Does empirical evidence clearly show that more experienced teachers with higher degrees are better? YES
Overview Background Models with inconclusive effects Ehrenberg & Brewer Monk & King Limitations of econometrics Models finding positive effects Goldhaber & Brewer Evertson & al. Darling-Hammond Policy Questions Case studies finding positive effects Policy implications
What do we mean by better? In most econometric models, better = positive effect of teacher characteristics (teacher experience and teacher degree level) on student achievement.
What are experienced teachers? Student teaching experience Continuous professional development Number of years in the classroom What are higher degrees? Bachelor s degree Education degree Master s degree
Overall Findings from Econometric Models PPE, teacher experience and teacher degree inputs are relatively unimportant predictors of outcomes and impact of inputs is inconsistent across studies Results are therefore inconclusive for policy-makers
Ehrenberg & Brewer Use High School and Beyond longitudinal study Look at effects of school and teacher characteristics on drop-out rates and gain scores
E&B Findings School and teacher characteristics influence drop-out rates Increasing % of teacher s with at least a master s degree or with 10 or more years of experience is associated with decreased drop-out rates for black students Student gain scores are higher when teachers come from more selective institutions (presumably more able )
Teacher Salary Variables Matter Increase in estimated return to holding post-graduate degree _ increase in % of teachers with such degrees Higher starting salaries _ increase in selectivity of colleges that teachers in a school attended
More E&B Findings The greater the percentage of teacher s with at least a master s degree: The lower white students gain scores The higher black students gain scores Negative association with Hispanic students gain scores
Monk & King Use Longitudinal Study of American Youth Explore effects of teacher subject-matter preparation on student performance gains Mathematics Life Sciences Physical Sciences Multilevel approach looks at Proximate teachers Immediate past teachers Non-proximate teachers
Limitations of Econometrics Why we should be cautious before drawing conclusions for policy Conclusions may differ depending on: Methodology Levels of aggregation and analysis Omitted variable bias Publication bias
More on Limitations From Goldhaber Readings Major deficiencies in educational production function studies (especially those from 1970s) Crude measures of teacher ability Measurement errors Lack of detailed controls for student background characteristics (previous academic achievement)
What does this mean? The conventional view that observable school inputs, and teachers in particular, do not positively impact student achievement rests on somewhat empirical grounds (Goldhaber and Brewer, p.507)
Goldhaber & Brewer 1988 National Educational Longitudinal Study (which ties detailed teacher and class-level information to individual students) Use both fixed and random effects Consider both observable and non-observable effects Controls for previous knowledge/ability Look at students who took mathematics achievement test in 8 th and 10 th grade
Model Specifications Dependent Variable = 10 th grade mathematics test score Explanatory variables grouped according to: 1. Family background variables 2. School variables 3. Teacher characteristic variables 4. Classroom variables
G&B Findings Teachers certified in mathematics, or with a BA or MA in mathematics, are identified with higher test scores Unobservable school and teacher effects do play significant role in determining outcomes (but are not jointly correlated with model s explanatory variables so no bias)
Evertson,, Hawley, and Zlotnik Looked at thirteen studies comparing regularly and provisionally certified teachers Eleven of these ranked regularly certified teachers higher in effectiveness Relationship between teacher training in science and cognitive student outcome is progressively higher in higher level courses
Darling-Hammond Reviews variety of studies relating teacher quality to student achievement and looks at: General academic ability and intelligence Subject matter knowledge Knowledge of teaching and learning Teaching experience Certification status Teacher behaviors and practices
Some General Findings Teacher education coursework combined with subject matter knowledge associated with positive performance Sustained professional development grounded in content-specific pedagogy Typically find higher ratings and greater student learning gains for teachers who have more formal preparation for teaching
Ferguson (1991) Analyzed 900 Texas school districts and controlled for student background and district characteristics Measure s of teacher s expertise (scores on licensing exam, master s degree, and experience) accounted for more of interdistrict variation in reading and math achievement gains in grades 1-11 than socioeconomic status
Strauss & Sawyer (1986) Looked at North Carolina teacher s Average scores on the National Teacher Examinations found to have strong influence on average school district test performance 1% increase in teacher quality_3-5% decline in percent of students failing the state competency exam
Armour, Thomas & al. (1989) Looked at demographically similar student populations at high and low achieving schools in New York City Differences in teacher qualifications (educational degrees, certification status, experience) accounted for 90% of total variation in average school-level student achievement in reading and mathematics
Wisconsin and Minnesota Wisconsin and Minnesota = high standard states with high enforcement BA in subject area required 18 weeks of student teaching Students typically scored at the top of the distribution on national assessments of reading and math
Louisiana Louisiana = lower standards and low enforcement No BA or minor in teaching subject required 6 weeks of student teaching Students typically score near the bottom of the NAEP distributions
Characteristics of High Performance States that repeatedly lead the nation in student achievement in reading and mathematics show the following characteristics: Have among the most highly qualified teachers in the country Longstanding investments in the quality of teaching
Darling-Hammond Study Also used Schools and Staffing Surveys (SASS) and NAEP to examine the national trends of teacher qualifications and student achievements Finds that teacher quality characteristics (certification status and degree in field to be taught) are very significantly and positively correlated with student outcomes
Ferguson The case of certification testing in Alabama in 1981 Test score gap between black and white children fell sharply In Texas, TECAT scores were found to be correlated with students math scores Basic test of literacy skills Teacher s measured skills are important determinants of students scores
Policy Questions How do we make sense of all of these findings, in light of their many limitations? How do we define teacher quality? How do we train better teachers? What makes a difference? Is it possible to find a method that will systematically lead to improvements, nationwide?
North Carolina & Connecticut Share relatively large high-poverty student populations Increased salaries and salary equity Initiated improvements (based on current knowledge on teaching and learning): pre-service teacher education licensing beginning teacher mentoring ongoing professional development
Results of Initiatives NC has shown largest student achievement gains in math and reading of any state CT also showed significant gains, despite increases in low-income and limited English proficient students
New Jersey vs. Connecticut (Demographically and economically similar) New Jersey Decreased teacher prep and licensing requirements Reduced amount of required coursework entry-level teachers Encouraged extensive hiring of alternate certification candidates Lower rates of mentoring and induction Higher rates of out-of-field teaching Connecticut Raised teacher salaries Equalized district ability to pay for qualified teachers Improvements in qualifications of teachers Rigorous licensing examinations Major in field required Masters in education required State-funded mentoring
Comparison of Outcomes NJ students did less well on NAEP 4 th grade math assessment in 1996 than CT In contrast to NJ, Connecticut has improved scores in recent years (Darling-Hammond, p.19)
Georgia vs. N. Carolina ( Demographically similar) Georgia Raised teacher salaries No investment in professional development Did little to increase expectation for preservice or inservice preparation Less likely to have mentoring opportunities for new teachers N. Carolina Raised teacher salaries Raised standards for teacher preparation and licensing Invested heavily in teacher development Increased education and subject matter requirements Rigorous teacher examinations for licensing Required national accreditation of education schools
Comparison of Outcomes North Carolina's students now perform substantially better on the NAEP assessments than those in demographically similar Georgia, which North Carolina lagged behind in 1990. (Darling-Hammond, p.19)
West Virginia vs. Virginia (Virginia more affluent) W. Virginia Higher teacher salaries Reduced amount of required coursework on teaching and learning in undergrad programs Encouraged extensive hiring of alternate certification candidates Less rigorous hiring standards Virginia Raised teacher salaries Improved teacher education and licensing requirements Teacher Ed. Programs must meet national accreditation standards Introduced professional development program Mentoring program
Comparison of Outcomes Very poor W. Virginia ranks as well in elementary math as Virginia, whose students are much more affluent (Darling-Hammond, p.20)
Potential Policy Implications Implementation and environment is key! Competence in subject-matter coupled with educational coursework Certification testing Standard setting and enforcement mechanisms Teacher education accreditation District hiring standards Improve incentive structure for teacher compensation Mentoring during the induction period for new teachers Professional development programs Collegial work settings to share best practices Strong liberal arts undergraduate education Great Expectations Initiative