Improving State Need Assessments of Secondary Science and Mathematics Teachers: Challenges, Possibilities, and Recommendations
Michael B. Allen
The urgency of present problems that schools may be having in adequately staffing their science and mathematics classes makes states’ assessments of their current need1 for science and mathematics teachers the obvious first step in addressing the inability of teacher supply to satisfy teacher demand. A current teacher shortage may not simply be a reflection of present realities, however, but also may be an indication of persistent trends or emerging patterns that need to be understood and confronted. In addition, the current status of teacher need reflects the impact of various state and district policies and practices on the teacher supply and demand situation.
The repertoire of responses available to address an identified immediate need for teachers is extremely limited, however. There is simply not time to make changes in policy, or even significant changes in practice, which might ameliorate a shortage problem. Moreover, effective policies cannot be designed only with an eye to current realities; they must look to the future and take into consideration anticipated changes in the situation the policies are intended to address. Thus, projections of the future need for science and mathematics teachers are the necessary basis for effective policy responses.
Projections of the future need for teachers are significantly more challenging, however, than assessments of current need. First, they face the inherent uncertainty of the future. Estimates of current need and identified differences in the profile of teachers between schools and districts (e.g., in age, experience, certification status, and attrition rates) are important in alerting educators and state officials to current staffing problems and inter-district inequities in teacher quality that may be chronic and likely to persist into the future. If current need data are consistent with historical trends, the data can be taken as an indication that the situations they describe can be expected to continue into the future. But demographic, economic, political, and other realities may in fact conspire to make future needs substantially different.
The attempt to account for such contingencies in projections of the future need for teachers introduces a level of complexity into the calculation that becomes the second challenge faced by future need forecasts. The effort to collect and synthesize more data about the future in and of itself exposes the resulting estimates to greater sources of error and inaccuracy. For example, the relatively straightforward identification of classes without adequately qualified teachers – the essential data point for estimates of current need – must be supplemented in projections of future need by an estimate of future teacher demand both statewide and, ideally, by district.
This future demand forecast is complex and contingent upon several uncertain and constantly-changing variables. In particular, it involves tricky demographic projections of the number of students who can be expected to enroll in science and mathematics courses in various subjects. Such projections require data on historical course-taking trends among the population of students, trends which are not linear but reflect the changing socioeconomic composition of students and differing (and evolving) course-taking patterns among the various student sub-groups. Furthermore, enrollment projections for science and mathematics need to consider the impact of any impending curricular revisions or changes in graduation requirements.
A projection of future teacher need, however, requires not only a forecast of future teacher demand but also an estimate of future teacher supply – i.e., of the number of adequately qualified teachers available to fill the courses to be offered. And this involves the synthesis and collection of still more often-slippery data. Some of these data, such as the age distribution and the attrition and retirement rates of teachers in a state are comparatively easy to collect. But other data, such as the number of teachers in the teacher “reserve pool” who are available to fill needed positions, are much more elusive and may require calculations based on complex econometric models that are idealized and thus imperfect approximations of human behavior. Moreover, some relevant factors, such as the labor market for teachers, are much more volatile than the factors on the demand side and thus much more difficult to predict with confidence.
Finally, projections of future teacher need integrally involve not only forecasts of supply and demand but also the complicated consideration of teacher quality. Truly satisfactory estimates must address any shortcomings in the qualifications of the current and/or projected supply of science and mathematics teachers – whether statewide or local. And they ideally should address not only the qualifications appropriate to the current requirements of the secondary school curriculum but also the qualifications necessary to handle what could well be more rigorous science and mathematics requirements and courses in the future.
These significant challenges should not be taken to imply, however, that efforts to predict the future need for science and mathematics teachers are fruitless or that the inherent unreliability of such estimates is so great that one forecasting method cannot be judged more trustworthy or useful than another. States willing to employ high quality data and sound statistical methods to undertake a rigorous and thorough analysis of their current and future need for teachers can be confident both that their projections have reasonable reliability and that policies developed in response to those projections have that much greater chance of success.
The express purpose of Projecting a State’s Future Need for Science and Mathematics Teachers is to offer guidance in generating rigorous forecasts of a state’s near-term and longer-term need for science and mathematics teachers. Although the guidance includes suggested steps that should be taken, it is as much an orientation to the theoretical and practical challenges involved as it is any sort of recipe. The assumed focus of the discussion is on need forecasts for a state’s public schools because it is a state’s or district’s direct responsibility to address the need for teachers in those schools. At a time, however, when many districts increasingly rely upon private schools – especially private charter schools – to provide educational services, the inclusion of those schools in a supply and demand forecast might be worth considering.2 The focus here is also limited to secondary science and mathematics because the need for teachers at the secondary level is more acute and the related policy issues are distinct from those in elementary education.
To develop a solid projection of a state’s future need for science and mathematics teachers requires that the state have reliable data on both teacher supply and teacher demand – ideally both for individual districts and for the state as a whole. The “Basic Data” are the minimum kinds of data required to develop a reliable first-order estimate of the state’s current unfilled need for teachers. The “Bonus Data” are data that, if available and reliable, will enable states to refine that first-order estimate. The purpose of collecting much of the data suggested here is to establish historical trends and baselines for future projections. Thus, any additional information (e.g., about emerging economic conditions, changes in student educational requirements, or significant initiatives to increase the production and retention of science and mathematics teachers) that suggests future deviations from historical patterns is extremely important to bring into the supply and demand calculations. Clearly, states and districts ultimately must make a need determination with the best data available, even if they do not meet the ideal for quality or scope.
1. To Determine Teacher Demand:
2. To Determine Teacher Supply:
In addition to the data related to developing rigorous forecasts of teacher supply and demand, a good analysis of a state’s currently employed teacher workforce can provide a few key data points that offer a preliminary indication (an “early warning”) of any emerging need for science and mathematics teachers that is likely to be particularly acute either across the state or in particular districts. One important caveat, however, is that data on a single year may be aberrant and not necessarily a sign of a trend; it is advisable, if possible, to collect these same data points over the past 3-5 years.
If a high proportion of teaching slots are filled with new hires – especially as a multi-year trend – this may be an indication of a serious turnover problem. This indicator is especially useful when comparing districts to one another or individual schools within a district.
This is a proxy measure of the quality of the teacher workforce. The higher the proportion of teachers with limited qualifications, the greater the concern about their quality. As with the proportion of new hires, this indicator is especially valuable in comparing the quality of teachers between districts or schools.
The greater the proportion of teachers nearing retirement in a state or district, the more the state or district needs to be concerned about recruiting and developing new teachers to fill the impending departures.
Although the starting point for states’ estimates of their need for science and mathematics teachers should be the assessment of current need,6 it is the projection of their need for the future that is most important in shaping related policies and practices. There is only a very limited range of actions a state or district can undertake to address a current shortfall of teachers because what’s required in response are very immediate “fixes” that cannot await the lengthy deliberations of the policy-making process. Responses to immediate needs for teachers can involve adjustments in the assignments of current teaching staff, consolidation or cancellation of classes, emergency hires that may involve requests for licensure waivers, perhaps the offer of financial incentives, but little more.
If, however, there is a shortfall (or significant surplus) of teachers that has to be addressed not only this year but over the next five years or longer, then it becomes relevant and important to develop policies and enact other more systemic changes in response. Addressing the factors that may contribute to problematic rates of teacher attrition, responding to the labor market for teachers by adjusting compensation or reducing opportunity costs (like lengthy preparation and certification requirements), implementing more aggressive teacher recruitment efforts, pursuing appropriate adjustments in the production capacity of the state’s teacher preparation programs – all of these and similar longer-term strategies may be appropriate for addressing a need for teachers projected into the future.
Such policy options make it clear that in responding to the projected need for teachers it becomes important to focus not only on the demand for teachers throughout the state but also on the supply. The available supply is not a fixed and unalterable constant, however, but a variable that can be manipulated. There may be limitations to how extensively it can be altered in the span of a few years; an ambitious new effort to recruit greater numbers of recent high school graduates into teaching, for example, would only begin to have impact after the first recruits begin to graduate college four years later. Moreover, economic uncertainties limit our ability to predict accurately what the supply will be at a future point in time; the economic downturn of 2007-2010, for example, has had a significant impact on teacher labor markets across the U.S. Nevertheless, teacher supply is malleable and responsive to our efforts to influence it – however inadequate or ill-timed those efforts ultimately may be.
Teacher demand, of course, is also not fixed. It is subject, on the one hand, to the fluctuating economic circumstances that influence birth and immigration rates and thus have an impact on the size and make-up of the student population that must be served. But it is also a function of deliberate and voluntary policy decisions, such as changes in class size limits or in the number of science and mathematics credits required for high school graduation or college admission, which have an impact on the number and kinds of courses to be made available for secondary school students and, consequently, on the number and expertise of the science and mathematics teachers who will be required to staff them. Beyond that, teacher demand is a function of the nature of schooling in the U.S., which may seem like an unalterable given but is in reality the product of decisions that more and more educators believe need to be revisited. It goes well beyond the 5-10 year timeline that is the assumed need target here to think about a fundamental restructuring of our public schools that may change the relationship between teachers and students, rely more heavily on technology to encourage greater student self-instruction and reduce dependence on the physical classroom, eliminate grade levels in favor of student competencies, and in other ways greatly alter the way we think about and calculate our need for teachers. But such possibilities certainly loom on the long-term education horizon and have already been realized to an extent in isolated schools and districts throughout the U.S.
All of this points to the fact that making reliable projections of a state’s future need for science and mathematics teachers is a much more complicated affair than developing current need estimates. And it is more complicated not only because of the additional factors that have to be considered and because of the ability of state educators and policymakers to influence the projections through their actions, but also because there are far more data to gather and digest and more complex statistical methods that can be brought to bear in generating forecasts from these data. Simply to project changes in the population over the next decade requires attention to the validity and reliability of the method for doing it. To project the student population and the impact on of its changing demographic make-up on the demand for courses in science and mathematics requires yet more sophisticated calculations. And to estimate the probable impact on the teacher supply of various economic factors, whether compensation incentives that may be under consideration or recession-related changes in the entire labor market, is more methodologically complex still.
Adding to the complexity of estimates of future teacher need, whether short term over the next five years or longer term over the next ten to twenty, is the fact that the consideration of teacher quality is even more in play than it is in estimates of current need. Not only must estimates of future need consider shortcomings in the qualifications of the current supply of teachers, both statewide and local. In addition, they should consider the qualifications that will be required of teachers in order to respond appropriately to future demand, which may include a greater need for teachers capable of teaching more sophisticated science and mathematics courses. And adequate responses to projected shortages go beyond finding ways to help ensure that the best available teachers will fill the vacancies identified; ideally, they involve efforts to improve the quality of the entire science and mathematics teacher workforce so that all available teachers will be well-qualified and effective and especially so that any imbalances in teacher quality between individual schools and districts in the state will be addressed and largely eliminated.
Finally, as we shall discuss in more detail in the Guide to Teacher Data, good projections of future teacher supply and demand have more sophisticated data requirements than estimates of current need. On the supply side, for example, longitudinal data that track teachers over time as they move through their careers are invaluable in estimating attrition and retention rates. These are even more helpful if they also include information about teachers’ salaries and job placements that can then be used, for example, to generate hypotheses about the impact of proposals to increase compensation or improve working conditions. Likewise, data about the production capacity and attrition rates of programs that prepare science and mathematics teachers are essential.
On the demand side, we previously noted the need to track population trends over time. In addition, historical data would be very helpful. It would be useful to have data, for example, on the effect of any previous increases in high school science and mathematics course requirements on the demand for teachers in order to forecast the impact of any new changes in requirements that may be in the offing. Similarly, historical data on course-taking patterns among different populations of students7 could be helpful in projecting how predicted changes in the make-up of the student population will affect the demand for teachers.
A fortunate few states8 have, or come very close to having, the ideal sets of data required to undertake sophisticated teacher supply and demand projections or the capacity to generate these data readily. These states have invested heavily in education data systems and have a robust data “warehouse” that centralizes different kinds of longitudinal (historical) data on teachers, including their education history, certification status and history, course and school assignment history, and retirement status. Centralized state data warehouses also hold detailed information on the educational histories of K-12 students, which makes it possible to identify historical enrollment trends in science and mathematics and the course-taking patterns of different student sub-groups. Moreover, a primary purpose of establishing a state data warehouse is to ensure the maximum accuracy of the data and the uniformity and compatibility of data from different sources.
Although there has been significant progress among states nationwide9 over the last several years in establishing high-quality centralized databases that hold detailed information on students’ educational careers from kindergarten through college, there is no similar national groundswell to strengthen states’ data on teachers. In most states, the data on teachers are of variable quality depending upon their source, with state-generated data (such as teacher certification and retirement information) tending to be more reliable and district and school-generated data (such as teacher assignment and shortage information) tending to be both less reliable and less uniform between districts. Data from different sources within a state may use different database platforms (i.e., basic database architecture) and different data definitions, thus limiting the state’s ability to aggregate data confidently and make accurate inter-district comparisons. Districts may differ, for example, in the course titles they use; a “pre-calculus” course in one district may be a close equivalent to a combined Algebra III and Trigonometry course offered in another district. Similarly, data sources may have different technical definitions of “out-of-field” teaching or different criteria for what constitutes a shortage of teachers in a particular subject. In addition, school and district administrative data can be tainted as the result of incentives, such as sanctions for non-compliance with No Child Left Behind, which encourage school and district administrators to mis-report particular data like the number of classes not taught by appropriately qualified teachers. This means that in states without a well-centralized data system it may be difficult to track the movement of teachers between districts or tabulate statewide student enrollment in specific science or mathematics subjects – in other words, difficult to derive any reliable statewide picture of teacher supply and demand.
If states want to be able to carry out more accurate statewide estimates of student enrollment in science and mathematics courses – both current and projected – it would be helpful to undertake an inventory of district course offerings in those subjects throughout the state in an effort to ascertain the true congruence of courses. Likewise, it would behoove states to implement some sort of unitary course titling system at the secondary level and to standardize other student-related and teacher-related data definitions.
States must weigh the costs and benefits of improving data systems to provide more sophisticated and useful data and facilitate reliable data analyses related to teacher supply and demand. Given the inherent uncertainty attaching to projections of future need – both estimates of demand and especially estimates of supply – there may be reluctance on the part of some states to invest in the data systems and expert personnel required to generate high-quality projections. We believe that the benefits far exceed the costs in the long-run, however, and we hope the discussion here helps to illustrate the advantages of good-quality data and a thorough supply and demand analysis.
A number of states10 have commissioned studies of their current and projected workforce needs with a view toward strengthening their position in the increasingly technological economy of the future. These studies typically identify various indicators of the science, technology, engineering, and mathematics (STEM) capacity of their current workforce and discuss the adequacy or inadequacy of their K-12 and post-secondary educational systems to meet the future STEM education needs they have projected. Such reports are not so much rigorous efforts to estimate long-term future needs for STEM-proficient workers (including teachers) as they are attempts to persuade policymakers and the public to strengthen the science and mathematics proficiency of a state’s students. We would hope states would be interested in augmenting their capacity to generate the high-quality data and the need projections based upon them that would facilitate efforts to ensure that their teacher workforce is adequate to the task.
It is the specific purpose of Projecting a State’s Future Need for Science and Mathematics Teachers to provide direction in developing reliable forecasts that can ground appropriate policies and practices to address an imbalance between teacher supply and demand. At their best, however, such projections cannot be as confident as estimates of current need. The future is inherently uncertain, and that uncertainty increases11 the farther into it one attempts to venture; even a five-year timeline invites the intervention of all sorts of unforeseen turns and unanticipated events that can wreak havoc on any forecast one might develop. Moreover, whereas current need estimates depend principally on descriptive data, projections into the future require inferences from that data. And the statistical methods used to produce those inferences or projections employ calculations that assume the continuity of various contingent historical trends and on mathematical models that make reasonable but not ironclad assumptions about human choices and behavior. Projections into the future also are more reliable over large populations than over smaller ones, meaning that statewide supply and demand estimates generally have a higher confidence level than local projections.
The guidance here is offered notwithstanding the very complex methodological nuances and issues that attend estimates of teacher supply and demand. We note some of these complexities in the ensuing discussion, but we ignore others in the interest of simplicity and the desire to provide some sort of reasonable, if ultimately imperfect, direction to those individuals who have the responsibility to develop or employ the kinds of estimates discussed. A more detailed discussion of methodological issues can be found in the Research Analysis unit of this project, which was written by Stephen Raphael and provides much of the theoretical underpinning for many of comments that follow below. And a still more rigorous discussion of the state of the art in teacher supply and demand assessments, which is illuminating even though somewhat dated, can be found in the 1992 National Research Council publication Teacher Supply, Demand and Quality: Policy Issues, Models, and Data Bases.
The projection of a state’s need for science and mathematics teachers, whether short-term or long-term, requires independent projections of teacher supply and teacher demand. The guidelines12 offered here for developing a basic projection of teacher demand consist of five suggested steps.
1. Gather current and, if possible, historical data (preferably for at least the last 5 years) on the identified statewide need each year for secondary science and mathematics teachers.
Although the focus of this specific section is on developing a statewide teacher demand estimate for the future, it is important to start with an estimate of current need. Producing an estimate along the lines outlined in the unit on Establishing a State’s Current Need for Teachers accomplishes the following:
Similarly, gathering reliable historical or longitudinal data is invaluable in identifying trends that can serve as the basis for future projections of both teacher demand and supply. It would be impossible, for example, to estimate population growth based on a single year’s population figures; longitudinal data on population changes are required in order to discern any long-term or emerging patterns of growth or decline that can then be used to predict how they may play out in the future. Likewise, it would be indispensable to know whether or not the average number of science and mathematics courses students take has been increasing or decreasing in recent years. Clearly, trends are not necessarily linear; an increase in course-taking patterns may have leveled off over time, or new graduation requirements may have intervened to increase course taking and thereby disrupt a previous trend. Barring disruptive events like new course requirements, economic cataclysms, or sudden changes in the population, however, the longer back historical data can be traced, the more reliable a future projection can be made on the basis of those data because anomalous peaks and valleys are more readily averaged out.
2. Determine current statewide student enrollment by specific subject (e.g., Algebra, Physics, and Biology) and course difficulty level (basic, college preparatory, Advanced Placement) in secondary science and mathematics courses, and if possible track historical enrollments over at least the past 5 years
Since accurate statewide enrollment data rely on local data, student enrollment data in science and mathematics courses for individual schools and districts must be available as the starting point for the statewide computation. Although districts should routinely collect and record this data, we noted in the previous discussion of The State of State Data that in the absence of a strong centralized statewide data system the quality and compatibility of district data are far from assured.
3. Project future statewide student enrollment over the next 5-10 years13 for secondary science and mathematics
It is not particularly difficult for states to project total student enrollment in the different grade levels K-12, though this calculation is always approximate because enrollments are a moving target due to population migration. Indeed, virtually all states14 undertake or have access to projections of the school-age population by grade – generally both statewide and by county or district. And many individual districts15 also develop their own, more detailed school-by-school projections.
It is much more difficult to forecast total student enrollment in specific mathematics and science courses and subjects, but this is ultimately necessary in order to accurately project the demand for teachers in those subjects. With adequate data as described in 2. above, a serviceable projection16 of enrollment in specific science and mathematics subjects could be constructed by projecting the trend in actual current and historical enrollments in science and mathematics courses into the future and adjusting for significant anticipated changes in the total student population in future years, including changes in the proportion of students from various ethnic or socio-economic backgrounds that might be associated with course-taking patterns.
4. Derive a first-order projection of teacher demand for each science and mathematics subject over the next five years by multiplying the target class size17 in science and mathematics subjects by the projected student enrollments in the subjects for the next five years.
This will actually yield a projection of the number of classes that must be covered in the various science and mathematics subjects. To project the number of teachers required, further divide the number of classes to be covered by the average teacher course load.18
5. Refine the first-order projection of teacher demand
The guidelines19 offered here for developing a basic projection of teacher supply consist of four suggested steps.
1. Gather baseline data on the teachers who are currently teaching science and mathematics courses throughout the state
The calculation of the number of science and mathematics teachers in a state is not uncomplicated. There is the issue, first of all, of whether all teachers teaching science and mathematics courses should be properly counted as science and mathematics teachers. As we note several times throughout this project and discuss in greater detail in the unit on Teacher Quality and Teacher Licensure, we would argue that only teachers who have the appropriate state credentials to teach specific science and mathematics subjects – including those alternate route and in-migrating teachers who have been legitimately granted temporary credentials – should be counted as part of the active science and mathematics teacher supply. Without this restriction, the supply of science and mathematics teachers becomes, if not infinite, then certainly so large and ill-defined that it threatens to make a meaningful analysis of supply and demand and the development of effective policies to address them impossible.
Secondly, the number of teachers is contingent upon the inclusion or exclusion of private schools from the supply calculation. We have chosen to focus on public education in these guidelines, which both narrows the definition of who is included as a teacher and reduces the number of teachers counted in the active supply. (It would increase the number of teachers in the reserve pool, however, because active private school teachers are potential public school employees.)
Still another complication is the fact that individual teachers may have multiple certifications in science and mathematics. Thus, a single individual who teaches both biology and physics, for example, or both natural science and mathematics, may be double counted in the state’s computation of its active science and mathematics teachers. One way of handling this would be to count teachers on the basis of their Full Time Equivalent (FTE) course load; a teacher who has a third of an assignment in mathematics and two-thirds in biology would count as one-third of a mathematics teacher and two-thirds of a biology teacher.
Ultimately, for all of these complicating factors21 states simply must attempt to make appropriate adjustments in their calculations and in the protocols they employ to derive them.
2. Project the number of the currently employed teachers in each mathematics and science discipline who are likely still to be teaching in the state22 (though not necessarily in the same school) in each of the next 5-10 years
Most states keep data on attrition and retirement rates of teachers over time. This is important because national data cannot capture the nuances of attrition in individual states, which include the impact of labor markets and retirement rules and incentives that vary from state to state and even from district to district. Ideally, state data will track rates of attrition over the years by teachers’ age and years of experience, and also by gender and race/ethnicity because attrition rates differ as a function of these factors. Some studies23 indicate, for example, that the longer a teacher’s initial stay in teaching, and the older a teacher is (at least up to the age of around 40) before taking an initial break, the more likely the teacher is to re-enter the classroom.
Hopefully, states also will track the teacher attrition and re-entry patterns not only in science and mathematics generally but more specifically in chemistry, physics, etc., although at least one recent study24 based on nationwide data found no difference between the general attrition rate of secondary school teachers and the rate in science and mathematics specifically. Two other studies,25 however, found that once high school science and mathematics teachers did leave teaching, they were the least likely of all teachers to re-enter the classroom.
Clearly, as the 2007-2010 economic downturn in the U.S. has demonstrated, economic realities have a significant impact on the extent to which such historical averages and trends can serve as a reliable basis for future teacher attrition or retirement projections.
3. Project the number of new science and mathematics teachers who are likely to be licensed and available to teach in each of the next 5-10 years
The two principal sources for newly licensed teachers in a state are the state’s teacher preparation programs (both traditional and alternate route programs) and the migration of teachers from other states or countries. Because it is the responsibility of a single agency in a state to grant new teacher licenses, it should be relatively easy to find out how many licenses are granted from year to year. Future projections can then be made on the basis of trends in the number of licenses issued over time – although such trends may not account for the expansion or addition of programs that prepare science and mathematics teachers or greatly expanded out-of-state recruitment efforts.
It is ultimately not sufficient, however, simply to project the number of new teacher licensees in science and mathematics over the next 5-10 years. Some of these teachers will leave the state to work elsewhere, some will teach in private schools, some will delay their entry into teaching, and some will pursue another career and never enter the K-12 classroom, at all. Thus, what is needed is an estimate of the number of new licensees who are likely to be available to teach in the next 5-10 years, and that is a far more difficult calculation.
One study from the National Center for Education Statistics (NCES) that offers some help in this direction is Alt, Henke, & Perry (2007), To Teach or Not to Teach? (pp. 31-36). The study involved a ten-year longitudinal survey of 1992-93 bachelor’s degree recipients, and the data provided make it reasonable to conclude that, at a minimum, approximately 13% of science, mathematics, and engineering graduates who received K-12 teacher licenses never taught. It might be possible, then, to subtract 13% of the number of newly prepared and certified science and mathematics teachers as a rough correction to the calculation of available new teachers in a given state, but the national data in the NCES study are not necessarily reflective of the labor market in individual states. Moreover, the study does not provide guidance in calculating the percentage of science and mathematics teachers who are delayed entrants, let alone what percentage of these delayed entrants should be considered candidates for teaching positions in any specific year.
In the end, all that such studies can do for any specific state is to provide some possible suggestions for how they might refine their estimates of the supply of science and mathematics teachers. And the studies clearly indicate that projections of teacher supply in individual states would benefit from careful state-level investigations – ideally in each local labor market – that can identify and quantify complex patterns of postponed entry and re-entry into teaching among science and mathematics teachers. These patterns not only become part of the basis for future projections but also can help in developing more effective recruitment and retention policies and practices.
Because of the many contingencies involved, it is also difficult to project the number of science and mathematics teachers who are likely to migrate into the state over the next 5-10 years and seek teaching licenses. Some states have aggressive out-of-state teacher recruitment programs, however, which may target and yield a reliable number of new teachers annually and also may be able to increase that number to respond to any projected shortfall.
4. Construct a first-order projection of the available supply of science and mathematics teachers statewide by adding the number of teachers estimated to be available over each of the next five years from the various sources and subtracting the number of teachers anticipated to be lost through attrition each year. This yields an actionable estimate that is based on the most dependable sources of teachers, though it lacks helpful additional refinements.
Beyond a first-order projection, which gives an approximate, statewide estimate of the available supply of science and mathematics teachers, it is ultimately desirable to develop a more sophisticated projection of the supply of teachers that includes other important factors and is more attuned to local realities. Several refinements would yield more precise and helpful supply projections – if data of adequate quality can be collected:
1. To the number of science and mathematics teachers who are currently employed, add the number of licensed science and mathematics teachers who applied for teaching positions in the current year (or year the data were collected) but were not hired.
If computed by district, the non-hired teachers can be counted as surplus even if they were eventually hired in another district or by private schools. If computed statewide, however, teachers can be counted as surplus only if they were not hired by any public school or district in the state. This will yield an estimate of the total available pool of science and mathematics teachers that counts only individuals who have demonstrated an actual interest in teaching – arguably a much more realistic estimate than one which attempts includes all licensed teachers in the state.
2. Calculate and subtract from the first-order supply projection the number of teachers who are currently teaching science and mathematics courses but lack the appropriate state credentials to teach secondary science and mathematics courses
This limitation already should have been included in the first-order supply projection. Its repetition here emphasizes the importance of ensuring that only adequately qualified teachers are included in the calculation of the supply of science and mathematics teachers.
There is, however, no universally accepted definition of an “adequately qualified” teacher. And there is an additional challenge in finding a set of readily usable data points that would permit the easy identification of teachers’ qualifications as adequate or inadequate in the kind of large-scale supply and demand analysis we are discussing here. For the sake of convenience and consistent with our focus in this project on teacher supply and demand, we would recommend that states consider a class as being taught by an “adequately qualified” teacher only if the teacher (a) is fully licensed or certified (i.e., not teaching on the basis of a temporary or emergency credential or waiver), or has demonstrated solid knowledge of his or her field and is enrolled in a teacher preparation program and pursuing a license; (b) is not teaching “out-of-field” – i.e., has the subject knowledge required by licensure or endorsement in the field to be teaching the class; and (c) is an actual employee of the school or district and not a temporary substitute.
To be sure, not all adequately qualified teachers are equally well-qualified or capable, and some teachers who are not technically qualified to teach a particular class may in fact be quite suitable for the task. A teacher who is a recent transfer from another state, for example, and has a temporary license because he or she needs to take a course or two to satisfy the new state’s licensure requirements is likely to do just fine. Moreover, the fact that a teacher is technically qualified to teach a class according to the criteria we have suggested does not guarantee that he or she will teach it well. As we discuss in the Teacher Quality and Teacher Licensure unit of the project, state licensure – especially licensure in the sciences – is a blunt instrument that does not necessarily ensure that teachers who meet the criteria for licensure and for teaching a particular subject are ultimately sufficiently well-qualified to be successful in teaching it. For purposes of conducting a statewide assessment of the adequacy of the teacher workforce, however, state licensure – if it is based on valid criteria and rigorously enforced – provides the analysis with an important quality control dimension.
3. To the extent possible, develop district-by-district supply projections.
The fact that there may be enough science and mathematics teachers available in the aggregate statewide does not mean that there will be enough teachers available for each school in each district. In some labor markets, there will be more than enough teachers to fill the positions available, while in other labor markets there will be too few. Teaching remains largely a locally-based26 profession, and many individuals simply will not accept teacher positions in communities where they do not wish to live or in schools at which they do not wish to teach even if there are no other teaching jobs available to them.
Good state-level projections will reflect these district variations, and a trend study of state projections based on solid data could make it possible to project the science and mathematics teacher supply for specific districts as a function of fluctuations in the overall state supply numbers. In the absence of such a trend analysis, or for districts that want even more accurate projections and a clearer understanding of the likely causes of and possible remedies for an estimated shortage of science and mathematics teachers, several kinds of specific data are required:
Excluded from the guidelines for developing teacher supply projection that were offered above are two other potential sources of teachers that are commonly cited27 in discussions of teacher supply but that simply involve too much speculation to incorporate them into estimates that could claim reliability:
1. The entire reserve pool of science and mathematics teachers – the total number of individuals in the state who are certified to teach science and mathematics and who are not currently teaching in the public schools
It is impossible to know how many of these individuals are truly potential entrants into teaching. Those most likely to enter will be the delayed entrants or re-entrants already included in 3. above as well as private school teachers, whom research indicates have much higher rates of attrition than public school teachers and who possibly could be enticed into the public schools – though likely28 only a small percentage. The possibility of further increasing the number of entrants from this pool depends upon labor market conditions – or deliberate market manipulations through various kinds of incentives – and possibly upon the introduction of policies that provide such incentives or that allow retired teachers, for example, to re-enter teaching without sacrificing their retirement benefits.
For a variety of reasons we cannot assume, as some scholars29 have suggested, that all science and mathematics teachers currently licensed in a state should be counted as being potentially available to teach:
2. The greater scientific and technological workforce, which includes many individuals who have strong educational backgrounds in science or mathematics
Most of these individuals will never go into teaching. It is possible, however, to target specific incentives at this population to entice some of them into the classroom and give them training in pedagogy. How many individuals can thus be wooed into teaching depends upon the size and nature of the incentives and the realities of the larger labor market. Surveys of this population may provide an indication of the size of the incentives that would be required and thus inform a deliberate effort to recruit a certain number into the classroom – assuming that available resources and state and district teacher compensation policies permit it. An interest in teaching expressed in a survey, however, may not translate into a willingness to teach in actual fact.
At the simplest level, calculating the future need for teachers is simply a matter of comparing projected demand against projected supply (step 1 below). Several additional steps may be considered, however, in order to enhance the accuracy and usefulness of the future need calculation
1. Determine the extent to which the projected statewide supply of teachers for the next 5-10 years matches the projected statewide demand for that same period
2. To the extent possible, reconcile statewide and district-specific projections of the future need for teachers
Some of this reconciliation already should have been accomplished in developing and refining the first-order projections of both demand and supply. District-level projections may need to be further adjusted, however, to reflect the specific district impact of factors such as changes to the science and mathematics curriculum, more rigorous high school requirements in science and mathematics, or new state expectations for teacher qualifications. At the same time, statewide projections should reflect some of the nuances of the supply and demand picture in individual districts as well as providing an overall picture of the state’s ability to meet its demand for teachers. This is necessary if state policymakers and education leaders are to identify the most fruitful avenues for developing appropriate responses and predict with any confidence the likely impact of various policy options under consideration. An overall statewide estimate may indicate, for example, that the current production capacity of new science and mathematics teachers comes close to meeting the aggregate demand, but specific districts may be unable to recruit the teachers they need from state preparation programs. Conversely, a statewide estimate may indicate a severe overall shortage of science and mathematics teachers, but some districts may in fact have an abundance of qualified applicants.
Although it is desirable to have as much consistency as possible between state-level and district-level projections of supply and demand, there are inevitable differences between them that make reconciliation difficult. Projections generated at the individual district level are likely to reflect a much more fine-grained analysis than the state-generated projection for specific districts. A good district estimate will make more use of detailed information concerning the specific sources, quality, classroom effectiveness, and attrition rate of the district’s science and mathematics teachers, as well as relevant information about the local labor market, the local reserve pool of teachers, and the diverse needs of individual schools. This is simply a level of detail that it is beyond the ability of states – and even many individual districts – to collect, let alone to incorporate into a statewide analysis.
Beyond the difference in the level of detail, however, there may be an inherent conflict between statewide and district level interests that can lead to divergent projections of the need for science and mathematics teachers. The perceived best interests of individual districts cannot be readily reflected in state-generated estimates intended to serve the overall interests of every district in the state. From a statewide standpoint, for example, a particular district may appear to be privileged in comparison with other districts in the state in terms of the quality and number of its available supply of teachers in a particular science or mathematics field. The district, however, may perceive a need to improve the quality of its teachers in that field still further, and thus it may seek to increase the size of its candidate pool despite the state’s perception that the district has no shortage. Likewise, high-school graduation requirements in specific districts may include more science and mathematics courses than the state mandates, and this also may figure into district-generated need projections but not into the state-generated projection. Or, a particular district may want to retain a smaller than mandated class size, and thus it may project a greater need for teachers than the statewide analysis projects it will have.
3. It is valuable to attempt to factor into the need projections the desire to have more applicants for science and mathematics teaching positions than there are positions available. A surplus of applicants allows for greater selectivity in hiring and thus increases the likelihood of hiring better quality teachers
It is difficult to determine how the quantitative estimate of teacher need should be adjusted to reflect the desire to have a surplus of applicants for science and mathematics teaching positions, especially on a statewide basis. In reality, some schools and districts will always have more applicants for open positions than will others, and applicants from districts with a surplus cannot necessarily be redirected to other districts experiencing a shortage. Thus, a careful district-by-district analysis would be necessary to determine how many additional teachers should be added to a need projection in order to ensure that most districts can hire more selectively.
Also important in calculating how great an applicant surplus is desirable would be some sort of determination of how different ratios of applicants per position are likely to enhance the quality of teachers hired and whether there is some “lowest effective ratio” that significantly improves the likelihood of hiring better teachers. There are few, if any, available empirical studies of the phenomenon that could provide useful guidance.
One example of a state wrestling with a related issue is Maryland. In looking at the ratio of STEM graduates to STEM jobs (not restricted to teaching) in other states that are its closest competitors in high technology industries, the state’s Bohannon Commission determined32 that in order to be competitive with its rivals Maryland needed to increase its ratio from two STEM graduates for every three STEM jobs to a ratio of 1:1. This ratio was not rigorously calculated, and it is unclear to what extent it applies to the teaching profession. It is a reminder, however, that estimates of teacher demand that err on the low side court not only the possibility that some schools and districts will be unable to meet their need for science and mathematics teachers but also the possibility that many of the teachers hired will be less qualified than is desirable.
1. We use the term “need” throughout this report to reflect the extent to which the supply of teachers meets the demand. “Supply” is used in this report to denote the available pool of teachers, and “demand” is used to denote the number of teachers required to staff the classes offered – and ideally, the number of classes that would be offered were the supply adequate to staff them.
2. The inclusion of the private school sector would require appropriate modifications of both supply and demand estimates. Private school teachers, for example, could no longer be considered part of the “reserve pool” for the public schools, even though there would continue to be migration of private school teachers into public schools. Similarly, teachers who leave public schools to teach in private schools would not properly be counted in state attrition rates if both public and private education is included in the analysis of teacher supply. Another issue if private schools are included involves the appropriate definition and count of the state’s science and mathematics teachers. Not all private schools require their teachers to have the credentials that would be required to teach in the state’s public schools.
3. The target class size may be a state-mandated or district-mandated upper limit. Target limits may differ from one district to another, and these may differ from the state target, thus complicating the statewide computation of demand.
4. If minority students in a particular state or district, for example, tend to take fewer science courses than white, non-Hispanic students, then we could anticipate a change in the demand for science courses – other things being equal – if the proportions of these groups in the total student population are projected to change.
5. It is virtually impossible to know the exact number of licensed teachers in a state, as many who have licenses on record will have moved away and others no longer in teaching are likely to be lost track of. In any event, the importance of data on the number of all licensed science and mathematics teachers is not to provide a hard figure that can be included in a calculation of total teacher supply but to gain an order of magnitude estimate of the “reserve pool” of licensed teachers on the assumption that, if the pool is large enough, effective policies and incentives might draw more of them into (or back into) teaching than those who currently become delayed entrants and re-entrants.
6. We use the term “need” throughout this report to reflect the extent to which the supply of teachers meets the demand. “Supply” is used in this report to denote the available pool of teachers, and “demand” is used to denote the number of teachers required to staff the classes offered – and ideally, the number of classes that would be offered were the supply adequate to staff them.
7. This is not intended to imply that we should rest content with the status quo, in which African-American and Hispanic students tend to take significantly fewer and less rigorous courses in science and mathematics than Anglo and Asian-American students. This pattern will not change easily, however, and it is virtually certain to persist to some degree into the short-term future.
8. Florida and Virginia are two states that have very sophisticated teacher data warehouses with information on teachers from the beginning of their preparation program through their certification and employment history.
9. This progress can be attributed to the emphasis in No Child Left Behind on measuring student achievement and to the specific efforts of the Data Quality Campaignand others, including, most recently, encouragement from the U.S. Department of Education’s Race to the Top.
10. Several of these studies can be accessed via the Internet. See, for example, the studies from Illinois and Kansas.
11. It is instructive to look at the ten-year Projections of Education Statistics released annually by the National Center for Education Statistics. NCES notes(see, for example, p. 18 of the 2017 Projections) that the accuracy of its predictions declines significantly the further into the future they project.. And the accuracy of projections of teacher supply declines more than twice as much as that of projections of student population – the main driver of teacher demand. Compare the “Accuracy of Projections” discussions on pp. 7 and 18.
12. A more detailed discussion of state-level teacher supply and demand projection can be found in Reichardt (2003). Teacher Supply and Demand in the State of Colorado (MCREL research report). Denver, CO: Mid-continent Research for Education and Learning. Our discussion is not directly based on Reichardt’s, but the two are consistent.
13. Enrollment can be projected more than 5-10 years into the future at a state’s discretion. The reliability of estimates deteriorates the further into the future they aim, however. The National Center for Education Statistics limits its projections to ten years and notes the significant decrease in accuracy of those projections over the 10-year period. (See, in particular, the “Accuracy of Projections” discussion on p. 7).
14. See, for example, the statewide projections for Maryland. The National Center for Education Statistics also projects ten-year student enrollment figures by state in its annual Projections of Education Statistics. The NCES projections are not by individual grade level, however.
15. See, for example, the projections for Eau Claire, Wisconsin.
16. As an example, suppose that the percentage of all students taking high school physics courses in a state has increased from 21% to 32% over the last five years, with enrollment among upper income students growing from 35% to 45%, among median income students from 25% to 34%, and among low-income students from 17% to 22%. Similar data might be obtained for students from different ethnic groups. Projecting these same trends into the future, taking into account the anticipated growth rates of these different populations, it becomes possible to derive a reasonable estimate of the student demand for physics courses into the future assuming all other factors (e.g., high school graduation requirements) remain the same.
17. The target class size may be a state-mandated or district-mandated upper limit. Target ratios may differ from one district to another and these may differ from the state target, complicating the statewide computation of demand.
18. This will yield an approximation that likely underestimates the actual number of qualified individual science and mathematics teachers required because some schools and districts will not be able to give teachers a full-time course load in these disciplines. This means that a FTE load in the demand estimate would in actuality sometimes have to be split among more than one teacher. On the other hand, some of those teachers may be doubly qualified in science and mathematics, thus reducing the actual number of individual teachers needed overall.
19. We noted previously the more detailed discussion of state-level teacher supply and demand projection that can be found in Reichardt (2003). In addition, a particularly illuminating discussion of the many considerations that must be brought to bear in analyzing a state’s supply of science and mathematics teachers can be found in The Critical Path Analysis of California’s Science and Mathematics Teacher Preparation System (2007). Beyond this practical example, the National Research Council, in its 1990 study Precollege Science and Mathematics Teachers: Monitoring Supply, Demand, and Quality, presents a sophisticated methodological discussion of the particular difficulties that attend efforts to estimate teacher supply.
20. Data about teachers’ licensure or certification status are useful in providing insight into the extent to which a state’s overall teacher workforce – and the workforce in specific districts – has the minimum basic qualifications. The fact that all teachers in a state, district, or school have a valid teaching license is not an assurance, however, that they teach only those classes for which their license and qualifications are appropriate.
21. For an insightful discussion of the issue of multiple certifications and similar complicating factors, see Boe, E.E., & Gilford, D.M. (Eds.). (1992). Teacher Supply, Demand, and Quality: Policy Issues, Models, and Data Bases. Washington, DC: National Academy Press, pp. 153ff.).
22. For an aggregate estimate of teacher supply for the entire state, changes of teacher assignments from one school or district to another in the state are irrelevant. The only relevant consideration is whether teachers remain in the state teacher workforce or whether they retire, step out, or move to another state. Clearly, for an estimate of the teacher supply at the district level, teacher transfers out of the district are of critical importance.
23. For a review of several studies that discuss teacher attrition and retention, see Allen, M. 2005. Eight Questions on Teacher Recruitment and Retention: What Does the Research Say? Denver, CO: Education Commission of the States. For a recent analysis of teacher attrition patterns in Illinois, see DeAngelis, K.J., and Presley, J.B. (2007). Leaving Schools or Leaving the Profession: Setting the Illinois Record Straight on New Teacher Attrition. Carbondale, IL: Illinois Education Research Council. For a more theoretical general discussion on teacher attrition, see Boe & Gilford (1992).
24. Ingersoll, R.M., & Perda, D. (2009, March). The Mathematics and Science Teacher Shortage: Fact and Myth. Philadelphia: Consortium for Policy Research in Education, p. 36.
25. Murnane, R.J., and Olsen, R.J. (1989). Will There Be Enough Teachers? The American Economic Review, 79, 242-246; and Alt, N.N., Henke, R.R., & Perry, K. (2007). To Teach or Not to Teach? Teaching Experience and Preparation Among 1992–93 Bachelor’s Degree Recipients 10 Years after College. Washington, DC: National Center for Education Statistics, p. 26.
26. See Boyd, D., Lankford, H., Loeb, S., & Wyckoff, J. (2002, October). Understanding Teacher Labor Markets: Implications for Equity.
27. See, for example, Hunt, J., & Carroll, T. G. (2003). No dream denied: A pledge to America’s children. National Commission on Teaching and America’s Future. Accessed at http://nctaf.org/wp-content/uploads/no-dream-denied_full-report.pdf. See also Ingersoll, R.M., & Perda, D. (2009, March). The Mathematics and Science Teacher Shortage: Fact and Myth. Philadelphia: Consortium for Policy Research in Education.
28. See Luekens, M.T., Lyter, D. M., Fox, E.E., & Chandler, K. (2004). Teacher Attrition and Mobility. Results from the Teacher Follow-up Survey, 2000-2001. Washington, DC: National Center for Education Statistics.
29. See, for example, Ingersoll, R.M., & Perda, D. (2009, March). The Mathematics and Science Teacher Shortage: Fact and Myth. Philadelphia: Consortium for Policy Research in Education.
30. See for example DeAngelis, K.J., and Presley, J.B. (2007). Leaving Schools or Leaving the Profession: Setting the Illinois Record Straight on New Teacher Attrition. Carbondale, IL: Illinois Education Research Council.
31. Conclusion based on data from the Common Core of Data for school year 2002-2003(NCES, 2005) and the Fiscal and Policy Note on House Bill 1254 (Maryland Department of Legislative Services, 2004, Revised.)
32. Related to the author on 11-05-09 by Ben Passmore, Director of Policy Research and Analysis for the University System of Maryland.
Join the Conversation