Before ridiculing the reporting too much (especially the MarketWatch rendering of it), it's helpful to read it in more context (p.10) even in the the Griffiths et al. paper. Even though in the intro on p.6 Griffiths et al. only make that singular quip that was quoted in the question), they later say (p. 10):
The National Autism Society (2016) estimated the
unemployment rate to be as high as 85%, while a study of 200 transition-age young
adults with ASD found that 81% were unemployed (Gerhardt & Lanier, 2011). A
small study of young adults with ASD and IQs above 50 found that only 11.76% were
employed (Howlin et al., 2004). One study showed that approximately half of young
adults with ASD worked for pay after high school (Roux et al., 2013). The same study
also found that the odds of ever having a paid job were higher for those who were older,
from higher income households, or who had better conversational or functional skills
(Roux et al., 2013).
What might account for the differences in unemployment rates reported for adults
with ASD? One explanation is that the variance may be due to differences in the
populations sampled. Another explanation is that both full- and part-time employment
are often included when calculating unemployment rates. In addition, individuals with
ASD are often given fewer hours of work than they would like. One study found that
74% of young adults with ASD who worked were doing so only on a part-time basis
(Gerhardt & Lanier, 2011). Baldwin et al. (2014) found that adults with Asperger’s
Syndrome, as compared to the general population, were more likely to work part-time.
Finally, adults with ASD tend to be underpaid compared to their peers without ASD
(Ballaban-Gil et al., 1996; Howlin et al., 2004; Roux et al., 2013).
It also is important to take into account the type of work included in the definition
of employment. One study found that 56% of the individuals considered employed
were working in day programs or sheltered workshops (Taylor & Seltzer, 2011).
Howlin et al. (2004) found that 12% of individuals with ASD worked in supported,
sheltered, or volunteer employment and that 62.5% were in some type of day program.
Research also has shown that young adults with ASD are more likely to be overeducated
and overqualified for their jobs, meaning that the work that they do is beneath
their capabilities (Baldwin et al., 2014). Further, they work in a limited number of
occupations (Roux et al., 2013). Finally, they are more likely to be underemployed
(Baldwin et al., 2014; Krieger et al., 2012; Nord et al., 2016; Paul et al., 2016; Richards,
2012; Roux et al., 2013; Scott et al., 2015; Shattuck et al., 2011).
Adults with ASD also face significant challenges in maintaining employment
(Baldwin et al., 2014; Lorenz & Heinitz, 2014; Richards, 2012; Roux et al., 2013).
When they experience conflict or stress at work, adults with ASD may quit or miss
work without prior notice (Richards, 2012). They also are more likely than their peers
without ASD to change jobs frequently and, as a result, to experience higher levels of
ongoing stress and financial concerns (Baldwin et al., 2014).
I.e. it's not a singular finding in that (high) range and it does indeed depend on what definition one adopts for both autism and employment. In general, as it's evident from the literature surveyed above, the studies involved small samples, which do raise some questions about generalization.
None of these speak of "85% of college grads affected by autism" as Market Watch has it (emphasis mine), so that (rendering) is clearly outlandish--if only am I to address the title question.
Now if want to compare those findings with those form the paper linked by the Special Olympics page... (Siperstein et al., 2013) you also have to keep in mind that they use more proper economics (BLS) terms, but they also discuss related works...
Although the unemployment rate is a well-known
and popular way to describe the employment landscape,
other statistics should be considered in conjunction to
fully understand the employment situation of adults
with disabilities. The unemployment rate only takes
into account those individuals who are in the labor
force (i.e., either working or out of work yet seeking
employment). For individuals with disabilities it is also
important to consider the employment rate, as this figure
takes into account all working-aged individuals, regardless of whether they are in or out of the labor force.
During the period of 2008– 2010 the estimated employment rate of adults with disabilities was extremely low, ranging from 39% to 34% compared with the much
higher rates of 79% and 76% for individuals without a
disability (Erickson, Lee, & von Schrader, 2010, 2011, 2012) [...]
It is difficult to derive
national estimates as to the employment rate of adults
with ID [intellectual disabilities], given the variation in the way disability is
defined (e.g. Taylor, Krane, & Orkis, 2010; Yamaki
& Fujiura, 2002); however, the American Community
Survey (ACS) includes the category of cognitive disability which provides a close approximation of the
employment rate. According to the ACS, presently, less
than one out of four adults with cognitive disabilities
are employed. In fact, estimates of the employment rate among these adults suggest that it declined slightly
over the past 5 years; in 2008 it was reported that
28% of adults with cognitive disabilities were employed
(Erickson et al., 2010), compared with the most recent
estimate of 23% (Erickson et al., 2012). Furthermore,
these numbers do not address the issue of underemployment. That is, when adults with ID are employed,
they are most often employed part time and are paid
a lower wage than their fellow workers without disabilities (Butterworth et al., 2012; U.S. Dept. of Labor,
2013).
Siperstein et al. (2013) themselves used BLS defs for their own study, but also some unofficial sub-categorizations :
The labor department defines employment as a
working-age adult currently holding a job, regardless of
the pay, hours, or type of work. [...]
The employment rate represents the proportion of
working-age people who are currently employed. The
employment rate of adults with ID aged 21–64 in the
present sample was 34%. [...] Of great import in
describing the employment of adults with ID is the setting.
Of the adults with ID in the sample, 18% were
competitively employed and 13% were employed in a
sheltered setting.
(BLS themselves probably has no def of "competitively employed". Also 18+13=31; another 3% were employed in an "other setting", according to a later table.)
Assuming those papers on ASD used not-in-the-workforce as their definition of "unemployment" (outside of pure economics papers, the confusion is made often enough) there isn't that much of gap between the findings on ASD (80-85%) and ID (75%). Actually, if you look carefully at the 1st quote Howlin et al., 2004 measured employment, not unemployment.
Gerhardt and Lainer (2011) (which is itself a review) actually have a decent summary of Howlin's paper:
For example, Howlin et al. (2004)
surveyed 68 adults (average age of 29 years) with autism
and a performance IQ of above 50 and found a majority
(58%) were rated as having poor or very poor outcomes.
Individuals with a performance IQ of at least 70 had significantly
better outcomes than those with a performance
IQ below 70. Within the normal IQ range outcome was
very variable and, on a case by case basis, neither verbal
nor performance IQ proved to be a consistent indicator of
positive outcome. With regards to employment status the
authors found that only 8 of the 68 individuals in the
sample were competitively employed; 1 was self employed
as a fabric printer; 14 worked in supported, sheltered or
volunteer employment; and 42 had ‘‘programs’’ or chores
through their residential provider. Additionally, more than
half (51%) were reported as having no friends.
Clearly these are not BLS defintions being used. (Also note 8/68 = 11.76%)
And actually Gerhardt and Lainer did not themselves conduct that other study which found 81% "unemployment", but rather the quote another study on employment.
In the United States, the University of Miami/Nova
Southeastern University CARD (2008) conducted an online
survey of approximately 200 families of transition-age
and adult constituents with ASD in southern Florida. [...]
only 19% of individuals with autism were
employed at the time of the survey with 74% of those
employed working less than 20 h per week; 85% still lived
with parents, siblings, or older relatives. Overall, outcomes
were generally regarded as being poor and access to necessary
and appropriate services severely limited.
If this needs saying again, using BLS standards, if those 81% who were not employed were also not looking for work, unemployment among them is 0%. Also this study didn't seem to have tried to define "employment" more carefully.
As much as another answer here lampoons the National Autistic Society for their methodology and "political motivation", I do not see a great discrepancy between what they report and other research (which is also of similar quality in terms of methodology). What NAS actually say:
We surveyed over 2,000 autistic adults, or people responding on their behalf, to ask them about their experiences of finding and keeping a job.
only 16% are in full-time paid work. Only 32% are in some kind of paid work (full and part-time combined), compared to 47% of disabled people and 80% of non-disabled people*
over three quarters (77%) who are unemployed say they want to work
four in ten say they've never worked.
[*] Both of these stats are from the Office for National Statistics (2016) Dataset: A08: Labour market status of disabled people (20 July 2016).
So there is actually a problem however in how NAS were cited in Griffiths et al. Perhaps what happened is that Griffiths et al. mis-interpreting that 16% and ignored the larger 32% which includes part-time work. (Or maybe NAS updated their page in the meantime, I don't know...) On the other hand, Griffiths et al. do pick from the other studies (e.g. from Howlin's) only the "competitively employed" and report only those as "employed". So maybe Griffiths et al. made a deliberate choice that they applied uniformly across the studies they looked at, i.e. to pick/report only the full-time/competitive employment as "employment"... Alas they were not explicit in this choice (and they also used "unemployment" simply as the complement of employment, even though that is a "no no" for an economist; in an economist's world/words "lack of occupation and willingness to work are not sufficient to classify people as unemployed".)
