Public Policy and the Shaping of Disability: Incidence Growth in Educational Autism
Dana Lee Baker
University of Missouri-Columbia
Citation: Baker, D. L.,
(2004, March 16). Public Policy and the Shaping of Disability: Incidence Growth in Educational Autism.
Education Policy
Analysis Archives, 12(11). Retrieved [Date] from
http://epaa.asu.edu/epaa/v12n11/.
Abstract
Autism has gained the attention of policy makers and public
administrators in recent years. The surge in prevalence, in tandem
with a growing social preference for community inclusion of
individuals with disabilities, strains a variety of policy
infrastructures. Autism and related disorders, which were first
described in 1943, were originally thought to be extremely low
incidence and usually coincident with mental retardation. In
accordance with the disability policy paradigm of the era, public
services for autism were provided predominantly in institutional
settings. Since then, however, autism and related disorders have
come to be understood as more common than was originally thought
and more rarely associated with mental retardation. In this
article, shift-share analysis is used to gain insight into how the
growth in autism incidence is being differentially experienced and
recorded within a single arena of policy across the United States.
The challenges associated with a sudden growth in supply (that is
the number of children with autism), while unique to autism in
some respects, include aspects that are similar for other
disabilities and in policy challenges in other arenas. Especially
since the implementation of the Government Performance Results Act
of 1996, there is increased pressure to create public policy
infrastructures that are anchored by clearly cut categorical
service delivery. If the categories themselves leave significant
room for interpretation and their use actually has a shaping
effect on the target population, then it is important to
administration and policy evaluation to understand how the effect
is playing out. |
The increased prevalence of autism spectrum disorders has come
to the attention of a variety of public agencies over the past
decade (Bertrand et al, 2001; Croen et al., 2000). A dramatic rise
in incidence of a developmental disability over a short period of
time is a pervasive administrative and policy challenge in and of
itself. However, many circumstances surrounding the surge in
autism make an effective public response elusive at best. These
circumstances include: largely uncertain (and highly contested)
causality; politicized and somewhat polarized treatments; an era
of increasing reliance on community and civil rights based policy
responses to disability; and, for the time being at least,
oftentimes unknown prognosis (Feinberg and Vacca, 2000, p.
130).
Such circumstances beg not only for more attention toward
autism on government policy agendas, but also beg the question of
potential patterns in observable growth in autism in light of
varying public infrastructures. In order to best address the
recorded growth in autism, it is important to directly consider
the hypothesis that the recorded growth is a matter of observation
as much as it is a matter of proliferation. Furthermore, since
autism and its related disorders are viewed by most as a continuum
and, to some degree, a construction, the patterns of growth that
cannot be immediately explained by obvious environmental or
socio-economic factors are of particular interest to policymakers
(Bargerhuff, 2003; Simpson, 2003; Tinge, 2002).
There have been efforts to reshape the public policy
infrastructure directed at autism in recent years (Feinberg and
Vacca, 2000). For example, states such as Maryland and Indiana
have Medicaid waivers directed specifically at autism.
Nevertheless, as the children that are part of the autism baby
boom have yet to reach their teenage years, the work necessary for
effectively adapting public infrastructures to respond to the
autism challenge is far from complete. As terms such as
educational autism—which sometimes serve to differentiate
what a school considers to be autism from the opinions of the
medical community--suggest, the definition and nature of autism
itself has not even been solidified to the general satisfaction of
stakeholders. Understanding how the incidence of autism is being
differentially recorded across the states is helpful to the
management of public policy challenges associated with autism. It
is also among the first steps on a path toward understanding any
potentially bi-causal relationships between developing public
structures and the incidence of autism as recorded in the
educational environments of individual states.
In this article, a shift-share analysis of the incidence of
autism and related disorders (hereinafter referred to as autism)
reported as part of the Individuals with Disabilities Education
Act (IDEA) is used to gain insight into the nature of the growth
patterns of autism as it is being experienced in the public
education system. Since the public education system most
comprehensively touches the lives of children in the United States
and the observed incidence of autism is currently highest among
young children, the public administration of autism happens most
frequently within the public schools. A better understanding of
the growth pattern of the recorded incidence of autism is likely
to help in the development of public policy that more
appropriately addresses society’s challenges associated with
the incidence of autism.
Autism and the Public Context
The word “autism” was first coined in 1911 by Eugen
Bleuler, a Swiss psychiatrist (Williams 2000) who described it as
a temporary disorder related to schizophrenia. However, Leo
Kanner, who studied a group of children with what came to be known
as early onset autism, more (in)famously reinterpreted autism as
non-temporary disorder and emphasized a connection to mental
retardation and the need for institutionalization (Kanner, 1943).
One of the causes often assigned to autism in the early years was
the supposed “refrigerator mother” phenomena.
Essentially, it was believed that children with autism chose to
withdraw into an internal world because they were burdened with
emotionally shutdown or cruel—and almost invariably
WASP--mothers who did not show them enough or appropriate
affection to allow the children to develop normally. As a result,
a common early treatment for autism was therapy for the
child’s mother.
This understanding of autism, combined with the prevailing
tenor of disability policy in the middle part of the twentieth
century that encouraged separation of individuals with
disabilities from the general public (O’Briend, 2003;
Jongbloed, 2003), meant that autism fell from most policy agendas.
However, during the last part of the twentieth century, a dramatic
change in the general perception of autism began to take place.
Autism came to be understood as a complex disorder that was not
caused solely by external factors, at least not external factors
are simple as having a parent that did not provide enough love
(Stokstad, 2001, Rutter, 2000). The causality of autism is now an
open question toward which significant resources and research time
are currently being directed.
A crucial component of the modern reformation of the
understanding of autism was the reconsideration of the question
“What exactly is autism?” During the 1990s,
international efforts were made to specify the definition of
autism. For example, the definition from the Diagnostic and
Statistical Manual of Mental Disorders, Fourth Edition
(DSM-IV), published in 1994, includes criteria in three
categories: qualitative impairment in social interaction;
qualitative impairments in communication; and delays or abnormal
functioning in either social interaction language as used in
social communications or symbolic or imaginative play with onset
prior to age three. In order to be diagnosed as having autism, a
person must have a set number of characteristics in these
categories from a defined list of possible symptoms.
According to Dr. Deborah Hirtz of the National Institute of
Neurological Disorders and Stroke, autism is “a complex,
life-long, developmental disability that results in difficulty
with social interactions, problems in communication, and
restrictive or repetitive interests and behavioral challenges.
There is considerable variability in the severity of the symptoms,
and intellectual function can range from profound mental
retardation to above mean performance on IQ tests” (Hirtz
2000). However, the specifics of the definition of autism are
hotly debated. For example, the Michigan Department of Education
reported in 2002 that there was a lack of agreement on the
proposed definition of autism and that despite the fact that the
original criteria were retained, further development should be
anticipated (Michigan Department of Education 2002).
Public policy that provides for services on the basis of
diagnosis categories is, therefore, a difficult administrative
match for autism because the service needs of children with autism
spectrum disorder vary dramatically between children with autism
spectrum disorders and during the life of a child with an autism
spectrum disorder over time (Feinberg and Vacca, 2000). Neither
the exact needs nor the expected prognosis can be easily estimated
on a case-by-case basis given diagnosis. By the same token, since
establishing accurate prognosis for individual children is nearly
impossible and the most effective treatment highly debated
(presumably because the treatments have not well understood
differential effects on different individuals), public policy that
provides services on the basis of individual demands or rights can
be equally and uniquely difficult in managing the social
challenges associated with autism.
This type of challenge, while unique to
autism in some respects, includes aspects that are similar in
other disabilities and in policy challenges in other arenas.
Especially since the implementation of the Government Performance
Results Act of 1996, there is pressure to create public policy
infrastructures that are anchored by clearly cut categorical
service delivery. If the categories themselves leave significant
room for interpretation and their differential implementation has
a shaping effect on the target population, then the distribution
of incidence reflects variance in the (broadly cast) environment,
including public infrastructures. Examining this potential is
especially important in policy arenas that have complex fiscal
federalism structures, such as is found in Medicaid and the
provision of special education under the Individuals with
Disabilities Education Act or Section 504 of the Rehabilitation
Act.
Autism Incidence
The incidence of autism was once believed to be 1 to 2 per
10,000 people. More recently, reported incidence has climbed
drastically—to around 1 in 500 in most estimates (Mandell et
al., 2002). However, specific estimates of the prevalence of
autism as recorded in the research and reported to the public
vary. For example, in 2001, Bertrand et al, who studied the
prevalence of autism in Brick Township, New Jersey, reported,
“the prevalence of all autism spectrum disorders was 6.7
cases per 1000 children. The prevalence for children whose
condition met full diagnostic criteria for autistic disorder was
4.0 cases per 1000 children, and the prevalence for PDD-NOS and
Asperger disorder was 2.7 cases per 1000 children” (Bertrand
et al., 2001). However, in a study of the incidence of autism in
children born in California between 1987 and 1994, it was found
that “a total of 5038 children with full symptom autism were
identified from 4,590,333 California births, a prevalence of 11.0
per 10,000. During the study period, the prevalence increased from
5.8 to 14.9 per 10,000, for an absolute change of 9.1 per
10,000” (Croen 2002). Outside of the academic literature,
the range of reported and suspected incidence is even wider.
The causality of the rise of incidence in autism is highly
debated and politicized. Class action suits, such as the one
discussed on vaccineautism.com, are arising that seek to place
blame on particular events or practices such as mercury poisoning
in infants during childhood vaccination. The most widely accepted
explanations are of complex causes: “recent research reports
show that autism spectrum disorders may actually be more common
than previously believed. General awareness and clinical knowledge
of these disorders have increased, and the criteria in the ICD-10
and the DSM-IV are also now more detailed” (Kielinen 2000).
As this quote suggests, there are two core cause groups—a
better professional understanding of autism or changes in the
(broadly defined) environment.
Shattock et al. describe four basic reasons that autism might
have a perceived increase in recorded incidence independent of any
actual increase in the raw rate of autism disorders in children.
Their reasons include: “the increased awareness and skill in
diagnosis which has developed, the changing diagnostic criteria,
the lack of appropriate and available records and the increased
number of associated disorders which may formerly have been
included within the ‘autism’ diagnosis”
(Shattock, 2001). Even though these authors are writing from the
United Kingdom, these types of issues are expected to arise by
those who are professionally or personally connected to autism
related issues in the United States as well.
Suggestions abound for reasons related to autism’s rise
and—to the extent that it has been noted—variance
within this recorded rise across space. Searching the web for
“autism incidence” using a basic search engine brings
up in excess of 21,000 hits. Autism tends to be popularly
intriguing for reasons including the fact that autism is much more
common in boys, with an incidence rate that is 3 to 4 times the
rate found in girls (Hirtz 2000, Miles 2003), that it was once
blamed on traumatic experience or perverse behavior on the part of
parents, and the way in which autism is manifested, particularly
in the case of the so-called savants.
These elements of fascination, in combination with the position
stakeholders find themselves in trying to manage a specific case
of autism in an era of rapidly shifting ground make the nature of
the growth of autism a crucial concern for policy development and
administration. Three major policy and administrative challenges
are: identification, the distribution and selection of treatment
options, and the creation of appropriate policy and administrative
goals that will effectively address the autism baby boom in the
long term. The effective management of these factors could be
expected to be easier with information about how autism is being
differentially recorded across the country.
Many observers have hypothesized that the reasons for this
recorded rise might be expected to have as much to do with
changing service systems and increased awareness as with an
epidemiological growth in the general population (DeFrancesco,
2001; Barbaresi, W.J., 2002). To the extent that the rise in
recorded incidence is the result of a change in broad based
professional practice and in public awareness, if the rise in
incidence is not occurring quite similarly across the country (if
not world), then the structure of the public policy and
socioeconomic conditions of states, as a defining region policy
arenas such as health and education, could be having a shaping
effect on the growth of autism. Especially because individuals
with autism and their families are currently expected to require
very different services from a variety of public agencies over the
course of a lifetime than are individuals without autism, it is
important to consider coincidences and correlations between
observed patterns of growth in recorded incidence of autism and
socioeconomic and political factors.
There is no centralized place or database to which all cases of
autism are reported. However, the incidence of autism in children is arguably
recorded most comprehensively by the public education system under
the provisions of the Individuals with Disability Education Act.
Autism is one of thirteen categories in which children with
disabilities are currently entitled to special education services
under the Individuals with Disabilities Education Act. In
accordance with this key structural policy of special education in
the United States, states and regions are required to report the
number of children with autism (and in twelve other disability
categories) served to the Office of Special Education and
Rehabilitation Services on a yearly basis. Whereas the categories
are defined at the federal level, states, regions and, to a
certain extent, individual districts, hold the responsibility to
define exactly which children will be included in counts within
individual school systems (Feinberg and Vacca, 2000).
The rise in autism incidence is catching the attention of
public administrators—perhaps particularly those involved in
special education service planning and delivery. In fact, in the
data appendix of the 2001 Report to Congress on the implementation
of IDEA, it is explained, “twelve states commented that the
increases in counts of students with autism were a result of
better diagnosis and identification of the disorder, continued
reclassification of students, and improved training in methods and
assessment of autism” (OSERS 2002). The twelve states are
Alabama, California, Colorado, Connecticut, Georgia, Indiana,
Kansas, Kentucky, Minnesota, Missouri, Washington and Wisconsin.
In conducting the shift-share analysis a hypothesis was that these
states would be those with both the highest rise in incidence and
the highest relative growth as compared to the growth in incidence
experienced by these states in the other disability categories.
Method
Shift-share analysis is most commonly used in regional
economics. In that context, “shift-share analysis produces
results that can be valuable for diagnosing, describing and
building understanding of major differences between the industry
pattern of employment growth locally and nationwide trends”
(Washington State University 2002). In the context of the
incidence of autism as recorded in the public education system,
this technique can be used to build understanding of the
differences between the pattern of growth in autism as compared to
the other diagnosis categories locally and in nationwide trends.
In this article, a shift-share analysis of special education
diagnosis categories are conducted using data reported by the
Office of Special Education and Rehabilitative Services (OSERS) to
Congress.
For the special education shift-share analysis, the local
regions of interest are the states. The aggregate level is the
national level. For this article, a shift-share analysis of the
changes in diagnosis incidences in autism as recorded by public
schools between the 1995-1996 and 2001-2002 school years was
conducted. This period is of particular interest. The DSM-IV
standards were developed and released during this era and the dawn
of widespread public attention to the perceived rise in incidence
in autism dates to at least the late 1990s (though perhaps up to
ten years earlier in some locations).
Shift-share analysis is fundamentally a technique of arithmetic
decomposition. In regional economics the purpose of shift-share
analysis is to allow for comparison of differences in growth in
selected industries in smaller regions (such as states or
localities) with one another and with a larger, encompassing
region (such as the nation). Through relatively simple
calculations, shift-share analysis produces two measures of
interest, which are typically called competitive and mix
components in regional economics.
The arithmetic decomposition in shift share proceeds as
follows. First, in traditional shift-share analysis, employment
data is collected on a chosen number of industries for two time
periods of interest for both the encompassing region (hereinafter
referred to as national) and smaller regions of interest
(hereinafter referred to as locality). The percent increase in
total employment at the national level is first calculated. For
each locality, the national growth share is then calculated. This
is the increase that would have been expected in a given industry
if that industry had grown at the regional level at exactly the
same rate as the overall, national employment growth rate. That
is, a number for each industry is calculated using the following
formula:
National Growth Share = (Industry Employment,
year 1) * (overall growth rate).
Not surprisingly, industries in given regions very rarely grow
at exactly the growth rate observed on the national level. The
national growth share is not generally observed in practice. In
this article, the national growth share shows the increase that
would have been seen in a given diagnosis category at the state
level if that diagnosis category had grown in the state at the
same rate as disability in general in the United States.
In traditional shift-share analysis, the expected growth in
employment in a particular industry is calculated, using the
national growth rate in that industry. This component, which is
called the industrial mix component in traditional shift-share
analysis, is calculated as follows:
Industrial Mix = (Local industry employment,
year 1) * (National Industry Growth Rate).
In the context of disability as explored in this article, this
is the number of additional individuals with a particular
diagnosis one would expect to see at the state level if the
category had grown at the same rate as the overall national
rate.
Finally, the local share or competitive regional shift is
calculated. This is the measure of particular interest in most
shift-share analysis. In traditional shift-share analysis this
demonstrates the extent to which factors unique to the local area
have caused growth or decline in regional employment of an
industrial group. It is calculated as follows:
Regional Shift = (Local industry employment,
year 1) * (Percent local growth in industry- percent national
growth in industry)
In the context of disability as explored in this article, the
result of this equation is the number of individuals (or lack of
individuals if the number is negative) with a particular diagnosis
attributable to a growth pattern unique to the state. Once these
three calculations have been performed, the results are examined
for growth patterns within and between states.
In the context of the administration of challenges and services
associated with autism, the traditional shift-share language is
somewhat awkward. Competition for children with specific
disability types does not typically take place in state education
systems in the same spirit as competition for businesses and
industries takes place between regional economies. Nevertheless,
the shift-share components are potentially very useful indicators
in the diagnosis of growth patterns in educational autism because
they provide a way to compare across states and between diagnosis
categories. Therefore, in applying this technique to recorded
incidence of disability it is helpful to employ language that
better describes the measures of interest in the disability
context. In this article “diagnosis mix” is
hereinafter used instead of “industrial mix” to
describe the expected growth in individual diagnostic categories
and “state-specific label growth” is used in place of
local share or competitive regional shift.
As is described above, in shift-share analysis national (or
another larger, encompassing region) growth patterns are used as a
reference point (Hoover and Giarantti 2002). At the national
level, disability incidence on the whole is growing at any given
time at a certain rate, but it is to be expected that the rate of
incidence of individual diagnoses will be growing at different
rates. That is, mental retardation incidence is not expected to be
growing at exactly the same rate as deaf-blindness, for example.
The diagnosis mix component shows how categories would have grown
at the more local level if the growth pattern at the national
level held uniformly in the localities. In the context of regional
economics, a region is said to have a favorable growth mix if
economic activity in a region is growing quickly (or more quickly)
in a set of industries that are also growing quickly at the
national level (Hoover and Giarantti 2002). Though the story and
its implications are more complex when it comes to comparing the
mix of slow and fast growing diagnosis in a state vis-à-vis a
nation, the diagnostic mix component is, of course, of interest in
public administration since a state that has a diagnostic mix
pattern that is different from the national trend will face unique
administration challenges (and, perhaps, opportunities),
especially in an arena as flush with federalist tension as special
education.
In regional economics, the competitive regional share or local
share component is understood by “imagining a case of a
region that has exactly the same mix of activities, as does the
nation (and) its percentage share is the same for all
activities” (Hoover and Giarantti 2002). A competitive
advantage, in the context of regional economics, is found in
regions that increase their share, or, as Hoover and Giarantti
explained, “if most activities grow faster in the region
than in the nation.” In the context of special education,
and specifically autism, the state-specific label growth component
examines how the growth in autism in individual states compare
relative to each other and to the national growth rate.
Results
The increase in incidence of all disabilities as recorded
through the system of special education was just over 15% between
the 1995-1996 and 2001-2002 school years. During this period,
autism grew faster than that in all fifty states, the District of
Columbia and Puerto Rico. In fact, of the twelve disability
categories that were recorded in these years, autism was the
fastest growing disability category in 32 states. The range of
incidence growth rate during this period was between 53.7 percent
(Puerto Rico) and 1,413.37 percent (Ohio) with a mean growth of
306.80 percent and a standard of deviation of 218.70 percent. The
states’ ranked growths are shown in Table 1 below.
Table 1. Ranked Autism Percent Growth
Rates
| State |
Rank |
Rate |
State |
Rank |
Rate |
State |
Rank |
Rate |
| Alabama |
36 |
201 |
Louisiana |
48 |
104 |
Ohio |
1 |
1413 |
| Alaska |
19 |
321 |
Maine |
16 |
364 |
Oklahoma |
24 |
283 |
| Arizona |
21 |
313 |
Maryland |
15 |
365 |
Oregon |
50 |
64 |
| Arkansas |
25 |
279 |
Massachusetts |
12 |
377 |
Pennsylvania |
32 |
227 |
| California |
18 |
333 |
Michigan |
43 |
168 |
Puerto Rico |
51 |
54 |
| Colorado |
4 |
573 |
Minnesota |
10 |
392 |
Rhode Island |
7 |
419 |
| Connecticut |
27 |
268 |
Mississippi |
40 |
185 |
South Carolina |
6 |
438 |
| Delaware |
47 |
118 |
Missouri |
31 |
229 |
South Dakota |
26 |
279 |
| Florida |
35 |
211 |
Montana |
42 |
170 |
Tennessee |
45 |
137 |
| Georgia |
9 |
394 |
Nebraska |
23 |
288 |
Texas |
38 |
193 |
| Hawaii |
17 |
352 |
Nevada |
5 |
517 |
Utah |
20 |
318 |
| Idaho |
30 |
233 |
N. Hampshire |
2 |
936 |
Vermont |
14 |
368 |
| Illinois |
11 |
379 |
New Jersey |
28 |
268 |
Virginia |
41 |
182 |
| Indiana |
29 |
250 |
New Mexico |
37 |
194 |
Washington |
3 |
650 |
| Iowa |
49 |
76 |
New York |
46 |
126 |
West Virginia |
39 |
188 |
| Kansas |
34 |
214 |
North Carolina |
44 |
151 |
Wisconsin |
8 |
397 |
| Kentucky |
13 |
373 |
North Dakota |
33 |
220 |
Wyoming |
22 |
303 |
As Table 1 demonstrates, the recorded growth of autism around
the nation was far from uniform across the country. As is
mentioned above, from a public administration and policy
standpoint, in the context of marble cake federalist special
education policy, to the extent the policies and administrative
react to prevailing growth rates, states with outlying growth
rates may have administrative and other policy related challenges.
When a 95% confidence interval is drawn around the mean growth
rate, the states that are found Ohio and New Hampshire are found
to have statistically significantly higher growth rates. No states
had statistically significantly lower growth rates (a state would
have had to have a decrease in the reported incidence of autism
for this to be the case).
The national growth rate for autism during this period was
almost 240%. It is not surprising that this national growth rate
is different from the mean growth rate since the populations of
states vary dramatically and, therefore, the change in growth in a
small state will have much less effect on the overall change in
growth than will a similar (or even smaller) change in a large
state. It is interesting to note, however, that the national
growth rate is less than one standard deviation away from the mean
state growth rate.
States that self identified in their 2002 reports that
increased incidence was due to better identification might be
expected to have both a reported incidence rate that was
relatively high when compared to other states and to have the
highest growth of all disability categories have taken place in
autism. However, neither of the outlying states—Ohio and New
Hampshire—were in this group. When the percentage growth
rates were examined, the ranks from highest growth to lowest of
the twelve states that self-identified as improving their
diagnoses mechanisms were: Alabama (36th); California
(18th); Colorado (4th); Connecticut
(27th); Georgia (9th); Indiana
(29th); Kansas (34th); Kentucky
(13th); Minnesota (10th); Missouri
(31st); Washington (3rd); and Wisconsin
(8th). As can be seen from these ranks, the states that
self-identified as more aggressively diagnosing autism were almost
as likely to be in the bottom half of the ranked growth rates and
in the top half. Furthermore, autism was the highest growth
category in only six (50%) of these self-identifying states (less
than the 62% of all states or regions). This evidence does not
support the hypothesis that exceeding rapid growth rate is caused
by institutionalized overenthusiastic discovery of new cases of
autism.
The diagnosis mix and state-specific label growth for each of
the states and regions is shown in Table 2. As is described above,
the numbers generated in the shift share analysis refer to the
number of cases of autism. Diagnosis mix refers to the number of
additional cases of autism one would expect in the school
system’s population if the state’s growth in autism
had exactly matched the national growth rate in autism. A larger
number, therefore, means that the state had a larger population of
children with autism in the mid-1990s. The state-specific label
growth refers to the number of cases of autism above or below what
would have been expected as an observed growth in autism once the
overall growth in autism at the national level is controlled for.
The state-specific label growth reports the absolute increase (or
decrease) in the number of recorded cases of autism once the
growth attributed to the national growth in autism has been
controlled for. States that did not grow at at least the national
rate would have a negative state-specific label growth. In other
words, for example, a state with a very negative number has much
less autism than would be expected given the number of cases they
began with and the growth experienced in autism nation wide.
Table 2. Diagnosis Mix (DM) and
State-Specific Label Growth (SSLG)
| State |
DM |
SSLG |
State |
DM |
SSLG |
State |
DM |
SSLG |
| Alabama |
672 |
-115 |
Louisiana |
1427 |
-866 |
Ohio |
453 |
2371 |
| Alaska |
119 |
43 |
Maine |
267 |
148 |
Oklahoma |
459 |
89 |
| Arizona |
730 |
241 |
Maryland |
1154 |
647 |
Oregon |
3887 |
-3045 |
| Arkansas |
457 |
81 |
Massachusetts |
1259 |
772 |
Pennsylvania |
2722 |
-157 |
| California |
6865 |
2852 |
Michigan |
3948 |
-1264 |
Puerto Rico |
755 |
-626 |
| Colorado |
179 |
266 |
Minnesota |
1488 |
1015 |
Rhode Island |
166 |
133 |
| Connecticut |
894 |
115 |
Mississippi |
363 |
-89 |
S. Carolina |
421 |
374 |
| Delaware |
302 |
-164 |
Missouri |
1331 |
-64 |
S. Dakota |
148 |
26 |
| Florida |
3121 |
-403 |
Montana |
164 |
-51 |
Tennessee |
1042 |
-476 |
| Georgia |
1116 |
771 |
Nebraska |
240 |
52 |
Texas |
5424 |
-1123 |
| Hawaii |
188 |
95 |
Nevada |
188 |
233 |
Utah |
388 |
136 |
| Idaho |
240 |
-7 |
N. Hampshire |
87 |
272 |
Vermont |
119 |
68 |
| Illinois |
1777 |
1109 |
New Jersey |
2149 |
269 |
Virginia |
1878 |
-481 |
| Indiana |
2088 |
97 |
New Mexico |
202 |
-41 |
Washington |
587 |
1079 |
| Iowa |
706 |
-516 |
New York |
6975 |
-3549 |
W. Virginia |
291 |
-67 |
| Kansas |
531 |
-62 |
N. Carolina |
2765 |
-1096 |
Wisconsin |
1013 |
712 |
| Kentucky |
484 |
288 |
North Dakota |
101 |
-9 |
Wyoming |
65 |
19 |
The numbers presented in Table 2 are not controlled for the
population size of the state or for the general population growth
experienced by the state during the time frame. It is arguable,
then, that except for very small deviations which may not be
(statistically) significantly different from zero, that from a
standpoint of public policy and public management, the most
telling aspect of the state specific label growth is whether the
number generated by the shift-share analysis is positive or
negative.
The examination of state-specific label growth (SSLG) and of
emergent patterns of growth demonstrated therein, also sheds light
on the growth pattern of autism. The median SSLG was 52 recorded
cases (with the mean, as would be expected, indistinguishable from
zero). The largest SSLG was in California, which had 2852 cases.
The four other states that had SSLGs in the top five were: Ohio
(2371 cases); Illinois (1109 cases); Washington (1079 cases); and
Minnesota (1015 cases). Autism grew the fastest of all the
disability categories in each of these states except
Washington.
The lowest SSLG was in New York, which had an SSLG of
–3549 cases. The four other states that had local shares in
the bottom five were: Oregon (-3045 cases); Michigan (-1265
cases); Texas (-1123 cases) and North Carolina (-1096 cases). As
can be seen from Table 1, in some cases these absolute growths
were also ranked in the same categories in terms of percentage
growth. Of these states, autism grew the fastest of all the
disability categories in Michigan, Texas and North Carolina. Other
states, however, such as California, Texas and New York had
changes that were large due to the size of the states population
rather than the size of the percentage change.
In addition to states with outlying growth rates mentioned
above, states or regions with growth rates closest to the mean,
that is the most average states, are interesting from a standpoint
of public policy and management. The states whose growth most
closely matched the states’ mean growth in autism included
Idaho, Indiana, Missouri, North Dakota and Pennsylvania. The
growth rate in each of these states was within 20 percentage
points of the national mean. Autism was the fastest growing
disability category in only two of these states—Idaho and
North Dakota.
When the growths of other disability categories in these states
are examined, we find that the state-specific label growth
patterns on the whole is not overly similar across these states.
We also find that the states were not average across all
disability categories. For example, Pennsylvania had a very high
state specific label growth in specific learning disabilities,
whereas Missouri had a quite low (and negative) SSLG in the same
category. This is interesting especially because learning
disabilities are another type of disability sometimes regarded as
potentially trendy. Also, North Dakota reported no children in the
multiple disabilities category, whereas Pennsylvania had a large
SSLG in that category as well. Given autism’s historical
connection with mental retardation, it is worth noting that all
but one of the average growth states (Indiana) had a negative
state-specific label growth in autism. Finally, since autism and
speech and language impairments are sometimes confounded or
combined, one might expect that states that are average in autism
would be similarly average in speech or language impairments.
However, as Table 3 shows, only North Dakota was close to average
in that category.
Table 3. State-Specific Label Growths in
States with Average Autism Growth
| |
Idaho |
Indiana |
Missouri |
North Dakota |
Pennsylvania |
| Specific Learning Disabilities |
413 |
3562 |
-1441 |
-793 |
15899 |
| Speech or Language Impairment |
749 |
-993 |
2027 |
71 |
-5795 |
| Mental Retardation |
-1078 |
883 |
-620 |
-122 |
-564 |
| Emotional Disturbance |
323 |
3464 |
-1364 |
345 |
1978 |
| Multiple Disabilities |
-41 |
190 |
43 |
N/A |
588 |
| Hearing Impairments |
-34 |
238 |
22 |
20 |
-346 |
| Orthopedic Impairments |
-46 |
258 |
-225 |
-10 |
-298 |
| Other Health Impairments |
-195 |
1611 |
2493 |
114 |
2084 |
| Visual Impairments |
23 |
60 |
85 |
6 |
-207 |
| Autism |
-7 |
97 |
-64 |
-9 |
-158 |
| Deaf-Blindness |
2 |
-47 |
-51 |
-50 |
21 |
| Traumatic Brain Injury |
-88 |
-147 |
-172 |
-17 |
-1612 |
Conclusion
State environment—including perhaps public
infrastructure--seems clearly to have had a role in the shaping of
the autism baby boom in the United States. Presumably in the
recording of any phenomena by agencies in the public
infrastructure (such as the system of formal education) there will
be always be some variation in growth rates. The range of growth
rates recorded by the public education systems as measured by
shift-share analysis is too large to be explained away through
pure chance or variation in a phenomenon of the physical
environment that remains unnoticed. Differences in the
implementation patterns of special education policy between states
are a far more likely causal element. As can be seen by the range
of growth rates, to the extent that identification by the school
district can be connected with appropriate educational (and, to
some degree other) services, the willingness on the part of states
to provide services for children with autism is perhaps remarkably
different in different states.
Furthermore, as can be seen from the shift-share results,
autism’s growth pattern as measured by the system of formal
education does not appear to be spatially correlated. Whereas the
states that are growing at close to the national rate are
basically midwestern, the states that grew most quickly or most
slowly have no such proximity. Neither can population or state
wealth explain the distribution of growth as measured in the shift
share analysis. After all whereas California experienced the
largest share of growth in educational autism during this era, New
York’s share indicated that educational autism grew much
less than was expected.
This lack of a preeminent environmental or regional causality
suggests that there is a relationship between the recorded growth
of autism and the public infrastructure. In his description of
shift-share analysis, Martin Sheilds states that among the many
questions to consider in the interpretation of results from
shift-share analysis, two of the most important are:
- Compared to other regions, does the community seem highly
competitive in any particular industry? What is the source of
this competitiveness?
- Does this information support popular perceptions? Or, does
the analysis uncover “surprising” areas of economic
growth? (Located online at: http://radburn.rutgers
.edu/lahr/509/)
This study has thus far focused on the first of these
questions, looking at a single diagnosis (the
“industry” for our purposes). As is mentioned above
there are several states that appear to be highly competitive when
it comes to the recording of autism in their educational system
and part of the source of this competitiveness is most likely
connected to the public infrastructure (but not to a reported
enthusiasm for diagnosing autism cases). As far as the second
question is concerned, the information leads several surprises,
both from the standpoint of growth in autism specifically and in
the way in which the development and administration of policy is
more generally understood. First of all, the popular perception is
that autism is growing very rapidly, but presumably relatively
evenly nationwide. Furthermore, the most oft discussed clustering
of autism is in Brick Township, New Jersey. New Jersey did not
rank among the top growths of states. A nationwide surge in
incidence is a much less complex (and arguably less troublesome)
occurrence than a surge with a magnitude that varies dramatically
from state to state.
From the standpoint of public policy and administration, these
findings call for a sustained look at the relationship between the
unfolding of social policy problems and the public infrastructure.
Shift-share analysis, after all, provides only a two-period
snapshot of growth that is continuous in nature. To the extent
that this variance in growth is due to street and state level
bureaucracy, public policy has a level of responsibility for the
shaping of the public challenge. Especially when this challenge is
so intimately connected to the development of children and to the
unfolding of the new conception of civil rights being forged
through modern disability policy, there should be more direct
examination of behavior within public infrastructures that
accounts for wide differences in observation and in understandings
of a federally defined public mission.
References
Barbaresi, W.J. et al. (2002). Explaining the
Apparent Increase in the Incidence of Autism in Olmsted County,
Minnesota, from 1976 to 1997. Journal of Developmental &
Behavioral Pediatrics. 23(5), p. 398-400.
Bargerhuff, Mary Ellen. (2003). Exploring the
Spectrum of Autism and Pervasive Developmental Disorders (Book
Review). Remedial and Special Education. 24(4), p.
255-256.
Bertrand, J., A. Mars, C. Boyle, F. Bove, M.
Yeargin-Allsopp and P. Decoufle. (2001). Prevalence of Autism in
a United States Population: the Brick Township, New Jersey,
Investigation. Pediatrics 108(5): 1155-61.
Bureau of Economic Analysis, U.S. Department of
Congress. (2002). Regional Accounts Data, Local Area Personal
Income. Located online at: http://w
ww.bea.doc.gov/bea/regional/reis/drill.cfm. Accessed on
September 17, 2002.
Braddock, David (editor). (2002). Disability at
the Dawn of the 21st Century and the State of the
States. Washington, D.C.: American Association of Mental
Retardation.
Croen, Lisa A., Judith K. Grether, Jenny
Hoogstrate and Steve Selvin. (2002). The Changing Prevalence of
Autism in California. Journal of Autism and Developmental
Disorders 32(3): 207-215.
DeFrancesco, Laura. (2001). Scientists Question
Rise in Autism. Nature Medicine. 7(6), 1.
DeFrancesco, Laura. (1994). Diagnostic and
Statistical Manual of Mental Disorders, Fourth Edition.
DeFrancesco, Laura. Exploring Autism. (2002).
Located online at: http://www.exploringautism.
org. Accessed on September 3, 2002.
Feinberg, Edward and John Vacca (2000). The
Drama and Trauma of Creating Policies on Autism: Critical Issues
to Consider in the New Millennium. Focus on Autism and Other
Developmental Disabilities. 15(3), p. 130.
Hirtz, Deborah. (2000). Testimony on ‘The
Challenges of Autism—Why the Increased Rates?’
Testimony before the House Committee on Government Reform. April
6, 2000. Located online at: http://www.hhs
.gov/asl/testify/t000406b.html. Accessed on September 10,
2002.
Hoover, Edgar M. and Frank Giarratani. (2002).
An Introduction to Regional Economics. West Virginia University:
Regional Research Institute. Located online at: http://www.rri.wv
u.edu/WebBook/Giarratani. Accessed on September 4, 2002.
Jongbloed, Lyn. (2003). Disability Policy in
Canada: An Overview. Journal of Disability Policy
Studies. 13(4), p. 203-210.
Kanner, Leo. (1943). Autistic Disturbances of
Effective Contact. Nervous Child. 2, p. 217-250.
Kielinen, M., SL Linna and I Moilanen. (2000).
Autism in Northern Finland. European Child and Adolescent
Psychiatry 9(3): 162-7.
Mandell, David S., et al. (2002). Race
Differences in Age at Diagnosis among Medicaid-Eligible Children
with Autism. Journal of the American Academy of Child and
Adolescent Psychiatry. 41(2), p. 1447-1454.
Michigan Department of Education, Office of
Special Education and Early Intervention Services. (2002).
Summary of Revised Rules for Special Education. Located online
at: http://www.michigan.gov/documents/MD
EP6_Rules Presentation SpeakingNotes_15687_7.pdf. Accessed on
September 11, 2002.
O’Brien, Ruth. (2003). From a
Doctor’s to a Judge’s Gaze: Epistemic Communities and
the History of Disability Rights Policy in the Workplace.
Polity. 35(3), p. 325-347.
Rimland, Bernard. (2002). The Autism Explosion.
Autism Research Institute. Located online at: http://
www.autism.com/ari/editorials/explosion.html. Accessed on
September 21, 2002.
Rutter, Michael. (2000). Genetic Studies of
Autism: From the 1970s into the Millennium. Journal of Abnormal
Child Psychology. 28(il): p. 3.
Shattock, Paul, Paul Whiteley, James Roger and
Lynda Todd. (2001). Incidence Rates in Autism: A Brief
Overview. Presented at Durham Conference.
Shields, Martin. (2003). Tool 4. Shift-share
Analysis Helps Identify Local Growth Engines. Located online at:
http://radburn.rutgers
.edu/lahr/509/ Shields_Shift_Share.PDF. Accessed on
February 3, 2003.
Simpson, David. (2003). Autism Spectrum Disorder
is Not as Certain as Implied. British Medical Journal.
326(7396), p. 986.
Stokstad, Erik. (2001). New Hints into the
Biological Basis of Autism. Science. 294(5540), p. 34.
Tinge, Bruce. (2002). Autism, Autistic Spectrum
and the Need for Better Definition. The Medical Journal of
Australia. 176(9), p. 412-414.
United States Census Bureau. (2000).
Statistical Abstract of the United States (selected tables).
Located online at: http://www.censu
s.gov/staab/www/ranks.html. Accessed on September 17,
2002.
Washington State University. (2002). Shift-share
Analysis of Employment Growth, 1969-2000. Located online at: http://niip.wsu
.edu/washington/ssharewa.htm. Accessed on September 4,
2002.
Williams, Robert Jr. (2000). Autism Through the
Ages Baffles Science. Located online at: http://ww
w.pediatricservices.com/prof/prof-26.htm. Accessed on
September 21, 2002.
About the Author
Dana Lee Baker
Assistant Professor
Harry S Truman School of Public Affairs
118 Middlebush Hall
University of Missouri-Columbia
Columbia, MO 65203
Email: bakerdan@missouri.edu
(573) 882-0363
Baker’s primary research interests are in disability and
children’s policy with a comparative focus. She is also
fascinated by the study of public policy particularly disability
policy and agenda setting. She earned her bachelor’s degree in
History and Religious Studies at Rice University and her Masters
of Public Policy at the University of Southern California. While
working on her Masters degree, Baker was a caseworker for the
Alliance for Children’s Rights in Los Angeles and interned with
the Los Angeles chapter of Physicians for Social Responsibility.
Baker did her Ph.D. work at the Lyndon Baines Johnson School of
Public Affairs at the University of Texas at Austin.
|
