Printer Friendly

Are special educators' career paths special? Reports from a 13-year longitudinal study.

This nation has a serious shortage of qualified special education and related services professionals. Projections of both student and professional demographic data indicate that over the coming years the shortages will reach crisis proportion and seriously impede the ability to provide students with handicaps the special education and related services they are guaranteed under Federal Law. (A Free Appropriate Education: But Who Will Provide It? Testimony to the Senate Subcommittee on the Handicapped and the House Subcommittee on Select Education, April, 1989)

* Public Law 94-142, The Education for All Handicapped Children Act (EHA), now known as the Individuals with Disabilities Education Act (IDEA), guarantees a free appropriate public education to all youngsters with educational, emotional, developmental, or physical disabilities. As EHA's authors recognized, fulfillment of this mandate requires an adequate number of qualified teachers who can deliver the prescribed specialized instruction and services. Since its implementation in 1977, however, despite EHA's provisions for a Comprehensive System of Personnel Development, school districts across the United States have reported difficulties attracting and retaining sufficient numbers of special educators. In 1991, the Office of Special Education and Rehabilitative Services (OSERS) estimated that our schools still needed 28,000 additional special educators (10% of the U.S. special education teaching force) to fill vacancies and to replace uncertified staff.

In response to these reports and to predictions of deeper shortages ahead, The Council for Exceptional Children and the National Association of State Directors of Special Education joined other professional groups during EHA's 1989 reauthorization hearings and testified that unless we train more special educators, "there will be a major deterioration of both the availability and quality of special education for our nation's children with handicaps" (A Free Appropriate Education, 1989, p. 3). Alongside familiar pleas for increased personnel development funds was a newer request--for better data and more research on the special education work force. The group called for the most basic information--data describing who teaches special education, who stops teaching special education, and what factors affect these career decisions. Although OSERS annually tallies the number of special educators working in the schools, even this federal office recognizes that such data tell us little about the supply of or demand for special educators (OSERS, 1989).

Special education is not the only area with this knowledge gap; we actually know very little about the career paths of any teachers. The dearth of knowledge is so severe that a 1987 panel of the National Academy of Sciences that examined the supply of and demand for all types of teachers, especially mathematics and science educators, could not find answers to such simple questions as: How long do teachers teach? Why do some teachers stay? Why do others leave? (National Research Council [NRC], 1987).

During the past 5 years, my colleagues and I have been using a statistical methodology relatively new for educational research--survival analysis--to address these and related questions for public school teachers with regular classroom assignments (Kemple, Murnane, Singer, & Willett, 1992; Murnane, Singer, & Willett, 1988, 1989; Murnane, Singer, Willett, Kemple, & Olsen, 1991; Singer & Willett, 1988). Survival analysis has enabled us to uncover many findings about regular educators' careers that had heretofore eluded documentation--that salary does make a difference; that the schools are losing academically talented teachers; and that teachers of chemistry and physics, but not teachers of mathematics, are at greatest risk of leaving.

In this article, I use survival analysis to explore the career paths of special educators--to show how their career decisions resemble, and differ from, the career decisions of regular educators. In a companion article, Singer (in press), I use survival analysis to determine whether those special educators who leave eventually return. The analyses are based on 13 years of longitudinal data describing 6,600 special educators who began teaching in the North Carolina and Michigan public schools between 1972 and 1983. By describing how long special educators teach and by identifying who among them chooses to stay and leave, we can begin to better understand who today's special educators are and who tomorrow's special educators might be.


Our ignorance about teachers' career paths in general, and special educators' career paths in particular, is powerfully illustrated by the simplicity of the National Center for Education Statistics' (NCES) teacher supply and demand model. Originally based on a single annual attrition rate of 6.0% (Metz & Fleischmann, 1974), recent refinements assume two new lower rates: 4.9% for elementary school teachers and 5.6% for secondary school teachers (Gerald, Horn, & Hussar, 1989).

Critics of the NCES model question whether one or two rates, (based, incidentally, on turnover between 1983 and 1984), can accurately describe attrition for all teachers over time (Darling-Hammond & Hudson, 1990; Haggstrom, Darling-Hammond, Grissmer, 1988; NRC, 1987). The NCES model assumes that the teaching force is static--that its composition remains stable over time and that attrition rates are constant across people and jurisdictions. Yet years of research show that the likelihood of leaving teaching differs by a teacher's age, gender, years of experience, academic achievement, level taught, subject specialty, and salary (see, e.g., Charters, 1970; Chapman, 1983, 1984; Chapman & Hutcheson, 1982; Heyns, 1988; Nelson, 1985; Murnane et al., 1988, 1989). Studies comparing teachers hired in different years also find that career paths change over time (Grissmer & Kirby, 1987; Mark & Anderson, 1978, 1985; Schlecty & Vance, 1981; Singer & Willett, 1988). Even NCES recognizes that its two new lower annual attrition rates may already be inaccurate: In its first Schools and Staffing Survey, only 4.1% of the nation's public school teachers left their jobs between 1987 and 1988 (Bobbitt, Faupel, & Bums, 1991).

From a special education perspective, a more serious problem with the NCES model is that it does not distinguish between regular and special educators. Without data germane to special education, OSERS routinely applies NCES' 6% attrition rate when projecting personnel needs, sometimes citing the rate as "normal" (OSERS, 1979, p. 56), sometimes using it without justification (OSERS, 1984, p. 35). Such extrapolation is understandable, if inappropriate, given the variability of special education attrition rates reported in the literature--ranging from 1% to 24% (Shofer & Duncan, 1986) to as high as 30% to 50% (Schrag, 1986; Smith-Davis, Burke, & Noel, 1984; Thomas, 1982).

So the core question becomes whether special educators' career paths resemble, or depart from, those of regular educators. Although few studies have addressed this question directly, several related lines of research suggest that they may well differ. First, teachers are not interchangeable; they are licensed to teach specific levels and subjects. Because the career paths of regular educators differ by subject specialty and school level (Chapman & Hutcheson, 1982; Charters, 1970; Grissmer & Kirby, 1987; Heyns, 1988; Murnane et al., 1988, 1989; see Whitener, 1965, for an exception), so, too, may they differ from the career paths of the special educators.

Noninterchangeability of teachers leads to a second distinction between the two groups: their labor markets. The 1975 passage of EHA increased the demand for special educators just when the demand for regular educators was decreasing. Between 1976 and 1988, as public school enrollment dropped by 10%, the number of public school students receiving special education services increased by 21% (to 4.5 million, or 11% of enrollment) (NCES, 1989, 1990). Enrollment declines deterred many college students from considering teaching, but those who did prepare to teach apparently responded to the changing needs of the student body. Between 1970 and 1980, as the number of college graduates licensed to teach dropped by half, special education's representation in the pool nearly quadrupled, from 4% to 15% (National Education Association, 1982, p. 23). Although the number of college degrees awarded in special education plummeted during the 1980s (National Clearinghouse for Professions in Special Education, 1988), the number of special educators employed in the schools continued to increase, up 66% between 1976 and 1986 (NCES, 1988; OSERS, 1989). The net result: Special educators now comprise 13% of the U.S. teaching force (NCES, 1990; OSERS, 1990).

Increased demand for special educators, coupled with decreased demand for regular educators, leads to a third disparity between the two groups: their age and teaching experience. Comparing nearly 1,000 special and regular educators in five metropolitan school districts, for example, the Collaborative Study of Children with Special Needs found that special educators were younger, with an average of 4 years less teaching experience (Singer & Butler, 1987; Singer & Raphael, 1988; Raphael, Singer, & Walker, 1985). Because younger teachers with less experience are more likely to leave teaching (Murnane et al., 1988, 1989), and a larger proportion of special educators fits this description, it is not surprising that attrition rates (computed for teachers of all ages) are higher for special educators (see, e.g., Bogenschild, Lauritzen, & Metzke, 1988; Cross, 1987; Parshall, 1990).

Some observers have marshalled a fourth argument for differences between the career paths of the two groups: purportedly higher levels of stress and burnout among special educators. Special education journals all too often offer compelling vignettes and noncomparative surveys describing stress and burnout in the profession (e.g., Chandler, 1983; Fimian & Santoro, 1983; Holland, 1982; Johnson, Gold, & Vickers, 1982; Lawrenson & McKinnon, 1982; Olsen & Matuskey, 1982; Raschke, Dedrick, & DeVries, 1988; Weiskopf, 1980; Zabel & Zabel, 1982) and how supervisors can alleviate or exacerbate the problem (e.g., Cherniss, 1988; Cook & Leffingwell, 1982; Fimian, 1986; Greer & Wethered, 1984). With persuasive rhetoric and little hard evidence comes the message that special educators are more stressed than regular educators.

Not only is there little evidence to support this claim, but the few comparative studies available suggest that special educators are no more stressed or burned out than are regular educators (Beck & Gargiulo, 1983; Bensky et al., 1980; Billingsley & Cross, 1991b; Crane & Iwanicki, 1986; Fimian, 1983; Presley, 1982; Sutton & Huberty, 1984). Although some of these studies could be criticized on methodological ground--slow statistical power, unreliable measures, and inadequate comparison groups--the available evidence does not support the conventional wisdom.

Even without invoking the argument of elevated stress, the other factors I have cited--nonnterchangeability of teachers, the separate labor markets, and the demographic differences between the two work forces--suggest that the career trajectories of special educators and regular educators may differ. To help evaluate whether such differences exist, this article describes the paths of more than 6,600 special educators who began their careers in Michigan and North Carolina. Two broad sets of questions guided this study: * How many years do special educators continue

to teach (on average) and in which years

of their careers are they most likely to leave?

Does the risk of leaving teaching differ by a

teacher's year of hire, personal and professional

background characteristics, job responsibilities,

or salary?

The answers to these questions, alone and compared with answers derived from previous research on regular educators' careers, provide new insights into who stays and who leaves special education.



Cooperation from the state departments of education and support from the National Science Foundation facilitated construction of two longitudinal data sets describing the career trajectories of 3,941 black and white special educators newly hired in Michigan between 1972 and 1978 and 2,701 black and white special educators newly hired in North Carolina between 1975 and 1983. Other than student teaching, none of these teachers had ever taught before in a U.S. public school. As Grissmer and Kirby (1987) noted, it is only by studying beginning teachers from their date of hire until their date of departure that we can understand who teaches and for how long. Special educators with prior classroom experience--either special or regular--may follow different career paths, having already learned what it is like to teach and having already left a teaching job at least once before.


Michigan and North Carolina are interesting settings for studying special educators' careers. These two large states have different educational infrastructures: Michigan's 1.6 million public school students are distributed among almost four times as many districts as the 1.1 million students in North Carolina. Per-capita income and per-pupil expenditures also differ: Michigan exceeds the national average and North Carolina falls far below (Murnane et al., 1991). Both states provide special education services to about 10% of their students, slightly below the 11% national norm. But special educators comprise almost 15% of Michigan's teaching force, compared with only 10% in North Carolina (NCES, 1990; OSERS, 1990).

Unlike states scrambling to comply with EHA, Michigan and, to a lesser extent, North Carolina had fairly stable special education enrollments after the law's passage. Between 1976 and 1987, as national special education enrollment increased by 20%, Michigan's remained virtually constant, whereas North Carolina's increased by only 11% (OSERS, 1979, 1989). In a 1984 needs assessment, Michigan was the only state reporting no special educator shortage; and North Carolina reported a need only for teachers of students with emotional disturbance (Smith-Davis et al., 1984). As a result of such stability, the career paths of this study's special educators may resemble those of future special educators hired in response to steady or slightly increasing special education enrollment.


Because of differences in the contents of state personnel files, each data set contains some overlapping and some unique information. Both provide data describing (a) the teachers' personal background (year of hire, age at hire, gender, and race); (b) the demographic characteristics of the school districts in which the teachers worked, as measured during the 1980 U.S. Census of Population (median family income, median years of education of head of household, percentage of the population that is black, percentage of owner-occupied homes, and percentage of children living in poverty); and (c) the teachers' salary each year on the job. The Michigan data set also indicates whether the teachers work only with elementary students and the types of disabilities that their students have (speech, learning, emotional, mental retardation, vision/hearing, or physical/multiple). Teachers providing support services (counseling, guidance, and so forth) are classified separately. Although the North Carolina data set does not include grade level and student-assignment data, it does include the teachers' scores on the general and professional knowledge subtest of the National Teacher Examinations (NTE).

Each teacher's employment duration is measured as the number of years he or she taught continuously in the state between his or her year of hire and the end of data collection (1985 in Michigan and 1986 in North Carolina). As in previous analyses of the regular educators hired in these two states, moves between districts within state are not treated as exits from the profession (Murnane et al., 1988, 1989, 1991). Lack of student descriptors precludes identification of transfers to regular education among teachers in North Carolina. To ensure comparability, regular education transfers in Michigan are not classified as exits either. As described later, the 8% of Michigan special educators who transferred to regular education were coded as serving another category of students at the time of the transfer, allowing separate characterization of this career move. For an interesting study of the reasons why special educators transfer to general education, see Billingsley and Cross (1991a).

Statistical Analysis

Most previous quantitative research on teacher career paths has suffered from serious methodological limitations (Singer & Willett, 1991; Willett & Singer, 1989, 1991). The core dilemma has been how to include teachers still teaching when data collection ended, for whom the outcome--employment duration--is necessarily unknown (that is, censored). Forty-four percent of the Michigan special educators and 5 1 % of the North Carolina special educators in this study were still teaching when data collection ended. For these teachers, all I know is that they taught for at least the number of years between their date of hire and 1985 (or 1986). Although some may have left soon thereafter and others will teach for years to come, there is no way of knowing how many years any of them will ultimately teach. Censored teachers provide much information, though, especially about the probability that teachers teach for periods longer than the length of data collection. Accurate analysis of career-path data must include censored teachers even though their eventual employment durations are unknown.

Survival analysis is one statistical technique that permits simultaneous study of the employment durations of teachers who have left and teachers who have not (Singer & Willett, 1991; Willett & Singer, 1991). Originally developed by biostatisticians studying how long people survive after surgery, survival analysis can be used to study how long it takes for any event to occur, even when the event under study is within the individual's control. In this study, I used a specific type of survival analysis--discrete-time survival analysis--that focuses not on employment duration directly, but on the probability that a teacher leaves in any particular year, given that he or she taught continuously until that year. This conditional probability, called the hazard rate, measures the risk of leaving teaching in any particular year among teachers who taught continuously through the immediately preceding year. The hazard rate can be estimated just like any statistical quantity. If many teachers leave during the first few years on the job, early hazards (risks) are high. If relatively few teachers who remain until their 7th year leave in Year 7, in contrast, the 7th-year risk is low. Plots of risk versus years of teaching completed describe the year-to-year changes in the risk of leaving over the teaching career.

Relationships between hazard and predictors can be explored by building hazard models. By using hazard models instead of more familiar regression models, a researcher can (a) appropriately analyze censored and noncensored data; (b) estimate effects of predictors, such as grade level, student assignment, and salary, which themselves vary over time; and (c) determine whether the effects of predictors remain constant across teachers' careers or whether they vary over time. This last feature permits assessment, for example, of whether salary is a more important predictor of the stay-or-leave decisions of beginning teachers, who have not yet made a long-term commitment to teaching, than of the stay-or-leave decisions of teachers who have already invested years in the job.

The results reported here are based on a sequence of discrete-time hazard models that linked the risk of leaving teaching, on one hand, to teacher and job characteristics on the other. Because the variables differed slightly across states, separate, but parallel, analyses were conducted. Initial models explored the relationship between hazard and teachers' year of hire, age at hire, gender, and race. Subsequent models examined the effects of student characteristics (in Michigan) after controlling for teacher characteristics. Final models explored the effects of NTE scores and salary, after controlling for student, teacher, and district characteristics. At every stage, I explored the main effects of each predictor and all possible interactions between predictors. This enabled me to determine, for example, not only whether NTE scores are associated with hazard (the main effect), but also whether the effect of NTE score differs by a teacher's gender, race, or student assignment (the interaction effect). All comparisons cited here are significant at the .05 level.

I summarize the analytic results using three related statistics: (a) fitted hazard functions for subgroups of teachers; (b) estimates of median employment duration, the number of years that pass before half of a given group of teachers leaves teaching; and (c) 5- and 10-year survival rates, the percentage of a cohort of teachers that remains 5 (or 10) years after hire. Median employment durations and 5- and 10-year survival rates summarize the cumulative effects depicted by fitted hazard functions; subgroups at greater risk of leaving teaching have shorter median employment durations and smaller percentages surviving.


How Long Do Special Educators Teach?

Despite working under different economic conditions and educational infrastructures, the career paths of beginning special educators in Michigan and North Carolina are strikingly similar (Figure 1). New special educators are most likely to leave teaching early in their careers. Twelve percent of the new special educators in Michigan and 13% of the new special educators in North Carolina leave by the end of the 1st year. Among those teachers in both states who continue, 11% leave by the end of the 2nd year; among those who continue after this point, 11% leave by the end of the 3rd year.

These similar annual risks cumulate so that by the end of 5 years, 43% of the special educators newly hired in each state are not teaching in that state. Because the risks of leaving teaching in the two states continue to resemble each other, the estimated median employment durations nearly coincide, at 6.6 years in Michigan and 7.2 years in North Carolina.

After the initial "hazardous" years, the risk of leaving teaching declines steadily over time. By the 8th year, less than 5% of the remaining teachers leave; by the 10th year, the percentage leaving is even lower. Because of these low later risks, the percentage of teachers teaching after 10 years--46% in North Carolina and 43% in Michigan--is only 10 to 15 points lower than the percentage of teachers teaching after 5 years (which is 57% in both states).

Is the Profile of Risk Changing Over Time?

The Michigan special educators under study were hired between 1972 and 1978; the North Carolina special educators were hired between 1976 and 1983. After examining the variation in risk of leaving by year of hire, I found that the Michigan teachers could be collapsed into two cohorts (1972-1974 and 1975-1978), whereas the North Carolina teachers could be collapsed into three cohorts (1976-1978, 1979-1980, and 1981-1983). In each state, the risk of leaving teaching does not differ between entry years within cohort; it does, however, differ between cohorts. Figure 2 presents hazard profiles by entry cohort in Michigan (top panel) and North Carolina (bottom panel). The effects of entry cohort are far from straightforward; indeed, the shape of the entire risk profile differs by entry cohort. In North Carolina and, to a lesser extent, Michigan, teachers in later cohorts are more likely than teachers in earlier cohorts to leave shortly after hire. In both states, for example, the risk of leaving teaching in the 1st year is consistently higher among teachers hired later.

Over time, however, the risk of leaving among members of later cohorts drops more precipitously than it does among members of early cohorts, so that experienced teachers in the later cohorts are less likely than experienced teachers in the early cohorts to leave teaching. In essence, the cohort differentials switch over time. Among teachers who have taught 5 years, for example, the risk of leaving teaching in both states is lower for teachers hired in the later cohorts. Although special educators hired well after the implementation of EHA have higher risks of leaving initially, they ultimately have lower risks of leaving after several years on the job.

Does a Teacher's Age or Gender Make a Difference?

The modal new special educator--a young woman (age 30 or younger)--represents 75% of the new hires in Michigan and 88% of the new hires in North Carolina. Young men (also age 30 or younger), the next largest group, account for 14% of the new hires in Michigan and 5% of the new hires in North Carolina. Mature women (those over 30) represent 10% of the new hires in Michigan and 6% of the new hires in North Carolina. Less than 2% of the beginning special educators in either state are mature men (those over 30).

Demographics are important for understanding the special education teaching force because members of the largest demographic groups are the most likely to leave. Begin with the effects of age: Young teachers are nearly twice as likely as mature teachers to leave. The magnitude of this effect is similar in both states and remains stable across the teaching career. Even among teachers who stay 5 years, for example, those who were 30 or younger when hired are nearly twice as likely to leave.

Now add the effects of gender; across the entire teaching career, women in both states are much more likely to leave (with the exception of the first 3 years on the job in North Carolina). Although the magnitude of the gender differential differs across the career in the two states, it sometimes nears a factor of 2 to 1, as in Michigan after several years of teaching.

These differential patterns of risk lead to dramatically different career trajectories. Table 1 presents estimated 5- and 10-year survival rates and median employment durations for special educators in each state by their age at hire and gender. (For each state, the model generating these estimates included all the teachers in all entry cohorts [with variables denoting cohort membership and interactions between cohort membership and time]. To increase the chances of estimating a precise median career duration in each state, the summary statistics presented (and all others presented subsequently] are computed for teachers hired in the earliest cohorts (1975-1978 in Michigan and 1976-1978 in North Carolina) because these groups have the longest data records.) [TABULAR DATA 1 OMITTED]

The modal group of teachers--young women--has particularly short careers, with median employment durations of 5.4 years in Michigan and 6.7 years in North Carolina. Although young men in North Carolina also have short careers on average, people in the remaining demographic groups in both states continue to teach for so long that I cannot estimate median employment durations for them--their median durations exceed the length of data collection. More than half the mature men and women are still teaching after 10 years in North Carolina and 12 years in Michigan. Although these groups comprise a small proportion of the special education teaching force, this small group is particularly stable.

Do Career Paths Differ by a Teacher's Race?

Eighteen percent of the public school students in Michigan and 30% of the public school students in North Carolina are black; 8% of the special educators newly hired in Michigan and 9% of the special educators newly hired in North Carolina are black. The disparity between the racial composition of the student body and the racial composition of the teaching force is of grave concern, and all available evidence suggests that it is increasing over time (Southern Education Foundation, 1990).

Although one explanation of the disparity might be differences in teachers' career paths, comparison of the stay-or-leave decisions of black and white special educators in these two states reveals few statistically significant effects. The sole race differential in Michigan varies by teacher gender: Black women have lower hazards than white women, but black men have higher hazards than white men. The sole race differential in North Carolina disappears after controlling statistically for NTE scores. In no case did the effects of other predictors--such as NTE scores or salary--differ by teacher race. In these two states, then, the career trajectories of black special educators parallel those of their white colleagues: They are just as likely to leave or stay and they respond to the same incentives and disincentives.

What Effects Do Teaching Assignments Have?

Special educators teach diverse types of students with diverse needs; not surprisingly, the risk of leaving teaching depends on a teacher's assignment. Table 2 presents the estimated median employment duration of Michigan special educators by the grade level and disabilities of the students taught. (The hazard model used to derive these estimates controls statistically for entry cohort, age at hire, gender, and race. These estimates are for young white women, the modal Michigan special educator. Because the effects of grade level and student assignment do not interact with these control characteristics, the relative differentials in estimated median employment duration for these groups of teachers mirror those for all other demographic groups).
Estimated Median Employment Duration (in years) for Michigan Special Education T
Assignment Elementary Secondary
Special education
 Learning disabilities 9.5 7.3
 Physical/multiple disabilities 8.6 6.7
 Mental retardation 7.8 6.1
 Emotional disturbance 6.8 5.5
 Speech disabilities 6.6 5.4
 Hearing/vision disabilities 6.1 5.0
Regular education(a) 5.8 4.7
Support services 4.9 4.0
(a) Eight percent of the special educators switched to
regular education after their 1st year of teaching.

Regardless of the type of students taught, special educators working in elementary schools exclusively (about half the teachers) stay an average of 1.6 years longer than their colleagues working in junior high and high schools. (And although not apparent in Table 2, the risk differential associated with grade diminishes over time; it is greatest during the 1st year of teaching and disappears entirely by the 6th.)

Even more striking are the effects of student assignment--median employment durations vary nearly twofold across groups. Teachers of students with learning disabilities are the most stable--they are least likely to leave in every year of their careers, staying a median of 9.5 years at the elementary level and 7.3 years at the secondary level. Teachers of students with physical and multiple disabilities or mental retardation are also fairly stable, staying only 1 to 2 years less, on average, than their colleagues working with students with learning disabilities. Teachers of students with emotional and speech problems have somewhat shorter careers, staying a median of 6 to 7 years at the elementary level and 5 to 6 years at the secondary level.

But the greatest risk of leaving is found among the remaining two groups of special educators--those providing support services and those who switch from special education (of any type) into regular education. Eleven percent of the special educators in Michigan began their careers as social workers, diagnosticians, psychologists, counselors, or consultants. In comparison to all other special educators, this group is at greatest risk of leaving the schools, staying a median of 4 to 5 years. The other group at great risk of leaving is those 8% of special educators who, after working exclusively with special needs students for at least 1 year, transfer to regular education. Such transfers lengthen their teaching careers only slightly, for they ultimately stay a median of 5.8 years total (combining all their years of teaching) if they work in elementary schools and 4.7 years if they work in secondary schools.

Are the Schools Losing Special Educators with High Standardized Test Scores?

The career durations of North Carolina special educators are strongly associated with their NTE scores: Among both beginning and experienced teachers, those with higher NTE scores are likelier to leave. One way to summarize this effect is by comparing the risk of leaving teaching for special educators with different NTE scores, say one with an average score (586, the sample median), one with a low score (492, the 10th percentile in the sample), and one with a high score (683, the 90th percentile in the sample). (For simplicity, the hazard model used to derive the estimates concerning NTE scores controlled only for entry cohort, age at hire, and gender. Adding controls for district characteristics and teacher salary had virtually no effect on the coefficients for NTE score.)

In every year of their careers, special educators with high NTE scores are nearly twice as likely as their colleagues with low scores to leave. This effect holds regardless of a teacher's year of hire, age at hire, gender, or race. Higher scoring teachers, whether young or old, male or female, black or white, are more likely to leave than are their lower scoring peers.

This large difference in risk by NTE score leads to a sizable difference in estimated median employment durations. Whereas over half the low-scoring teachers stay more than 10 years, the average mid-scoring teacher stays 7.7 years, and the average high-scoring teacher stays only 5.0 years. The net result is that only 36% of the high-scoring teachers continue to teach for 10 years, compared with 47% of the mid-scoring teachers and 57% of the low-scoring teachers.

Do Dollars Make a Difference?

In both Michigan and North Carolina, special educators working in school districts offering comparatively high salaries are less likely to leave than are special educators working in districts offering comparatively low salaries. This effect persists after controlling statistically for all available background characteristics of the teachers and their school districts, including NTE score in North Carolina and grade level and student assignment in Michigan.

Although better-paid teachers stay longer, the effect of salary differs by state. Figure 3 presents fitted hazard profiles for three prototypical teachers in each state: one in an average salary stream, one in a low salary stream (making $2,000 less per year than average, in 1988 dollars), and one in a high salary stream (making $2,000 more per year than average, in 1988 dollars). (In both states, the hazard models underlying these estimates control statistically for entry cohort, age at hire, gender, and district characteristics. The Michigan model also includes controls for the grade level, type of student taught, and teacher race, while the North Carolina model also includes controls for the teacher's NTE score. To derive the hazard profiles presented in Figure 3, I substituted in average values for all the teacher and district covariates, except for year of entry and student and grade assignment [in Michigan]. The Michigan curves are for elementary school teachers working with students with learning disabilities. Because salary did not interact with any other characteristics in the model, the relative effects presented in Figure 3 are the same for all subgroups of teachers.)

First note that Michigan special educators are less sensitive to salary than are their peers in North Carolina; the vertical distances between the three Michigan profiles are smaller than are the vertical distances between the three North Carolina profiles.

Next note that the shapes of the risk profiles differ across states; in Michigan, they begin apart and converge over time, whereas in North Carolina, they begin together and widen over time. These differences in shape arise from differences between states in the functional form that best relates hazard to salary. In Michigan, the best fitting models are quadratic; the $2,000 differential has a larger effect among teachers at the lower end of the salary scale than it does among teachers at the higher end. Because beginning teachers earn less than experienced teachers, the effect of salary appears to diminish over time; in actuality, the small quadratic effect of salary is constant over time. Regardless of experience, poorly paid teachers in Michigan are somewhat more likely to leave than are their better paid colleagues.

In North Carolina, in contrast, the effect of salary does differ over time--experienced teachers are more sensitive to salary. Among beginning special educators in North Carolina, the risk of leaving teaching is the same whether a teacher makes $2,000 more than average or $2,000 less. As special educators gain experience, the effects of salary surface; after several years of teaching, the effects are quite pronounced. So although the effect of salary differs across states, in both cases, dollars do make a difference.


Over the next 15 years, the demand for all teachers--regular and special--will increase as student enrollments grow, and as more and more current teachers retire (Gerald et al., 1989; U.S. Department of Labor, 1989). Whether this increased demand will be met with an adequate supply depends on the career decisions of hundreds of thousands of current and future teachers. This article presents some of the first large-scale empirical evidence describing the career decisions of beginning special educators, documenting when they are most likely to leave teaching, who among them is most likely to go, and whether job assignments, teacher test scores, and salaries make a difference. The question I now consider is whether their career paths resemble, or depart from, those of their regular education colleagues.

The Big Picture: How Do Special and Regular Educators' Careers Compare?

In broad outline, regular and special educators follow strikingly similar career paths. Both are most likely to leave teaching during their early years on the job, when they first confront the daily demands of classroom life. Teachers who survive this initial "hazardous" period typically teach for many years to come. This pattern of declining risks, exhibited by the special educators in both Michigan and North Carolina, has been replicated in most studies of regular educators' careers (see, e.g., Charters, 1970; Grissmer & Kirby, 1987; Haberman & Rickards, 1990; Mark & Anderson, 1978, 1985; Schlecty & Vance, 1981; Theobold, 1990; Whitener, 1965).

The general similarity in the career paths of special and regular educators is to be expected. After all, most beginning teachers in both fields have had very little classroom experience. While some find their first years of teaching exciting and satisfying, others find it frustrating and overwhelming. Those who cannot cope, leave. As both types of teachers learn how to manage a classroom, the burdens can diminish, allowing the rewards to grow (Lortie, 1975; Rosenholtz, 1989).

But the underlying question is really whether special educators have longer or shorter careers, and unfortunately, because of variability in regular educators' careers, the answer to this question remains equivocal. This study shows that beginning special educators in both Michigan and North Carolina have a median employment duration of 7 years. While regular educators hired in Michigan between 1972 and 1975 and in North Carolina between 1974 and 1978 (corresponding closely to the earliest cohorts studied here) have similar patterns of risk, their levels of risk differ so markedly that their estimated median employment durations--4.6 years in Michigan and 11.3 years in North Carolina--are grossly disparate (Murnane et al., 1991). So in Michigan, special educators stay an average of 2 years more than regular educators, whereas in North Carolina, they stay an average of 5 years less.

Without an incontrovertible explanation of this reversal across states, any generalization is arguable. Perhaps the best explanation is rooted in differences between regular and special educators' responses to local economic conditions. Michigan suffered greatly during the late 1970s and early 1980s; declining enrollments may have led many regular educators to quit in anticipation of layoffs, whereas steady numbers of students with disabilities may have allowed special educators to feel more secure. Following this argument, the substantially shorter careers of North Carolina's special educators--which are virtually identical to the careers of Michigan's special educators--may be more typical of special/regular differentials throughout the country. But without data from other states, this argument, although plausible, is hardly incontrovertible. Therefore, the comparability of the special educators' career paths in these two states suggests a generalizable finding about them; but the variation in the career paths of regular educators makes conclusive comparisons difficult to draw.

Changes over time in the career paths of special educators complicate the matter. More recently hired special educators in both states have higher initial risks and lower later risks than do their colleagues hired years ago. One clearly relevant factor is the 1975 passage of EHA. As the demand for special educators increased during the mid-to-late 1970s and early 1980s, some college students may have chosen special education simply because jobs were plentiful (Murnane & Schwinden, 1989). After confronting the challenges and responsibilities of teaching children with disabilities, some of these teachers may have realized their error and quickly left (for a description of a similar pattern, see Lauritzen & Friedman, 1991). This early exodus may have eliminated teachers without a real commitment from the pool of continuing teachers, leaving those more attracted and dedicated to the field. In the years ahead, as special education enrollment growth more closely parallels that found elsewhere in the schools, special education may lose its magnet field status. But on the basis of the evidence presented here, it is difficult to say whether these future special educators will have longer or shorter stays in the classroom than their future regular education colleagues.

Are Demographic Effects the Same Among Regular and Special Educators?

As most researchers studying regular educators have found, I too, found that young women--the modal group of new teachers throughout the public schools--are far more likely to leave teaching than are members of all other demographic groups (Charters, 1970; Heyns, 1988; Murnane et al., 1988, 1989, 1991). Although many of these young women leave only temporarily, returning after having children or after moving to follow a spouse (Berry, 1988; Haberman & Rickards, 1990), they nevertheless have particularly short initial spells in the classroom.

These demographic differentials, which appear throughout education, have an especially great impact in special education because young women comprise such a large fraction of this teaching force: 88% in North Carolina and 75% in Michigan. The comparable regular education percentages in these two states are 78 and 59 (Murnane et al., 1988, 1989). Disproportionate representation of the highest risk group may, in fact, explain some of the astronomical attrition rates of special educators reported in the literature (e.g., Shofer & Duncan, 1986; Smith-Davis et al., 1984).

The impact of young women's short spells on overall special education attrition rates may diminish in the years ahead as the composition of the special education teaching force changes. Although the change is coming slowly, it has already begun. Comparing the earliest and latest cohorts of special educators hired in North Carolina, for example, the percentage of mature women tripled from 3 % to 9%. Regular education is experiencing similar demographic changes (Murnane et al., 1991). This leads to the possibility that future attrition rates computed across all special educators may drop as older teachers comprise a larger fraction of the new teaching force and as special educators hired in response to EHA reach the stable years of mid-career.

Do Special Educators Respond to Opportunity Costs?

As researchers studying the careers of regular educators have found, a teacher's particular job assignment also makes a difference. In this study, special educators working with elementary school students stayed an average of 1 to 2 years longer than their colleagues working with junior high and high school students. The literature on regular educators' careers reveals inconsistent effects of grade level. Although some studies have found similar differentials (e.g., Heyns, 1988; Murnane et al., 1988, 1989, 1991), others have not (e.g., Chapman & Hutcheson, 1982; Charters, 1970).

More striking, and consistent with the regular education literature, is the variation in career paths associated with the type of education delivered. I found that special educators who teach students with learning disabilities, physical disabilities, mental retardation, and, to a lesser extent, emotional disturbance are the least likely to leave. Those working with students with speech, hearing, or vision disabilities are somewhat more likely to go. The group at greatest risk of leaving is those special educators providing support services, including counseling, guidance, and diagnostic evaluations.

Job assignment differentials may stem from at least two distinct sources. One may be the varying levels of stress found among teachers of students with varying needs. Although special educators, on average, seem no more stressed than regular educators, their stress levels differ by classroom setting (Bensky et al., 1980; Crane & Iwanicki, 1986) and students taught (Fimian, Pierson, & McHardy, 1986; Johnson et al., 1982; Meadow, 1981; Zabel & Zabel, 1982). Special educators working with students with speech, hearing, or vision disabilities or providing support services may be especially susceptible to stress and burnout, working, as they do, with large numbers of students for intermittent and brief sessions.

But a more likely explanation of these risk differentials lies in the economic concept of opportunity costs, the salary a teacher foregoes by not pursuing the best career alternatives outside of education. The ramifications of high opportunity costs for teachers have been noted by many researchers studying the supply of and demand for regular educators, especially mathematics and science teachers (e.g., Darling-Hammond & Hudson, 1990; Kershaw & McKean, 1962; Levin, 1985; Murnane et al., 1988, 1989, 1991; Rumberger, 1985). Chemistry and physics teachers, for example, who command particularly high starting salaries in business and industry, have the shortest careers among all regular educators; and elementary school teachers, whose skills are less marketable outside the schools, have the longest (Murnane et al., 1989, 1991).

Similar variations in opportunity costs appear in special education. Those providing services to students with speech, hearing, or vision disabilities or general support services throughout the schools may be the best able to obtain jobs elsewhere--in hospitals, nursing homes, or private practice. Those working with children with learning disabilities, mental retardation, and emotional disabilities, like many of their regular education colleagues, may have fewer options (Grossman, 1982). So for special educators with good options outside of teaching, the "costs" of continuing to teach may be too high.

Further support for the opportunity cost hypothesis comes from the strong relationship between special educators' NTE scores and their length in stay in teaching. Many studies find that regular educators with high standardized test scores are more likely to leave the profession (Murnane et al., 1989, 1991; Nelson, 1985; Schlecty & Vance, 1981; for a special education study, see Frank & Keith, 1984). Although the NTE may not measure teaching effectiveness as well as other more direct assessments (Haney, Madaus, & Kreitzer, 1987), it can measure a component of opportunity cost because NTE scores correlate highly with scores on the standardized tests used to screen applicants for entry to relatively high-paying fields (Ayers & Qualls, 1979; Wexler, 1975). Teachers with high NTE scores may have better access to jobs outside the schools, increasing the "costs" of staying in education. Those who see the costs as too great to ignore--teachers with especially high test scores--leave.

This raises the related issue of the role of teachers' salaries in the stay-or-leave decision. If teacher salaries were more competitive with those offered in other sectors of the economy, the opportunity costs of teaching would diminish. But teaching salaries are far from competitive. In 1981, for example, mathematics majors who became teachers earned 40% less than their peers who entered business or industry (Murnane et al., 1991). Although the average special educator might not command a very high salary outside the schools, those with particularly low teaching salaries, or those whose teaching salaries fall far behind inflation, have higher opportunity costs.

So it comes as no surprise that special educators who earn comparatively low salaries are more likely to leave than teachers earning comparatively high salaries. Although the salary effects are smaller than the effects of NTE scores and job assignments, and they vary across states and the teaching career, the general point remains: Salary makes a difference to special educators. Special educators who are paid more stay longer; those who are paid less leave more quickly. When studying regular educators, my colleagues and I have found that their career paths also differ depending on their salaries (Murnane et al., 1989, 1991). This has led us to suggest that salary increases might be one way to help reduce the number of school districts that find themselves repeatedly hiring new, inexperienced teachers to replace those who leave after 1 or 2 years.

The collective findings that job assignments, standardized test scores, and salaries predict variation in career paths suggest strongly that special educators, like their regular education colleagues, assess the "costs" of continuing to teach. Each year, our schools lose large numbers of special (and regular) educators who conclude that the costs are too high. And these may be teachers we desperately want to keep, those who provide support and consultation services to students with disabilities throughout the schools and those who score well on standardized tests. If we want to encourage these teachers to stay in education, we will have to decrease their opportunity costs.

But it is also true that many experienced special educators, who have already invested several years of their professional lives in schools, continue to teach for years to come. So, too, do many teachers working in elementary schools and many teachers working with students who have learning disabilities, mental retardation, and physical disabilities. Although the question of whether special educators, on the whole, are at greater risk of leaving teaching than their regular education colleagues remains, it is clear that many special educators do stay in the schools for long periods of time.


A free appropriate public education: But who will provide it? (1989, April). Testimony to the Senate Subcommittee on the Handicapped and the House Subcommittee on Select Education. Ayers, J. B., & Qualls, G. S. (1979). Concurrent and predictive validity of the National Teachers Examinations. Journal of Educational Research, 73, 86-92. Beck, C. L., & Garguilo, R. M. (1983). Burnout in teachers of retarded and nonretarded children. Journal of Educational Research, 76, 169-173. Bensky, J. M., Shaw, S. F., Gouse, A. S., Bates, H., Dixon, B., & Beane, W. E. (1980). Public Law 94-142 and stress: A problem for special educators. Exceptional Children, 47,24-29. Berry, B. (1988). Labor market choices and teacher reform: Policy options for the public schools of the twenty-first century. Teaching and Teacher Education, 4, 71-81. Billingsley, B. S., & Cross, L. H. (1991a). Teachers' decisions to transfer from special to general education. The Journal of Special Education, 24, 496-511. Billingsley, B. S., & Cross, L. H. (1991b). Predictors of commitment, job satisfaction, and intent to stay in teaching: A comparison of general and special educators. The Journal of Special Education, 25, 452-471. Bobbitt, S. A., Faupel, E., & Burns, S. (1991). Characteristics of stayers, movers, and leavers: Results from the teacher follow-up survey, 1988-89. Washington, DC: Author. Bogenschild, E. G., Lauritzen, P., & Metzke, L. (1988). A study of teacher attrition. Milwaukee, WI: Marquette University, College of Education. Chandler, H. N. (1983). The loneliness of the special education teacher. Journal of Learning Disabilities, 16, 126-127. Chapman, D. W. (1983). A model of the influences on teacher attrition. Journal of Teacher Education, 34, 43-49. Chapman, D. W. (1984). Teacher retention: The test of a model. American Educational Research Journal, 21, 645-658. Chapman, D. W. & Hutcheson, S. M. (1982). Attrition for teaching careers: A discriminant analysis. American Educational Research Journal, 19, 93-105. Charters, W. W., Jr. (1970). Some factors affecting teacher survival in school districts. American Educational Research Journal, 7, 1-27. Cherniss, C. (1988). Observed supervisory behavior and teacher burnout in special education. Exceptional Children, 54, 449-454. Cook, J. M. E., & Leffingwell, R. J. (1982). Stressors and remediation techniques for special educators. Exceptional Children, 49, 54-59. Crane, S. J., & Iwanicki, E. F. (1986). Perceived role conflict, role ambiguity, and burnout among special education teachers, Remedial and Special Education, 7, 24-31. Cross, L. H. (1987, April). Teacher crisis: Myth or reality revisited. Paper presented to the annual meeting of the American Educational Research Association, Washington, DC. Darling-Hammond, L., & Hudson, L. (1990). Precollege science and mathematics teachers: Supply, demand, and quality. In C. B. Cazden (Ed.), Review of Research in Education, 16, (pp. 223-264). Washington, DC: American Educational Research Association. Fimian, M. J. (1983). A comparison of occupational stress correlates as reported by teachers of mentally retarded and nonmentally retarded handicapped students. Education and Training of the Mentally Retarded, February, 62-68. Fimian, M. J. (1986). Social support and occupational stress in special education. Exceptional Children, 52, 436-442. Fimian, M. J., Pierson, D., & McHardy, R. (1986). Occupational stress reported by teachers of learning disabled and nonlearning disabled handicapped students. Journal of Learning Disabilities, 19, 154-158. Fimian, M. J., & Santoro, T. M. (1983). Sources and manifestations of occupational stress as reported by full-time special education teachers. Exceptional Children, 49, 540-543. Frank, A. R., & Keith, T. Z. (1984). Academic abilities of persons entering and remaining in special education. Exceptional Children, 51, 76-77. Gerald, D. E., Horn, P. J., & Hussar, W. J. (1989). Projections of education statistics to 2000. Washington, DC: National Center for Education Statistics. Greer, J. G., & Wethered, C. E. (1984). Learned helplessness: A piece of the burnout puzzle. Exceptional Children, 50, 524-531. Grissmer, D. W., & Kirby, S. N. (1987). Teacher attrition: The uphill climb to staff the nation's schools. Santa Monica, CA: Rand. Grossman, R. P. (1982). Economic standing of the LD profession. Journal of Learning Disabilities, 15, 134-138. Haberman, M. & Rickards, W. H. (1990). Urban teachers who quit: Why they leave and what they do. Urban Education, 25, 297-303. Haggstrom, G. W., Darling-Hammond, L., & Grissmer, D. W. (1988). Assessing teacher supply and demand. Santa Monica, CA: Rand. Haney, W., Madaus, G., & Kreitzer, A. (1987). Charms talismanic: Testing teachers for the improvement of American education. In E. Rothkopf (Ed.), Review of research in education, (Vol. 14., pp. 169-238). Washington, DC: American Educational Research Association. Heyns, B. (1988). Educational defectors: A first look at teacher attrition in the NLS-72. Educational Researcher, 17, 24-32. Holland, R. P. (1982, Winter). Special educator burnout. Educational Horizons, 5 8-64. Johnson, A. B., Gold, V., & Vickers, L. L. (1982). Stress and teachers of the learning disabled, behavior disordered, and educable mentally retarded. Psychology in the Schools, 19, 552-557. Kemple, J. J., Murnane, R. J., Singer, J. D., & Willett, J., B. (1992). Why are there fewer and fewer Black teachers? Manuscript submitted for publication. Kershaw, J. A., & McKean, R. N. (1962). Teacher shortages and salary schedules. New York: McGraw-Hill. Lauritzen, P., & Friedman, S. J. (1991). Teachers for children with emotional/behavioral disorders: Education's greatest challenge. Preventing School Failure, 35, 11-16. Lawrenson, G. M., & McKinnon, A. J. (1982). A survey of classroom teachers of the emotionally disturbed: Attrition and burnout factors. Behavioral Disorders, 8, 41-49. Levin, H. M. (1985). Solving the shortage of mathematics and science teachers. Educational Evaluation and Policy Analysis, 7, 371-382. Lortie, D. C. (1975). Schoolteacher: A sociological study. Chicago: University of Chicago Press. Mark, J. H., & Anderson, B. D. (1978). Teacher survival rates: A current look. American Educational Research Journal, 15, 379-383. Mark, J. H., & Anderson, B. D. (1985). Teacher survival rates in St. Louis, 1969-1982. American Educational Research Journal, 22, 413-421. Meadow, K. P. (1981). Burnout in professionals working with deaf children. American Annals of the Deaf February, 13-22. Metz, S. A., & Fleischmann, H. L. (1974). Teacher turnover in public schools, Fall 1968 to Fall 1969. Washington, DC: U.S. Department of Health, Education and Welfare. Murnane, R. J., & Schwinden, M. (1989). Race, gender, and opportunity: Supply and demand for new teachers in North Carolina: 1975-1985. Educational Evaluation and Policy Analysis, 11, 93-108. Murnane, R. J., Singer, J. D., & Willett, J. B. (1988). The career paths of teachers: Implications for teachers supply and methodological lessons for research. Educational Researcher, 17, 22-30. Murnane, R. J., Singer, J. D., & Willett, J. B. (1989). The influences of salaries and opportunity costs on teachers' career choices: Evidence from North Carolina. Harvard Educational Review, 59, 325-346. Murnane, R. J., Singer, J. D., Willett, J. B., Kemple, J. J., & Olsen, R. J. (1991). Who will teach?: Policies that matter. Cambridge: Harvard University Press. National Center for Education Statistics. (1988). Digest of education statistics, 1988. Washington, DC: Author. National Center for Education Statistics. (1989). Digest of education statistics, 1989. Washington, DC: Author. National Center for Education Statistics. (1990). Digest of education statistics, 1990. Washington, DC: Author. National Clearinghouse for Professions in Special Education. (1988). The decline in special education degrees conferred Washington, DC: National Association of State Directors of Special Education. National Education Association. (1982). Teacher supply and demand in public schools, 1981-82. Washington, DC: Author. National Research Council. (1987). Towards understanding teacher supply and demand: Priorities for research and development. Interim report, Panel on Statistics on Supply and Demand for Precollege Science and Mathematics Teachers. Washington, DC: National Academy Press. Nelson, F. H. (1985). New perspectives on the teacher quality debate: Empirical evidence from the National Longitudinal Survey. Journal of Educational Research, 78, 133-140. Office of Special Education and Rehabilitative Services. (1979). Second annual report to Congress on the implementation of P.L. 94-142. Washington, DC: Author. Office of Special Education and Rehabilitative Services. (1984). Sixth annual report to Congress on the implementation of P.L. 94-142. Washington, DC: Author. Office of Special Education and Rehabilitative Services. (1988). Tenth annual report to Congress on the implementation of P.L. 94-142. Washington, DC: Author. Office of Special Education and Rehabilitative Services. (1989). Eleventh annual report to Congress on the implementation of P.L. 94-142. Washington, DC: Author. Office of Special Education and Rehabilitative Services. (1990). Twelfth annual report to Congress on the implementation of P.L. 94-142. Washington, DC: Author. Office of Special Education and Rehabilitative Services. (1991). Thirteenth annual report to Congress on the implementation of the Individuals with Disabilities Education Act. Washington, DC: Author. Olsen, J., & Matuskey, P.V. (1982). Causes of burnout in SLD teachers. Journal of Learning Disabilities, 15, 97-99. Parshall, L. (1990). A study of the special education workforce in Michigan. Washington, DC: Decision Resources Corporation. Presley, P. H. (1982, April). Teacher burnout in special education--Myth or reality? Paper presented at the annual meeting of The Council for Exceptional Children, Houston, TX. Raphael, E. S., Singer, J. D., & Walker, D. K. (1985). Per pupil expenditures on special education in three metropolitan school districts. Journal of Education Finance, 11, 69-88. Raschke, D., Dedrick, C., & DeVries, A. (1988, Fall). Coping with stress: The special educator's perspective. TEACHING Exceptional Children, 21, 10-14. Rosenholtz, S. J. (1989). Workplace conditions that affect teacher quality and commitment: Implications for teacher induction programs. The Elementary School Journal, 89, 421-439. Rumberger, R. (1985). The shortage of mathematics and science teachers: A review of the evidence. Educational Evaluation and Policy Analysis, 7, 355-369. Schlechty, P. C., & Vance, V. S. (1981). Do academically able teachers leave education? The North Carolina case. Phi Delta Kappan, 63, 106-112. Schofer, R. C., & Duncan, J. R. (1986). A study of certain personnel preparation factors in special education. The Journal of Special Education, 20, 61-68. Schrag, J. (1986). Implementation of P.L. 94-142 and its accomplishments, problems and future challenges: A state education agency perspective. In H. J. Prehm (Ed.), The future of special education (pp. 75-110). Reston, VA: The Council for Exceptional Children. Singer, J. D. (in press) Once is not enough: Former special educators who return to teaching. Exceptional Children. Singer, J. D., & Butler, J. B. (1987). The Education for All Handicapped Children Act: Schools as Agents of Social Reform. Harvard Educational Review, 57, 125-152. Singer, J. D., & Raphael, E. S. (1988). Per pupil expenditures for special education: to whom are limited resources targeted? Final report for Grant No. G008630147, U.S. Department of Education, Office of Special Education. Singer, J. D., & Willett, J. B. (1988). Detecting involuntary layoffs in teacher survival data: The year of leaving dangerously. Educational Evaluation and Policy Analysis, 10, 212-224. Singer, J.D., & Willett, J. B. (1991). Modeling the days of our lives: Using survival analysis when designing and analyzing longitudinal studies of duration and the timing of events. Psychological Bulletin, 110, 268-290. Singer, J. D., & Willett, J. B., (in press). It's about time: Using discrete time survival analysis to study the duration and the timing of events. The Journal of Educational Statistics. Smith-Davis, J., Burke, P. J., & Noel, M. M. (1984). Personnel to educate the handicapped in America: Supply and demand from a programmatic viewpoint. College Park: Institute for the Study of Exceptional Children and Youth, University of Maryland. Southern Education Foundation. (1990). Pipeline: Finding, educating and retaining teachers. Atlanta, GA: Author. Sutton, G. W., & Huberty, T. J. (1984). An evaluation of teacher stress and job satisfaction. Education, 105, 189-192. Theobold, N. D. (1990). An examination of the influence of personal, professional, and school district characteristics on public school teacher retention. Economics of Education Review, 9, 241-250. Thomas, M. A. (1982, December). State education agency support for mental retardation programs. Education and Training of the Mentally Retarded, 17, 283-287. U.S. Department of Labor. (1989). Projections of occupational employment. Monthly Labor Review, 112, 48-62. Weiskopf, P. E. (1980). Burnout among teachers of exceptional children. Exceptional Children, 47, 18-23. Wexler, N. (1975). Concurrent validity of the National Teacher Exams. Washington, DC: National Institute of Education. Whitener, J. E. (1965). An actuarial approach to teacher turnover. Unpublished doctoral dissertation, Washington University, St. Louis. Willett, J. B., & Singer, J. D. (1989). Two types of questions about time: Methodological issues in the analysis of teacher career path data. International Journal of Educational Research, 13, 421-437. Willett, J. B., & Singer, J. D. (1991). From whether to when: New methods for studying student dropout and teacher attrition. Review of Educational Research, 61, 407-450. Zabel, R. H., & Zabel, M. K. (1982). Factors in burnout among teachers of exceptional children. Exceptional Children, 49, 261-263.

JUDITH D. SINGER is an Associate Professor of Education at the Harvard University Graduate School of Education, Cambridge, Massachusetts.

This work was completed while I was on a Spencer Post-Doctoral Fellowship from the National Academy of Education. I thank Richard J. Murnane for first sparking my interest in teacher career paths and for providing the data sets on which this research is based and John B. Willett for his continuing support and collaboration. Please address correspondence to Judith D. Singer, Associate Professor of Education, Harvard University, Graduate School of Education, Larsen Hall, Cambridge, MA 02138, (617) 495-1961.
COPYRIGHT 1992 Council for Exceptional Children
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Singer, Judith D.
Publication:Exceptional Children
Date:Dec 1, 1992
Previous Article:Relationships among states' fiscal and demographic data and the implementation of P.L. 94-142.
Next Article:20/20 Analysis: taking a close look at the margins.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters