Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Determining Supply: Individual and District Activities Each year between 5 and 10 percent of the nation's public school teachers leave the profession. Some leave permanently; some leave tem- porarily; and they leave for a variety of reasons to take a different job, to pursue further education, to start a family, etc. What this means, however, is that every year between 100,000 and 200,000 replacements are needed to fill those vacancies, although the actual number is not known. From where do these teachers come? The panel attempted to answer that question, particularly for secondary school science and mathematics teachers, by looldug at three sources of evidence: (1) state and national models of teacher supply and demand; (2) in-depth case studies of classroom teacher recruitment in a number of school districts; and (3) insights obtained through a conference on professional personnel systems in large school districts. In this chapter we first discuss what constitutes supply-continuing and new science and mathematics teachers and their incentives and decisions about teaching along stages in their career paths. We then look at supply from the district viewpoint, from which widely varied policies for recruiting, screening, and selecting teachers cause variations in the adequacy and the quality of the supply of teachers available to different districts. THE COMPONENTS OF SUPPLY The supply of teachers for the coming school year is a relationship between the number of qualified individuals who would be willing to teach and such incentives as the salaries, benefits, retirement programs, working conditions offered by school districts, and other alternative career opportu- nities. Ideally, it would be desirable to have a behavioral model of supply 60
DETERMINING SUPPLY 61 that would take into account the interaction and interdependence of a- wide range of variables and could help answer such questions as how many teachers can be expected to quit in response to a change in retirement policy, or how many former teachers can be expected to reenter if salaries are raised by a certain amount. Policy makers frequently ask questions about the likely impacts of various education policy actions and socioeconomic forces on prospective teacher supply and demand. 1b address such questions requires a capacitor to project supply and demand under varying assumptions about future circumstances. In turn, this capability requires the development of models that are both behavioral and dynamic. By this we mean models that capture relationships between variables in the environment and the behavior of actors in the educational system, and in particular capture relationships between changes in circumstances and subsequent changes in the numbers and kinds of people interested in obtaining teaching positions or in the numbers and kinds of teachers demanded by school systems. Before such models can be developed, additional research on the relation between incentives and supply and between variables in the en- vironment and supply, as well as additional data to support the models, will be needed. The national and state models examined by the panel are projection models based on extrapolations of current conditions or histor- ical trends, although some use refinements such as age and field-specific attrition rates in projections of continuing teachers and consideration of a broader range of new supply sources. In practice, these models try to estimate the number who will be available from each of the two major components of supply: continuing teachers teachers who are teaching this year and will continue to teach next year in the same location and new entrants. There is a continuous flow of teachers into and out of the teach- ing force, as shown in Figure 3.1. This diagram can apply to the nation, a state, a school district, or to special groups of schools such as rural or inner city schools, or to special types of teachers such as science teachers, mathematics teachers, or minority teachers. Continuing Teachers The most important element of teacher supply during a given year is the retention of people returning from the prior year. ~ obtain that component of teacher supply, we need to know the attrition between the two years. However, the attrition rate is a complex function depending on the various incentives that cause teachers to retire, to move to another school, or to leave teaching for other careers including homemaking. In practice, the method typically used in current models involves making an assumption about attrition rates, sometimes adjusted for trend and sometimes not. For
a) o ~ - en a) z it in z - ~ in a) a) ~ - . - o In ~ - E o In - - LD A) z tic CO a) a) Or o IL ._ a) Cal CO 62 coIn _~ CO .~ __ 111LL 3. 3 a' ZZ a' =~ CD Z o >,= a) c) 0 , ~ _ - a, ~ ~ a) ~ _ ~ ~ ~ LL] C: In - ._ Cal - C) - o Cal - {:: . . V) V Lo
DETERMINING SUPPLY 63 many years the model used by the National Center for Education Statistics for the intermediate set of projections used an attrition rate of 6 percent based on survey data that were collected by NCES in 1968 (Metz and Fleischman, 1974~. In 1987 NCES used an estimated turnover rate of 7.5 percent for elementary teachers and 6.5 percent for secondary teachers (CES, 1987a:46~. The results of the Schools and Staffing Survey (SASS) should provide a basis for more accurate attrition rates. In state models, the supply due to retention can be estimated in a more satisfactory way, since states generally have information on the attrition rate for teachers in the state for the prior year and assume that the rate will be the same in the current year. However, attrition levels vary from state to state and over time. Attrition for states is different from national attrition, since the former includes teachers who move to other states and continue in teaching. Some states also have the information to compute more refined attrition rates, such as for different age groups of teachers and for different subject fields. (Evidence from these states shows that as the teaching stock ages, the average attrition rate will change.) Interestingly, attrition rates in these states for teachers of science and mathematics are not noticeably different from rates of teacher attrition in other fields. Tables 3.1 and 3.2 show retention rates for public school teachers in the states of Illinois and New York for mathematics, science, and the total for all subjects. (As we point out later, the lack of difference in rates by subject may be due to the influences of general enrollment declines during the early 1980s.) TABLE 3.1 Retention Rates for Illinois Public School Teachers 1977-1984 (Percentage Retained in Consecutive Years) Secondary Grades (9-12) Year PrimaryMathematicsScience All Grades Subjects (PreK-8) 1977-1978 90.591.792.1 90.8 1978-1979 90.491.792.3 90.2 1979-1980 91.692.390.5 91.9 1980-1981 91.892.592.5 92.1 1981- 1982 92.693.093.8 93.3 1982-1983 92.594.694.5 93.0 1983-1984 93.5n.a.n.a. 93.4 Note: Data are for downstate schools only (i.e., all school districts except the Chicago Public Schools). This table shows retention rates of teachers; however, the source publication shows attrition rates, i.e., 100 = the retention rate. Source: Illinois State Board of Education (1983: Tables 2 and 3; 1985b: Table 8).
64 PRECOLLEGE SCIENCE AND ~4THEMATICS TEACHERS TABLE 3.2 Age-Specific Retention Rates of New York Public Secondary School Teachers (grades 7-12) in 1984 (Percentage Retained from 1983) Age All Secondary Mathematics Science Subjects Under 35 90.6 89.9 89.2 35-39 94.1 95.6 94.4 40-44 94.9 95.6 94.8 45-49 g5.8 95.8 95.2 50-54 91.9 93.0 92.0 55-59 84.2 84.0 83.5 60 and over 73.2 69.1 70.2 Total all ages 92.4 92.7 91.6 Source: New York State Education Department (1985a). New Entrants The more difficult part of modeling teacher supply consists of pre- dicting the potential willingness of people who were not teaching last year to enter the teaching force. In Figure 3.1 we have labeled all sources of teacher supply other than continuing teachers as "new entrants" or "reentrants." Major categories under the heading of new entrants include newly certified persons, persons with previous teaching experience and certification (i.e., reentrants people who come from the so called reserve pool of teachers), persons hired through some alternative or emergency certification procedure, and in-migrants. The major categories can be broken down into yet finer components. Newly certified persons may be either newly certified graduates of teacher training programs or newly certified graduates with other majors. Experi- enced teachers may have been on leave or layoff, they may have entered other careers (including homemaking); they may have been teaching as substitutes; they may have resigned for long-term health reasons; or they may be in-migrants. In-migrants are teachers who were teaching last year, but not in the particular jurisdiction or subject field for which the supply is being estimated. In some states virtually any college graduate, with or without teaching certification or experience, can be counted in the supply
DETERMINING SUPPLY 65 of new entrants; these states permit certification on the basis of testing, permit hiring on an emergency certification basis, or use an apprentice teaching program. In the first phases of our deliberations, we discovered that the major proportion of new [hires each year did not come from new college graduates, but rather from the corps of experienced returning teachers. Although the percentages varied across subject areas, level, and location, in general it was found that less than half of new hires were new college graduates (National Research Council, 1987c:27~. For example, Able 3.3 shows that the proportion of new hires who were new college graduates was less than 30 percent in each of six types of urban-suburban-rural districts. The National Education Association's (NEA) surveys of American public school teachers found a decline over the years in the proportion of new entrants who came directly from college (NEA, 1987e:24~. From the data provided, the percentage of new hires who had been in college the previous year can be computed to be 67 percent in 1966, decreasing to 17 percent in 1986. These findings are important in light of the fact that the supply-demand model used by the National Center for Education Statistics until 1987 based its estimates of teacher shortage on the assumption that all new hires would be new college graduates. Following the publication of the panel's interim report, NCES discontinued this practice. WHAT INFLUENCES AN INDIVIDUAL TO TEACH? From an examination of teacher supply, the panel has concluded that the answer to the question "Who will teach science and mathematics in the nation's schools?" is heavily influenced by the incentives offered to teachers, former teachers, and potential teachers. This conclusion follows from the results of a long history of studies showing that the supply of skilled labor for particular occupations is sensitive to financial incentives (see, for example, Harris, 1949; Arrow and Capron, 1959; Freeman, 1971~. This section summarizes what is known about the role that particular incentives play in the career decisions of teachers, former teachers, and potential teachers. We discuss in subsequent sections the extent to which the important incentives play a role in teacher supply and demand models, or could play a role in improved models. Although this report is concerned with science and mathematics teach- ers, most of the literature on teacher supply does not distinguish between these teachers and other elementary and secondary school teachers. Con- sequently, we must look to the broader literature for evidence on incentives and teachers' responses to them.
v · - ~ Q CO v so ._ o :~ D a:. L. ._ h Be: I_ o 0 ~4 0 CD 00 1 ~ a:' 00 00 r4 a, S" _ o `'S o 0 ~ CO ~ V C ._ Q I ~ go . SO En o, - E33 o m ~ E~ ~ ~ ~O O O ~ . - .= ~O N ~ a N N ~C ~Ct) ~F U ~U) 0 ~C ~06) N ~ .o ~ a) ~F ~ CO O ~ ~ ~ O O _ .= ~N 0 ~ a _ ~N ~N ~N _t ~N 03 N e~ ~: C P ~ . N ~ ~ O . ~ . - ~_ ~_ ~N z v ~ a ~_ _ _ _ ~N C N _ _ _I _~ .o ~C ~ ~ O V ~ ~1 ~ 4,) ~ 4' bO 4' ~C ~ ·- .~ O ~ ~ a)0 {D CO ~ C~ U~ ~ O O O ~ ~ ~ .. . . . . . . . . . . . . N ~4 ~ ~ (D CO tD O O O 10 ~ c~ _N _ N _ ~ N CO ~ ~ 03 N N e~ e ~oo ~ ~ ~ ~ co ~ ~ _ oo u, u, .. . . . . . . . . . . . . 000 CD tD C9 ~ CC CO C9 ~ ~F ~ O ~ _ N _ N _ ~N 03 _ _I N N N N _ ~_ 3 :^ t ~o ~oo ~o ~N C" CS ~- dJJ 0 · ~CO C ~N ~1O 14 0) Ut _1 00 _i ~C~ Z ~P V N N N N N N N N 0, N N N N . E ~ ~ · E ~, g;; ~ ~ ~ ~ ,,, ', ~ ~ ~ ~ ~ m v P~ cq 0 ~ OO 4, O VE~ 66 - 00 00 - a o . . {6 O O O u: N - L. o ID ._ .~ V ~ ~I CQ
DETE~INING SUPPLY 67 Before turning to this evidence, we want to make clear that this discussion should not be interpreted as implying that financial incentives are the only factors that influence teachers' and potential teachers' career decisions, or even that they are the most important influences. Teachers enter teaching for a variety of reasons to work with children, to experience the satisfaction of helping others, to have a schedule similar to their own children's schedule. Other reasons for entering teaching were given by some of the newly hired teachers interviewed in the panel's case studies. They entered teaching because of particular experiences they had in the past teaching opportunities during college that were rewarding or an outstanding individual high school teacher who served as a role model, for example. Teachers also leave teaching for a variety of reasons" to pursue another occupation, to follow a spouse whose job has been relocated, to engage in full time childrearing. For most teachers and potential teachers, a moderate change in salary, say $2,000 to $4,000, probably does not influence the decision about whether to enter teaching or how long to stay in teaching. However, a critical question is whether such a moderate-sized salary change would influence the career decisions of enough college graduates to have a marked influence on supply. That question is addressed here. It is also important to keep in mind the unit of analysis that provides the focus for particular studies of the determinants of teacher supply. For example, a number of studies report that recruitment efforts by individual school districts have been successful in expanding the quantity and quality of applicants for teaching positions. Presumably the reason is that these efforts have made particular school districts seem especially attractive to a significant portion of the pool of potential teachers. It does not follow, however, that active recruitment policies by all school districts would ~m- prove the quantity or quality of science and mathematics teachers in our schools. Instead, these policies are likely to influence only the distribution of the available supply of teachers among different districts. The impli- cation of this example is that, when evaluating the evidence on responses to incentives, it is important to consider the extent to which the incen- tives alter the quantity and quality of the pool of science and mathematics teachers available to the nation's schools, or whether they influence only the distribution of the available supply among districts. This section focusing on individuals is organized according to what might be called the steps in the pipeline that place teachers in schools: 1. College students' decisions about occupational preparation; 2. The decision about whether to enter teaching; 3. Teachers' decisions about how long to stay in teaching; 4. Former teachers' decisions about whether to return to teaching;
68 PRECOLLEGE SCIENCE AND MATHEMATICS TEACHERS Teachers' decisions about moving from one state to another; and Teachers' decisions about when to retire. College Students' Occupational Preparation Decisions Over the last 15 years, the percentage of American college students training to become teachers has declined precipitously. One indicator of this is the proportion of graduating seniors majoring in education. This proportion has fallen from 22 percent in 1971-72 to 9 percent in 1985- 86 (National Center for Education Statistics, 1988b:210~. This indicator is suspect, however, because an increasing proportion of college students training to teach also major in a particular discipline, for example, math- ematics or biology. As a result, the trend in the number of education majors may provide misleading information about the trend in the number of college students preparing to teach. Unfortunately, no reliable national data exist on the number of individuals obtaining teacher certification each year. This makes it necessary to turn to individual states for information on the number of new certificants. Able 3.4, which provides information on the number of individuals obtaining certification in New York and North Carolina in selected years between 1974 and 1985, illustrates the dramatic decline in the number of individuals obtaining teacher certification in these states. In each state the number of new certificants in 1985 was less than half of the number of new certificants in the mid-l97Os. There are two related reasons why the number of college students training to teach declined dramatically over the last 15 years. The first is the decline in the number of teaching positions available for newly certified teachers a response to enrollment declines. For example, the number of new teachers (that is, teachers without previous teaching experience) hired by public school districts in Michigan declined from more than 6,000 in 1973 to fewer than 700 in 1984. In addition, many beginning teachers lost their jobs as fiscally strapped school districts reduced staff in response to enrollment declines. Since the probability of obtaining a teaching position is a critical factor influencing college students' decisions about whether to train to teach, the decline in this probability was an important factor contributing to the decline in the proportion of college students preparing to teach. A second factor was the decline in teaching salaries relative to salaries offered by business and industry. As depicted in Figure 3.2, teaching salaries fell relative to salaries in business and industry during the late 1970s. Thus, as papers by Manski (1987) and Zarkin (1985) have shown, the combination of the decline in the probability of obtaining a teaching position and the decline in the competitiveness of teaching salaries were strong signals to college students to pursue occupations other than teaching.
DETERMINING SUPPLY TABLE 3.4 Number of People Obtaining Teacher Certification in New York and North Carolina, 1974-1985 Year New York North Carolina 1974 34,770 1975 6,538 1976 24,039 6,413 1977 5,673 1978 5,105 1979 4,684 1980 16,348 3,852 1981 3,145 1982 3,095 1983 3,071 1984 17,275 2,997 1985 16,002 2,830 69 Sources: New York State Education Department (1988~; Murnane and Schwinden (1989:9, Figure 1~. The Decision to Enter Teaching One of the surprising facts about the operation of the teacher labor market is that one-third to one-half of college graduates who obtain teacher certification never teach-or at least do not teach in the state where they obtain certifications One explanation is that teacher certification has traditionally been relatively easy to obtain in most states and, as a result, many college students obtain certification even though they have little interest in teaching. A second explanation is that the decline in the number of teaching vacancies during the l970s left many newly certified graduates without job offers in teaching. The fact that a large proportion of graduates certified to teach do not teach raises the interesting question of who enters teaching and who does not. Recent work by Murnane and Schwinden (1989) indicates that the answer varies across subject specialties and race. They found that the National Teachers Examination (NTE) scores of white certificants trained 1 National data on who is certified to teach are not available, and state-level data provide no information on cenificants who leave the state.
70 PRECOLLEGE SCIENCE AND lL4THE~lTICS TEACHERS $32,000 $28,000 ~- Is cn a) cr: $24,000 $20,000 $16,000 O Physics ~ Chemistry ~ Teachers 0 Mathematics ~ Biology ~ Humanities ) 0~ ~ ~ 1 1 1 1 ~ 1 1 1 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 Year it, ¢\~ ~ ~q Pa i ~ V FIGURE 3.2 Average starting salaries for teachers and for business and industry ($1987) by college graduates' field of college major. Sources: College Placement Council (1988~; National Education Association (1987a). in chemistry, mathematics, and English were important predictors of the probability of entry into teaching. The entry probabilities for white certifi- cants in these areas with scores at the 90th percentile were 10-17 percentage points higher than the enter probabilities for white certificants in these areas with scores at the 10th percentile. (NTE scores made the most difference in the probability of entry for certificants specializing in mathematics.) For white certificants with other subject specialties, the NTE score was not an important predictor of the probability of entry. The likely explanation for this pattern concerns the opportunity cost of becoming a teacher that is, what one gives up if one decides to teach. As presented in Figure 3.2, between 1968 and 1987 starting salaries
DETERMINING SUPPLY 71 in business and industry for college graduates trained in chemistry or mathematics were considerably higher on average than starting salaries for graduates trained in biology or the humanities. If graduates in chemistry and mathematics with high scores on the NTE (a standardized test whose scores are positively correlated with scores on other standardized tests, such as the Scholastic Aptitude Test and the Graduate Record Exam) are more likely to receive offers of high-paying jobs in business and industry than are graduates with low scores, this would explain the negative relationship between NTE scores and probability of entry into teaching for certificants in these fields. (This explanation also implies that job offers in business and industry for graduates majoring in English are more attractive than job offers for graduates majoring in history an assumption that we cannot test.) Although we have not demonstrated that salary differences by field affect career choice, the key point here is that the evidence supports. the proposition that a college graduate's choice of occupation depends on relative salaries. Murnane and Schwinden's study found that there was no negative relationship between NTE score and probability of entry into teaching for black college graduates certified to teach. The reason may be that black graduates faced less attractive job opportunities in business and industry than did white graduates, or that they were more place-bound than were white graduates. The implications that one draws from the negative relationship between NTE score and the probability of entry into teaching for white college graduates trained in chemistry and mathematics depend on one's assessment of the relationship between NTE score and teaching effectiveness. If NTE score were a strong predictor of teaching effectiveness, the results would imply that the profession of teaching is losing a high proportion of the most promising potential teachers. However, in a review of the history of NTE scores, Haney, Madaus, and Kreitzer (1987) find little if any correlation between teachers' NTE scores and other measures of teacher effectiveness, such as supervisor's ratings. What does follow from this evidence is that higher teaching salaries may be a necessary condition for recruiting teachers who have the skills to do well on standardized tests (e.g., to score above the 10th percentile) and who have subject specialties such as science and mathematics that are highly rewarded in business and industry. Where to Teach One of the unique characteristics of public education in the United States is that hiring is done quite independently by 15,000 school districts that establish their own salary schedules (usually through bargaining with local teachers' unions) and design their own recruitment and screening procedures. As a result of differences in salaries, working conditions, and
72 PRECOLLEGE SCIENCE AND MATHEMATICS TEACHERS recruiting practices, there is significant variation in the ability of local school districts to attract and retain skilled science and mathematics teachers. As a result, one must be careful when using the term teacher supply. It is quite possible that districts that pay high salaries, have good working conditions, and have aggressive recruitment practices may be successful in attracting skilled teachers at the same time that districts without high salaries, and districts in which working conditions are difficult, cannot attract skilled teachers at all. Indeed, this is the inference that the panel has drawn from the case studies on district hiring practices and from the conference with personnel directors of large, urban school districts. Unfortunately, relatively little is known about the incentives that are important determinants of school districts' ability to recruit skilled teachers. There is little documentation about the extent to which school districts' recruiting and screening strategies influence their ability to hire skilled teachers, as opposed to less skilled teachers. Typically, even with low salaries, difficult working conditions, and poor recruitment practices, a district can find adults to stand in front of classrooms; however, they are unlikely to be skilled teachers. In other words, the adjustment mechanism concerns quality, not quantity. Consequently, studies that examine only whether districts have applicants for teaching positions, without paying close attention to the skills of the applicants, do not provide reliable information about the influences of school district salaries, working conditions, and recruiting practices on the ability to staff the schools with skilled teachers. How Long to Stay in Teaching One element of teacher supply that has an important influence on the demand for new teachers is the length of time that teachers already in the schools stay in teaching. As Grissmer and Kirby (1987) have pointed out, even a small change in the percentage of teachers who leave teaching from one year to the next (the attrition rate) has a dramatic change on the demand for new teachers. As explained in the panel's interim report, the national teacher demand model used by the National Center for Education Statistics assumes implicitly that: the attrition rate is constant over time; the attrition rate is not influenced by changes in teacher salaries; and the attrition rate does not vary among subject specialties. Recent research examining the factors that influence the length of time that individual teachers stay in teaching call all of these assumptions into question. There is a long history of studies showing that attrition rates follow a U- shaped distribution (see Grissmer and Kirby, 1987, for a list of references). Young, inexperienced teachers tend to have very high attrition rates often as high as 20 percent in the first year. The probability that a teacher leaves teaching declines with experience. Attrition rates are very low for teachers
DETERMINING SUPPLY 73 with more than five years of experience. Finally, attrition rates begin to climb again as teachers near retirement age. A series of studies employing data from Michigan, North Carolina, and Colorado (Murnane and Olsen, 1989a, 1989b, 1990, Murnane et al., 1988, 1989) have demonstrated that teaching salaries are an important determinant of the length of time that teachers stay in teaching. The evidence implies that a $1,000 annual increase in salary (in 1987 dollars) is associated with an increase of one to two years in the median length of time that teachers stay in teaching.2 These studies also show that high school teachers tend to stay in teach- ing for shorter durations than elementary school teachers do-a pattern present in all three states. The studies also find some differences in the career paths of secondary school teachers with different subject specialties. For example, Figure 3.3, which is based on a sample of North Carolina teachers who began their careers in the late 1970s, shows that chemistry and physics teachers tended to leave teaching sooner than did secondary school teachers with other subject specialties. It is important to point out, however, that the sizes of the differences in career paths by subject specialty vary across sample and time period. This is illustrated in Table 3.5, which displays predicted median first spell lengths in teaching for teachers with different subject specialties. As explained in Murnane and Olsen (1990), the predicted survival functions for teachers with different subject special- ties are based on a model in which length of first spell was modeled to be a function of age at entry, gender, subject specialty, NTE score, annual salary expressed in 1987 dollars, and a dummy variable for the district in which the teacher started his or her career. Notice that in the two states in which it is possible to differentiate chemistry-physics teachers from biology teachers, the latter group has a higher median first spell length. This suggests that, in discussing teacher supply, it is important not to treat science teachers as a homogeneous group. This quantitative evidence is also supported by the comments of several personnel directors who were interviewed as part of the panel's mini case studies. The frequent comment was that there were plenty of strong candidates for biology positions but, in some districts, a shortage of strong applicants for teaching some other sciences, especially physics. A number of recent studies have examined attrition rates by subject specialty using an approach different from the research described above, which is based on the analysis of longitudinal data on teachers' careers. These other studies develop estimates of age-specific or experience-specific 21he methodology Murnane and Olsen use controls for time-invariant district-specific charac- teristics, even unmeasured ones. In effect, their model includes a dummy variable for every district.
74 PRECOLLEGE SCIENCE AND MATHEMATICS TEA ClIERS 100 ~ 90 _~ 80 70 ~ Elem. ~ English ~ Soc. Std. ~hv ~Biology ~ 60 C/) 50 40 30 20 10 o - · Math _ L_ I I I I I I I I 1 2 3 4 5 6 7 8 9 10 11 12 Years of Teaching Completed FIGURE 3.3 Predicted survival functions by subject specialty, for a sample of teachers in North Carolina. Source: Murnane and Olsen (1990~. attrition by comparing the rosters of teachers employed in a state or in a school district in two consecutive years (Grissmer and Kirby, 1987~. The analytic strategy is to calculate the proportion of teachers with particular characteristics (for example, mathematics teachers between the ages of 30 and 34) who were teaching in the first year, but not in the second. An advantage of this approach is that it is possible to calculate attrition rates using very recent data. One significant research puzzle is that the studies using longitudi- nal data tend to find greater differences in attrition rates by subject area than do the studies comparing cross-sections of teachers for two consec- utive years. The likely explanation concerns the timing of the data. The research comparing cross-sections tends to be based on data from the mid-1980s, when declining school enrollments led to involuntary attrition in many school districts. These involuntary quits, which tend to be based on seniority, may mask differences in voluntary quits that are sensitive to opportunity cost. The studies using longitudinal data tend to be based on the careers of teachers who started to teach in the 1970s. These teachers may have acquired enough seniority by the time enrollment declines set in to be relatively free from involuntary layoffs. Thus, the attrition patterns observed in the studies based on longitudinal data may be less influenced by the consequences of enrollment declines. If this explanation is correct, then the studies based on longitudinal data studies that show significant differences in first spell lengths by subject specialty may predict attrition patterns in the 1990s better than the studies based on more recent data.
DETERMINING SUPPLY 75 TABLE 3.5 Predicted Median First Spell Length (in Years) of Teaching: Samples of Teachers from Three States State North Michigan Colorado Carolina Period Sample Began Teaching 1975-79 1972-75 1979, 1982 Sample Size (8,462) (7,785) (1,377) Teaching specialty Elementary school 13.5 16.4 6.6 Mathematics 7.9 7.4 4.4 Social studies 6.6 7.6 3.4 English 5.7 7.3 3.1 Biology 5.6 9.6 6.0 Chemistry-physics 4.1 4.9 a a Chemistry-physice teachers cannot be distinguished from biology teachers in the Colorado data. Source: Murnane and Olsen (1990~. The reason is that secondary school student enrollments (age 14-17) will increase in most parts of the country in the 1990s; as a result, involuntary layoff; will be rare, and in some regions there may be a teacher shortage. Whether to Retum to Teaching Until recently, the national teacher supply and demand model used by NCES assumed that newly minted college graduates provide the only source of teacher supply available to fill new vacancies. Recent data from several states suggests that this assumption may be seriously out of line with the current situation. For example, 75 percent of the individuals newly hired by Connecticut school districts for the 1986-87 school year had prior teaching experience and were returning to teaching after a career interruption (Connecticut State Department of Education, 1987~. The
76 PRECOLLEGE SCIENCE AND MATHEMATICS TEACHERS analogous number for New York State for the 1984 85 school year is 70 percent (New York State Education Department, 1987~. These data raise the question of whether the "reserve pool," consisting of individuals fully certified to teach but not currently teaching, will be an important source of supply in the years ahead. The question is difficult to answer since there is little systematic knowledge about the size of the reserve pool, its composition, and the factors that influence members of the reserve pool to return to teaching. Studies based on data from North Carolina, Michigan, and Colorado throw some light on this question by examining whether teachers who left teaching within the first five years after entry3 returned to the classroom after a career interruption (Murnane and Olsen, 1989b; Murnane et al., 1988, 1989~. The evidence indicates that approximately one-third of ele- mentary school teachers return to the classroom after a career interruption. The return rate for secondary school teachers varied by field from 10 to 30 percent, with teachers of mathematics, chemistry, and physics having the lowest return rates. Among high school teachers in Michigan, teachers of chemistry and physics were the least likely to return to the classroom. In North Carolina, secondary mathematics teachers were the least likely to return. This evidence is quite consistent with the notion that former teachers' career decisions are sensitive to relative salaries. Those former teachers with subject specialties that paid relatively high salaries in busi- ness and industry (shown in Figure 3.4) were much less likely to return to teaching than teachers with subject specialties that paid lower salaries in business and industry. Thus, the limited evidence currently available suggests that the reserve pool is less likely to be a significant source of supply of chemistry, physics, and mathematics teachers in the future than it will be a source of teachers in other fields, especially elementary education. But to the panel's knowledge, the size of the reserve pool is unknown. Whether to Move to a Different State As a result of demographic trends that include migration and differ- ential fertility rates, some parts of the country experience shortages of teachers while other parts of the country do not. One logical solution to this problem is migration of teachers from areas of oversupply to areas of excess demand. Clearly some mobility exists, but the design of state pension systems for teachers is a significant deterrent to relocation. As Bernard 3The data bases Murnane and his colleagues examined include 12 years of longitudinal infor- mation on teachers' careers. In examining the return rate of teachem, they focused on teachers who ended a first spell of teaching within five years, in order to provide a significant period of time for teachers to return after a career interruption.
DETERMINING SUPPLY field in which college graduates were trained ~200 ~ ~ Math ~Physics ~ Chemistry In 180 - ~ Humanities ~Biology ~ K-12Teaching =~ 160 t ~_ ~ no ~ 120 Be 100~= 1 1 · · · · · o 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 Year 77 FIGURE 3.4 Starting salaries in industry as a percentage of starting salaries in teaching. Sources: College Placement Council (1988~; National Educational Association (1987a). Jump (1986) has explained, the design of most state pension systems is such that a teacher who moves from one state to another ends up with a significantly smaller pension than do teachers who continue to teach in the same state. A second deterrent is the character of state certification systems. While a significant amount of reciprocity among states exists, it is not always possible for a teacher certified in one state to obtain certification automatically in another state. In fact, recent efforts to stiffen certification requirements in individual states may reduce the amount of reciprocity. The designs of pension systems and certification systems influence the balance between teacher demand and supply because they influence the extent to which a critical equilibrating mechanism, movement of teach- ers from areas of teacher surplus to areas of teacher shortage, operates. Changes in pension portability rules, a policy recently discussed by the Na- tional Governors' Association (1988), may have a significant influence on the rate of interstate mobility of teachers, and consequently, may influence the balance of teacher supply and demand. When to Retire The conditions that influence the timing of retirement for teachers are similar to those for individuals in other occupations. Teachers tend to retire when they feel financially able to do so; when incentives for early retirement are considered worthwhile; and when increases in benefits for continuing another year or two are minimal (Taylor, 1986~. The most common provision for maximum retirement benefits occurs at 60 years of age and 30 years of service. States provide for early retirement below these maximum levels with penalties for either years of service, age,
78 PRECO' if EGE SCIENCE AND MATHEMATICS TEACHERS or a combination of both (Ginsberg et al., 1989~. States vary from these provisions, with some having easier and some stricter provisions. Five states require only 25 years of service for maximum benefit; 10 states require that teachers be employed 30 years and be 65 years of age to obtain maximum retirement payment; 7 states provide maximum payment for teachers 55 years of age with 25 years of service. Between 1979 and 1982, special early retirement incentives were offered by some school systems and states (Wood, 1982~. Until recently, studies seemed to indicate that teachers tend to retire at the earliest age allowed; this would mean approximately age 60 Taylor, 1986~. In South Carolina, a study asked teachers if they would retire if there were a hypothetical change in the provisions of the retirement law. The current provisions permit retirement after 30 years of service, or at age 65, or at age 60 with a penalty. The change would permit retirement after 25 years of service or at age 55. The responses showed this change would increase separations by 5.5 percent above those under the current law, and a change to retire at any age after 25 years of service would increase separations by 16.6 percent above those under the current law, or 10.5 percent above the level with age 55 (Ginsberg et al., 1989~. However, teachers in some states deviate from this pattern. For example, in 1983, approximately 20 percent of the teachers employed in Maryland were eligible to retire. Only about half of these teachers had retired by 1988. Their decisions were undoubtedly heavily influenced by the state's policy with respect to provision of the employee's share of health benefits and cost-of-living salary adjustments a policy that did not have an age cap. This combination of incentives to stay beyond retirement age is unusual 41 states have an age cap on provision of health benefits and 20 place an age cap on cost-of-living benefits. Policies about provision of health and cost-of-living benefits obviously influence decisions to retire or delay retirement. Many factors other than age can affect the decision to retire. When a large wage increase is expected, some teachers will delay retirement to obtain a higher average salary. A change in options for health plans can hasten or impede retirement, especially when retirees can not change plans after separation. When a large cost-of-living adjustment in retirement benefits is expected, retirements can increase. During a period of large cost-of-living increases, such as in the late l970s, retirements can decrease because of worry about having sufficient retirement funds. Since these factors affect all teachers, the combined influence may lead to a bunching of retirements. Although separation rates for retirement are known by the state re- tirement systems (Kotlikoff and Smith, 1983), this information has not been used either for strengthening supply-demand statistics related to science and .
DETERMINING SUPPLY 79 mathematics teachers or for research. ~ obtain a better understanding of the retirement decision, research is needed on the relationships among separation rates, individual reasons for early or late retirement, external variables, and incentives and disincentives for remaining in the teaching force or retiring. In short, although we know the effect of separation rates on states, we also need information about their effect on individual school districts and on the teaching force by level and by discipline or subject taught. Conclusions The basic theme of this section is that the number of teachers in a given field willing to work in a given location depends on a number of incentives, including the availability of teaching positions, salaries, opportunity COSt salaries, working conditions, certification rules, and pension rules. Changes in any of these incentives will influence supply. For two reasons, however, the panel is skeptical about the feasibility of improving projection models by incorporating the influences of these incentives. First, there is a great deal of variation across school districts and across states in salaries and working conditions, and it is difficult to incorporate this variation in an aggregate model. Second, while a number of studies show that teachers' career decisions depend on salaries, the coefficients indicating the sizes of the impacts are quite sensitive to sample definitions and estimation techniques. In other words, research is not sufficiently developed to provide reliable estimates of the response coefficients that could be included in projection models. For these reasons, the panel does not recommend at this time the development of projection models of teacher supply and demand that include responses of teachers and potential teachers to changes in incentives; however, the panel does urge support for the development of better behavioral models to measure the sensitivity of teacher supply to incentives (see Chapter 6~. The panel also recommends that resources be devoted to monitoring trends in the levels of key incentives that influence teacher supply. Timely information about changes in the proportion of newly certified teachers who obtain teaching positions, the competitiveness of teacher salaries relative to opportunity cost salaries, the amount of reciprocity in certification across states, and the portability of teacher pensions may provide policy makers with early warnings about likely changes in the supply of teachers. A number of research issues related to the decision about when to retire and its effect on teacher supply need investigation. Separation rates for retirement are known by state retirement systems and could be used in research relating these rates to individual reasons for separations, to external shock variables, and to incentives and disincentives for retention
80 PRECOQ;EGE SCIENCE AND MATHEMATICS TEACHERS and retirement. Related research issues include the effect of separation rates on individual school districts and on the teaching force for different fields of study. HOW DOES A DISTRICT MESH SUPPLY WITH DEMAND? School district policies, practices, and constraints exert considerable influence on an individual's decision to apply for a position, take a position, remain in a position, or leave. In teacher supply and demand models, school districts are treated as black boxes. No models incorporate information on school districts' recruitment, screening, and hiring processes in the structure used to generate predictions of teacher supply and demand. As a result, the design of the models implicitly assumes that variation in these practices do not have a marked impact on the ability of individual school districts to attract skilled math and science teachers. While the variation in practices does not matter in using models to project the supply of teachers in a state or in the country, it does suggest that the models do not provide reliable information about the supply of skilled teachers available to individual school districts. Recent case study evidence (Berry, 1984; Wise et al., 1987) has called this assumption into question by pointing out that there is considerable variation among school districts in recruiting, screening, and hiring practices and that these practices may have a marked influence on the ability of school districts to hire skilled teachers. For this reason, the panel wanted to learn more about the school district practices that affect the supply and demand of science and mathe- matics teachers. We were concerned with the flow of teachers through and within the school districts. We also hoped to obtain insights concerning variables at the school or district level that affect demand and supply. ~ pursue this goal, the panel first commissioned the development of detailed case studies of the recruitment, selection, and retention of science and mathematics teachers in six districts, which varied in size, student clientele, enrollment trend, wealth, and location. Second, the staff, in supporting the panel's activities, conducted supplementary case studies focusing on supply and demand issues affecting science and mathematics teachers in 24 districts. These supplementary studies used telephone conversations with personnel directors and a follow-up mail survey to collect more detailed information about the hiring of science and mathematics teachers in each district. Third, the panel convened a conference of the personnel directors of seven of the nation's largest public school districts, which represent over 5 percent of the public school enrollment in the United States. Topics for discussion in the day-and-a-half-long meeting included: effective recruiting strategies, experiences with the reserve pool, recruitment during the school
DETERMINING SUPPLY 81 year, suppIy-demand models, and information system design and use. (See Appendix A for more detail.) In both the case studies and the conference of personnel directors, differences among the districts were as apparent as the commonalities. Observations made by the school district officials during the course of these three activities are noted throughout the report. This section summarizes the lessons the panel has learned from these activities concerning the variation in school district practices and the per- ceptions of how practices influence school districts' success in recruiting skilled math and science teachers. The section is organized by topics cor- responding to the following elements of the hiring process: determining needs, soliciting applicants, screening applicants, and making offers. The section emphasizes school district practices because they provided the fo- cus for the case studies and the conference with the personnel directors. However, it is critical to keep in mind that determination of who teaches in the schools depends not only on these practices, but also on applicants' responses to these practices. Determining Needs Knowing how many new teachers of each subject at each grade level will be needed in the coming year is a critical first step in planning a hiring strategy. Yet, for many districts, it is extremely difficult to collect this information in a timely fashion. Some reasons are detailed below. Uncertainly About Student Enrollments Student enrollments are the primary determinant of the demand for teachers. Based on comments from the personnel directors of seven large school districts, it appears that projecting future enrollments accurately is difficult to do, especially in districts experiencing significant in-migration or out-migration. Since there are many such school districts, the panel infers that many districts do not have reasonably accurate projections of enrollments for a given year until the students actually appear in September. When finances preclude flexibility in the ratio of the number of students to teachers a situation present in most of the districts included in the case studies-teachers cannot be hired in anticipation of enrollment increases. This inability to offer firm contracts to strong applicants in late spring, when many applicants desire commitments of employment, hinders many districts' efforts to hire skilled teachers.
82 PRECO! :~;EGE SCIENCE AND MATHEMATICS TEACHERS Uncertainty About Budget for the Next School Year Another related problem is that in the spring when hiring takes place, districts may not know the budget for the next school year. According to union contracts, this forces the district to inform large numbers of teachers in March that they will not be employed the following year. Then when the budget is assured, they find that some of the people who receive notices have found other work. Thus, the district must look for new people. Internal Transfer Queues In many districts, contracts with teachers specify a formal procedure under which teaching vacancies are made available to teachers already employed by the district, before they can be filled by a newly hired applicant. Completing the steps of the internal transfer process often takes several months. Until the process is completed, the personnel office cannot be sure of the identity of the school in which a vacancy will ultimately be present, or even of the teaching specialties that will be needed. Delays in Reporting Resignations For a number of reasons, teachers may delay reporting that they plan to resign their positions. One reason is that some contracts specify that teachers employed by the district on the date on which a new contract is signed are eligible for certain fringe benefits included in the new contract, such as improvement in health benefits. As a result, teachers wait until a new contract is signed, which often runs into the summer months, before resigning. Another reason for late resignations is that some school princi- pals will ask teachers who intend to resign to withhold formal notification so as to subvert the internal transfer process. Principals do this to gain control over who fills the vacancy. One consequence of this practice, however, is a wave of resignations in late summer, when it is difficult to find qualified applicants. Attrition During the School Year While most suburban districts and smaller-sized districts tend to con- centrate their recruiting on finding strong applicants in late spring to fill vacancies expected for the following September, many urban districts hire teachers throughout the year to fill unanticipated vacancies resulting from teacher resignations and unexpected enrollment growth. In fact, the per- sonnel directors from several urban districts reported that as many as half of the teachers they hire are asked to start teaching during the school year, rather than in September.
DETERMINING SUPPLY 83 Soliciting Applicants School district personnel directors use a variety of strategies to recruit applicants for teaching positions. These include recruiting at nearby col- leges and universities, relying on informal networks of information about individuals not currently teaching who are interested in returning to the classroom, and, in one urban district, recruiting graduate students to teach part time. These are a few examples of the many strategies that personnel directors described as effective for finding applicants. The need to recruit varies greatly among school districts, and large differences are observed in ratios of applicants to vacancies reported by districts in the same labor market area. For example, in the Washington, D.C., metropolitan area, the District of Columbia has great difficulty at- tracting applicants; it reports about three applicants for each teaching job. By contrast, suburban Montgomery County reports a 13 to 1 ratio; and Prince George's Counpr 8 to 1 (Sanchez, 1989~. The wide array of recruit- ing strategies employed reflects such differences in the ability to attract applicants. School systems may advertise and make trips to job fairs or colleges where they have successfully recruited in the past. If personnel officials feel there is a particular shortage, special early offers may be made. At times, to eliminate a particular shortage, special incentives, such as a bonus, may be offered. What seems clear from personnel administrators is that many school systems are searching nationally, or at least beyond their local or state borders, for persons in similar fields. One year the quest may be for science and mathematics teachers; another year it will be for reading teachers; still another year, it may be for early childhood teachers. Lately a widespread need has been for special education teachers and for teachers of the same ethnic backgrounds as those of the students in the district. According to the personnel directors of large school districts who shared their experiences with the panel, recruiting generally was restricted to known sources, because experience had taught recruiters that persons unfamiliar with the climate, housing costs, student populations, or culture were unlikely to remain in their systems. While there was enormous variation in the way personnel directors found applicants, some patterns emerged. First, almost all personnel di- rectors indicated there was no shortage of qualified applicants for teaching positions in science or mathematics at this time. (Most districts did re- port shortages of minority applicants and applicants for special education positions.) Exceptions were a few cases of an inadequate supply of appli- cants to teach physics. Several respondents commented that the supply of qualified applicants for each vacancy in biology was considerably greater than the ratio of qualified applicants to number of vacancies in chemistry
84 PRECOLLEGE SCIENCE AND MATHEMATICS TEACHERS or physics. Second, most respondents indicated that a large percentage of their applicants were individuals with previous teaching experience. While it is hazardous to make inferences about patterns in recruiting strategies from an unrepresentative sample, the following patterns seemed to be present in the survey information. School districts that paid relatively low salaries, particularly districts in rural areas, relied especially heavily on attracting applicants who had grown up in the area and were eager to return home. Districts that paid high salaries and offered attractive working conditions found that many of their applicants were teachers currently employed in nearby districts. Urban districts that needed large numbers of new teachers each year were more likely than other districts to engage in national recruiting strategies. While some personnel directors of large districts indicated that they did find national recruiting worthwhile, they were quick to point out that hiring applicants from regions of the country with very different climates often led to very high turnover rates. As a result, they had learned to concentrate their recruiting efforts on geographical areas that had supplied a relatively large number of applicants in the past, and had found it particularly fruitful to recruit in areas in which teachers were being laid off as a result of declining enrollments and budget cutbacks. Screening Applicants The strategies used to screen applicants for teaching positions, in- cluding who does the screening and how it is done, varied considerably from district to district. Our discussions with school districts indicated that recruitment of new teachers by large school systems with diverse student populations was often hindered by the fact that recruiters could not specify the school to which the applicant would be assigned. Many persons would find such a school system desirable if they could teach in a given section of the school system or in a specified school. Since recruiters could not make those promises or could not make those promises soon enough in the recruitment period, candidates were lost to the school system. In other school systems of various sizes, however, the school and the position was able to be specified early in the screening process. Ib provide a sense of the variation in screening applicants, we describe the screening practices in two school districts in the Northeast-the sub- ject of a recent Harvard University doctoral dissertation (Shivers, 1989~. (The districts are not identified because a commitment of confidentiality was made to the study districts.) The first district is an ethnically and socioeconomically diverse community adjacent to a central city, and has a long-standing reputation of providing excellent education. The second is an urban district with a history of budget problems and difficulties in raising
DETERMINING SUPPLY 85 the achievement of a clientele containing a large proportion of low-income and minority children. In the first district, the screening process is extremely decentralized. The central personnel office weeds out unpromising applicants and passes on to school principals a list of promising candidates. In the words of the acting superintendent, who had been personnel director for a great many years (as quoted in Shivers, 1989~: In a nutshell, our aim here is to use the central staff to do a paper screening of the candidates, to do some initial interviewing, and then to forward as quickly as possible as many reasonable candidates as possible to the building principal or the curriculum coordinator [department chair] at the secondary level . . . and then to let them do the selection. That is to say that we sort of send out a group of people with the Good Housekeeping Seal of Approval that are a general batch. And from that general batch the principal should choose. At the high school level, the school principal delegates to the depart- ment chair authority to choose among applicants. The logic underlying this practice is that chairs are responsible for the quality of instruction offered in their departments and for evaluating teachers. Consequently, they should be responsible for hiring the teachers who will provide the instruction to students. Using open-ended questions, chairs interview each candidate sent from the central personnel office. They also call references. They are not obliged to choose among the candidates sent to them. If none seems satis- factory, they can ask the personnel director to find other candidates. They can also use their own informal networks, such as professional associations and experiences with substitute teachers, to find candidates. Chairs indicated that they do not attempt to hire teachers fitting one mold. Rather they look for candidates who know their subjects, demonstrate evidence of teaching skills, and also do something special, so as to maximize the probability that they will appeal to a subset of the school's diverse student population. One indication of the diversity that is sought is that in one year during the early 1970s, when 115 teachers were newly hired, there were 87 different graduate backgrounds, 25 states, and 7 countries represented (Shivers, 1989~. One attribute that chairs do seek in applicants is some (but not too much) teaching experience. One chair summarized this priority by stating that she did not want a "person with a B.N degree and no experi- ence .... This is too complicated a school to take children to teach children. If I had my druthers they [great candidates] would have had two years experience somewhere else so they would have made their really bad [teaching] mistakes somewhere else" (Shivers, 1989~. Although the superintendent is formally responsible for hiring teachers, in practice the authority is delegated to principals, who in turn delegate it to chairs. In
86 PRECOLLEGE SCIENCE AND hl`4THEM`4TICS TEACHERS fact, after the chairs complete their interviewing and choose the candidates they want to hire, they convey their choice to the winning candidate and also call the candidates who are not chosen. Thus, the decentralization goes beyond advice-giving. In practice, the department chairs choose the teachers. In the second district studied in this doctoral dissertation, the screening process is markedly different. Principals and building-level department chairs have only a minor role. The superintendent and central office assistants play the major role in determining who will be hired. As In the first d~stnct, the process begins with a central office screening of the credentials of applicants. Potentially acceptable candidates are asked to come for interviews. It is at the interview stage at which the process In this d~stnct is so different from the one-on-one interviews between candidates and building-level department chairs that were used in the first d~stnct. Shivers (1989) describes the interview process used In the second district as follows: All new teacher candidates are asked to gather at the same time at the . . . high school gymnasium to be interviewed. Interview panels, which are put together by the personnel director and by department heads, include three to six interviewers [typically including] the appropriate district-level department chair, a secondary school principal, a building- level department chair and another teacher from the department, and one or two central office administrators .... Teachers are called one by one to face a panel of interviewers who are seated at a table on the gymnasium floor, out of earshot of the candidates waiting in the stands .... Lists of questions are prepared beforehand by the respective district- level department chairs. Before the interviews, panel members choose from the list the five or six questions that their panel will ask. The same questions in the same order must be asked of each candidate .... During the interview, the panel members rate each answer as positive, negative or neutral .... Panel members are not permitted to respond to the candidate's answers, and no follow-up questions are permitted .... The strict procedure for interviewing was developed in response to concerns voiced by the affirmative action office . . . that there be no preferential treatment of candidates for teaching jobs. After all candidates have been interviewed, the respective teams rate their candidates as highly recommended, recommended, or not recommended. Generally, they do so by consensus. The ratings are sent to the assistant superintendent for personnel who then checks references of recommended candidates .... The superintendent or assistant superintendent may interview top candidates after they have been recommended by the panels. At this stage the superintendent will make the final selection.
DETERMINING SUPPLY 87 Clearly, the screening processes in the two districts studied by Shivers (1989) are extremely different. The experience of applying for a teaching position differs greatly in the two districts. What cannot be known from Shivers' work is whether the screening practices influence who ultimately is hired. This could occur because the department chairs in the first district look for skills that are different from those sought by the superintendent in the second district. It could also occur because the screening processes influence the size and quality of the applicant pool or the rate at which po- tentially effective teachers accept job offers. Informal networks of college placement officers, college faculties, or students may have a great deal of information about how districts screen applicants and the effects on appli- cants. Unfortunately, there has been virtually no systematic research about potential applicants' responses to differences in recruiting and screening practices. The case studies commissioned by the panel revealed considerable variation in screening practices among districts both in the degree of centralization and in the relative roles played by paper credentials, test scores on standardized written tests (which some districts administer as part of the screening process), and interviews. While the case studies do not provide a basis for describing the distribution of screening practices among the nation's 15,000 school districts, they do verify that practices vary enormously. They also raise the question of the extent to which variation in these practices influences the ability of school districts to hire teachers who are effective in teaching math and science to students. Making Offers The case studies revealed enormous variation in the types of offers made to candidates whom school districts would like to employ. Dimensions of the variation include timing, specificity concerning the nature of the position, and salary. Timing Personnel officers in some school districts, especially well-financed, growing districts, are authorized to offer binding contracts to strong can- didates before the exact number and composition of vacancies are known. Several personnel directors suggested that this practice facilitates their re- cruitment efforts by allowing them to recruit aggressively in colleges and universities during the spring months and to sign up promising candidates before other districts had ascertained the number and nature of their vacan- cies. Other personnel directors told about the other side of the coin, losing promising candidates because their districts prohibited offering contracts
88 PRECOLLEGE SCIENCE AND AL4THEMATICS TEACHERS until firm information on vacancies was available, which often took until late summer. Specificity In the first of the two districts that were studied by Shivers (1989), being offered a contract meant that the candidate knew a great deal about the position: the school building, the subjects to be taught, the grade levels, and the name of the department chair. In the second district, a teaching contract meant only a commitment to salary. Not only did the newly hired teacher not know the building or the classes to be taught, but also the new teacher did not know the date on which this information would be available. The case studies commissioned by the panel indicated that the exam- ples described by Shivers are not particularly unusual. Epically, in smaller districts, candidates are told more about the details of their teaching po- sition than in larger districts. However, in some large districts Stan at the school site play a significant role in the screening process and, in these districts, candidates are often hired to teach in a particular school. Salary In the 24 districts included in the panel's supplementary case studies, the starting salary for a candidate with a B.N and no teaching experience ranged from $14,420 to $26, 061. The starting salary for a teacher with an M.N and the maximum amount of experience that the district rewarded ranged from $25,956 to $47,941. Some of the differences in salary scales were responses to differences across communities in the cost of living. However, the comments of the large district personnel officers indicated that the salaries they could offer played a significant role in the ability of school districts to attract a strong applicant pool and to capture the most capable candidates from the pool. Another important aspect of hiring practices revealed by the case studies is that the formal salary schedule in many districts does not totally determine the salary offered to a newly hired teacher. For example, Dade County offers a $1,000 signing bonus (paid in the first check of the second contract year) to new hires in shortage areas. The first of the two districts Shivers studied sometimes convinces especially strong candidates in short- age fields to sign contracts by giving credit in terms of steps on the salary schedule for practice teaching and for experience outside teaching. Current contracts in Boston and Rochester include specific language allowing the district to do the same thing.
DETERMINING SUPPLY 89 Not all variation from the salary schedule involves pay increases. The second of the two districts Shivers studied frequently hires teachers as permanent substitutes instead of as regular contract teachers. This provides an annual savings of $9,000 to the fiscally troubled district. In this district, attempts are made to attract strong candidates in science and mathematics by offering them regular teaching contracts instead of positions as full-time substitutes. Unfortunately, no information is available on the impact of this practice on the district's ability to attract strong candidates. Who is Hired The case studies revealed enormous variation in the practices school districts use to recruit. screen, and hire teachers. It is not possible from these studies to determine the extent to which the variation in practices influences the ability of districts to attract strong candidates. In fact, the case studies revealed that there is not even a common definition of a strong candidate. The remarks of personnel directors suggest that districts' constraints and practices do matter. For example, the notes contain many comments about losing candidates either because salaries were not competitive or because the district could not make a firm contractual offer, while another district could. Some districts can hire early, and those that can have a better choice of candidates. Maintaining close ties with a local teaching credential program also helps bring strong candidates. Moreover, there is the distinct possibility that school district practices matter less in the late 1980s than they will in the l990s. The reason concerns the potential change in the overall balance between teacher supply and demand. With the exception of a few fiscally constrained urban districts, most districts included in the case studies reported an adequate number of qualified candidates for each vacancy in mathematics and science. Most districts also reported that many applicants were experienced teachers, and that they filled a large proportion of vacancies with experienced teachers. This reliance on older or experienced applicants raises the question of whether the responses of personnel directors in 1987 and 1988 provide reliable predictions of the adequacy of the supply of qualified math and science teachers in the years ahead. In the late 198Os, the demand for new secondary school teachers is relatively low because high school enrollments are not growing. At the same time, the reserve pool of individuals certified to teach but not currently teaching appears to be quite large, in part because it contains many individuals from the large cohorts born at the tail end of the post-World-War-II baby boom. The l990s will be characterized by growing demand for science and mathematics teachers, both because of modestly growing secondary school enrollments and of increasing numbers of resignations from an aging teaching force. At the same time, the size
go PRECOLLEGE SCIENCE AD ~THE~TICS T~CHE~ of the reserve pool may decline, both because the number of individuals in the 30-40 age group will be smaller and because the anticipated general labor shortage will bring about more competition for all skilled workers. In an environment characterized by a shortage of qualified applicants for teaching positions, school district practices in recruiting, screening, and hiring teachers may have a considerable impact on the distribution of qualified teachers across school districts. Conclusions The evidence on school district hiring practices has two implications for understanding teacher supply and demand. First, as the demand for new hires increases in the 1990s due to increases in both student enrollment and teacher recruitment rates, recruiting, screening, and hiring practices are likely to have a much greater impact on a school district's ability to attract skilled science and mathematics teachers than is the case in the late 1980s. Districts that are able to offer attractive salaries and working conditions, to recruit aggressively, and to make offers in a timely fashion will be much more successful in attracting skilled teachers than districts that cannot. As a result, the variation in practices that currently exists may result in significant disparity in the quality of new hires attracted to different school districts. A particularly disturbing aspect of this prediction is that districts serving large numbers of disadvantaged children tend to have hiring practices that do not contribute to attracting skilled teachers. District practices through which seniority rules may restrict new hires to the least desirable school in the district or which introduce a long waiting period before vacancies can be opened to outside applicants are disincentives, as is uncertainty of initial school assignment. Consequently, increased competition for skilled teachers is likely to result in an additional factor contributing to the set of reasons that such children tend not to receive high-quality education. A second implication of the qualitative evidence described in this section is that increasing teachers' salaries, although perhaps a necessary condition for attracting more skilled teachers to individual school districts, is not a sufficient condition. For example, it is unlikely that significant salary increases in the second district Shivers studied would lead to improved school faculties unless screening practices are reformed. One might argue that these two implications drawn from the case studies are not particularly relevant to assessments of the adequacy of teacher supply and demand models. This would be correct if the goal of the models is seen in terms of assessing the overall balance between the demand for teachers and the supply of teachers. However, to the extent that the models are used to measure how well all school districts are able to provide qualified mathematics and science teachers to all children, then
DETERMINING SUPPLY 91 differences in the practices school districts use to recruit, screen, and hire teachers are extremely important. SUMMARY We have taken a close look at the effects of incentives to pursue a teaching career in science or mathematics on the supply of teachers. The factors influencing the individual's choices are beyond what the normal projection model can capture. Although better behavioral models are needed to measure the sensitivity of teacher supply to incentives, a number of research issues will have to be investigated before the models can be developed. In addition, although the teacher supply and demand models con- sidered in the panel's interim report do not use school districts as units of analysis, many decisions are made at the district level that affect sup- ply and demand. As our case studies and interviews with school district personnel directors have shown, school districts vary greatly in the initia- tive they exert to fill their demand for teachers of subjects experiencing shortages. Individual maneuverability in recruiting and special or external circumstances affecting a district are key factors that influence a district's science and mathematics supply and demand situation and these factors may outweigh those factors that can be quantified for modeling. These realities are central to the workings of supply and demand for science and mathematics teachers and should be monitored to the extent possible, as described in the following chapter and in Chapter 6. Chapter 6 concludes by recommending a series of conferences with a sample of school districts held on a regular basis, to discuss these factors and explore ways of rec- ognizing them in statistical and descriptive reports on teacher supply and demand.
