Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 225
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. V DATA BASES Part V addresses issues entailed in developing and maintaining data bases supportive of TSDQ modeling and research. Relevant data bases at the state, regional, and national levels are identified, reviewed, analyzed, and compared in terms of quality, comprehensiveness, timeliness, and inclusion of longitudinal information. Supplementary material describing relevant national data bases is contained in Appendix B, and the identification of variables relevant to TSDQ models and research contained in these national data bases is reported in Appendix C. In addition, the paper by Murnane and the discussion by Grissmer illustrate the use of state data bases to investigate important TSDQ problems. The information contained in teacher data bases is also considered in relation to the information needs of policy makers concerned with teacher work force issues. Finally, the adequacy of existing data bases and the need for improved or additional data bases is examined.
OCR for page 226
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. This page in the original is blank.
OCR for page 227
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. State Data on Teacher Supply, Equity, and Qualifications ROLF K. BLANK State departments of education have been active in building and improving management information systems. An important part of most state systems is a data file of teachers and other professional educators in public elementary and secondary schools. States typically collect and report information on characteristics of teachers in response to a state law requiring such data. But state information systems are designed for a multitude of purposes, such as reporting data to federal agencies, monitoring the quality of teachers hired by local districts, and analyzing teacher supply and demand in the state. The Council of Chief State School Officers (CCSSO), the national association of state superintendents and commissioners, has analyzed the types of data on teachers that are available from state education information systems. The CCSSO State Science/Mathematics Indicators Project has developed a system of indicators of the condition of science and mathematics education that are partly based on state-collected data, along with information about teacher characteristics (Blank, 1986; Blank and Dalkilic, 1990). The CCSSO Education Data Improvement Project has worked with states and the National Center for Education Statistics to standardize the definitions for state-collected teacher data and to improve the quality of state data (CCSSO, 1988). Information gathered from states through these two projects form the basis for this paper. The characteristics of teachers that are included in state information systems vary widely, and the definitions used to collect these data also vary among states (Blank and Espenshade, 1988; CCSSO, 1988). The uses of state data on teachers for producing statistics on teacher supply and demand
OCR for page 228
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. also vary widely (National Research Council, 1987). As of 1987, 34 states had developed state reports on teacher shortages or supply and demand, while 26 of them indicated that such a report is provided annually (Blank and Espenshade, 1988). If state data are to be considered as a source for national analyses of the teaching force or for projections of teacher supply and demand, it is important to consider the availability, quality, and utility of state teacher data. This paper addresses the following four questions concerning state data on elementary and secondary teachers in public schools: What is the breadth of teacher data are available from state information systems? What is the quality and timeliness of the state data? To what extent are state data on teachers linked through one or more data files so that analyses can be conducted? How can state data be used for national-level analyses? BREADTH OF STATE DATA ON TEACHERS CCSSO conducted a survey of all state departments of education in 1987 to determine the availability of state-collected data on a variety of possible indicators of science and mathematics education (Blank and Espenshade, 1988). The results were used to select a small set of indicators that were subsequently developed as state-by-state and national indicators (Blank and Dalkilic, 1990). In 1988 the CCSSO Education Data Improvement Project collected information from state departments of education on the definitions used to collect and report on characteristics of education staff (CCSSO, 1988). The information was used to consider revisions in procedures and data elements reported to the National Center for Education Statistics in the Common Core of Data. Information on availability of state data from the two CCSSO projects was integrated to produce a summary of the number of states that collect data on elements of teacher supply, demand, and quality. The delineation of these three categories of teacher data and the identification of desired variables for measuring supply, demand, and quality followed the analysis and recommendations of two National Research Council reports (1987, 1990). The results of the NRC survey on ''availability of state data on public school professional personnel'' (1991) were used to check the results of the CCSSO surveys. Table 1 lists the number of states that collect data on demographic characteristics of teachers, measures of teacher quality, and elements of teacher supply and demand. CCSSO surveys indicate that 49 states have an automated data file on the characteristics of teachers (and other professional
OCR for page 229
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. TABLE 1 State Data on Teachers: Number of States that Collect Data on Elements of Teacher Supply, Demand, and Qualifications as of Spring 1987 (N = 50 states and District of Columbia) Number of States Teacher Assignment by: District 49 School 48 Grade level of assignment 43 Subject of current assignment 48 Demographic Characteristics of Teachers: Date of birth (age) 40 Sex 47 Race/Ethnicity 44 Teacher Qualifications: Education attainment (degree status) 47 Academic major (bachelor's degree) 40 Certification type 41 Subject/field of certification 40 Years of professional experience: 40 Years of teaching experience 25 Years in current district 36 Years in other district 23 Years in current assignment 13 Teacher Demand: Pupil-teacher ratio 37 Pupil-teacher ratio by subject 22 Enrollment projections 36 Emergency/provisional certificates 41 Positions vacant or withdrawn, or filled with non-certified teacher or substitute 31 Teacher salary (contract or base) 47 Teacher Supply: New college graduates in education 33 New graduates with non-education majors and certified to teach 20 In-migration of teachers from other states 27 Re-entrants into teaching 21 Entrants from other occupations 17 Continuing teachers/teacher attrition 32 Continuing teachers in new subject/field 27 Teachers retiring 37 New Hires: Occupation prior year 10 Location of occupation prior year 7 SOURCE: State Science/Mathematics Indicators Project, 1987 (unpublished data). Council of Chief State School Officers, Washington, D.C.
OCR for page 230
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. education personnel). All of the variables listed in Table 1 are generally not included on the same data file. There are three kinds of data files on teachers in most states: (a) current teacher file (including demographics, district and school, current assignments, education attainment, contract salary, and experience); (b) certification file (including type of certification and subjects or fields of certification); and (c) state retirement system file (including year of birth and year entering the system). Most states can link the files through the teacher social security number or other identifier. The items listed in Table 1 under Teacher Demand and Teacher Supply may be from several data sources at the state level, including student membership counts by grade and higher education data, as well as from computations or analyses of data on the current teacher file, such as identifying the source of new hires. In 1987, 34 states reported having developed estimates of teacher shortages or supply and demand, but states were not asked to identify which of the data elements were used in these estimates. Thus, the list of elements by number of states in Table 1 shows the potential of state data for analyses of teacher supply, demand, and quality. Over 40 states collect data on teachers' demographic characteristics, education attainment, and certification status. Cross-tabulations of data on combinations of these variables are difficult to accomplish in some of the states. For example, the Science/Math Indicators Project found that only 30 of the 40 states could report the number of teachers assigned in mathematics and science by their certification status in the subject(s) they were teaching. QUALITY AND TIMELINESS OF TEACHER DATA State departments of education collect data on characteristics of teachers for their current teacher file through several different data collection approaches. One approach is an annual survey form developed by the state for each individual teacher in the state. This form asks for new or updated demographic information, current teaching assignments and student enrollments by school period, and information on education attainment and salary. This type of form is used by about 20 states, including California, New York, Connecticut, Ohio, Minnesota, South Carolina, Alabama, and Virginia. A second approach is a district or school form that lists teachers on the current state file and asks the district or school to update the existing data or provide new data on teacher demographics, subject and grade assignments, and education attainment and salary. Over half the states use this method of collecting data on teachers. A third approach is a student-based computerized information system in which information about teacher assignments and teacher characteristics is linked to a statewide data base on each student. Student records and schedules and teacher data are relayed on computer files from the school to the
OCR for page 231
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. district to the state. This system is in place in Florida and Hawaii and is being developed in Georgia, Texas, North Carolina, and other states. Certification information is collected and maintained in a separate certification file in most states. A majority of states have the current teacher file and certification file linked by social security number and annually check the certification status of each teacher. Some states have only recently computerized their certification file and some states do not cross-check assignments and certification. There are a variety of issues in assessing the quality of state data on teachers. One issue is the response rate, i.e., the proportion of teachers in classrooms for which data are reported to the state. The Science/Math Indicators Project asked states to report on response rates and data editing procedures. About one-third of the states collect data directly from teachers, and response rates for those states were from 98 to 100 percent. However, both these states and states that collect data through schools and districts must depend on lists of current teachers and new teachers that are provided by schools or school districts. A second issue is the reliability of the data that are reported by teachers and stored on state files. States reporting data on science and mathematics indicators were asked to report on data editing procedures in 1989–90. The large majority of states use computer edits, logic checks, and external validity comparisons. However, there are states that use few editing procedures and the quality of the data could be questioned. Another issue is the comparability of teacher data between states due to differing definitions of teacher characteristics. The Education Data Improvement Project analyzed state definitions for 22 data elements concerning professional staff and found some variation for all the data elements. For example, variation in the definition of teacher age was very slight. The definitions and categories of types of teaching certificates are different in almost every state, although all but three states' definitions could be placed in three categories: regular/standard, probationary, and temporary/provisional/emergency. CCSSO worked with groups of state representatives to develop consensus definitions for data elements recommended to the Common Core of Data. Then, differences in definitions were reported to states so that data collection could be standardized, or, in some cases, a crosswalk could be designed for analyzing a state's data in relation to the consensus definition. Another criterion of completeness of teacher data is the response rate for each requested data element. The 1989–90 data reported on science and mathematics indicators also provides information on this question. A total of 34 states reported on the age of teachers assigned in high school science and mathematics. Among these states, I percent of data on teacher age was missing. One state had 8 percent missing data on teacher age. A total of 32
OCR for page 232
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. states reported data on the race/ethnicity of science and mathematics teachers, and a total of I percent of these data were missing. One state (Hawaii) had 5 percent missing data on race/ethnicity. All of the states with teacher files collect or update the data at least annually. Most states ask that data be reported early in the fall of each school year. The National Center for Education Statistics requests that Common Core of Data on education staff be reported as of October 1. States enter, edit, and clean the data between October and March. Some states do not have the data ready until June. For the CCSSO Science/Mathematics Indicators Project, 43 states reported data on teacher characteristics for the preceding school year by the end of July. LINKED TEACHER DATA FROM STATES Table 1 reports the number of states that collect and file data on elements of teacher quality, demand, and supply. An important question for any researcher or policy analyst is whether these data can be linked together for purposes of analysis. For many analyses, the ideal situation is to analyze teacher characteristics at the teacher level, such as teacher assignment by teacher age. Some analyses could be with data aggregated at the school, district, or state level. For example, in analyzing teacher demand and supply, it is useful to compare the number of new hires in a state who are first-time teachers with the previous year's number of college graduates who were certified to teach. If possible, however, it is most desirable to have data on teachers that can be analyzed at the teacher level. To gain some perspective on the availability of state data on teachers that are linked at the teacher level, Table 2 shows a state-by-state listing of teacher characteristics and the extent to which the data are linked. The first column indicates that 44 states have data on teacher assignments by field or subject (e.g., subject=science, field=biology). Of these states, 36 states can analyze assignment by age, 40 states assignment by sex, and 33 states assignment by race/ethnicity. The second column shows that 30 states can determine the number of teachers assigned to a subject/field that are certified in the subject/field. The third column indicates that 35 states have data on student enrollments in secondary-level courses and 19 states (with links between teacher and student data) can analyze the proportion of students in a course taught by a teacher certified in their assigned field. EXAMPLE OF USES OF STATE DATA FOR NATIONAL ANALYSES The CCSSO Science/Mathematics Indicators Project gave high priority to developing three types of indicators of teacher work force: (a) supply,
OCR for page 233
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. TABLE 2 State Data Linked to Individual Teachers (Fall 1989) State Teacher Assignment by Field/Subject by Age (A). Sex (S), Race/Ethnicity (R) Teacher Assignment by Certification Field Secondary Course Enrollment Alabama Field by A,S,R Yes Yes, link Alaska No No No Arizona Subject by S,R No No Arkansas Field by A,S,R Yes Yes, link California Field by A,S,R Yes Yes, link Colorado Subject by A,S,R Yes No Connecticut Field by A,S,R Yes Yes, link Delaware Field by A,S,R Yes Yes District of Columbia No No Yes Florida Field No Yes, link Georgia No No No Hawaii Field by A,S,R No Yes, link Idaho Field by A,S,R Yes Yes Illinois Field by A,S,R Yes Yes, 5 yrs. Indiana Field by A,S,R No Yes Iowa Field by A,S,R No Yes, link Kansas Field by A,S,R No Yes Kentucky Field by A,S,R Yes Yes, link Louisiana Field No Yes Maine Field by A,S,R No No Maryland Subject by A,S,R Yes No Massachusetts Field No No Michigan Field by A,S,R No No Minnesota Field by A,S Yes Yes, link Mississippi Field by A,S,R Yes Yes, link Missouri Field by A,S,R Yes Yes, link Montana Field by A,S,R Yes Yes Nebraska No No Yes Nevada Field by A,S,R Yes Yes, link New Hampshire Field by S No No New Jersey Field by A,S,R Yes No New Mexico Field by A,S,R Yes Yes New York Field by A,S Yes Yes, link North Carolina Field by A,S,R Yes Yes, link North Dakota Field by A,S,R Yes Yes, link Ohio Field by A,S,R Yes Yes, link Oklahoma Field by A,S,R Yes Yes Oregon Field by A,S Yes No Pennsylvania Field by A,S,R Yes Yes Rhode Island Field by A,S,R Yes No South Carolina Field by A,S,R Yes Yes, link South Dakota Field by A,S Yes No Tennessee Field by A,S Yes Yes
OCR for page 234
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. State Teacher Assignment by Field/Subject by Age (A), Sex (S), Race/Ethnicity (R) Teacher Assignment by Certification Field Secondary Course Enrollment Texas Field by S,R No Yes, link Utah Field by A,S,R Yes No Vermont No No No Virginia Field by A,S,R Yes Yes, link Washington No No No West Virginia No No No Wisconsin Field by A,S,R No Yes, 3 yrs. Wyoming Field Yes Yes Total Field/subject = 44 Yes = 30 Yes = 35 Age = 36 Sex = 40 Race = 33 SOURCE: State Science/Mathematics Indicators Project, 1990 (unpublished data), Council of Chief State School Officers, Washington, D.C. (b) equity, and (c) qualifications. Another priority area for state indicators of science and mathematics is school conditions that affect teaching and learning. The CCSSO plan for state-by-state indicators of science and mathematics is based on cross-sectional data that can be compared by state and tracked over time. Some desirable indicators of teacher quality that require more complex data or qualitative measurement were not selected, such as state-by-state projections of teacher supply and demand and quality of instruction in the classroom. Other possible indicators of teacher quality, such as degree level and years of experience, were not selected because there is less evidence of a relationship to outcomes in science and mathematics. States reported data on teachers using a common reporting system designed by the Science/Mathematics Indicators Project. CCSSO also conducted state-by-state analyses of the Schools and Staffing Survey of NCES. Some of the indicators are summarized in the following sections. A full report, entitled, State Indicators of Science and Mathematics Education: 1990 is available (Blank and Dalkilic). Indicators of Current Teacher Supply States reported data on the total number of teachers assigned to teach science, math, and computer science in grades 9–12 as of October 1, 1989.
OCR for page 235
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. The state teacher numbers are universe counts based on data collected through state information systems. Proportion of Teachers with Primary and Secondary Assignments The CCSSO state data reporting plan requested the number of teachers with primary and secondary assignments in each of eight subjects. The operational definition of primary assignment is a teacher assigned to one subject for 50 percent or more of teaching periods; and secondary assignment is a teacher assigned to one subject less than 50 percent of teaching periods. The state data show that 89 percent of high school teachers of mathematics have their primary assignment in mathematics. Only slightly over half (53 percent) of all teachers of chemistry have their primary assignment in chemistry, and three-fourths of all teachers of physics have their primary assignment in another field (24 percent in physics). In many schools, physics is taught by a teacher with primary assignment in chemistry or earth science. States vary in the proportion of teachers with primary assignments in science and math. For example, teachers of mathematics in Connecticut (95 percent) and Illinois (96 percent) are almost all teaching mathematics as their primary assignment, while California (68 percent primary assignment) and Utah (69 percent primary assignment) have about one-third of teachers of mathematics who have their primary assignment in another subject. Higher numbers of teachers with secondary assignments are probably due to population growth (such as in California) as well as increases in state course requirements. States with more small, rural districts, such as Arkansas, Oklahoma, and North Dakota had fewer teachers with primary assignments in any of the science fields, while states with a greater proportion of urban and suburban districts, such as Connecticut, New York, and Pennsylvania, have more teachers with primary assignments in the science fields. Age of Science and Mathematics Teachers Although the state science and mathematics indicators do not include detailed projections of teacher supply and demand, the age distributions of current science and mathematics teachers provide useful information on possible shortage fields as teachers near retirement age. The average percentage of teachers over age 50 is 20 percent in mathematics, 20 percent in biology, 23 percent in chemistry, and 23 percent in physics. The average percent of teachers under age 30 is 13 percent in mathematics, 12 percent in biology, 12 percent in chemistry, and 11 percent in physics. The age distributions of mathematics and science teachers vary widely by state in all fields. The percentage of mathematics teachers over age 50
OCR for page 274
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. science and math teachers not teaching in their main area, or conversely, teachers not having their main certification in science or math may be teaching these subjects. To the extent that (a) entering cohorts of science and math teachers teach other subjects or (b) individuals who were not initially certified in science and math are teaching these subjects, then the attrition rates for the state and national data could be different. These discrepancies emphasize the need to perform attrition analysis within a common analytic context in which similar methodologies, similar controls, similar time periods, and similar definitions of attrition can be utilized. This common analysis needs to include data from several states and national data from the SASS and TFS. Otherwise these kinds of conflicting findings will arise over and over again, when common analysis could eliminate almost all of these conflicts. There is a need to develop a complete supply and demand analysis for math and science teachers that includes identifying the sources of supply (returning, migrating, and inexperienced teachers), the paths into and out of teaching math and science, national return rates, and the effects of upcoming retirement eligibility and early retirement offers. We have not yet included these individual groups in supply and demand models to assess the potential for additional shortages. This analysis is best done by combining data from several states, along with the simultaneous use of SASS and TFS data. SASS and TFS provide nationally representative data on math and science teachers, but sample sizes for newly entering teachers are rather small. In addition, supply and demand analysis is best done using long time series data whereby trends in key variables can be analyzed. Since at the present time SASS data will cover only a short time period, state data are best used to analyze these trends. The state data are based on very large sample sizes and on moderately long histories that allow better entering cohort analysis. Supply and demand analysis also needs to model the early stage of teaching careers most accurately since this is where attrition rates are high and more unpredictable. Finally, more research is needed to discover the causes of the significant attrition differences among types of science and math teachers. One hypothesis is that it simply reflects differences in outside job opportunities (Murnane et al., 1989). The hypothesis is that mathematics teacher training may provide less job transferability than that of science training. This is especially true of those teaching lower levels of mathematics. Another hypothesis is that laboratory teaching as opposed to classroom teaching is inherently harder, and there is greater sensitivity to the quality of equipment and facilities. Survey responses (Weiss and Boyd, 1990) from science and math teachers show differences in their sensitivities to working conditions. Science teachers—but not mathematics teachers—rate facilities
OCR for page 275
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. and equipment and ''time for hands-on instruction'' as key aspects of their dissatisfaction with teaching. Finally, differences in attrition rate by subject may reflect variation in gender proportions in each area and their sensitivities. Men are more sensitive to lack of administrative support and low salaries than women (Weiss and Boyd, 1990). This sensitivity would be heightened by higher outside opportunities. Any teaching specialty with more men—other things being equal—would probably have higher attrition in situations in which where higher outside wage opportunities exist. Minority Teachers Declining proportions of minority representation in the overall teacher force could occur at a time when minority student enrollments are rising. The cause of lower minority teacher proportions may be more attributable to lower minority college enrollments and choice of education as a career rather than lower proportions of minority certified applicants obtaining jobs (Murnane and Schwinden, 1989). The studies of state data have not generally focused on the question of differences in attrition rates among racial/ ethnic groups. Analyses of SASS and TFS generally show white and black rates to be similar, but lower attrition for Hispanic teachers (Bobbitt et al., 1991). More specific analysis of minority supply and demand is required. Hispanic teachers especially merit analysis given the recent immigration trends. However, the sample sizes generally are small except in certain states. Two states stand out as having large enough populations of both black and Hispanic teachers to justify an analysis: New York and Texas. Fortunately, both of these states have fairly good data, and it has been edited and prepared for analysis (Texas by RAND and New York by the University of Massachusetts). Some joint analysis of these two larger states with the SASS and TFS data would provide a much clearer picture of minority teacher supply and demand behavior. Teacher Early Retirement and Teacher Demand The teaching force is unbalanced with respect to age and experience. Younger teachers—those under 35—are a smaller portion of the teaching force than at any time in the last 25 years, and half of all teachers are over 42, making them retirement eligible at age 55 within 13 years at the latest. An important supply and demand question is how soon these retirements will occur, and thus when replacement will be needed. Current retirement patterns show a strong tendency for teachers to stay until 62 or 65. If this is the case, then demand for new teachers will increase more slowly. Budget problems in states could make early retirement offers very attractive—in fact epidemic. Replacing older teachers with younger teachers significantly
OCR for page 276
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. reduces education costs, even with somewhat increased retirement costs. Retirement costs are not generally paid out of operating budgets, which makes early retirement even more attractive to educational administrators. Research is needed on the precise patterns of teacher retirement and the effects of early retirement offers on the decision to leave teaching. Massive early retirement could increase demand for younger teachers significantly. The federal government could also provide states with research on the effects of different types of early retirement offers. Research is also needed on quality issues inherit in early retirement offers. Is the tradeoff of younger for older teachers likely to increase or decrease teacher quality? This research would again best be done in the context of combining data from several states with SASS and TFS data to find whether retirement patterns are similar across states. However, the state data may be especially valuable here because early retirement plans will be state specific. The effect of early retirement plans on retirement age could be obtained using time-series state data and/or comparing retirement behavior across states. Alternately, future SASS and TFS questions or a separate survey of teachers with regard to early retirement plans and the factors behind retirement decisions might be important. The Declining Reserve Pool Perhaps the most ominous trends for future shortages is the fact that the supply of returning and migrating teachers will be declining in future years. We currently depend on these teachers to fill about 50 percent or more of vacancies in any year. If there are fewer returning and migrating teachers, then we will need more younger teachers. Returning and migrating teachers will decline because of simple demographics. Teachers who return to teaching leave teaching most often between 25–35 years of age. Teachers over age 40 leave teaching less frequently. So as the average age of the teaching population increases, there will be a smaller reserve pool of teachers. Teacher migration also peaks during the 25–35 age span. Since there are going to be fewer teachers in this age span as the teaching force ages, there will be fewer returning and migrating teachers to fill vacancies. We need more research on the patterns of returning and migrating teachers to determine the precise decline in these pools over the next 5–15 years. Data available from several states could be readily utilized to explore these patterns and the subsequent decline in the reserve pool. SASS and TFS data will have only limited utility here because they will not capture the longer time period in which many teachers return. Estimating the changing return rates of teachers would enable better estimates to be made of the reserve pool, and the timing of much stronger demand for new teachers. If this occurs about the time of massive early retirements, a problem in supply
OCR for page 277
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. could result. So this research needs to be combined with the early retirement research to determine the relative timing of the two phenomena. Supply and Attrition of High-Quality Teachers Research has generally established the lower entrance rate and higher attrition rate of teacher graduates who score well on aptitude tests (Vance and Schlechty, 1982; Murnane et al., 1989; Manski, 1987; Murnane and Schwinden, 1989). The latter study however distinguishes between white and black applicants and shows that black applicants show the opposite effect, namely that higher NTE scores lead to increased chances of entry into teaching. There have been many programs within colleges to attract better students, and better induction programs into teaching may lower attrition rates (Hudson et al., 1991). But research is needed to discover differences in quality in teachers who stay and leave and the role of salary and working conditions in these decisions. There are several approaches that could be tried—case studies, surveys, and analysis of state data that include teacher test scores or teacher evaluations. FUTURE RESEARCH DIRECTIONS Much of the work of the last few years has focused on the collection, editing, and initial analysis of very large data bases. The SASS and TFS surveys have been designed and fielded. These surveys are an invaluable addition to the study of teacher supply and demand. The analysis of this data base is only beginning, and other waves of data will eventually be added to the 1987–88 and 1988–89 data. SASS and TFS provide the only nationally representative data on teachers available for this purpose. The sample size of the survey is large and adequate for many purposes, although analysis of entering cohorts (or any age group such as retirees) within specialty categories can be a problem. The main drawback for analyzing teacher supply and demand issues is that it contains no time-series data, and therefore trends in key variables cannot be tracked. There are three research centers that have invested in analysis of state data: Harvard University, the University of Massachusetts, and RAND. Harvard University has analyzed data from Michigan and North Carolina. The University of Massachusetts has data from eight New England states, and RAND has data from Indiana and Texas. A sizable investment has been made simply to assemble these data and prepare them for analysis. While some analysis has been done and published, this analysis has not yet tapped the potential of the data bases, partly because the initial costs of preparing the
OCR for page 278
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. data was underestimated by all, and fewer resources were then available for actual research. State data has the advantage of large samples (Texas has 180,000 teachers and New York has 240,000). The combined state data bases that have been prepared for analysis contain the records of approximately I million teachers—about 40 percent of U.S. teachers. These large sample sizes are critical when studying the behavior of age subgroups of teachers such as minority and science and math teachers. Data drawn from surveys is simply too expensive to allow for the sample sizes necessary to study many subgroups. In addition, state data are the only available data from which time-series and long-term longitudinal data can be constructed. Such data, as opposed to data collected at a single point in time, are critical to assessments of teacher supply and demand. State and SASS/TFS data are basically complementary, and joint analysis can enhance considerably the analysis of several educational policy issues. I will provide two examples in which joint analysis can considerably improve our understanding of issues. The first is the question of attrition of science and math teachers described above. As reported, several conflicting results arose from analysis of the several data bases. It will be difficult to determine if these differences are an artifact of the analysis method or the time period or represent real differences among states and national samples in teacher behavior. However, joint analysis of data will be able to address many of these issues and avoid many unnecessary conflicts. A second example is the analysis of early retirement behavior among teachers. The SASS data will be able to determine age-specific retirement rates that are nationally representative. However, since variation in pensions and early retirement offers exist at the state level, the sample sizes of retirement-eligible teachers within states that make such offers will be too small to determine how offers affect behavior. However, state data with its large sample sizes can track retirement rates over long time periods and reveal differences in retirement rates when offers are made. These data can also be used for interstate analysis of retirement rates and the relationship of these rates to differences in pension rules and amounts. Research with data from several states is showing both similarities and differences in teacher supply and demand factors. There is similarity in age distributions and in some general attrition patterns among age groups and subject areas. There appear to be differences in (a) dependence on returning and migrating teachers to fill vacancies, (b) attrition levels, (c) capacity to generate new teachers, (d) proportions of minority teachers, and (e) early retirement programs and retirement patterns. These differences can arise from local or district policies, from state policies, and from differences in behavioral characteristics among teachers in different states. Understanding these differences is important because teacher supply
OCR for page 279
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. and demand factors, and the potential for shortages, will vary by state. It is also important because some states and districts have better policies and practices that can be identified and exported. Certain questions can also be better addressed by certain state data than others partly because data are available from some states for longer periods. Also, some states have implemented programs such as early retirement offers, while others do not. The critical point here is that the analysis of state and national data needs to be done within an integrated framework that involves the following: Framing the same hypothesis as the basis for the research, Estimating attrition and other factors with the same model specifications, Using common definitions of variables across states, Estimating for similar time periods whenever possible, and Recognizing key differences across states that can explain differences in results. If analyses of state and national data are done within this framework, then key national trends can be distinguished from within state trends. Trends that are supported by analyses of data from several states are more persuasive than from one state only. Moreover it is usually impossible to compare results derived from independent teacher supply and demand models because of different model specifications and variable definitions. This type of analysis would also allow us to identify key differences caused by particular state policies. A research consortium also has significant cost advantages. Such analyses accomplished jointly would be less costly than if done independently and would result in a more interpretable set of findings. Exchange of models, ideas, hypotheses, and data handling techniques would benefit all research groups and would yield information of greater utility for education policy makers. REFERENCES Bobbitt, S.A., E. Faupel, and S. Burns 1991 Characteristics of Stayers, Movers, and Leavers: Results from the Teacher Followup Survey, 1988–89. NCES 91-128. Washington, D.C.: U.S. Department of Education, National Center for Education Statistics. Grissmer, D. W., and S. N. Kirby 1991 Patterns of Attrition Among Indiana Teachers: An Executive Summary . R4167-LE. Santa Monica, California: RAND. 1992 Patterns of Attrition Among Indiana Teachers. R-4076-LE. Santa Monica, California: RAND. Hudson. L., D.W. Grissmer, and S.N. Kirby 1991 Entering and Reentering Teachers in Indiana: The Role of the Beginning Teacher Internship Program. R-4048-LE. October. Santa Monica, California: RAND.
OCR for page 280
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. Manski, C.F. 1987 Academic ability, earnings, and the decision to become a teacher: Evidence from the National Longitudinal Study of the High School Class of 1972. In D. A. Wise, ed., Public Sector Payrolls. Chicago, Illinois:University of Chicago Press. Murnane, R.J., and R.J. Olsen 1989a Will there be enough teachers? American Economic Review Papers and Proceedings 79:242-246. 1989b The effects of salaries and opportunity costs on duration in teaching: Evidence from Michigan. Review of Economics and Statistics 11:347-352. Murnane, R.J., and M. Schwinden 1989 Race, gender, and opportunity: Supply and demand for new teachers in North Carolina, 1975–1985. Educational Evaluation and Policy Analysts 11:93-108. 1990 The effects of salaries and opportunity costs on length of stay in teaching evidence from North Carolina. Journal of Human Resources 25:106-124. Murnane. R.J., J.D. Singer, and J.B. Willet 1989 The influences of salaries and "opportunity costs" on teachers' career choices: Evidence from North Carolina. Harvard Educational Review 59:325-346. Vance, V.S., and P.C. Schlechty 1982 The distribution of academic ability in the teaching force: Policy implications. Phi Delta Kappan 64:22-27. Weiss, I.R., and S.E. Boyd 1990 Where Are They Now? Chapel Hill, North Carolina: Horizon Research, Inc.
OCR for page 281
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. General Discussion The general discussion focused primarily on data bases relevant to teacher supply, demand, and quality (TSDQ) organized at three different levels: state, regional, and national. The characteristics of such data bases established at each level were considered, including information sources, methods used, and their strengths and limitations. The following paragraphs summarize the discussion pertaining to the five state data bases, national data bases, the relationship between data bases and modeling, and funding for data base development and maintenance. Existing state data bases relevant to TSDQ typically have been derived from state administrative records containing information about teachers, students, and schools. States rarely use sample surveys of teachers to obtain TSDQ data. Several factors cause development of a teacher data base within a particular state to be difficult, time-consuming, and expensive. These factors include the sheer volume of detailed information available from multiple in-state administrative records, accessibility of data from these sources, and differences in definitions and time periods covered. Once assembled, data bases support the investigation of a wide range of important TSDQ issues because of three characteristic strengths. First, state records contain a wealth of detailed data, often much greater detail than can be gathered reasonably by sample surveys. Second, these records are maintained year-by-year, thereby permitting examination of trends over time and permitting cohort studies of teachers as their careers develop. Third, state records incorporate entire populations of teachers, students, and schools. In contrast, surveys are usually based on samples of these populations, and limited sample sizes often do not permit the disaggregation needed for de-
OCR for page 282
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. tailed analyses. Thus, at their best, state data bases provide opportunities for detailed, longitudinal analyses of TSDQ phenomena in teacher populations. Unfortunately, state data bases are beset by a number of limitations. One is that teachers of a single state, or a combination of a few states, cannot provide national estimates. Furthermore, variations among states in definitions of variables as well as policies pertaining to teachers limit the generalizability of state-level findings. For example, definitions of second-year algebra and state policies pertaining to such factors as teacher certification, benefits packages, and collective bargaining vary a great deal among states. The conference participants discussed approaches to address the problem of differences in definitions. One is that cross-walks are sometimes successfully devised to bridge such differences. Another is that NCES is revising and developing, at the urging of many states, its handbook of definitions of teacher variables. Beginning with state and SASS definitions, the objective is to develop consensus definitions among the states and NCES and to use these in data base development. Of course, the definitions of many variables already are consistent, and much TSDQ research with data bases from different states has yielded consistent findings. Another major problem is the quality of many state data bases relevant to TSDQ. In this context, quality includes the dimensions of data accuracy; breadth of variables covered; completeness of data for variables covered; timeliness of data; and difficulty in linking, and consistency within a state of, definitions across different sets of administrative records that contain the basic information used to create teacher data bases. It was generally agreed that most states will have to expand a great deal of effort to establish high-quality teacher data bases. However, this is a costly enterprise, which many states cannot afford. A final limitation of state data bases is their inability to provide information on teachers migrating out of state. When a teacher discontinues teaching in the public school sector of a particular state, state records typically do not indicate whether the teacher left the profession, migrated to a school in another state, or transferred to a private school. Without this information, attrition studies lack precision and may yield misleading interpretations. Turning next to the regional level, it was recognized that the only regional TSDQ data base currently operational has been developed for New York and the New England states by the Massachusetts Institute for Social and Economic Research (MISER).1 In cooperation with these states, MISER has taken the lead in assembling seven useful state data bases by extracting information available from multiple sources within each of the seven states constituting its region. However, MISER has not generated "original" teacher data through sample surveys or other means.
OCR for page 283
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. With funding from federal, state, and private sources, MISER worked closely with the seven states of the northeastern region to improve and expand considerably the teacher data bases for each state, and has begun to create linkages among them for analytic purposes. These data bases probably go well beyond what the cooperating states would have developed on their own initiative. This regional interest and activity has had several beneficial results. First, it has stimulated these states to improve their collection and management of teacher data. As one commentator noted, this unlikely result suggests a new data base principle, namely, that "bad data begets good data," at least when comparisons and competition among states can occur. Second, the regionalization of the data has permitted the study of cross-state migration of teachers. Third, the generation of comparative state-by-state teacher data has stimulated intense interest among the chief state school officers and has led to policy changes such as regionalizing teacher credentialing. With respect to the national level, a number of data bases useful for analyses of TSDQ issues now exist. Only one or two of these data sets are derived from data collected and reported by state or local education agencies. Most national data relevant to TSDQ have been generated by sample surveys with questionnaires—the prime example being the Schools and Staffing Survey (SASS), and its longitudinal component the Teacher Followup Survey (TFS), of the National Center for Education Statistics (NCES). Since national and state data bases (including the one regional data base) differ greatly in their data sources and methods, their strengths and limitations also remarkably differ. For the most part, the strengths of national data bases relevant to TSDQ are the limitations of state data bases, and vice versa. At their best, national data bases, such as SASS, include a great deal of information about probability samples of teachers for each state and for the nation as a whole. In addition, the surveys define variables uniformly across all units sampled and use standardized procedures. Therefore, these surveys provide a national perspective of the teaching force, opportunity for state-by-state comparisons, and a basis for analysis of teacher migration among states. National data bases, however, are as yet very limited in the extent to which they can support time-series analyses for teachers. In addition, practical limitations on the length of survey questionnaires do not permit the level of detailed information gathering that is typical of administrative records from which state data bases are derived. Finally, resources available for national surveys limit sample sizes, thereby either excluding many important cross-tabulations of data or producing estimates with large standard errors. Since the best currently available state and national data bases are complementary in their respective patterns of strengths and limitations, there is
OCR for page 284
Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. great advantage not only in having both types, but also in being able to link data bases at these two levels. From the perspective of NCES, there is considerable potential for linking SASS with state data bases, and it is important to work toward achieving this dual strategy. The reciprocal relationship between state data bases and MISER's regional TSDQ model was also discussed. It is readily apparent that data limitations (both quality and availability) constrain the variety of analyses demonstrated to be important by the model. However, this very fact was helpful in identifying data gap problems with the quality of state data. When these limitations became apparent, some states were stimulated to improve their state data collection and management. Conferees recognized the limitations of funding data base development. States need to make major improvements in their teacher data collection and management. This is expensive. Several participants argued that all state money available for teacher data base development should be devoted to improving within-state data, and that none of it should be allocated to development of regional data bases. Even if this principle were adopted, most states probably would not have sufficient resources to produce high-quality data bases. It was therefore suggested that NCES, or some other federal agency, provide states with capacity-building grants. While this might seem to exclude the funding of regional data bases, it was pointed out that resources available to MISER to develop data in the Northeast were devoted primarily to working with individual states to improve and expand their data, and that this was carried much further than would have occurred had there not been a regional presence. NOTE 1. In addition, the Southern Region Education Board (SREB) has explored the feasibility of developing such a data base for the Southern region of the nation. In a report delivered at the conference, Lynn Cornett and Robert Stoltz, both of SREB, stated that 11 SREB states were interested in cooperating in an effort to create a regional TSDQ data base derived from the records of each state. A preliminary survey of these 11 interested states found that 7 had excellent data for this purpose, one had satisfactory data, and three did not seem to have the minimum essential data available in readily usable form. SREB staff prepared a cost analysis for the project, but it is unlikely that it will start in the near future. There are several reasons for the delay: lack of state funds due to budget shortfalls, lack of high-level strong advocates within state education departments, lack of an immediate crisis involving TSDQ issues, and lack of external pressure on state education departments to produce TSDQ data and analyses. In view of these constraints, SREB is exploring the feasibility of an incremental low-cost approach to initiating the development of a TSDQ data base and model for the southern region of the nation.
Representative terms from entire chapter: