Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 37
2
Determining Teacher Demand
In this chapter we are concerned with the demand for new teachers,
specifically teachers new to a particular job. Projections of the demand
for new teachers require the projection of a minimum of three data ele-
ments: student enrollment, pupil-teacher ratios, and teacher attrition rates.
Demand projections for segments of the teacher population, such as sec-
ondary school teachers of mathematics and science, require the projection
of these data elements for the specific segments. For example, projecting
the demand for mathematics teachers requires, at a minimum, projections
of enrollment in mathematics classes, the expected size of mathematics
classes, and the attrition rate of mathematics teachers. The necessary
data for teacher demand projections vary in availability and reliability.
Pupil-teacher ratios vary with staffing patterns, class sizes, teaching loads,
course requirements, and course-taking patterns in science and mathemat-
ics. Meaningful projections of the consequences of course-taking patterns
or teacher attrition are typically less available than are projections of future
student enrollment. Statewide enrollment projections are more reliable
than those for local school districts. In general, the smaller the subset
projected, the lower the reliability.
STUDENT ENROLLMENT
The two main topics discussed in this section describe methods of
projecting student enrollment, based on either public-school enrollment
data or population data. The main features and limitations of both sources
of data are discussed. We then turn to the other key components of
estimating demand for teachers pupil-teacher ratios and attrition.
37
OCR for page 38
38
PRECOLLEGE SCIENCE AND MATHEMATICS TEACHERS
Enrollment Projections Based on Student Enrollment Data
Projections Of student enrollment, one of the three elements neces-
sary for projecting the demand for new teachers, are the easiest. K-12
enrollment projections are widely available. Most states and many local
districts produce enrollment projections, particularly for the public schools,
although some states produce them for both public and private schools.
These projections typically follow a standard "cohort survival" methodolog I,
which uses observed enrollment ratios between grades to move ("survive")
classes ("cohorts") forward to the next level. If the state or school system
has low or constant levels of migration, then reliable projections by grade
are produced.
The National Center for Education Statistics (NCES) produces one-
year projections by state and nationwide projections for 10 years into the
future. Until 1988, NCES projected enrollments were for the public schools
only; projections for private schools began in 1989. NCES employs a mixed
model in which participation rates for kindergarten, grade 1, and special and
ungraded classes are calculated by applying recent public school enrollment
data, collected in its annual survey of the states, to age-specific population
estimates produced by the Census Bureau. The resultant rates are then
applied to projected populations of the appropriate ages (e.g., 5-year-olds
for kindergarten, 6-year-olds for first graders) to arrive at levels of future
enrollment for those grades and classes. Retention or grade progression
rates from the NCES annual survey are used to calculate grades 2 through
12. The NCES method of projecting enrollments is described more fully in
Part III of the panel's interim report (National Research Council, 1987c),
which discusses the components of the NCES model and those of six states.
The sources of data used by NCES are also described in the Guide to
Sources portion of the Digest of Education Stai~si~cs (National Center for
Education Statistics, l9~b:358-3803.
Enrollment-based changes in the demand for public school teachers
below the national level are usually better obtained directly from actual
enrollment projections for the state or locality in question than from pro-
jections of the school-age population. The reason for preferring enrollment
projections at the subnational level is the difficulty of projecting internal
and foreign migration for subnational populations. Although birth and
death data for population projections are quite accurate, migration esti-
mates are less certain, especially as they must be allocated by age. Both the
uncertainty and the effect of migration are greater for subnational aggre-
gations than for the entire country, yet no data are collected for interstate
or intrastate movement.
Enrollment projections, however, are typically based on annual cen-
suses of the school population that are taken for administrative purposes,
OCR for page 39
DETERMINING TEACHER DEMAND
39
including the allocation of state support. The bases for enrollment projec-
tions are therefore firm, and the grade progression ratios can be updated
annually as circumstances change. The changing circumstances include net
migration, which is picked up by the grade progression ratios.
Geographic Differences
in Projected K-12 Enrollment
There are great differences among regions of the United States, and
the localities within those regions, in prospective public school enrollment
change over the next 10 to 20 years. The nationally projected growth,
and expected eventual decline, in the 5- to 13-year-old population and the
current nationwide decline and subsequent slow growth in the 14- to 17-
year-old population are far from evenly distributed among states or within
states.
A well-known compendium of individual state enrollment projections
that shows the dramatically different demographic expectations among
states is produced by the Western Interstate Commission for Higher Ed-
ucation (WICHE) in cooperation with Teachers Insurance and Annuity
Association and the College Board. WICHE produces projections of num-
bers of high school graduates using K-12 enrollment data provided by the
states and a cohort survival methodology. The authors do not attempt to
integrate the individual state projections into a valid national projection by
making explicit assumptions about migratory movements among the states,
although they do sum the projections into regional and a national totals.
However, recent migratory movements are embedded in the observed pro-
gression ratios for each state, which are used to move the enrollments
forward into the future. The WICHE projections illustrate the potential
for sharply divergent demographic futures in the different regions of the
nation and within states between 1986-91 and 2003-4.
In WICHE's 1988 set of projections (WICHE, 1988) the number of
high school graduates nationally has formed a "roller coaster" pattern since
the late 1970s, a pattern that will continue through the next decade. The
general roller coaster pattern reflects past birth patterns in the United
States, but it differs from region to region and from state to state. The
regional differences, the document's foreword explains, are due to "the
mobility of the population, varying economic conditions, and growth in
minority populations." The number of high school graduates is projected
to decline for the North Central and Northeast states, while the West,
South, and South Central regions are projected to have little decline in the
mid-199Os and then will experience substantial growth Cables 2.1 and 2.2~.
OCR for page 40
J
x U: 11
'
h
04 ~-P
1o
47
-
.o -
~ce
. - ~
o
o
e~
1
CD
oo
o ~
.- --
-
oo
'e
L.
-
o
o
c)
co
~c
b.
· -
-
o
oo
o
o
. -
· - ~
o
e~
m
L~
0
V
O O
O C~ ~ CO 00
O O O O ~
. . . . .
O
~4
o
U:
CD
C~
x ~ 1 1 _
~ ~ ~ 8
~ ~, ~
o
X Co ~ ~ _
_ , . _
X U. ~ ~ _
_' - , , ~
o
e~ ~ ~ a,
o oo oo CO
~ ~ U:
u: oo e
~ C9 CO
CO ~ ~ U:
~ c~ eq c
oo C~ ~ o
00 0 ~ a,
.. . . .
~ ~ o o
CD~ 00 ~ 00
O _ _1 t_
CO~ CO CO oo
UtUt ~ ~ oo
COCD CD tD u:
oC~ ~ CO CO
oo o o C~
.....
~o
C~
o
~o
C~
C9
-
oo
o ~ U: CO
CO ~
O ~ c~ CD
(D ~ c~
oO
~ ~ ~ cD
o~ u, ~ ~
o o o o
. . . .
oo u' o
o
o
- - -
~ oo ~
~ c~ ~
~ oo co
C a, Co O °. ° ° °
-4
oo
o.
oo
~i
o
40
oo CO o oo
oo C5) ~ _
CO ~ CO
_ . - C~
oo
~ U: Ut ~
CD - ~ 00 ~ O
°° `rs °° °° °° '5)
~ ,. . .
~ ~ oo C~
oo ~ oo ~ oo oo
~`e ~ G) ~ C)
~
OCR for page 41
oo e~ ~ c~ co
~ ~ ~ ~ ~ ~ o o
. . . . . ~ . .
o o o o o o ~ -
e~ ~ u, a) ~ u, 0 ~ ~oo c~ e~ CD
a~ u~ 0 ~ (D ~ ~ C ~c~ ~ ~ C~ ~
- 0 (53 Oo _~ ut ~ co ~ e~ V °
00 0 CO ~ CD CO O CO U) ·- C~
oo ~ ~ oo c~ ~ e~
~3 U~ CD ~ ~ ~ ~ Ch
- - - - - - - - - - - - - -
co ~ co ~ oo
oo oo oo oo ~ ~ oo oo ~
o o o o o o o o o
~ o ~u,
00 C53 ~
. . ..
o o oo
O ~ 00 ~ ~ ~ 00 00 CD~ ~ CC ~P
00 CO ~ 00 C5) ~ ~ ~ CD 00
co ~ 0 oo 0 e~ oo 0 ~ co
- - - - - - - - -- - - --
CO C 7 CD CO ~ O ~ U:~ CO O C~CO
ee co e~ ~ ~ co ~ ~ oo ~ ~e~
U~ U~ U) u0 U~ U: U) U~ U~U: U: U~ CDCO
oo e~ c~ CD 00 t-u3 CO CO ~e~
oo ~ oo CO
. . . . . . . . .. . . ..
O 0 0 0 0 0 0 0 00 0 0 00
~ CO U: ~ CD ~ r~ a, oot~ ~ e~ cs
O oO (D ~ ~ e~ co 0 0~ ~ _ cor~
~ O (D ~ c~4 ~ O ~ ~CO oo CO ~e~
_ _ _ _ _ . _ _ __ _ _ __
co a) oo ~ ~ co ~ 0c~ ~ e~ oo
~ ~ C~ CD CO <53 _ O~ ~ ~ ~
CO CD C9 CD ~ CD CD t~ ~CO <9 C~ ~CD
0 ~ oo oo e~ ~ ~
0 ~ C~ CS) 0 0 0
~ 0 0 0 ~ _ _
~ ~ e~ a, ~co
_4 _~ _ _ _~
. .. . . ..
_ ~_ _1 _
CO ~ ~ ~ ~ CD e~ ~ ~ u: ooo
e~ co _ oo 0 co ~ ~ ~ ~ ~ CD~
oo ~ ~ oo co e~co ~ CD COO
________._____
(D ~ co CO U) e~ ~ CD~ e~ 0 ~
oo ~ ~ ~ 0 ~ ~ ~ ~ oo ~ ~_
r~ ~ oo oo oo oo oo00 ~ oO ~a
oo 0 c~ ~o O o: a) co ~ u~ oo ~
0 0 ~ _ _ e~ coco co c~ ~r
. . . . . . . . .. . . ..
O ~ ~ ~ ~ ~ ~ ~ ~ ~ _ ~_1
0 ~ o' co ~ e~ 0 0 e~ ~ ~ ~0
c~ U3 oo u, ~ e~ ~ o' u:~ ~ ~ ~oo
0O ~ ~ O 00 ~ CO CO ~ ~ O 00C9
- - ~ - - - - - - -- - - --
CD ~4 ~ ~ CO ~ CO O a,
o ~ ~ - eo~ u: ~ c~o
U~ U: u~ ~ ~ COC9 CO CD CD
~ -
cO
. .
oo
oo
C53
-
-
o
· -
cS
v
~:
. -
o
. -
~3
~o
· -
o
v
~d
~a
~;
s"
U3
, ~ ~ co ~ ~ ~ o ~ ~ ~
..
1 1 1 1 1 1 1 ' 1 1 `_ 1 1 1 1
0 ~ e~ oo ~ u3 co ~ °° ~ C~ ° ~ e4 o~
C5) Ch ~ ~ ~ C5) ~ ~ ~ ~ O O O O ~
C5) ~ ~ O O O O O
)
41
OCR for page 42
42
PRECOLLEGE SCIENCE AND MATHEMATICS TEACHERS
TABLE 2.2 Projected Proportion of United States High School Graduates, by
Region, 1986-2004
l
South/ Total
South North North- United
West Central Central east States
1985-86 0.18 0.30 0.27 0.25 1.00
1986-87 0.18 0.30 0.27 0.24 1.00
1987-88 0.19 0.30 0.27 0.24 1.00
1988-89 0.19 0.31 0.27 0.23 1.00
1989-90 0.19 O.31 0.27 0.23 1.00
1990-91 0.19 0.32 0.27 0.22 1.00
1991-92 0.20 0.32 0.26 0.22 1.00
1992-93 0.20 0.32 0.27 0.22 1.00
1993-94 0.21 0.32 0.26 0.22 1.00
1994-95 0.21 0.32 0.26 0.21 1.00
1995-96 0.21 0.32 0.26 0.21 1.00
1996-97 0.22 0.32 0.26 0.21 1.00
1997-98 0.22 0.31 0.26 0.21 1.00
1998-99 0.23 0.31 0.25 0.21 1.00
1999-2000 0.23 0.32 0.25 0.21 1.00
2000-01 0.23 0.32 0.24 0.21 1.00
2001-02 0.24 0.32 0.24 0.21 1.00
2002-03 0.24 0.31 0.23 0.21 1.00
2003-04 0.24 0.31 0.23 0.21 1.00
Source: Western Interstate Commission for Higher Education (1988:133.
Local-Area Projections
The striking differences in enrollment-based demand projections among
the states are mirrored within states by differences among localities. Since
teacher labor markets have important local components (as the panel's
case studies suggest), it would be useful to be able to produce enrollment
projections for local areas. One barrier to local enrollment projections is
the reliability of small-population projections that are needed to estimate
enrollment in kindergarten and the first grade. Fertility patterns can vary
locally, and births and deaths may be reported for an area different from
OCR for page 43
DETERMINING TEACHER DEMAND
43
school district boundaries. More important, the smaller the population, the
greater the potential influence of hard-to-predict migration on future size
and distribution.
The difficulty of making reliable small-area projections is undoubtedly
one reason that few school districts appear to make projections beyond the
next year, if that far. However, standard enrollment projection techniques
would be adequate to give general magnitudes of change in all but the
smallest and least stable districts for 5 to 10 years into the future, longer
for the secondary level. State departments of education could make a
considerable contribution by encouraging school districts to project their
enrollment for 5 to 10 years in the future, providing technical guidance
in developing projections and coordinating their efforts. Properly done,
such projections could be combined in order to approximate likely levels of
enrollment-generated demand within teacher labor markets. Of course, es-
timation of attrition-generated demand and subject-specific demand would
require additional projection efforts.
Population Projections A Proxy for Enrollment Projections
Although the cohort survival method of projecting enrollment is widely
used in the education community, population projections reveal some highly
interesting trends that could influence the demand for teachers.
Using Census Population Projections
to Estimate K-12 Enrollment Demand
At the national level, projections of the population by age provide a
very good proxy for enrollment projections, especially if the interest is in
total enrollment demand and not just demand for public school teachers.
This is particularly true for the population age 5 to 13, which has close to
100 percent attendance, virtually all of it in grades K through 8.
Population projections have the advantage of greater simplicity than
enrollment projections, since assumptions about movement from grade to
grade or from public to private schools do not have to be made. National
projections of the population are updated by the Census Bureau every
several years, more often if the underlying assumptions prove incorrect.
The biggest disadvantage to using population projections as a substitute
for enrollment projections is that population projections take no account
of possible changes in dropout rates, an important element of enrollment
projections for the secondary level. Using population projections as a proxy
for K-12 enrollment, in particular for grade 9-12 enrollment projections,
makes the implicit assumption that dropout rates will remain constant.
OCR for page 44
44
PRECOLLEGE SCIENCE AND MATHEMATICS TEACHERS
Since our present discussion is limited to a general overview of the
likely forces of change in the demand for teachers, our remarks on the
likely contribution of nationwide enrollment change to teacher demand are
based on population projections.
The national projections for the 5- to 17-year-old age group should
prove moderately accurate through the year 2000, as shown in Table 2.3.
Reliability declines toward the end of the l990s, when the projections for
the age group begin to depend more on projections of births and less
on children already born. However, the fertility assumptions used in the
projection have been close to actual fertility so far, and there is little
reason to expect large changes in the fertility rates in the next few years.
The other factor that could lead to the divergence of the actual numbers
from those projected international migration is unlikely to cause major
discrepancies at the national level in the period and at the ages shown
in the table. The size of the U.S. population relative even to high levels
of migration, and the typical concentration of migration in the early adult
years, dampens the effect on the school-age population, at least in the short
and medium run. Interstate migration is, of course, not relevant to national
projections.
Recent projections by the Census Bureau show 12 percent growth for
the school-age population in the United States between the middle l980s
and the end of this century (Table 2.3~. The 5- to 17-year-old age group is
projected to grow by more than 5 million. However, as a result of past birth
patterns, the increase will not be distributed equally across the age group.
Until the end of this century, most of the growth will occur at the younger
ages and will affect the elementary grades. The number of children age 5
to 13 is projected to increase by nearly 5 million, or 17 percent, by 1999.
After 1999 this age group is projected to decline. By contrast, the number
of young people age 14 to 17, the secondary school-age group, is projected
to decline 12 percent in the 5 years between 1985 and 1990, a reduction of
1.8 million. Thereafter, the number of secondary school-age children will
grow slowly but is not expected to regain the 1985 level until 1997. More
rapid growth is projected for the early years of the next century.
The projected demographic changes in the school-age population will
have opposing potential effects on the nationwide demand for teachers,
increasing it at the elementary level and reducing it at the secondary level.
There could even be a reduction in the absolute number of secondary school
teachers employed over the next few years. However, change in the number
of students is only one of the elements in the calculation of demand for
teachers. Teacher attrition and pupil-teacher ratios are the other important
factors in demand, although pupil-teacher ratios are probably as much
dependent on enrollment change as they are an independent factor. The
OCR for page 45
DETERMINING TEA ClIER DEMAND
TABLE 2.3 Projections of the United States School-Age Population to the Year
2000 (in thousande)
Age 5-13
Number
(000)
Age 14-17
-
Index
(1985
= 1.00)
Number
(000)
Age 5-17
Index
(1985
= 1.00)
Number
(000)
Index
(1985
= 1.00)
1985 29,654 1.00 14,731 1.00 44,385 1.00
(actual)
1986 29,922 1.01 14,588 0.99 44,510 1.00
1987 30,358 1.02 14,237 0.97 44,595 1.00
1988 30,954 1.04 13,662 0.93 44,616 1.01
1989 31,523 1.06 13,160 0.89 44,683 1.01
1990 32,189 1.09 12,950 0.88 45,139 1.02
1991 32,777 1.11 12,964 0.88 45,741 1.03
1992 33,400 1.13 13,087 0.89 46,487 1.05
1993 33,900 1.14 13,260 0.90 47,160 1.06
1994 34,193 1.15 13,714 0.93 47,907 1.08
1995 34,435 1.16 14,082 0.96 48,517 1.09
1996 34,598 1.17 14,513 0.99 49,111 1.11
1997 34,681 1.17 14,848 1.01 49,529 1.12
1998 34,668 1.17 15,027 1.02 49,695 1.12
1999 34,566 1.17 15,214 1.03 49,780 1.12
2000 34,382 1.16 15,381 1.04 49,763 1.12
Source: Bureau of the Census (1984a:43-74~.
45
demand for teachers of specific disciplines, of course, depends on student
choice (or changing graduation requirements) as well.
The Changing Demographic Profile of the School-Age Population
Education planners and social observers have devoted considerable
attention to the changing demographics of the American population and to
projections of large continued changes. The demographic changes referred
to are usually changes in ethnic composition and family circumstances,
especially increased proportions of children in single-parent families and/or
OCR for page 46
46
PRECOLLEGE SCIENCE AND MATHEMATICS T=CHE~
especially increased proportions of children In s~ngle-parent families and/or
with working mothers. Increased poverty levels among children are often
also at issue. The conclusion, sometimes stated explicitly and sometimes
left to the reader, is that these changes require urgent social and political
attention.
These changes in the relative size of elementary and secondary
school-age populations, in racial/ethnic composition, in income and pov-
erty- are interesting and relevant considerations attendant on the demand
for teachers. They do not emerge from enrollment projections, but rather
through population projections. We first tale about broad population fac-
tors, demographic trends that influence demand. Then we return to the
use of enrollment projections in models of teacher supply and demand.
.
Projecting Changes in Race and Ethnicit~r. Expected change in the eth-
nic distribution of school-age children in America is of interest for our
discussion only insofar as youngsters of the different categories may be
expected to have differential patterns of enrollment in mathematics and
science courses or require different strategies of teaching than are used
currently. However, the link between ethnic and racial identity and school-
related characteristics or needs is not a clear one, particularly over the long
run. Unless there is reason to believe that racial and ethnic groups will
retain currently observed particular needs over the long run, projecting the
racial and ethnic distribution of the school-age population, or of school
enrollment, is of little utility for planning curricular or other change for
science and mathematics.
Very often, the effect of racial and ethnic change, especially the effect
of the projected increases in the proportion of the population of Hispanic
and Asian origin, is confounded with the effects of migration for example,
an increase in the number of students with limited ability to speak English
or from families with the low educational levels characteristic of Latin
America and much of Southeast Asia. This confusion is bemusing in a
country that has seen the children and grandchildren of poor, illiterate
immigrants from Southern and Eastern Europe people viewed as forever
unassimilable 75 years ago-become thoroughly assimilated Americans.
Changes in racial and ethnic distribution per se may be the least
reason for expectations of changed enrollment patterns in science and
mathematics or for planning changes in curricular and teaching strategies.
However, because of the great public interest in ethnic change, we explore
the feasibility of racial and ethnic projections and consider the results of
recent projections.
As noted earlier, standard population projections require a base popu-
lation and assumed rates of fertility, mortality, and migration for each age.
The decennial census counts the black population with reasonable accuracy,
OCR for page 47
DETERMINING TEACHER DEMAND
47
at least for most public policy analysis purposes. Vital statistics are virtually
universally collected by race, including birth and death records. Estimating
migration by race is more troublesome, but sufficient data exist to develop
estimates of base populations by race between censuses and to develop
assumptions for projections. It is feasible to project the black population
and the white population, although, as with all subcategory projections,
such projections will tend to be somewhat less reliable than projections
of the total population. Other racial groups are more difficult to estimate
because of small numbers and, in the case of Asians, very high- rates of
foreign migration.
Projections of nonracial ethnic populations are more difficult. On the
whole, ethnic identity has been gathered only sporadically in either the
census or in vital statistics. Even were a group to arrive all at once, thereby
providing a clear base population, and subsequently maintain accurate birth
and death records, the accuracy of any long-range projection would be in
doubt because of the likelihood of intermarriage and the lack of an agreed-
on definition of ethnic identity for the children. The reasons underlying
the questionable feasibility of ethnic projections also raise the question of
their meaningfulness for social or educational policy planning.
Because of the interest in the rapid increase in population from Mexico
and Central and South America, numerous projections of the Hispanic pop-
ulation have been produced. The Census Bureau first asked respondents
to identify themselves as Hispanic or non-Hispanic in the 1980 census.
The states with the bunk of the U.S. Hispanic population began to ask
Hispanic identity for birth and death certificates around 1980 as well. The
data collection efforts since 1980 provide a base population for projecting
Hispanics, as well as fertility and mortality rates. Migration estimates can
also be made, although with only very modest reliability. Given the im-
portance of migration for determining the size of the Hispanic population,
this is a decided disadvantage. More problematic for long-range projection
purposes is the lack of a socially agreed-on ethnic identity for the children
of marriages in which only one partner is identified as Hispanic.
Results of Racial and Ethnic Projections. The following discussion ex-
amines the results of ethnic and racial estimates and projections for the
national population by age. In the last two decades, the proportion of
American youngsters from non-Hispanic white backgrounds has decreased
nationwide and the proportion from Hispanic and from nonwhite back-
grounds has risen. This shift is projected to continue, although, as shown
in Table 2.4 and described below, the change will be relatively modest at
the national level.
In 1970 the census recorded that 13.5 percent of the population age
5 to 17 was black The proportion rose to 14.8 percent in 1980 and by
OCR for page 49
DETERMINING TEACHER DEED
49
which are real but less dramatic nationally than might be supposed from the
level of popular interest. Providing an adequate education for immigrants
and their children is a crucial concern for the schools, especially in areas
of heavy influx. It is less clear how important the ethnic shifts, in and of
themselves, will be to the schools in the future.
Family Structure and Changes in Poverty Rate for Children. Other shifts
in the characteristics of school-age children may be of far more importance
to science and mathematics enrollment, as well as to the curriculum that
teachers should be prepared to teach, than changes in racial and ethnic
distribution. These include changes in family structure, specifically the
increase in the proportion of children in single-parent families, and the
increase in the proportion of children living in poor families.
Data as of 1985 showed that 16 percent of all white children, 43 percent
of all black children, and 40 percent of children of Spanish origin were
reported to be living in poor families (Bureau of the Census, 1986: 22~. The
rate of family poverty among all children rose during the 1970s and early
1980s, wiping out gains made in the 1960s. The poverty rate for children
was 20.1 percent in 1985, compared with a low of 13.8 percent in 1969
(Bureau of the Census, 1986:22~.
During roughly the same period 1970 to 198~Current Population
Reports revealed that the percentage of children living in single-parent
families doubled (Bureau of the Census, 1984b:4~. In 1984, 22.6 percent of
children under 18 were living with one parent, compared with 11.9 percent
in 1970 (Bureau of the Census, -1984b:4~. In 1984, over half of all black
children lived with only one parent, compared with one-sixth of all white
children. Among children of Spanish origin, one of every four lived with
one parent.
As has been relentlessly demonstrated in innumerable studies, poor
children, so many of whom live in one-parent families, are at risk of school
failure because of multiple disadvantages, which may include the lack of
adequate housing, or any housing at all; frequent moves from school to
school; less than sufficient food; inferior medical care; a total lack of dental
care; exposure to criminal behavior in deteriorating neighborhoods; and the
stress that accompanies the struggle of the adults in the family to survive.
Projections of the proportion of children living with one parent can
be made with moderate reliability using current data; the proportion liv-
ing in poverty requires assumptions about the economy as well and are
therefore less easy to construct, or at least less easy to construct with any
reliability. The nature of the effect of family poverty and family structure
on academic achievement and, more specifically, on the demand for science
and mathematics instruction is less well understood. The subject is of great
social importance, given the large numbers of children involved, and we
OCR for page 50
so
PRECOLLEGE SCIENCE AND MATHEAL4TICS TEACHERS
hope that the links will be better understood in the future. However, at
present, not enough is known to be useful in constructing projections of
demand for science and mathematics teachers.
1b summarize, there are interesting and relevant trends that emerge
from population projections, which should be included in any statistical
description of the changing demand for teachers. However, for purposes
of projecting demand for precollege science and mathematics teachers, the
education community generally finds enrollment-based projections more
useful
Research Areas Related to Student Enrollment
Methods employed in current teacher demand models, specifically the
cohort survival methodology used to project enrollments, are relatively
reliable. However, for longer-term projections, particularly at the high
school level, and for specific subjects within science and mathematics fields,
enrollment projections are less reliable due to the impact of changes in the
behavior of students, parents, and school systems. The utility of demand
models for addressing-policy issues concerning science and mathematics
education over the long term would be greatly enhanced by the development
of more dynamic, behaviorally responsive models. We discuss three types
of behavioral responses that need to be understood to develop more useful
models of teacher demand. They are determinants of course selection by
students, determinants of parental and student preferences for public and
private schooling, and changes in dropout rates that can be expected in
response to social, economic, and educational changes. We discuss them
in the order of their priority as we assess the relative importance of each
topic to teacher demand projections and the relative gains that could be
expected from research.
As the panel's interim reported stated (National Research Council,
1987c:49), research on the determinants of course selection by students is
critical to the development of useful projections for broad subject cate-
gories, including science and mathematics, at the high school (and possibly
middle school) level. This is an area about which we know very little. Many
factors can influence students' choice of courses, including high school grad-
uation requirements, college entrance requirements, government (including
federal and state) support for science and mathematics education that mo-
tivates schools to encourage enrollment in these subjects, and fashions
or tastes on the part of students and their parents and peers for certain
subjects.
Given that most current models focus on public school demand (al-
though the National Center for Education Statistics model develops sep-
arate public and private school projections), another important area for
OCR for page 51
DETERMINING TEACHER DEMAND
51
research concerns the determinants of parental and student preferences
for private and public schooling (National Research Council, 1987c:49~.
Nationwide, private elementary and secondary school enrollment was 11.5
percent of the total in 1980, but had grown to 12~5 percent by fall 1987
(NCES, 1988b:9~. Changing preferences for private school enrollment, a
topic about which almost nothing is known, can importantly affect public
school demand. Particularly in today's educational climate, when private
schools are perceived by some parents to offer a more attractive educational
environment than public schools, research into the factors that influence
the choice of type of school is needed.
One type of response that affects demand projections at the high
school level is the dropout rate (National Research Council, 1987c:50~. We
know a good deal from previous research about why students drop out of
school. Work is needed, however, on changes in dropout rates that can
be expected in response to a variety of social, economic, and educational
changes. For example, the changing ethnic composition of the school-age
population in many areas of the country may dramatically affect dropout
rates in those areas. Increased high school graduation requirements may
increase dropout rates as a side effect of raising educational levels for those
who stay in school.
PUPIL-TEACHER RATIOS
Enrollment change does not translate immediately into a corresponding
proportional change in the demand for teachers. As a recent RAND report
assessing teacher supply and demand explained, "adjustments are made to
pupil-teacher ratios to smooth the effects of rapid enrollment changes, to
accommodate established school staffing patterns and budgets, and to take
into account existing contractual agreements with teachers, in the case of
enrollment declines" (Haggstrom et al., 1988:37~.
A small change in pupil-teacher ratio can cause a significant change
in the projected demand for teachers. Changes in pupil-teacher ratios can
be caused by a number of factors at the school, district, or state level:
changes in school budgets; staffing patterns, class sizes, or teaching loads;
graduation or program requirements; and course offerings. A layering of
school, district, and state policies may add to the complexity of factors that
change the ratio and the demand for teachers. Even though these factors
are complex, they should be identified and discussed briefly as components
of teacher demand (Haggstrom et al, 1988:37-38~.
OCR for page 52
52
PRECO' [FETE SCIENCE AND MATHEMATICS TEACHERS
Components of Teacher Demand and Related Data
Changes in school budgets can cause changes in staffing practices, class
sizes, and teaching loads. The district's allocation of its budget among its
many needs for staff, materials, and services affects the pupil-teacher ratio
generally as well as specific programs or subjects. In one of the panel's case
studies of a school system in a western state, the district filed for bankruptcy
following a teachers' strike. This led district officials to make conservative
estimates of the number of students expected to be enrolled- and hence
the number of teachers needed. An overestimated enrollment could cost
the district roughly $1,000 per student, it was thought. The tendency to
underestimate enrollment and therefore the number of teaching positions
has had various effects: raising pupil-teacher ratios, straining teaching
loads, or eliminating such support as department chairmanships or resource
teachers in disciplines such as mathematics.
Implementation of a school finance formula that changes a district's
proportion of local discretionary resources can also affect pupil-teacher
ratios. The panel's case studies found that, although some school districts in
a southeastern state had enough local discretionary funds to hire additional
teachers (part time or full time) in computer science or other subjects,
other districts in the state had very little. Local discretionary money if the
district knew the amount far enough in advance-could be used to sign an
early contract with a talented candidate for a mathematics resource teacher
or an elementary science teacher, for example. Loss of that opportunity
could mean leaving the position unfilled. These examples suggest how the
budget can directly affect the ability to hire and can substantially affect the
pupil-teacher ratio for certain subjects, and ultimately general pupil-teacher
ratios.
Changes in stations pattems, class sizes, or teaching loads may be pro-
mulgated by a district rule or policy change, by state policy-or by both,
as when a district rule extends beyond a state requirement. These can
cause an immediate change in pupil-teacher ratios and in the demand for
teachers. A district requirement to employ a full-time guidance counselor
in every elementary school, without full additional funding to do so, could
strain staffing patterns elsewhere in the school and indirectly push up the
pupil-teacher ratio. The Schools and Staffing Survey (SASS), recently ini-
tiated by NCES and first fielded in 1988, collects information on staffing
patterns, class sizes, and teaching loads. The second SASS survey will be
conducted in 1991 and at regular intervals thereafter. As a time series
of data becomes available, it will be possible to monitor changes in these
variables over time. SASS includes a teacher demand and shortage ques-
tionnaire for public school districts and private schools, as well as a school
administrator questionnaire for public school principals and private school
OCR for page 53
DETERMINING TEA ClIER DEMAND
53
heads. It also includes a teacher questionnaire and a teacher-follow-up
survey: a one-year follow-up survey of the sample teachers who have left
teaching and some who have remained. This ongoing, integrated survey ef-
fort has been designed to provide the most comprehensive data on teacher
demand and supply available to date. Although we mention the survey
frequently as a potential data source, as with all new surveys, the extent to
which it will live up to its expectation cannot be known until policy makers
and the research community have used the data in a variety of analyses.
Course requirements in science and mathematics for graduation- (usually
established by the state) clearly affect pupil-teacher ratios. As an example
of such policies and responses that can cause changes in pupil-teacher
ratios, most states and school districts have increased their graduation
requirements since 1980 (NCES, 1988c), often adding additional science
and mathematics course requirements. The Center for Policy Research in
Education (CPRE) reports that since 1983, 42 states have added course
credit requirements in science, mathematics, or both (CPRE, 19894. NCES
has found, through its Fast Response Survey System, that the state re-
quirements are often exceeded by the requirements of individual districts
(personal communication, M. Papageorgiou, NCES, June 7, 1989~. Unless
more teachers are assigned or hired to teach science and mathematics, the
pupil-teacher ratio for these subject areas clearly increases.
Some nationally collected data on high school graduation requirements
are available on a regular basis. The Education Commission of the States
(ECS) and the Council of Chief State School Officers publish information
periodically on state-mandated high school graduation requirements. They
track mathematics and science as general categories, however, listing only
the number of courses or years of science and mathematics that are required
for graduation.
SASS includes an item in the teacher demand and shortage question-
naire for public school districts on high school graduation requirements, by
subject (physical and biological sciences, mathematics/computer science).
It asks for changes in these requirements between 1987 and 1988. Future
SASS results will reveal changes in requirements over longer time spans.
Course offerings and enrollments also influence pupil-teacher ratios.
Whereas requirements clearly help determine what courses high school
students take and the demand for teachers of those subjects an impor-
tant constraint is whether the required courses are actually offered. For
example, only a few schools offer a complete range of college-preparatory
mathematics and science courses; a physics course might be offered only
every other year. And very few students are enrolled in the most advanced
courses.
Course offerings and enrollments by school, school system, and state
emerge as an important variable. Course offering data could serve as an
OCR for page 54
54
PRECOLLEGE SCIENCE AND MATHE~4TICS TEACHERS
excellent indicator of science and mathematics demand by either students
or state requirement. Course offerings and enrollments would contribute
toward indicating change in demand over time, by size of school and
school system, by other relevant school district characteristics, and by state,
especially when state graduation requirements for science and mathematics
courses have changed. Course offerings and enrollment would also permit
analysis of the degree of school response in terms of teacher assignment.
The extent to which teachers need to teach more than a single subject
could be noted. Some data related to course offerings and enrollments by
school system and school are being gathered nationally.
The 1985-86 National Survey of Science and Mathematics Education
(Weiss, 1987) provides the most recent comprehensive data on course of-
ferings. Data on course offerings are also included in (1) the longitudinal
study High School and Beyond, (2) in the National Educational Longitudi-
nal Study of 1988 (NELS:88) (for middle schools and junior high schools),
and (3) for students age 9, 13, and 17 in the National Assessment of
Educational Progress (NAEP) in those years when science or mathematics
achievement are assessed. While data from these surveys are disaggregated
by specific subject areas, they are conducted infrequently.
The data on course offerings and course enrollments are "plagued
with inconsistencies," according to a recent report on elementary and
secondary science education (Office of Technology Assessment, 1988:42~.
Course titles often are not a reliable basis for comparisons among schools,
states, or years. Moreover, some advanced courses are offered not by the
high school but by the community college, and there are no national data
available on this practice. National data do not show how often a physics
course is given. Nor do we know how many sections of a given course
are taught. These data may change, as well, from year to year in a single
school.
NELS:88 asks middle schools and junior high schools for data on nu-
merous courses and whether they are offered. In a more detailed format,
questions on course offerings are included in the SASS teacher question-
naire. Given a probability sample of teachers by fields (as is the case for the
teacher sample), it should be possible to estimate the prevalence of course
offerings and trends in course offerings (including science and mathematics
offerings) at national and regional levels.
Finally, course offerings could serve as a basis for drawing samples
to test for varied working conditions, recruitment patterns, and range of
initial assignments possible among school systems. Vacancies matched to
schools classified by offerings might indicate conditions of low retention.
The SASS questionnaire of local education agencies (LEAs) asks district
administrators for the total number of positions that are either vacant, filled
OCR for page 55
DETERMINING TEACHER DEMAND
55
by a substitute, or withdrawn for lack of a suitable candidate; this total is
disaggregated by subject.
Research on Determinants of Pupil-Teacher Ratios
A closer analysis of often interrelated factors that influence pupil-
teacher ratios is a rich area for further research. As noted earlier, in
most models pupil-teacher ratios are estimated in a relatively-arbitra~y
way. But numerous factors operate and interact to cause changes in pupil-
teacher ratios for science and mathematics subjects and in general. And we
suspect that certain types of dynamics in teacher markets (e.g., declining
enrollments or increased school budget) may be associated with declining
pupil-teacher ratios. Other conditions (e.g., surging enrollments, budget
cutbacks) are associated with rising pupil-teacher ratios. Since that ratio
is so critical to an assessment of the demand for teachers, research on its
determinants is needed.
I-he factors that can change pupil-teacher ratios affect adjustments over
both short-term and long-term periods, although short-term adjustments
differ from longer-term ones (National Research Council, 1987c:50~. For
example, a shortage situation may result in a marked increase in pupil-
teacher ratios until the school system has had time to implement responses,
such as extended recruitment or hiring teacher aides.
A common practice in projecting the demand for teachers is to project
the increase in enrollment and to divide it by the current pupil-teacher
ratio to calculate the number of additional teachers needed to provide for
the enrollment increase (National Education Association, 1987f:14~. But
an assumption that using the current, general pupil-teacher ratio reflects
accurately the number of teachers is too simplistic; enrollment changes
affect pupil-teacher ratios in more indirect ways. For example, research
suggests that, when enrollments decline, teacher unions may be willing to
forgo salary increases to keep current teachers employed (Cavin et al.,
1985; noted in Haggstrom et al., 1988:42~. Despite enrollment declines,
school boards may decide not to lay off staff in science and mathematics
if they have had difficulty hiring them in the past, or if they feel it will be
hard to find qualified teachers of certain subjects in the future (Prowda
and Grissmer, 1986:12~.
Moreover, supply-demand projections for precollege science and math-
ematics teachers will be far more meaningful if both enrollments and
pupil-teacher ratios are disaggregated by subject area. In Connecticut, for
example, secondary enrollments have been declining, but the demand for
secondary science and mathematics teachers is steady or may increase be-
cause of increased graduation requirements in these subjects, coupled with
OCR for page 56
56
PRECO!:LEGE SCIENCE AND MATHEMATICS TEACHERS
decisions made (influenced by budgetary considerations) to decrease class
sizes (Prowda and Grissmer, 1986:12~.
There is reason to believe that pupil-teacher ratio is a dependent as
well as an independent factor in the creation of demand. In periods of
enrollment growth and teacher or financial shortages, the ratio (or class
size) can be increased. When demand slackens, if all "surplus" teachers are
not let go, then the ratio drops. A history of these coping responses would
be extremely useful in developing better assumptions about pupil-teacher
ratios for demand models than the assumption that the current, general
pupil-teacher ratio is adequate. The development of demand models for
science and mathematics teachers would require similar information on the
history of class size responses to teacher shortages or surpluses within these
disciplines, a formidable undertaking but necessary for really well-defined
models.
In sum, we continue to recommend, as we did in our interim report,
that research be conducted on the determinants of pupil-teacher ratios,
including research on adjustment lags as enrollments change and on how
changes in demand for courses contribute to changes in these ratios and in
the demand for teachers of science and mathematics.
TEACHER ATTRITION RATES
The third major element in the construction of teacher demand models
is the rate at which teachers leave their jobs. It should be noted that
teacher attrition is largely a supply phenomenon, reflecting the decisions
of individual teachers. In Chapters 3 and 4 we treat attrition as a supply
variable, but here it is natural to think of it as resulting in a demand for
new teachers.
One part of the leaving rate-retirement is fairly easy to model, if
data on the teaching force by age are available and if something is known
about the typical ages at which teachers retire. Rates of attrition for other
reasons are much less easy to determine. Some of this information exists in
school records, but it must be gathered and put into forms usable by those
developing models. In aggregating attrition data gathered from school
districts it is important to avoid double counting, since what is attrition to
one district might be a new hire to another.
The Connecticut model of teacher supply and demand revealed how
important age is in estimating attrition rates (Prowda and Grissmer, 1986~:
"We have noted high early career attrition rates, low mid-career attrition,
and high attrition around 60 and 65 years old" (p.l). It is likely that
attrition rates change over time, reflecting the numbers of teachers hired
in a given year or period (high attrition rates may be observed soon after).
A RAND study of teacher attrition found a similar U-shaped pattern of
OCR for page 57
DETERMINING TEACHER DEMAND
57
attrition in Illinois, Michigan, New York, and Utah (Grissmer and Kirby,
1987:36-38~. Higher attrition rates were found among newly hired teachers
than among other groups, including those eligible to retire.
In addition to reflecting age or years of experience, calculating attrition
for mathematics and science teachers requires gathering the necessary data
by specific field, a considerably more demanding task than obtaining the
data for the entire teaching force. There are studies that have gathered
discipline-specific data (Murnane and Olsen, 1989a, l990b; Grissmer and
Kirby, 1987), but these data have not often, if at all, been used in the
development of teacher demand models (or supply models in which they
would also be of use). As part of the fifth follow-up of participants in
the National Longitudinal Study of 1972, completed in 1986, a Teaching
Supplement Questionnaire was sent to sample members who were teach-
ers, former teachers, and those who had been trained to teach but had
not taught. Information from the survey included detailed professional
and personal histories that could be used for analyses of attrition patterns
during the early part of their careers, from 1977 to 1986. Heyns (1988)
analyzed the data, but results were not reported by field of discipline. The
SASS questionnaires were designed to provide national data on teacher at-
trition by field. The SASS public school questionnaire asks for the number
of teachers, by field, who left in the previous year and their destinations.
However, due to low response rates for these items, researchers have to de-
pend on the SASS teacher follow-up survey of former teachers (conducted
in 1988-89) for estimates of attrition by field. NCES staff is exploring alter-
native ways of obtaining better attrition data. The follow-up survey, which
also asks for the destinations of leavers, is expected to provide national
attrition rates by field.
Attrition has important demographic elements, in part because so much
of it is caused by retirement. Since something is known of the demographic
profile of the teaching force, it is possible to estimate the likely general
trend of attrition in the future, which is almost certain to be on the rise.
In much of the country, low rates of new hires over the past 10 years and
reductions in force, with the newest teachers being the ones let go, have
left a relatively senior task force. Able 2.5 shows an upward trend from
1976 to 1986 in the proportion of current teachers who are age 40 and
over (from 34.6 to 51.3 percent) and who are age 50 and over (from 15.5
to 21.2 percent). A rise in the rate of retirements will, of course, increase
attrition. In addition, evidence noted earlier points to particularly high
levels of attrition in the early years of teachers' careers.
In order to forecast attrition adequately, more information is needed
not only on the distribution of teachers by age, but also by disciplinary
area and level of preparation, and on the current attrition levels within
those categories. It would also be useful to have a better understanding
OCR for page 58
58
PRECOLI EGE SCIENCE AND Mi4THEMA TICS TEACHERS
TABLE 2.5 Age Distribution of U.S. Public School Teachers, 1961-1986
1961 1966 1971
1976 1981 1986
Years
Mean
Median 41 36 35
42 39 38 36
33
Under age
30
Age 30 39 a
Age 40_49 a
Age 50
a 33.9% 37.1%
22.8 22.8
17.5 17.8
and over a 25.8 22.3
39 ~ 41
37 40
37.1% 18.7% 11.0%
28.3 38.8 37.7
19.1 23.1 30.1
15.5 19.4 21.2
a Subgroup data not available.
Source: National Education Association (1987e:73~.
Of why attrition differs for the different categories, if it does, so that more
reliable assumptions can be developed for projection models. These issues
are discussed in more detail in Chapters 3 and 4, which focus on supply.
SUMP
The demand for teachers, as we have indicated, depends on enrollment
changes, both generally and in mathematics and science courses. The
demand for teachers also depends on changes in pupil-teacher ratios,
caused by changes in staffing patterns, class size, teaching loads, course
requirements, and course offerings in mathematics and science. In addition,
a school district's demand for new science and mathematics teachers in a
given hiring season also depends on the number of vacancies in those
subjects. The number of vacancies results not only from the creation of
new positions, but also from teacher attrition, a component of supply.
In general, the panel considers the data available for projecting de-
mand to be more adequate than data for projecting supply. The task of
projecting enrollment-driven demand for science and mathematics teachers
is relatively straightforward.
OCR for page 59
DETERMINING TEACHER DEMAND
59
There is a small number of significant gaps, however, in data related
to demand, and the panel recommends collecting data to fill these gaps.
Forecasting the demand for science and mathematics teachers particularly
could be improved by better data on the following variables:
Course-taking behavior in high school. Data on state-mandated high
school course requirements, collected regularly over time and by science
or mathematics subject, could suggest changes in demand for teachers of
various types or levels of courses. School district requirements often exceed
state requirements, but with both state and district data we could- begin to
trace how changes in course requirements stimulate changes in demand for
secondary science and mathematics teachers.
Changes in course o~enngs in science and mathematics. Changes in
course offerings can change the demand for science and mathematics
teachers and can identify the need for teachers with special skills,. for
example, ability to teach advanced placement physics.
Enrollment changes disaggregated into science and mathematics course
enrollments. Better data on this aspect of course-taking behavior, in con-
junction with changes in course requirements, would strengthen the demand
component of projection models.
Data on attrition for reasons other than retirement by field. Attrition by
retirement is relatively well known. For other types of attrition, further
analysis of the NLS-72 follow-up of teachers and former teachers should
provide more insight into patterns of attrition during the early years of a
teaching career (Heyns, 1988~. The best source for obtaining new nonre-
tirement attrition data will be SASS, which has recently experimented with
questions on attrition by field, although the item response rate was low.
High priority should be given to collecting attrition data because they are
essential to both demand and supply models. The panel urges that NCES
redesign the SASS questions on attrition and subject them to a thorough
pilot test before using them. When combined with other SASS data on
teachers, the attrition data could help answer questions such as: Among
mathematics teachers and science teachers who leave earlier in their ca-
reers, how many had taught advanced courses? Introductory-level courses?
In high school or middle school? The demand created by such patterns
will thus be better known, and a closer fit may be possible in filling the
demand.
In addition, research is suggested on the behavioral factors that influ-
ence the demand for teachers, particularly teachers of science and mathe-
matics in the higher grades, for use in development of improved models for
longer-term projections. Among the research areas noted are the behav-
ioral determinants behind course selection, factors that influence dropout
rates, influences on parents' choice of public versus private schools, and the
relationship between demand for certain courses and pupil-teacher ratios.
Representative terms from entire chapter:
enrollment projections
