teaching or to new learning standards, nor did they invest in improvin g schools of education or ongoing professional development. Few districts in either s tate require teachers to hold a degree in the field to be taught and full state certificat ion as a condition of hiring. As Figures 1-3 show, student achievement in mathematics has be en flat in these states while achievement in reading has declined. Since 1996, Ge orgia has launched an ambitious series of reforms through its P-16 Council to upgra de the quality of teacher preparation and professional development and to raise licensing standards, as well as to recruit high
19 of 44 ability students to teaching. Future analyses might examine whether these moves have made a difference. There are competing hypotheses that could explain these across-state differences in achievement trajectories. One could speculate th at student testing and curriculum changes are not in themselves powerful enough refor ms to overcome the depressing effects on teaching quality of low standards for te acher education, licensing, and hiring, and the resulting large numbers of under-prepared t eachers. On the other hand, one can argue that variables like student poverty and langu age background, rather than conditions that might influence the quality of teac hing, are the determining factors in student achievement and that the critical differenc es between highand low-achieving states are differences in their student populations It is interesting to compare the student achievement levels and trajectories for some of these states in comparison to geographicall y proximate states with similar student populations that have taken very different approaches to teaching policy. While the comparisons in Table 1 are only suggestive, the y demonstrate that student achievement cannot be assumed to be only or primari ly a function of demographics. Although the states that have aggressively pursued investments in teacher knowledge and skills have equal or higher levels of student p overty than nearby states that pursued other, distinctively different reform strategies, t heir students now achieve at higher levels. Even though all of these states increased t eacher salaries during the 1990s, those that insisted on higher standards for teacher educa tion and licensing realized gains that were not realized by states that maintained or lowe red their standards for entering teaching.Table 1 State teacher salaries, student poverty, and studen t achievement NAEP 4th grade mathematics scores, 1996 NAEP Score, 1996 Gain from 1992 % of students in poverty Teacher Salaries Minimum Maximum Connecticut 232 + 5 18.6 $28,195 $56,189 New Jersey 227 + 0 14.6 $28,424 $58,208 North Carolina 224 +11 18.4 $20,077 $38,733 Georgia 215 + 0 18.5 $20,065 $42,134 West Virginia 223 + 8 22.0 $21,466 $36,378 Virginia 223 + 2 12.6 $23,098 $38,328 Data on student achievement and poverty status from NAEP 1996 Mathematics Report Card for the States, Washington, D.C.: U.S. Departm ent of Education, 1997, pp. 28, 139. Data on teachers' salaries from NCES, America's Tea chers: Profile of a Profession, 1993-94, Washington, D.C.: U.S. Department of Educa tion, 1996, Table A6.2.
20 of 44 For example, with their industrialized ur ban areas and affluent suburbs, Connecticut and New Jersey are demographically and economically similar states, although Connecticut has noticeably higher rates of student poverty. Despite a more affluent student population, New Jersey's students did less well than those in Connecticut on the NAEP 4th grade mathematics asses sments in 1996, and, in contrast to Connecticut's students, they have not improved i n recent years. Whereas Connecticut raised teachers' salaries and equalized districts' abilities to pay for qualified teachers, New Jersey decreased its requirements for teacher p reparation and licensing at the end of the 1980s, reducing the amount of education coursew ork for entry into teaching to a maximum of 18 undergraduate credit hours and encour aging the more extensive hiring of alternative certification candidates prepared in a short summer program. These less-prepared teachers are primarily hired in low-w ealth city school districts that have had radically lower revenues and salary schedules t han other parts of the state. While New Jersey's average teachers' sala ries are the highest in the country, even higher than Connecticut's, New Jersey's salary incr eases were not tied to improvements in the qualifications of teachers or to equalizatio n in districts' ability to pay for qualified teachers. New Jersey also lacks the rigorous licens ing examinations, requirements for a major in the field and a masters in education, and state-funded mentoring for beginning teachers that Connecticut enacted in 1986. Compared to Connecticut, New Jersey has much lower rates of beginning teachers receiving me ntoring and induction, much lower proportions of districts insisting on rigorous hiri ng standards, much lower proportions of teachers receiving professional development, much l ower rates of teachers holding full certification plus a major in the field, and much h igher rates of out-of-field teaching in every subject matter field (Appendix B, Tables 1-5, Darling-Hammond, 1997a). In the same fashion, North Carolina's stu dents now perform substantially better on the NAEP assessments than those in demographically similar Georgia, which North Carolina lagged behind in 1990. Although the states raised salaries during the 1980s and early '90s to comparable levels, Georgia did not ra ise standards for teacher preparation and licensing or invest heavily in teacher developm ent at the same time. While North Carolina increased both the education and subject m atter requirements for teacher preparation, introduced rigorous teacher examinatio ns for licensing, and required national accreditation for all of its education sch ools during the 1980s, Georgia did little to increase expectations for either preservice or i nservice preparation during those years. In addition to having had more extensive training t o meet certification standards, North Carolina teachers are much more likely than their p eers in Georgia to have had mentoring as beginning teachers and professional de velopment opportunities as veterans. And very poor West Virginia now ranks as well in elementary mathematics as its neighbor Virginia, whose students are much more aff luent. Virginia, with its higher cost of living, pays its teachers more. However, West Vi rginia's efforts to raise salaries were accompanied by efforts to improve teacher education and licensing standards. All of West Virginia's teacher education programs must now meet national accreditation standards--a much higher set of requirements than t hose in Virginia, which lowered standards for education programs and licensing duri ng the 1980s to among the lowest in the country. Like New Jersey, Virginia reduced the requirements for coursework on teaching and learning in undergraduate programs, wh ile West Virginia raised its standards. West Virginia introduced an ambitious pr ogram of professional development even before it launched its new curriculum framewor ks in the mid-1990s, and enacted a mentoring program for beginning teachers. Despite i ts relative wealth, Virginia hires many more unlicensed new teachers than West Virgini a and its districts are less likely to
21 of 44insist on rigorous hiring standards. These kinds of contrasts can be seen in m any comparisons of geographically proximate, demographically similar states that have taken different approaches to the issue of teacher investments over the last decade. Policies that jointly raise salaries and standards may offer particularly high leverage on t eaching quality. It is interesting to note that, like states that introduced testing with out making investments in teaching, those that have raised salaries alone, without rais ing standards for preparation and licensing or investing in professional development, seem not to have realized the benefits of improved student outcomes. While intere sting, these observations of individual state cases could be idiosyncratic. An i mportant question is whether similar patterns exist when viewed from a national perspect ive.A National View of Teacher Qualifications and Stude nt Achievement To examine further the relative contribut ions of teaching policies and student characteristics to student achievement, this analys is uses data on public school teacher qualifications and other school inputs available fr om the 1993-94 Schools and Staffing Surveys (SASS) and data on student achievement and student characteristics from the 1990, 1992, 1994, and 1996 assessments in reading a nd mathematics administered by the National Assessment of Educational Progress. Th ese data are the basis for regression analyses of school resource variables on student ac hievement scores to examine whether teacher quality indicators, as well as other school inputs, are related to student achievement at the state level, after controlling f or such student characteristics as poverty and language background. The Database The 1993-94 SASS database includes linked surveys of 65,000 teachers (52,000 public and 13,000 private); 13,000 school principals (9,500 public and 3,500 private); and 5,600 school districts. SASS is designed to provide reliable estimates of the characteristics of schools and educators at the national and state levels. It also includes information from individual teachers, scho ol principals, and districts about salaries and compensation policies, induction polic ies, school climate and context variables (e.g., time to work with other teachers, teacher involvement in decision-making), professional development support, teachers' views of teaching, and their plans to remain in the profession. These anal yses use the following data derived from the public school teachers' questionnaire: dat a on teachers' qualifications (teachers' degrees, majors, certification status), teaching as signments, and average class size. Also included in the analysis are data from the public s chool district questionnaire on district hiring policies (whether districts require, as a co ndition of hiring, full certification, graduation from an approved teacher education progr am, or a college major or minor in the field to be taught) and salary schedules (minim um and maximum salaries) as reported by district officials. Salary schedule dat a are more appropriate for gauging attractions to teaching than average salary data, w hich do not control for differential levels of experience and education across states. A ll of the SASS data were aggregated to the state level. Teacher quality variables constructed fro m the SASS data include the proportion of "well-qualified teachers," defined as the propor tion holding state certification and the equivalent of a major (either an undergraduate majo r or masters degree) in the field taught. For elementary teachers, the equivalent of a major is an elementary education degree for generalists who teach multiple subjects to the same group of students or a degree in the field taught for specialists (e.g. re ading, mathematics or mathematics education, special education). The proportion of te achers who are "fully certified"
22 of 44includes teachers with standard or regular certific ation and new teachers on probationary certificates who have completed all requirements fo r a license except for the completion of the probationary period (usually 2 or 3 years of beginning teaching). The proportion of teachers who are "less than fully certified" inc ludes teachers with no certificate and those with provisional, temporary, or emergency cer tification. Additional data on each state, including policies regarding teacher education and licensing (number of weeks of student teaching requ ired, presence of a professional standards board, percentage of teacher education in stitutions that are NCATE accredited), were collected directly from states an d professional associations (see Darling-Hammond, 1997a, Appendix A). State school s pending data (current per pupil expenditures) are from the Common Core of Data (NCE S, 1995). Data from the National Assessment of Educ ational Progress (NAEP) include state average achievement scores for students in mathemat ics at the 4th grade level in 1990 and 1996 and at the 8th grade level in 1992 and 199 6, as well as data on state average achievement scores for students in reading at the 4 th grade level in 1992 and 1994 (Campbell, Donahue, Reese, & Phillips, 1996) and st udent poverty rates (Reese, Miller, Mazzeo, & Dossey, 1997). Limitations There are a number of limitations that pertain to the data set and the analyses. First, the NAEP data derive from tests th at do not measure all of the valued outcomes of schooling held by parents, teachers, an d schools. They cannot represent everything that schools do or should do. In additio n, state scores and changes in average scores on these measures are sensitive to differenc es in the population of students taking the tests, including decisions about which students will be excluded from testing and differences across states in the extent to which po pulations are represented in school (as a function of schoolage population characteristic s, dropout rates and patterns, and other variables). Finally, the level of aggregation necessa rily influences the interpretations of results. Aggregating data to the state level produc es different results than one would find if one looked at similar kinds of data at the indiv idual student, teacher, school, or district level. The direction of the differences cannot be p redicted with certainty (Ferguson and Ladd, 1996). While, on one hand, the possibility of greater variability or noise exists in disaggregated analyses, it is possible that omitted variables may bias the coefficients of school input variables upward when the data are agg regated to the district or state level (Hanushek, Rivkin, and Taylor, 1995). Although the results of more and less aggregated specifications can be consistent (for example, Ferg uson and Ladd's (1996) Alabama analysis found comparable influences of teacher qua lity and class sizes on student achievement when measured at the student and the di strict levels), this may not always occur. In particular, the size of relationships fou nd between variables measured at the state level cannot be assumed to represent the effe ct sizes one would find in a classroom level analysis. For the purposes of assessing broad policy influences at the state level, it is nonetheless reasonable to examine state-level da ta as a gauge of major trends when other confirming and disconfirming evidence is avai lable to supplement the analysis. The Findings All analyses include public schools and teachers o nly. Although the sample includes all states participating in state N AEP and thus is not a representative sample from which one would draw population inferen ces, I report p-values as an aid to readers who wish to use them to interpret the relat ive sizes of relationships and the probabilities of a Type I error. Before constructin g the multivariate analyses, initial bivariate correlations of school resource variables and student demographic variables with state average student test scores were conduct ed to examine the relationships
23 of 44among variables and to select variables for inclusi on in the multivariate equations. These analyses confirmed several findings reported elsewh ere: Student characteristics such as poverty, non-Englis h language status, and minority status are negatively correlated with stud ent outcomes, and usually significantly so These student characteristics are also significan tly and negatively correlated with the qualifications of teachers; tha t is the less socially advantaged the students, the less likely teachers are to hold full certification and a degree in their field and the more likely they are to have en tered teaching without certification. Student characteristics are generally not significa ntly correlated with state per-pupil spending or with teachers' salary schedul es with the exception that salary schedules are higher in states with larger p ercentages of minority and LEP (limited English proficient) students. Salary level s show an insignificant, negative relationship with levels of student poverty. Teacher quality characteristics such as certificati on status and degree in the field to be taught are very significantly and positively correlated with student outcomes Characteristics such as education level (percenta ge of teachers with master's degrees) show positive but less strong rel ationships with education outcomes. Per pupil spending (measured as current expenditure s) shows a significant positive relationship with student outcomes in 4th grade reading in both years, but no relationship with student outcomes in mathem atics This may be because the spending measure incorporates resources spent n ot only on teacher salaries and professional development but also on class sizes an d other resources that may especially support students in the early grades as they are learning to read. Although salaries and spending are strongly related to one another (p < .01), teacher salary levels, unadjusted for cost of livin g differences, are not correlated with student outcomes when aggregated to the state level. Other school resources, such as pupil-teacher ratio s, class sizes, and the proportion of all school staff who are teachers, sh ow very weak and rarely significant relationships to student achievement wh en they are aggregated to the state level Partial correlations confirm a strong, si gnificant relationship of teacher quality variables to student achievement even after control ling for student poverty and for student language background (LEP status) in (see Ta ble 2 and Figure 4). The most consistent highly significant predictor of student achievement in reading and mathematics in each year tested is the proportion o f well-qualified teachers in a state: those with full certification and a major in the fi eld they teach (r between .61 and .80, p<.001). The strongest, consistently negative predi ctors of student achievement, also significant in almost all cases, are the proportion s of new teachers who are uncertified (r between -.40 and -.63, p<.05) and the proportions o f teachers who hold less than a minor in the field they teach (r between -.33 and -.56, p <.05). General spending and salary variables, along with class sizes, are not signific antly related to achievement once student characteristics are taken into account. It should be noted, however, that this analysis did not take into account cost-of-living d ifferentials that may affect both salaries and spending levels; controlling for such different ials could produce a different set of results with respect to these variables.
24 of 44 Table 2 Partial Correlations (controlling for student pover ty) between Selected Teacher Quality Variables and Student Achi evement on the National Assessment of Educational Progress Grade 4 Math, 1992 Grade 4 Math, 1996 Grade 8 Math, 1990 Grade 8 Math, 1996 Grade 4 Reading, 1992 Grade 4 Reading, 1994 % of teachers well-qualified (with full certification anda major in their field) .71*** .61*** .75*** .67*** .80*** .75*** % of teachers out of field (with less than a minor in the fieldthey teach) -.48**-.44**-.32-.42**-.56**-.33* % of all teachersfully certified .36*.20.38*.28.57***.41* % of all teachers less than fully certified -.36*-.23-.33*-.28-.55***-.50* % of new entrants to teaching who are uncertified(excluding transfers) -.51**-.39*.43**-.38*-.44**-.47** % of all newly hired teachers uncertified -.40**-.41**-.53***-.49**-.59***-.63*** Per pupil spending.188.8.131.52.24.27Pupil: teacher ratio .03 .22 .09 .12 .08 .08 Class size-.03.21-.04-.00.08.13*p<.10 **p<.05 ***p<.01
25 of 44 Figure 4. Partial Correlations (controlling for stu dent poverty) between Selected Teacher Quality Variables and Student Achievement o n the National Assessment of Educational Progress Ordinary least squares regression analyse s were performed to create the most parsimonious specification of a hyperplane of best fit with student achievement data. Because of the small sample size ( n = 44 states participating in the state NAEP), the number of independent variables in each equation wa s minimized to preserve the necessary degrees of freedom (see Table 3). Variabl es were selected according to three criteria: to examine relationships often tested in other studies, to maximize explanatory power, and to avoid problems of multicollinearity. Teacher quality variables included the percentage of all teachers with full certification and a major in the field and the percentage of uncertified newly hired teachers, bec ause these exhibit large influences on achievement, and the percentage of teachers with ma ster's degrees, because this is a frequently examined teacher quality variable. Class size was also included because it is commonly found to influence achievement. Spending a nd salary variables were not included in the final estimations because they show ed little relationship to student achievement in preliminary estimates. Because the p ercentage of minority students is
26 of 44Analysis of Policy Relationships Clearly, in any analysis such as this, th e variables that can be measured are only proxies for the actual conditions or traits that ma y matter to student learning. In this case, a large number of variables associated with teacher quality appear to bear a significant relationship to student achievement. These include various ways of measuring state certification status (the proportions of teachers w ith full certification, less than full certification, and no certification) and disciplina ry preparation (e.g., a major or minor in the field to be taught). Given the differences in l icensing standards and teacher education programs across states, these proxies are fairly cr ude ones; nonetheless, they seem to indicate that teachers' knowledge, skills, and prep aration matter for student achievement. The findings are similar to those of several other studies described earlier (Ferguson, 1991; Ferguson and Ladd, 1996; Fetler, 1999; Fuller 1999; Strauss and Sawyer, 1986) in finding much stronger influences on student achieve ment of variables measuring teacher knowledge and skills than of variables like teacher experience, class sizes, or pupil-teacher ratios, which are generally found to have noticeable but smaller effects on student achievement where data are aggregated to th e school or district levels. The strength of the "well-qualified teach er" variable may be partly due to the fact that it is a proxy for both strong disciplinary kno wledge (a major in the field taught) and substantial knowledge of education (full certificat ion). If the two kinds of knowledge are interdependent as suggested in much of the literatu re, it makes sense that this variable would be more powerful than either subject matter k nowledge or teaching knowledge alone. It is also possible that this variable captu res other features of the state policy environment including general investments in, and c ommitment to, education, as well as aspects of the regulatory system for education, suc h as the extent to which standards are rigorous and the extent to which they are enforced. Recall that some states require teachers to acquire a subject matter major as well as extensive education training in human development and learning and in the methods o f teaching in their field, while other states require much less extensive preparatio n in the content area as well as teaching and learning. In addition, some states are vigilant in enforcing their certification standards while others are not.Teaching Standards and Other Policy Strategies Finally, there may be unmeasured correlat ions between the extent to which states enact and enforce high standards for teachers and t he extent to which they have enacted other policies that are supportive of public school s. Although it does not appear that teaching standards are strongly related to investme nts regarding class sizes or to overall education spending, it is possible that there are o ther factors influencing student achievement which generally co-exist with teacher q uality and which were unmeasured in these estimates. Since most of the states which ran ked among the highest-scoring on the NAEP examinations are strong local control states t hat have traditionally not exerted much control over school decision making, there are relatively few policy areas in which they have been active. Perhaps the relative lack of policy intervention is itself a support for student learning, leaving educators free of reg ulations that might force greater attention to procedures than learning. Another poss ibility is the influence of these states' small school and district sizes, a factor that has been identified in much research as contributing to student learning (for reviews, see Green & Stevens, 1988; Howley, 1989). In another analysis, Feistritzer (1993) has pointed out that most of the top-scoring states
27 of 44on NAEP have very small average school sizes relati ve to national norms. One area in which policies have not been positively correlated, however, is the extent to which states engaged in statewide student testing in the 1980s and the extent to which they enacted high standards for teachers. Amo ng the 12 highestscoring states in 8th grade mathematics in 1996 (10 of which had part icularly high licensing standards in the form of subject matter and teaching coursework requirements), none had mandatory statewide testing programs in place during the 1980 s or early 1990s. Only two of the top 12 states in 4th grade mathematics had statewide te sting programs in place prior to 1995. By contrast, among the 12 lowest-scoring states (8 of which had particularly large rates of out-of-field and uncertified teachers), 10 had exte nsive student testing programs in place prior to 1990, some of which were associated with h ighly specified state curricula and an extensive menu of rewards and sanctions. There are several possible interpretation s of the almost inverse relationship between statewide testing policies and both teachin g standards and student performance: It may be that states with low student performance and less qualified teachers were more likely to seek education improvements through stude nt testing strategies and curriculum controls. It may also be that states have tended to ward different theories of reform, with some investing more in testing and others in teachi ng. It is possible that regional differences in education investments and centraliza tion happen to be correlated with policies regarding both testing and teacher investm ents (with Southern states that tend to score lowest investing heavily in curriculum and te sting controls, while Northeastern and North Central states invest more in teacher educati on and less in curriculum controls). The lack of apparent relationship between testing programs and student achievement might be because, without other investm ents to improve teaching and schooling, tests alone do not transform learning. A nother possibility is that the kinds of basic skills tests and curricula enacted in many st ates during the 1980s were at odds with the NAEP assessments which increasingly seek to mea sure higher-order skills and performance abilities. It may be worth noting that most of the high-scoring and fast-gaining states discussed earlier instituted cu rriculum and testing reforms in the mid-1990s that were linked to the national student standards that guide NAEP and were much more performance-oriented than the basic skill s tests that predominated in state assessment systems of the 1980s. While there is lit tle evidence yet of the effects of these assessment programs on student learning, policy ana lysts may want to watch to see whether the types of tests matter for broad student outcomes as well as whether and how the supports that do or do not accompany testing pr ograms (professional development, funding equalization, investments in additional sup ports for students ranging from early childhood education to special services of various kinds) make a difference. Policies that May Influence Teachers' Qualification s Another set of questions has to do with w hether there are particular policy strategies used by states or districts that are ass ociated with the preparation and hiring of better qualified teachers. The SASS data set and ad ditional data collected directly from states allowed us to examine several policies in th is regard. Teacher education accreditation National data from the National Association of State Directors of Teacher Education and Certificat ion and from the National Council for the Accreditation of Teacher Education provided the percentage of teacher education institutions that were accredited by NCATE. NCATE-a ccreditation might lead to higher overall standards for teachers because NCATE standa rds revisions in 1988 and 1993 required higher admissions standards, evidence of g reater subject matter preparation, and
28 of 44 stronger rationales for the content of education co ursework than those often emphasized by state approval systems. Standard setting and enforcement mechanisms The state survey tracked the presence of a state professional standards board fo r teaching, analogous to the boards that govern other professions, which might enact and enf orce higher standards. Since any policies for teacher education adopted by such a bo ard would require several years to take broad effect, the enactment of a standards board pr ior to 1990 is the measure we used for examining influences on teacher qualifications in 1 994. District hiring standards SASS data provided the percentage of school distri cts in each state requiring each of the following as condi tions for hiring: full state certification, graduation from an approved teacher education progr am, and a college major or minor in the field to be taught. There was wide variation ac ross the states in the degree to which districts looked for evidence of these kinds of tea cher qualifications as part of the hiring process. Many more fine-grained variables, such as the content of licensing standards and the nature of teacher education programs, could not be tested with these data. Nonetheless, the results suggest some interesting a ssociations. As shown in Table 4, the strongest predictor of the percentage of well-quali fied teachers (that is, teachers with both a major and full certification in their field) is t he percentage of teacher education institutions in a state that meet national accredit ation standards through NCATE (p < .05).Table 4 Relationship Between Professional Accreditation And Teacher QualificationsVariable/(Beta coefficient) % ofwell-qualifiedteachers % ofwell-qualified English teachers % ofwell-qualified Math teachers % of mathteachers out-of-field % ofEnglish teachers out-of-field % ofcolleges NCATE accredited.42*.49**.36*-.37*-.37**p<.05 **p<.01 The proportion of NCATE-accredited instit utions is also significantly and negatively correlated with the proportion of Englis h and mathematics teachers who are "out-of-field" (i.e., have less than a minor in the field they teach). This may be because institutions that are NCATE-accredited must demonst rate that their students have the opportunity to acquire a base of content knowledge deemed acceptable by the subject matter associations that review applications as wel l as pedagogical knowledge in their field. Thus, these institutions may, as a group, ha ve less variability than others in establishing reasonably high standards for discipli nary knowledge as well as knowledge of how to teach the discipline. It may also be that states in which professional accreditation is more widespread also happen to hav e other policies or practices in effect that support the preparation and hiring of well-qua lified teachers.
29 of 44 As shown in Table 5, the extent to which districts maintain rigorous hiring standards (i.e., the percentage of districts requir ing full certification, graduation from an approved teacher education program, and a college m ajor or minor in the field to be taught) is a highly significant predictor (p < .001 ) of the proportions of teachers who are uncertified. It is also a strong predictor of the p roportions of new and veteran teachers who are fully certified. Since teachers' certificat ion status is also related to state demographics, these variables were regressed agains t hiring standards along with student poverty, percent minority, and percent LEP students The relationship between hiring standards and teacher certification status continue s to be highly significant after controlling for student poverty, race, and language status.Table 5 Correlations between Teacher Qualifications and District Hiring Standards (Pearson r)District Hiring Standards(Percent of districts requiring full certification, graduation from an approved teacher education program, and a college major or minor in the field to be taught as a condition of hiring)% of new teachers who are fully certified.28**% of all teachers who are fully certified.33**% of newly hired teachers who are uncertified-.51** % of all teachers who are uncertified-.66****p<.05 **p<.01 ***p<.001Table 6 Relationship between Teacher Qualifications and District Hiring Standards(Controlling for Student Poverty, Minority Status, and Language Status)Variable/Beta Weight/(t value) % all teachers fully certified % new teachers fully certified % all teachers uncertified % new teachers uncertified District hiring standards**** .393 (2.51)* .339 (2.16)* -.636 (-4.73)*** -.502 (-3.19)** Professional Standards Board -.173 (-1.20) -.080 (-.48) % students in poverty -.148 (-.64) -.063 (-.27) .172 (.94) -.108 (-.51) % students LEP .226 (1.23) .374 (2.02) .105 (.63) .045 (.23)
30 of 44 % students minority .125 (.58) -.112 (-.43) -.352 (-1.66) -.105 (-.42)*p<.05 **p<.01 ***p<.001****Percent of districts requiring, as a condition of hiring, full certification, graduation from an approved teacher education program, and a college m ajor or minor in the field to be taught This suggests that enforcing standards is both a state and local job. In a quasi-profession like teaching, there is a complex interplay between the standards adopted by states and the ways in which local schoo ls and districts manage their hiring processes, sometimes in accord with and sometimes i n violation of state standards. A minority of states enforce their teacher licensing standards in the inviolable fashion with which standards for doctors, lawyers, architects, a nd other professionals are enforced. These other professions use professional standards boards established by each state as standard-setting and enforcement bodies. Depending on the degree of authority and autonomy used as defining characteristics, 12 to 18 states have established such boards for teaching. As shown in Table 7, the presence of a pr ofessional standards board prior to 1990 proves to be significantly related to district hiri ng standards, a relationship that holds up after controlling for student characteristics. In a ddition, as Table 8 indicates, the presence of a standards board is significantly associated wi th the proportions of certified and uncertified teachers. This relationship may work th rough the influence such a board exerts over district decisions about hiring qualifi ed personnel, as suggested above. Districts often hire unqualified teachers even thou gh fully prepared teachers are available if state agencies do not prevent them from doing so This can occur as a function of cumbersome hiring procedures, patronage, lack of re cruitment effort or incentives, or efforts to reduce salary costs (NCTAF, 1996). Depen ding upon how they are structured, some standards boards may have more authority and/o r more commitment to prevent the hiring of unqualified teachers than some state agen cies do. In agency interviews, for example, a staff member of a highly effective state standards board described how the board examines the candidate qualifications as well as the district's advertising, selection, and hiring practices and applicant pool in any case where a district requests permission to hire staff on an emergency or temporary license. Ve ry few requests for hiring of unqualified personnel are ultimately granted, and d istrict hiring practices are often revised and improved in the process of the review. In other states, agency officials described routine, blanket approvals of district requests for emergency hiring even in situations where districts had just laid off large numbers of qualified teachers or had qualified applicants in the applicant pool. These officials g enerally felt they did not have the resources or the authority to investigate or stem p ractices they felt were illegal and widespread.Table 7 Correlations (Pearson r) of Presence of a Professio nal Standards Board with District Hiring Standards and Teacher Qualific ations% of districts requiring graduation from an approve d teacher education program .25*
31 of 44 % of districts requiring a college major or minor i n the field to be taught .23* % of districts requiring full certification, gradua tion from an approved program, and a college major or minor .30** % uncertified teachers -.27** % fully certified teachers .21* % fully certified new teachers .21* # of weeks required for student teaching.25**p<.05 **p<.01 Table 8 Relationship between Professional Standards Board P resence and District Hiring Standards District hiring standardsProfessional Standards Board .411 (2.49)** % students in poverty .132 (.58) % LEP students -.429 (-2.20)* % minority students .067 (.26)*p<.05 **p<.01 These relationships between the presence of standards boards and teacher education or hiring practices, although statistical ly significant, are quite modest (correlations in the .2 to .3 range), suggesting th at many other variables are at play here as well. It is certainly true that some states enact a nd enforce high standards for teaching without the presence of standards boards, while som e standards boards do not pursue their mission with the same vigor as others. Where they exist, however, such bodies often appear to bring greater consistency of effort and a ttention to the issues of preparation and qualifications.Conclusions and Implications This analysis triangulates data from surv eys of state policies, case study analyses of state policymaking, and quantitative examination of the distribution of state achievement scores and resources, taking student characteristic s into account. Some findings are
32 of 44particularly noteworthy. First, while student demog raphic characteristics are strongly related to student outcomes at the state level, the y are less influential in predicting achievement levels than variables assessing the qua lity of the teaching force. Second, when aggregated at the state level, teacher quality variables appear to be more strongly related to student achievement than class sizes, ov erall spending levels, teacher salaries (at least when unadjusted for cost of living differ entials), or such factors as the statewide proportion of staff who are teachers. Among variables assessing teacher "qualit y," the percentage of teachers with full certification and a major in the field is a more po werful predictor of student achievement than teachers' education levels (e.g., master's deg rees). This finding concurs with those of other studies cited earlier. It is not surprising t hat masters degrees would be relatively weaker measures of teacher knowledge, given the wid e range of content they can include, ranging from specialist degrees in reading or speci al education that are directly related to teaching to fields like administration and others t hat have little to do with teaching. Other measures of certification status (e.g., the percent of teachers uncertified, the percent with full certification) are also strong correlates of s tudent achievement. Finally, certain policy strategies associated with standard-setting at the state and local level--NCATE-accreditation of teacher education ins titutions, district hiring standards, and, to a lesser extent, state professional standar ds boards--appear to be related to teacher qualifications in the field. While the triangulation of data from seve ral sources lends some confidence to these findings, they should be viewed with caution. Like all studies that draw inferences from broad state trends and correlational data, there ar e many variables in play at any given time and many possible explanations for any phenome non observed. While this article presents a range of competing explanations for stud ent achievement trends (student background, curriculum and testing policies, school funding and equalization, school and class sizes), it could not fully test all of these explanations. This remains for other researchers to pursue. In addition, other data and other methodologies could shed further light on these questions. Adding information about parent education levels might make a difference in the measurement of student background ; adding data about school and district size (from the Common Core of Data) and cu rriculum and testing approaches (from the NAEP background surveys) might shed great er light on school factors that matter; and adjusting salary and spending data for cost of living differentials might allow a better evaluation of fiscal influences. By including estimates of the proportions of staff who are underqualified (and who tend to cluster in less advantaged schools and dist ricts), this study's estimates tapped some of the local variability in resources made ava ilable to children. However, because state data on average class sizes and other school resources ignore wide variations in teaching and learning conditions that may be very i mportant at the district, school, and classroom levels, these estimates cannot fully capt ure the effects of such variables. Average class sizes, for example, vary relatively l ittle across states but vary substantially within states and districts. Thus, effects of this variable are much more likely to be perceived with more disaggregated data. By merging district, school, and teacher files, the SASS data can allow for the use of Hierarchical Linear Modeling techniques, which would be a useful tool for further exploring relati onships between teaching and schooling variables at the school, district, and state levels Nonetheless, the findings of this study, in conjunction with a number of other studies in recent years, suggest that states intere sted in improving student achievement may be well-advised to attend, at least in part, to the preparation and qualifications of the teachers they hire and retain in the profession. It stands to reason that student learning
33 of 44should be enhanced by the efforts of teachers who a re more knowledgeable in their field and are skillful at teaching it to others. Substant ial evidence from prior reform efforts indicates that changes in course taking, curriculum content, testing, or textbooks make little difference if teachers do not know how to us e these tools well and how to diagnose their students' learning needs (for a review, see D arling-Hammond, 1997b). Like other studies cited earlier, this re search indicates that the effects of well-prepared teachers on student achievement can b e stronger than the influences of student background factors, such as poverty, langua ge background, and minority status. And while smaller class sizes appear to contribute to student learning, particularly in fields like elementary reading, the gains occasione d by smaller classes are most likely to be realized, as they were in the Tennessee experime nt, when they are accompanied by the hiring of well-qualified teachers. The largescale hiring of unqualified teachers, as was the case in California's recent class size reductio n initiative, would likely offset any achievement gains that could be realized by smaller class sizes. Another implication of this study is that states may impact the qualifications of the teachers through policies that influence the hiring standards of school districts (e.g., incentives and sanctions from the state level that encourage the hiring of well-qualified individuals), the accreditation of teacher educatio n institutions (e.g., encouragement or requirements for the use of NCATE standards or othe rs of equivalent rigor), and the bodies that establish and enforce teaching standard s (e.g. establishment of professional standards boards or assurance of adequate capacity and authority for state agencies to uphold high standards for teaching). Although this study used fairly crude mea sures of teacher knowledge and skills such as certification status, college major, and ma ster's degrees, policymakers should be aware that there are much more fine-grained distinc tions to be made among types of state certification standards, teacher education programs professional development offerings, and education requirements that make a difference t o the teachers' abilities and their students' outcomes. Reforms underway to create more thoughtful licensing systems, more productive teacher education programs, and more eff ective professional development strategies are producing evidence of the stronger e ffects on teaching and learning of approaches that strengthen teachers' abilities to t each diverse learners with a keen diagnostic eye and a wide repertoire of strategies supporting mastery of challenging content (for a review, see NCTAF, 1996; Darling-Ham mond 1997a). Over the next decade, federal, state, and local policymakers inte rested in helping students meet higher learning standards may want to consider how investm ents in teacher quality, along with other reforms, can assist them in achieving their g oals.NotesThis research was funded in part by the Office of E ducational Research and Improvement (OERI) of the U.S. Department of Educat ion through the Center for the Study of Teaching and Policy, which is housed a t the University of Washington and includes Stanford University, Teachers College, Columbia University, and the University of Michigan. The research was initiated while the author was a fellow at the Center for Advanced Study in the Behavioral Sci ences with the support of the Spencer Foundation. The views represented in this a rticle are those of the author alone, and do not represent those of any sponsor. 1. National Center for Education Statistics, Schools and Staffing Survey, 1993-94: State by State Data Washington, DC: U.S. Department of Education, 199 6, Table 3.5. Additional tabulations performed by the Nation al Commission on Teaching 2.
34 of 44and America's Future.The INTASC standards, developed by a consortium of more than 30 states and professional associations under the auspices of the Council of Chief State School Officers, are based on knowledge of effective learn ing and teaching and on the student learning standards developed by professiona l associations such as the National Council of Teachers of Mathematics. The IN TASC standards for beginning teacher licensing are compatible with the more advanced standards of the National Board for Professional Teaching Standa rds, which define and assess accomplished teaching among veteran teachers. 3.ReferencesAdams, R.D., S. Hutchinson, and C. Martray (1980). A developmental study of teacher concerns across time. Paper presented at the Americ an Educational Research Associaton Annual Meeting. Boston, Mass.Andrew, M. & Schwab, R.L. (1995). Has reform in tea cher education influenced teacher performance? An outcome assessment of graduates of eleven teacher education programs. Action in Teacher Education, 17 43-53. Andrews, J.W., Blackmon, C.R., and Mackey, J.A. (19 80). Preservice performance and the National Teacher Examinations. Phi Delta Kappan, 61 (5), pp. 358-359. Armour-Thomas, E., Clay, C., Domanico, R., Bruno, K ., & Allen, B. (1989). An outlier study of elementary and middle schools in New York City: Final report. NY: New York City Board of Education.Ashton, P. & Crocker, L. (1987, May-June). Systemat ic study of planned variations: The essential focus of teacher education reform. Journal of Teacher Education, 38 2-8. Ayers, J.B., and Qualls, G.S. (Nov/Dec 1979). Concu rrent and predictive validity of the National Teacher Examinations. Journal of Educational Research, 73 (2), pp.86-92. Barnes, S., Salmon, J. and Wale, W. (1989), "Altern ative Teacher Certification in Texas." Presented at the annual meeting of the American Edu cational Research Association, March. ERIC Document No. 307316.Begle, E.G. (1979). Critical Variables in Mathemati cs Education. Washington, D.C.: Mathematical Association of American and National C ouncil of Teachers of Mathematics.Begle, E.G. and Geeslin, W. (1972). Teacher effecti veness in mathematics instruction. National Longitudinal Study of Mathematical Abiliti es Reports No. 28. Washington, D.C. Mathematical Association of America and Nation al Council of Teachers of Mathematics.Bents, Mary, and Richard Bents (1990). Perceptions of Good Teaching Among Novice, Advanced Beginner and Expert Teachers. Paper presen ted at the Annual Meeting of the American Educational Research Association, Boston, MA. Berliner, D.C. & Biddle, B.J. (1995). The manufactured crisis: Myth, fraud, and the
35 of 44attack on America's public schools Reading, MA: Addison-Wesley. Berliner, D.C. & Tikunoff, W. J. (1976). The Califo rnia Beginning Teacher Study. Journal of Teacher Education, 27 24-30. Bowles, S., and Levin, H.M. (1968). The determinant s of scholastic achievementÂ—An appraisal of some recent evidence. Journal of Human Resources, 3 3-24. Brown, C.A., Smith, M.S., and Stein, M.K. (1995). L inking teacher support to enhanced classroom instruction. Paper presented at the annua l meeting of the American Educational Research Association. New York, New Yor k. Byrne, C.J. (1983). Teacher knowledge and teacher e ffectiveness: A literature review, theoretical analysis and discussion of research str ategy. Paper presented at the meeting of the Northwestern Educational Research Association, Ellenville, NY. Campbell, J.R., Donahue, P.L., Reese, C.M., Phillip s, G.W. (1996). NAEP 1994 reading report card for the nation and the states. Washingt on, D.C.: U.S. Department of Education.Carroll, J.B. (1975). The teaching of French as a foreign language in eig ht countries. NY: John Wiley and Sons.Carter, K., and Doyle, W. (1987). Teachers' knowled ge structures and comprehension processes. Pp. 147-160 in J. Calderhead (Ed.), Exploring Teacher Thinking London: Cassell.Connecticut State Department of Education Division of Research, Evaluation, and Assessment (1991). Research Bulletin, School Year 1 990-91, No. 1 (Hartford, CT: Bureau of Research and Teacher Assessment).Cooper, E. & Sherk, J. (1989). Addressing urban sch ool reform: Issues and alliances. Journal of Negro Education, 58 3, 315-331. Cohen, D.K. and Hill, H. (1997). Instructional Poli cy and Classroom Performance: The Mathematics Reform in California. Paper presented a t the Annual Meeting of the American Educational Research Association, Chicago, IL. Coleman, J.S., Campbell, E.Q., Hobson, C.J., McPart land, J., Mood, A.M., Weinfeld, F.D., York, R.L. (1966). Equality of educational op portunity. Washington, DC: U.S. Government Printing Office.Council on School Performance (1997). Teachers with advanced degrees advance student learning. Atlanta: Council for School Performance, Georgia State University. Darling-Hammond, L. (1992). Teaching and knowledge: Policy issues posed by alternative certification for teachers. Peabody Journal of Education, 67 3, 123-154. Darling-Hammond, L. (1997a). Doing What matters mos t: Investing in quality teaching. NY: National Commission on Teaching and America's F uture. Darling-Hammond, L. (1997b). The right to learn: A blueprint for creating school s that
36 of 44work San Francisco: Jossey Bass. Darling-Hammond, L.; Hudson, L.; and Sheila Kirby ( 1989). Redesigning Teacher Education: Opening the Door for New Recruits to Sci ence and Mathematics Teaching. Santa Monica:The RAND Corporation.Darling-Hammond, L., Wise, A.E. & Klein, S. (1995). A license to teach: Building a profession for 21st century schools Boulder: Westview Press. Darling-Hammond, L., Wise, A.E. and Pease S.R. (198 3). Teacher evaluation in the organizational context: a review of the literature. Review of educational research, 53 285-237.Denton, J.J., and L.J. Lacina (1984). Quantity of p rofessional education coursework linked with process measures of student teaching. Teacher Education and Practice 39-64.Denton, J.J., and Peters, W.H. (1988). Program asse ssment report: Curriculum evaluation of a non-traditional program for certifying teacher s. Texas A &M University, College Station, TX.Doyle, W. (1985). Recent research on classroom mana gement: Implications for teacher preparation. Journal of Teacher Education, 36 3, 31-35. Doyle, W. (1986). Content representation in teacher s' definitions of academic work. Journal of Curriculum Studies, 18 365379. Druva, C.A., & Anderson, R.D. (1983). Science teach er characteristics by teacher behavior and by student outcome: A meta-analysis of research. Journal of Research in Science Teaching, 20 ,5, 467-479. Erekson, T.L. and Barr, L. (1985). Alternative cred entialing: Lessons from vocational education. Journal of Teacher Education, 36 3, 16-19. Evertson, C., Hawley, W., & Zlotnik, M. (1985). Mak ing a difference in educational quality through teacher education. Journal of Teacher Education, 36 ,3, 2-12. Feiman-Nemser, Sharon, and Michelle B. Parker (1990 ). Making Subject Matter Part of the Conversation or Helping Beginning Teachers Lear n to Teach. East Lansing, MI: National Center for Research on Teacher Education.Feistritzer, C.E. (1993). Report Card on American E ducation: A State-by-State Analysis, 1972-73 to 1992-93. Washington, D.C.: National Cent er on Education Information. Fetler, M. (1999). High school staff characteristic s and mathematics test results. Education Policy Analysis Archives, 7 9 [Entire issue]. (Available online at http://epaa.asu.edu/epaa/v7n9.html)Ferguson, R.F. (1991, Summer). Paying for public ed ucation: New evidence on how and why money matters. Harvard Journal on Legislation, 28 ,2, 465-498. Ferguson, R.F. and Ladd, H.F. (1996). How and Why M oney Matters: An Analysis of
37 of 44Alabama Schools. Pp. 265-298 in Helen Ladd (Ed.) Holding Schools Accountable Washington, D.C.: Brookings Institution.Ferguson, P. and Womack, S.T. (1993). The impact of subject matter and education coursework on teaching performance. Journal of Teacher Education, 44, 1, 55-63. Fuller, E. J. (1999). Does teacher certification ma tter? A comparison of TAAS performance in 1997 between schools with low and hi gh percentages of certified teachers. Austin: Charles A. Dana Center, Universit y of Texas at Austin. Glass, G.V; Cahen, L.S.; Smith, M.L.; and Filby, N. N. (1982). School class size: Research and Policy. Beverly Hills, CA: SAGE Publications. Glassberg, S. (1980). A view of the beginning teach er from a developmental perspective. Paper presented at the American Educational Researc h Association Annual Meeting. Boston, MA.Goebel, Stephen D., Karl Ronacher, and Kathryn S. S anchez (1989), An Evaluation of HISD's Alternative Certification Program of the Aca demic Year: 1988-1989. Houston: Houston Independent School District Department of R esearch and Evaluation. ERIC Document No. 322103.Goldhaber, M. and Brewer, D. (1999). Does teacher c ertification matter? High school certification status and student achievement. Unpub lished manuscript. Gomez, D.L., and Grobe, R.P. (1990). Three Years of Alternative Certification in Dallas: Where are We? Paper presented at the Annual Meeting of the American Educational Research Association, Boston, MA.Good, T.L. (1983). Recent classroom research: Impli cations for teacher education. In D.C. Smith (ed.), Essential knowledge for beginning educators Washington, D.C.: American Association of Colleges for Teacher Educat ion. Good, T.L. and J. E. Brophy. (1986). Educational Psychology, 3rd Edition NY: Longman.Grady, M.P., Collins, P. & Grady, E.L. (1991). Teac h for America 1991 Summer Institute Evaluation Report. Unpublished manuscript.Greenwald, R., Hedges, L.V., and Laine, R.D. (1996) The effect of school resources on student achievement. Review of Educational Research, 66 361-396. Grissmer, D. and Flanagan, A. (1998). Exploring Rap id Achievement Gains in North Carolina and Texas. Washington, D.C.: National Educ ation Goals Panel. Grossman, P. L. (1989). Learning to teach without t eacher education. Teachers College Record, 91 2, 191-208. Guyton, E. and Farokhi, E. (1987). Relationships am ong academic performance, basic skills, subject matter knowledge and teaching skill s of teacher education graduates. Journal of Teacher Education, 38, (Sept-Oct.), 37-42.
38 of 44Haberman, M. (1995, June). Selecting 'star' teacher s for children. Phi Delta Kappan, 76 777-781.Haney, W. (1999). Supplementary report on Texas Ass essment of Academic Skills Exit Test (TAAS-X). Boston: Center for the Study of Test ing, Evaluation, and Educational Policy.Haney, W., Madaus, G., & Kreitzer, A. (1987). Charm s talismanic: testing teachers for the improvement of american education. Pp. 169-238 in E.Z. Rothkopf (Ed.) Review of Research in Education Vol. 14. Washington, D.C.: American Educational R esearch Association.Hansen, J. B. (1988). The Relationship of Skills an d Classroom Climate of Trained and Untrained Teachers of Gifted Students. Unpublished doctoral dissertation, Purdue University.Hanushek, E.A. (1971). Teacher characteristics and gains in student achievement: Estimation using micro data. The American Economic Review, 61 (2), 280-288. Hanushek, E.A., Rivkin, S.G., and Taylor, L.L. (199 5). Aggregation bias and the estimated effects of school resources. Working pape r 397. University of Rochester, Center for Economic Research.Hawk, P.; Coble, C.R.; and Swanson, M. (1985). Cert ification: It Does Matter, Journal of Teacher Education, 36 3, 13-15. Hellfritzch, A.G. (1945). A factor analysis of teac her abilities. Journal of Experimental Education, 14 166169. Hice, J.E.L. (1970). The Relationship between Teach er Characteristics and First-Grade Achievement. Dissertation Abstracts International, 25 1, 190. Johanson, G.A. and Gips, C.J. (1992). The hiring pr eferences of secondary school principals. The High School Journal, 76 (Oct./Nov.), 1-16. Jordan, H.R., Mendro, R.L., & Weersinghe, D. (1997) Teacher effects on longitudinal student achievement: A preliminary report on resear ch on teacher effectiveness. Paper presented at the National Evaluation Institute, Ind ianapolis, IN. Kalamazoo, MI: CREATE, Western Michigan University.Kentucky Institute for Education Research (1997). T he preparation of teachers for Kentucky Schools: A survey of new teachers. Frankfo rt, KY: Kentucky Institute for Education ResearchKlitgaard, R.E. and Hall, G.R. (1974). Are there un usually effective schools? Journal of Human Resources, 10 3, 90-106. Knoblock, G.A. (1986). Continuing professional educ ation for teachers and its relationship to teacher effectiveness. Unpublished dissertation. Michigan State University. Dissertation Abstracts International, 46 2, 3325A, University microfilms no. AAC8529729.
39 of 44Kurtz, M. (1999). Schools, teens battle barrier of 9th grade. American-Statesman May 23, 1999.LaDuke, D.V. (1945). The measurement of teaching ab ility. Journal of Experimental Education, 14 75-100. Lenk, Harriet Anne (1989). A Case Study: The Induct ion of Two Alternate Route Social Studies Teachers. Unpublished doctoral dissertation Teachers College, Columbia University.LuPone, L.J. (1961). A comparison of provisionally certified and permanently certified elementary school teachers in selected school distr icts in New York State. Journal of Educational Research, 55 1, 53-63. Lutz, Frank W., and Jerry B. Hutton (1989). Alterna tive teacher certification: Its policy implications for classroom and personnel practice. Educational Evaluation and Policy Analysis, 11 3, 237-254. McNeil, J.D. (1974). Who gets better results with y oung children -experienced teachers or novices? Elementary School Journal, 74 447-451. Mitchell, N. (1987). Interim Evaluation Report of t he Alternative Certification Program (REA87-027-2). Dallas, TX: DISD Department of Plann ing, Evaluation, and Testing. Monk, D. H. (1994). Subject matter preparation of s econdary mathematics and science teachers and student achievement. Economics of Education Review, 13 2, 125-145. Monk, D. H. and King, J.A. (1994). Multilevel teach er resource effects in pupil performance in secondary mathematics and science: T he case of teacher subject matter preparation. Pp. 2958 in R.G. Ehrenberg (Ed.), Choices and consequences: Contemporary policy issues in education Ithaca, NY: ILR Press. Mosteller, F. (1995). The Tennessee study of class size in the early school grades. The Future of Children, 5 2, 113127. Murnane, R.J. (1985, June). Do effective teachers h ave common characteristics: Interpreting the quantitative research evidence. Pa per presented at the National Research Council Conference on Teacher Quality in Science an d Mathematics, Washington, D.C. Murnane, R.J., and Phillips, B.R. (1981). Learning by doing, vintage, and selection: Three pieces of the puzzle relating teaching experi ence and teaching performance. Economics of Education Review, 1fR, 4, 691-693. National Center for Education Statistics (1994). Da ta compendium for the NAEP 1992 reading assessment of the nation and the states: 19 92 NAEP trial state assessment. Washington, D.C.: U.S. Department of Education.National Center for Education Statistics (n.d.) NAE P 1992, 1994 national reading assessments Â– Data almanac Â– Grade 4: Teacher quest ionnaire weighted percentages and composite proficiency means (Public school). (Avail able online at http:// www.Nces.ed.gov/nationsreportcard/y25alm/almanac.sh tml )
40 of 44National Association of State Directors of Teacher Education and Certification (NASDTEC)(1997). Manual on certification and prepar ation of educational personnel in the United States and Canada, 1997-98.National Center for Education Statistics (NCES) (19 95). Digest of Education Statistics, 1995. Washington, D.C.: U.S. Department of Educatio n. National Commission on Teaching and America's Futur e (NCTAF) (1996). What Matters Most: Teaching for America's Future. New York: Auth or. Natriello, G., Zumwalt, K., Hansen, A., and Frisch, A. (1990). Characteristics of Entering Teachers in New Jersey. Revised version of a paper presented at the 1988 Annual Meeting of the American Educational Research Associ ation. Olsen, D. G. (1985). The quality of prospective tea chers: Education vs. noneducation graduates, Journal of Teacher Education, 36 5, 56-59. Penick, J.E., and Yager, R.E. (1983). The search fo r excellence in science education. Phi Delta Kappan, 64 8, 621-623. Perkes, V.A. (1967-1968). Junior high school scienc e teacher preparation, teaching behavior, and student achievement. Journal of Research in Science Teaching, 6 4, 121-126.Pflaum, S.W. and Abramson, T. (1990, March). Teache r assignment, hiring, and preparation: Minority teachers in New York City. The Urban Review, 22 17-31. Quirk, T.J., Witten, B.J., and Weinberg, S.F. (1973 ). Review of studies of concurrent and predictive validity of the National Teacher Examina tions. Review of Educational Research, 43, 89-114. Reese, C.M., Miller, K.E., Mazzeo, J., Dossey, J.A. (1997). NAEP 1996 Mathematics Report Card for the Nation and the States. Washingt on, D.C.: National Center for Education Statistics, U.S. Department of Education.Rosenholtz, S. J. (1986). The organizational contex t of teaching. In Learning to Teach University of Illinois at Champaign-Urbana.Rosenshine, B. and Furst, N.F. (1973). The use of d irect observation to study teaching. In R.M.W. Travers (ed.), Handbook of Research on Teaching (2nd Edition) Chicago: Rand McNally.Rostker, L.E. (1945). The measurement of teaching a bility. Journal of Experimental Education, 14 5-51. Rottenberg, Claire J. and David C. Berliner (1990). Expert and Novice Teachers' Conceptions of Common Classroom Activities. Paper p resented at the Annual Meeting of the American Educational Research Association, Bost on, MA. Roupp, R.; Travers, J.; Glantz, F.; and Coelen, C. (1979). Children at the Center: Summary Findings and their Implications. Cambridge, MA: Abt Associates.
41 of 44Sanders, S.L., Skonie-Hardin, S.D., and Phelps, W.H (1994, November). The effects of teacher educational attainment on student education al attainment in four regions of Virginia: Implications for administrators. Paper pr esented at the Annual Meeting of the Mid-South Educational Research Association.Sanders, W.L. & Rivers, J.C. (1996). Cumulative and residual effects of teachers on future student academic achievement. Knoxville: Uni versity of Tennessee Value-Added Research and Assessment Center.Schalock, D. (1979). Research on teacher selection. In D.C. Berliner (Ed.), Review of research in education, Vol. 7, Washington, D.C.: American Educational Research Association.Skinner, W.A. (1947). An investigation of factors u seful in predicting teaching ability. University of Manchester. Master of Education thesi s. Skipper, C. E. and Quantz, R. (1987). Changes in ed ucational attitudes of education and arts and science students during four years of coll ege, Journal of Teacher Education, 38 MayJune, 39-44.Soar, R.S., Medley, D.M., and Coker, H. (1983). Tea cher evaluation: A critique of currently used methods. Phi Delta Kappan, 65 4, 239-246. Stoddart, T. (1992). An alternate route to teacher certification: Preliminary findings from the Los Angeles Unified School District Intern Prog ram. Peabody Journal of Education,67 3. Strauss, R. P. and Sawyer, E.A. (1986). Some New Ev idence on Teacher and Student Competencies. Economics of Education Review, 5 1, 41-48. Summers, A.A., and Wolfe, B.L. (1975, February). Wh ich School Resources Help Learning? Efficiency and Equality in Philadelphia P ublic Schools. Philadelphia, PA: ED 102 716Taylor, J.K. and R. Dale (1971). A Survey of Teache rs in the First Year of Service. Bristol: University of Bristol, Institute of Educat ion. Texas Education Agency (1993). Teach for America Vi siting Team Report. Austin: Texas State Board of Education Meeting Minutes, App endix B. Vernon, P.E. (1965). Personality Factors in Teacher Trainee Selection. British Journal of Educational Psychology, 35, 140-149. Walberg, H.J., Waxman, H.C. (1983). Teaching, learn ing, and the management of instruction. In D.C. Smith (Ed.), Essential knowled ge for beginning educators. Washington, DC: American Association of Colleges fo r Teacher Education and ERIC Clearinghouse on Teacher Education.Wiley, D. & Yoon, B. (1995). Teacher reports of opp ortunity to learn: Analyses of the 1993 California Learning Assessment System. Educational Evaluation and Policy Analysis, 17 3, 355-370.
42 of 44 Wise, A.E.; Darling-Hammond, L.; and Berry, B. (198 7). Effective Teacher Selection, From Recruitment to Retention. Santa Monica, CA: RA ND Corporation. Wright, S.P.; Horn, S.P.; and Sanders, W.L. (1997). Teacher and classroom context effects on student achievement: Implications for te acher evaluation. Journal of Personnel Evaluation in Education 57-67.About the AuthorLinda Darling-HammondSchool of EducationStanford University Email: firstname.lastname@example.org Linda Darling-Hammond is Charles E. Ducommun Profes sor of Education at Stanford University and executive director of the National C ommission on Teaching and America's Future. Her research, policy, and teachin g focus on teacher education and teaching quality, school restructuring, and educati onal equity. Among other writings, she is author of The Right to Learn, which received the Outstanding Book Award from the American Educational Research Association in 1998.Copyright 2000 by the Education Policy Analysis ArchivesThe World Wide Web address for the Education Policy Analysis Archives is epaa.asu.edu General questions about appropriateness of topics o r particular articles may be addressed to the Editor, Gene V Glass, email@example.com or reach him at College of Education, Arizona State University, Tempe, AZ 8 5287-0211. (602-965-9644). The Commentary Editor is Casey D. C obb: firstname.lastname@example.org .EPAA Editorial Board Michael W. Apple University of Wisconsin Greg Camilli Rutgers University John Covaleskie Northern Michigan University Alan Davis University of Colorado, Denver Sherman Dorn University of South Florida Mark E. Fetler California Commission on Teacher Credentialing Richard Garlikov email@example.com Thomas F. Green Syracuse University Alison I. Griffith York University Arlen Gullickson Western Michigan University Ernest R. House University of Colorado Aimee Howley Ohio University
43 of 44 Craig B. Howley Appalachia Educational Laboratory William Hunter University of Calgary Daniel Kalls Ume University Benjamin Levin University of Manitoba Thomas Mauhs-Pugh Green Mountain College Dewayne Matthews Western Interstate Commission for HigherEducation William McInerney Purdue University Mary McKeown-Moak MGT of America (Austin, TX) Les McLean University of Toronto Susan Bobbitt Nolen University of Washington Anne L. Pemberton firstname.lastname@example.org Hugh G. Petrie SUNY Buffalo Richard C. Richardson New York University Anthony G. Rud Jr. Purdue University Dennis Sayers Ann Leavenworth Centerfor Accelerated Learning Jay D. Scribner University of Texas at Austin Michael Scriven email@example.com Robert E. Stake University of IllinoisÂ—UC Robert Stonehill U.S. Department of Education David D. Williams Brigham Young UniversityEPAA Spanish Language Editorial BoardAssociate Editor for Spanish Language Roberto Rodrguez Gmez Universidad Nacional Autnoma de Mxico firstname.lastname@example.org Adrin Acosta (Mxico) Universidad de Guadalajaraadrianacosta@compuserve.com J. Flix Angulo Rasco (Spain) Universidad de Cdizfelix.email@example.com Teresa Bracho (Mxico) Centro de Investigacin y DocenciaEconmica-CIDEbracho dis1.cide.mx Alejandro Canales (Mxico) Universidad Nacional Autnoma deMxicocanalesa@servidor.unam.mx Ursula Casanova (U.S.A.) Arizona State Universitycasanova@asu.edu Jos Contreras Domingo Universitat de Barcelona Jose.Contreras@doe.d5.ub.es Erwin Epstein (U.S.A.) Loyola University of ChicagoEepstein@luc.edu Josu Gonzlez (U.S.A.) Arizona State Universityjosue@asu.edu Rollin Kent (Mxico)Departamento de InvestigacinEducativa-DIE/CINVESTAV Mara Beatriz Luce (Brazil)Universidad Federal de Rio Grande do Sul-UFRGS
44 of 44 firstname.lastname@example.org email@example.com firstname.lastname@example.orgJavier Mendoza Rojas (Mxico)Universidad Nacional Autnoma deMxicojaviermr@servidor.unam.mxMarcela Mollis (Argentina)Universidad de Buenos Airesmmollis@filo.uba.ar Humberto Muoz Garca (Mxico) Universidad Nacional Autnoma deMxicohumberto@servidor.unam.mxAngel Ignacio Prez Gmez (Spain)Universidad de Mlagaaiperez@uma.es Daniel Schugurensky (Argentina-Canad)OISE/UT, Canadadschugurensky@oise.utoronto.ca Simon Schwartzman (Brazil)Fundao Instituto Brasileiro e Geografiae Estatstica email@example.com Jurjo Torres Santom (Spain)Universidad de A Coruajurjo@udc.es Carlos Alberto Torres (U.S.A.)University of California, Los Angelestorres@gseisucla.edu