Vous êtes sur la page 1sur 23

731810

research-article2017
AEQXXX10.1177/0741713617731810Adult Education QuarterlyPatterson

Article
Adult Education Quarterly
2018, Vol. 68(1) 41­–62
The Forgotten 90%: © The Author(s) 2017
Reprints and permissions:
Adult Nonparticipation sagepub.com/journalsPermissions.nav
DOI: 10.1177/0741713617731810
https://doi.org/10.1177/0741713617731810
in Education journals.sagepub.com/home/aeq

Margaret Becker Patterson1

Abstract
Despite a highly developed U.S. adult education system, 90% of adults aged 20 years
and older considered the least educated did not participate recently in formal or
nonformal education. What are nonparticipants’ characteristics, learning backgrounds,
and skill levels? What predicts their likelihood of not participating in recent formal or
nonformal education? The author analyzed 2012/2014 Program for the International
Assessment of Adult Competencies–USA data. Situational deterrents of increasing
age, parental education, low income, and work and family responsibilities contribute
to nonparticipation. Dispositional deterrents include health and disability challenges,
low social trust, and difficulties relating new ideas to real life. Institutional deterrents
are education costs and little work schedule flexibility. Supports reported by
nonparticipants are liking to learn new things, use of computers, and getting
information from television and people they trust. Results from Program for the
International Assessment of Adult Competencies–USA analyses inform adult and
postsecondary educators and policy makers on what happened to—and how to
reach—the forgotten 90%.

Keywords
adult education, nonparticipation, recruitment, retention, postsecondary education,
PIAAC, deterrents, skill levels, health, disabilities

Introduction
In the wake of the 2008 U.S. recession, the need for adults to be prepared for family-
sustaining careers is acute (Reder, 2010). Carnevale, Smith, and Strohl (2013) project
that two thirds of 54.8 million jobs the U.S. economy creates by 2020 will require

1Research Allies for Lifelong Learning, Vienna, VA, USA

Corresponding Author:
Margaret Patterson, Research Allies for Lifelong Learning, 2710 Chanbourne Way, Vienna, VA 22181,
USA.
Email: margaret@researchallies.org
42 Adult Education Quarterly 68(1)

workers with some postsecondary education (PSE). To gain skills for these new jobs,
career-ready adults can no longer end initial education with a secondary credential or
less (“the least educated”; Zhang, Guison-Dowdy, Patterson, & Song, 2011).
The least educated include high school (HS) graduates with low skills or disabili-
ties, HS early leavers, and immigrants (Patterson & Paulson, 2016). Nearly 29 million
adults did not complete HS, according to the 2012 Program for the International
Assessment of Adult Competencies–USA’s (PIAAC-USA) Survey of Adult Skills
(Patterson & Paulson, 2016). Of those aged 25 to 65 years (i.e., beyond traditional PSE
entry age), 10% participated in formal education in the year before PIAAC.
Additionally, approximately one million annually leave HS early (“Diplomas Count
2013,” 2013) and an estimated seven million cannot read English well or at all
(Patterson & Paulson, 2014).
Logical questions educators and policy makers might ask are as follows: What
about the other 90%? What are nonparticipants’ backgrounds and which deterrents do
they face—and what might trigger them to reengage in education? What are their skill
levels? How do a propensity to graduate from HS and deterrent covariates predict their
likelihood of not participating in recent formal or nonformal education? The author
analyzes 2012/2014 PIAAC-USA data to address these questions.

Conceptual Framework and Literature Review


Researchers know little about the forgotten 90%, since nonparticipants are seldom
included in studies of PSE barriers (Quigley, 2006). Skill level is a factor; 2012
PIAAC-USA data indicate significantly large skill gaps between nonparticipants and
participants, and gaps are widest for least educated adults (Patterson & Paulson, 2014).
In what ways do skill levels differ? In a review of national assessments of adult
skills from 1985 through 2003, Smith (2009) pointed not only to stagnant literacy
skills across time but also to skill differences by education attainment and age. Smith
attributed stagnant skill levels in part to an aging population and increases in non-
English speaker immigration. More recently, initial 2012 PIAAC-USA findings indi-
cate substantial differences by education attainment and age (Organization for
Economic Cooperation and Development [OECD], 2013; Patterson & Paulson, 2014).
Increasing numbers at nontraditional ages, 25 years and older (Ryu, 2010) are in PSE.
Least educated older adults pursuing PSE tend to persist at higher rates than young
adults, even though their PSE graduation rates are minimal (Zhang et al., 2011). In
2012, PIAAC-USA, least educated nonparticipants are middle-aged, on average, with
the majority having dependents. The least educated sometimes face deterrents associ-
ated with visual or hearing difficulties and tend to report a high rate of learning dis-
abilities (Patterson & Paulson, 2014).
Another disparity is by income. Autor (2014) notes an earnings “inequality”
between U.S. postsecondary and HS graduates has more than doubled in 30 years.
This inequality leads to “literacy classism,” in which least educated adults having the
most “need to know” tend to be marginalized (Quigley, 2017). Those in poverty have
the least access to learning (Patterson & Paulson, 2014).
Patterson 43

Figure 1.  Conceptual model.

A conceptual model for nonparticipation is displayed in Figure 1. The model begins


with two pathways from initial education to a high-skilled U.S. workforce. In the first
pathway, adult learners participate in education or workplace training; in the second,
they become nonparticipants.
Often, adults have an incomplete initial education, in that they left school early
(Zhang et al., 2011). Others completed initial education, yet find themselves in need of
basic skills, English language skills, or HS equivalency credentials that permit further
education. Approximately 10% with the need take the narrow path to adult education.
The forgotten 90% find themselves along a broader path facing numerous potential
deterrents. Factors in adult nonparticipation can be divided into three clusters of deter-
rents: situational, institutional, and dispositional (Quigley, 2006). Situational deter-
rents result from circumstances as adults balance multiple roles in their lives (Reder,
1999; Ross-Gordon, 2011) or deal with health conditions or disabilities (Patterson,
2014). Needing child care, for example, can deter participation (McAnnaney, 2009;
Patterson, 2014). Institutional deterrents occur when educational or employment pro-
cedures, policies, or practices prevent participation. Examples include lack of infor-
mation, geographic inaccessibility, inconvenient course times, and prohibitive tuition
rates. Dispositional deterrents refer to barriers involving learners’ self-perceptions and
attitudes. Examples include low confidence, negative past schooling experiences, or
fear of math (Quigley, 1997; Zhang et al., 2011).
44 Adult Education Quarterly 68(1)

Employment can also contribute deterrents (Bergson-Shilcock, 2017; Patterson,


2013). Jobs lost in the 2008 recession affected least educated most and were frequently
replaced with jobs requiring PSE in fields like health care (Carnevale et al., 2013).
Questions arose from 2012 PIAAC-USA findings as to whether not recognizing a con-
nection between learning and career prospects deters participation (Kis & Field, 2013),
or even contributes to a “vicious cycle” of minimal learning and fewer career opportu-
nities (OECD, 2013; Patterson & Paulson, 2014).
Institutional barriers form a second cluster of deterrents. Postsecondary participation
may never happen if adults cannot navigate enrollment, program selection, or financial
aid processes (Fike & Fike, 2008). Potential participants, such as first-generation or
immigrant learners, may cope with cultural stereotypes, immigration problems, and
language barriers (Bergson-Shilcock, 2017; Spellman, 2007). Schleicher (2013) noted
limited employer support for learning—either with PSE costs or with flexible work
schedules. Furthermore, employer support is sorely lacking for least educated, who
tend to need it most (Bergson-Shilcock, 2017; Patterson & Paulson, 2014).
Other adults face dispositional deterrents. They may not see themselves as being in
the workforce or having career options. They may perceive their skills as enough to get
by in life or may discount enhancing skills to keep their jobs (Smith, 2009). They may
lack confidence or feel discouraged from pursuing education. For younger adults,
entering PSE depends on the encouragement of parents (Patterson, 2014). Most non-
participants in PIAAC-USA did not have strong parental PSE role models, and fre-
quently, their parents may not have graduated from HS (Patterson & Paulson, 2014).
Lacking parental role models, they may turn to siblings, or even children, for encour-
agement and support (McAnnaney, 2009; Patterson, 2014).
To summarize, the least educated face numerous deterrents. To lessen earnings
inequality (Autor, 2014), reduce literacy classism (Quigley, 2017), and remain eco-
nomically vibrant, the United States needs to raise skill levels of the entire population
(Carnevale et al., 2013; Reder, 2010), including the forgotten 90%.

Methodology
To investigate nonparticipation of least educated adults, this article applies quantita-
tive techniques to a large-scale sample, PIAAC-USA. The article investigates what
happens to least educated nonparticipants—and how it happens.

Research Questions
1. In PIAAC-USA 2012/2014, how do the characteristics and backgrounds of
least educated compare by participation status?
2. What are their assessed skill levels for literacy, numeracy, and problem-solving
via technology rich environments (PSTRE)?
3. How do matched characteristics, backgrounds, and skill levels contribute to
their propensity to graduate from HS?
4. How do propensity scores and deterrent covariates predict their likelihood of
not participating in recent formal or nonformal education?
Patterson 45

Sample and Analyses


PIAAC-USA 2012/2014 contains information on nonparticipants’ educational back-
ground, deterrents, and characteristics, along with skill levels in literacy, numeracy,
and PSTRE. The PIAAC data set contains U.S. data from OECD’s international work-
ing-age adult survey and assessment of skills in 2012, as well as data from PIAAC’s
2014 U.S. National Supplement Household Study data released in late 2016. The orig-
inal PIAAC-USA (2012) surveyed and assessed 5,010 adults ages 16 to 65 years; 2014
supplemental data extends the sample to more than 8,000 adults and includes key
subgroups: unemployed adults (ages 16-65 year), young adults (ages 16-34 year), and
older adults (ages 66-74 year).
PIAAC data collection relied on a complex sampling design to ensure representa-
tiveness of the population (Hogan et al., 2016). Adults took surveys and assessments
on laptop computers. They completed both an extensive Background Questionnaire
and assessments in literacy, numeracy, and PSTRE. Assessment scores were estimated
using 10 plausible values per content domain. Scores ranged from 0 to 500 and were
classified into one of five levels. Levels for literacy and numeracy were as follows:
below Level 1 (0-175), Level 1 (176-225), Level 2 (226-275), Level 3 (276-325), and
Levels 4/5 (326-500). Levels for PSTRE were below Level 1 (0-240), Level 1 (241-
290), Level 2 (291-340), and Level 3 (341-500; Hogan et al., 2016).
PIAAC files were assembled from public-use files that perturbed and categorized
individual data for confidentiality. Weights were applied so that each respondent rep-
resented an accurate proportion of 203 million adults and standard errors would reflect
variability estimated in that population. More detail on sampling, weighting, back-
ground questionnaire administration, and assessments is available in the PIAAC tech-
nical report (Hogan et al., 2016).
PIAAC data were analyzed in IDB Analyzer software (available for download through
the PIAAC Gateway website), SPSS 23, and R software. IDB Analyzer software allowed
for means comparison of plausible values for literacy, numeracy, and PSTRE assess-
ments, and for categorical analyses (primarily cross-tabulations). The OECD International
Data Explorer was used to calculate mean score differences in literacy, numeracy, and
PSTRE assessments. The reference group in all analyses, unless otherwise stated, is adults
ages 20 to 74 years with low skills who indicated learning formally or nonformally in the
year before PIAAC. To determine if group differences were practically meaningful, effect
sizes were calculated with a 95% confidence threshold of twice the standard error for
percentage differences, and Cohen’s d for mean differences.
Nonparticipants (n = 2,214) pursued no formal education in the 12 months (FE12) or
nonformal education in the 12 months (NFE12) prior to PIAAC participation (FE12 = 0
and NFE12 = 0). Adults are further filtered by age (20 to 74 years, in 5-year increments,
to exclude young adult age groups traditionally in school) and to two initial education
levels: those with less than HS (LHS, i.e., ISCED ≤ 1, 1, or 2) as highest education com-
pleted, and those graduating HS (i.e., ISCED = 3). Although adults may have taken post-
secondary coursework or nonformal education previously, they did not do so in the year
before PIAAC and had no postsecondary credential, diploma, or degree. The comparison
group (n = 1,894) consists of adults who reported pursuing formal or nonformal education
46 Adult Education Quarterly 68(1)

in the year before PIAAC participation (i.e., FE12 = 1 or NFE12 = 1), were ages 20 to 74
years, and were in either initial education level. Further PIAAC-USA variables represent-
ing deterrents and activities are described in the Findings section.
To address the third research question, data were matched and analyzed. The author
analyzed available predictors using a propensity score analysis (PSA) approach (Bai,
2015; Hahs-Vaughn, 2017; Schneider, Carnoy, Kilpatrick, Schmidt, & Shavelson,
2007). PSA is useful in reducing selection bias when comparing group differences in
situations in which randomized control trials are not feasible (Bai, 2015). Through
matching, PSA seeks to balance the distributions of observed covariates between the
proposed “treatment” group (in this article, HS graduates in the PIAAC-USA sample)
and “control” group (i.e., LHS adults) as if adults had been randomly assigned to either
condition. Variables for PSA matching were numeracy skills (the sum of 10 plausible
values for assessed numeracy) and those associating with HS graduation: age group
(5-year increments), gender, annual income (in quintiles), highest parental education,
residence density (urban or not), and U.S. region (southern or not). Scores were matched
with replacement using Nearest Neighbor procedures in the MatchIt package of R, with
a 2:1 matching ratio to improve match capability, and applied PIAAC sampling weight
(Hahs-Vaughn, 2017; Randolph, Falbe, Manuel, & Balloun, 2014). All HS adults and
all but 75 LHS adults were successfully matched (n = 4,033). The PIAAC sampling
weight was applied both in matching and in logistic regression analysis to acknowledge
PIAAC’s complex sampling design and enhance accuracy of inferences (Hahs-Vaughn,
2015). Replicate weights were also employed in logistic regression analysis. Once
scores were matched, PSA models were statistically evaluated by percentage of bias
reduction and inspection of absolute standardized differences in means.
Propensity scores and individual covariates were entered into logistic regression anal-
yses to predict nonparticipation (vs. other status). Covariates include propensity to com-
plete HS, monthly earnings (in deciles), age group (20-74 years), health status, employment
status, gender, how adults primarily got information on current events (i.e., from the
Internet, from television, or from family/friends/coworkers), level of incomplete educa-
tion, weekly hours worked, flexibility in work scheduling, learning strategies, measures
of social trust, U.S. region (southern or not), residence density (urban or not), primary
reasons for not taking nonformal education, computer use (ever used a computer vs. not),
and presence of hearing, visual, or learning difficulties. p Values were set at .10 and Wald
statistics greater than one were employed to permit a maximum number of predictors.
Odds ratios were calculated as effect sizes to identify practical significance.

Descriptive Findings
In response to the first research question, one in four nonparticipating adults had an
initial education level of LHS (26%); three in four completed HS (74%). Initial educa-
tion rates for adults in a comparison group (i.e., 20- to 74-year-old adults without
postsecondary degrees that did participate in learning) indicate 9 in 10 completed HS
(89%) and 1 in 10 were LHS (11%).
Patterson 47

Demographics and Degree Completion


Among nonparticipants, Southern United States was overrepresented, and the
Midwestern and Western regions were underrepresented. Regionally, 44% of nonpar-
ticipants live in the South; approximately 17% live in the Northeastern United States,
20% in the Midwest, and 19% in the Western United States. Approximately one third
each live in cities or rural areas, one fourth in suburbs, and one tenth in towns.
Residence density of nonparticipants varies by region, however. Half of nonpartici-
pants in the Midwest and 40% of nonparticipants in the South live in rural areas. In the
West, nearly 9 in 10 nonparticipants live in a city (46%) or a suburb (41%).
In the comparison group, proportionately fewer adults were from the South (37%),
about the same were from the Northeastern United States (15%), and proportionately
more were from the Midwest (24%) and West (25%). Residence density is similar for
the comparison group, with the exception that Midwestern and Southern rural propor-
tions are smaller than for nonparticipants. The Western percentages of city residents
(41%) and suburban residents (46%) are similar for the comparison group.
Age of nonparticipants (20-74 years) and the comparison group in 10-year incre-
ments is displayed in Figure 2. Nonparticipants are significantly older than the com-
parison group, d = .74. Proportions of nonparticipants and comparison group who are
middle-aged (i.e., ages 30-59 years) are similar. Nonparticipants, however, have a
smaller percentage of young adults, aged 20 to 29 years (11%), than the comparison
group (36%) and a higher proportion of older adults, aged 60 to 74 years (30%), than
the comparison group (12%).
Both nonparticipants and the comparison group have approximately half men and half
women overall. The proportions of nonparticipant men and women tend to stay balanced
with advancing age, yet in the comparison group the proportion of women gradually
increases with age. A middle-aged group of nonparticipants is an exception. In their early
50s, LHS nonparticipants are more often male (61%) than female. The LHS men in their
early 50s also had high rates of unemployment (36%) and disability (19%). When asked
why they did not participate in nonformal education, no prominent reason was given.
A total of 3 in 10 nonparticipants (30%) report not completing a degree program
which they started (regardless of degree type), less than in the comparison group
(47%). Unfinished degree types for nonparticipants are presented in Figure 3. Nearly
half of nonparticipants not completing a program left a postsecondary certificate or
diploma unfinished (45%). One in five left an associate degree unfinished (18%). In
the comparison group, 41% did not complete a postsecondary certificate or diploma.
However, the comparison group had a lower rate of leaving HS early (7%) and higher
rates of leaving more advanced degrees unfinished (23% left an associate degree and
29% left a bachelor degree or higher program incomplete).

Deterrents to Nonformal Education


In the year before PIAAC, 18% of nonparticipants wanted to pursue nonformal educa-
tion but did not. Three top reasons they gave, as displayed in Figure 4, were cost, work
obligations, and family responsibilities.
48 Adult Education Quarterly 68(1)

Comparison Group
36 18 18 18 12

Percent in Age
Group

Nonparcipants 11 17 20 23 30

0 20 40 60 80 100
20-29 Years 30-39 Years 40-49 Years 50-59 Years 60-74 Years

Figure 2.  Age distribution of nonparticipants in learning, PIAAC (2012/2014).


Note. PIAAC = Program for the International Assessment of Adult Competencies.

Figure 3.  Unfinished degree type percentages of nonparticipants, PIAAC (2012/2014).


Note. PSE = postsecondary education; PIAAC = Program for the International Assessment of Adult
Competencies.

When disaggregated by education attainment, nonparticipants indicate low interest


in nonformal education—15% of LHS and 20% of HS nonparticipants did so. LHS
adults who indicated interest cite family responsibilities as a reason for not participat-
ing (25%) at a proportionately higher percentage than HS adults (17%). HS nonpartici-
pants see cost as a reason at a higher rate (25%) than LHS nonparticipants (18%). Both
groups were equally “too busy at work” (19%).
In contrast, 37% of the comparison group, or twice the rate of nonparticipants, wanted
to pursue nonformal education the year before and did not, and this percentage did not
differ by LHS/HS status. Although cost concerned both groups, nonparticipants empha-
size family responsibilities more (19%) and the comparison group emphasizes work
responsibilities more (30%) as reasons for not learning nonformally (see Figure 4).
Patterson 49

Figure 4.  Top reasons for not pursuing nonformal education, PIAAC (2012/2014).
Note. PIAAC = Program for the International Assessment of Adult Competencies.

Families
More than three fourths of nonparticipants (77%) have a partner or spouse, at a signifi-
cantly higher rate than the comparison group (63%). Nonparticipants live in house-
holds of three people (M = 3.0, SD = 1.5), and half of households with children have
no more than two children (18% have one child and 33% have two). Rates are similar
in the comparison group: mean household number is 3.2 (SD = 1.4), 21% have one
child, and 33% have two children. Median annual income level is similar for both
groups; approximately 60% of nonparticipants and 56% of the comparison group have
an annual income at or below the median, which represents 200% of poverty.
Four in five nonparticipants are born in the United States (82%) and are native
English speakers (80%). Comparison group rates are significantly higher, 87% born in
United States and 84% native speakers. A total of 19% of nonparticipants are first-
generation immigrants, a significantly higher rate than 13% in comparison group.
Nonparticipant parents’ backgrounds model education positively (see Figure 5);
nearly half (49%) have a father or a mother who graduated from HS and completed
some PSE or earned a college degree. Still these rates are much lower than those of the
comparison group, 71% for mothers and 70% for fathers.
Figure 5 also shows approximately two in five nonparticipants have neither a father
or mother who completed HS. As seen in Figure 6, this tendency is largely genera-
tional. Of nonparticipants whose parents did not complete HS, 61% were at or above
the median age (50-54 years). Approximately two thirds (66%) of nonparticipants with
at least one parent having a postsecondary degree were younger than the median age.

Health and Disabilities


Nonparticipating adults claim a median “good” health, yet 30% report health challenges
(i.e., “fair” or “poor” health). The median health status for the comparison group is “very
50 Adult Education Quarterly 68(1)

0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%

39%
neither parent completed high school 19%

37%
father completed high school and some PSE 48%

40%
mother completed high school and some PSE 49%

9%
mother completed bachelor degree 22%

11%
father completed bachelor degree 22%

Nonparcipants Comparison Group

Figure 5.  Parents’ education background, PIAAC (2012/2014).


Note. PSE = postsecondary education; PIAAC = Program for the International Assessment of Adult
Competencies.

Figure 6.  Percentage of adults younger than age 45 by their parent’s highest education level,
PIAAC (2012/2014).
Note. LHS = less than high school; HS = high school; PSE = postsecondary education; PIAAC = Program
for the International Assessment of Adult Competencies.

good,” and only 15% report health challenges. The incidence of nonparticipant disabili-
ties is 15% hearing difficulties, 20% vision difficulties, and 10% learning disabilities. In
comparison group, rates of vision and hearing difficulties (9% and 11%, respectively)
are significantly lower, yet learning disabilities rates are comparable (9%).
Patterson 51

Table 1.  Ranked Industry Classifications of Nonparticipants’ Employers, PIAAC


(2012/2014).

Industry Nonparticipants (%) Comparison group (%)


1. Construction 6 5
2. Retail 5 8
3. Food and beverage service 4 7
4. Building and landscape service 4 3
5. Health care 2 5
6. Motor vehicle trade and repair 2 2
7. Education 1 3
8. Public administration and security 1 6

Note. PIAAC = Program for the International Assessment of Adult Competencies.

Of those who report not working, one third (33%) do not pursue work because of
long-term illness (the top reason). Another 4% are temporarily ill, bringing the total
percentage of unemployed adults affected by health to 37%, higher than the 15% com-
parison group rate. Illness particularly affects the ability of middle-aged adults to find
work; many reporting long-term illness (43%) are 35 to 54 years old and 39% are 55
to 65 years.

Employment
About half of nonparticipants are employed (53%) at a substantially lower rate than
comparison group, 81%. Predominantly employed in the private sector (88%), nonpar-
ticipants tend to work in small businesses employing 50 or fewer (59%) rather than in
large companies (250+ employees; 18%). In contrast, 76% of comparison group works
in the private sector, 55% in small businesses, and 22% in large companies (250+
employees). Indeed, 18% of nonparticipants work in skilled occupations, lower than
the 32% of comparison group members doing so. In semiskilled occupations, 33%
work in white-collar occupations (vs. 39% in comparison group), and 32% in blue-
collar occupations (higher than the 19% in comparison group). The top eight industries
employing them are presented in Table 1; approximately one fourth of nonparticipants
and two fifths of comparison group work in these industries. Nonparticipants work
most often in construction or retail, whereas comparison group members tend to work
most often in retail, food and beverage service, or public administration and security.
While the top occupation for both groups is sales workers, in nonparticipants’ top
five occupations are (in rank order) personal service workers, building and related
trades, metal and machinery trades, and drivers and mobile plant operators. The cor-
responding occupations for the comparison group, after sales workers, are personal
care workers, business and administration associate professionals, and drivers and
mobile plant operators.
Three fourths of nonparticipants (78%) and comparison group (76%) are satisfied
with their current job. Nonparticipants work an average 40 hours per week, and 27%
52 Adult Education Quarterly 68(1)

Figure 7.  Nonparticipant hours worked, PIAAC (2012/2014).


Note. PIAAC = Program for the International Assessment of Adult Competencies.

works more than 40 hours per week (see Figure 7). Rates for the comparison group are
similar.
One in five (21%) nonparticipants are self-employed, significantly more often than
adults in comparison group (11%). Managerial responsibilities for nonparticipants are
minimal; 24% manage other employees (vs. 30% in comparison group).

Learning at Work
Two thirds of employed nonparticipants have at least some flexibility in how they
sequence work tasks (67%) and accomplish their work (66%). Three fourths have
some flexibility in pacing the speed of work (75%). As in comparison group, about
half (46%) have some flexibility in how work hours are structured, while 54% does
not. The comparison group has significantly more flexibility in sequencing work tasks
(74%), accomplishing work (75%), and in pacing the speed of work (78%).
When working, nonparticipants report a median of weekly learning by doing.
However, they learn from their coworkers or supervisors and learn to keep up to date
a median of monthly. Learning at work tends to occur more frequently in comparison
group; the median for learning in all three areas is weekly.

Unemployment
For nonparticipants, rates of unemployment (12% vs. 8% in comparison group) and
permanent disability are high (13% vs. 2% for comparison group). More than four
fifths of unemployed nonparticipants give one of three reasons for not seeking work
(see Figure 8): long-term illness (33%), caring for family (17%), and retirement (37%,
97% at age 55 or later). Nearly all nonparticipants not seeking work because they care
for family are women (95%), and 39% are aged 45 to 65 years, implying many may
Patterson 53

Nonparcipants

Figure 8.  Top reasons unemployed adults gave for not looking for work, PIAAC
(2012/2014).
Note. PIAAC = Program for the International Assessment of Adult Competencies.

serve as caregivers for multiple generations. The comparison group’s reasons for not
seeking work, with much less long-term illness, are as follows: being a student (10%),
caring for family (19%), and long-term illness (13%).

Learning Strategies
Virtually all nonparticipants (90% to 92%), regardless of employment status, believe
they could understand and speak English well or very well. Additionally, 87% believe
they could read English and 83% write English well or very well. They report posi-
tively on learning strategies. Approximately two thirds report a “high extent” of liking
to learn new things (64%) and looking for additional information (66%), as shown in
Figure 9. In comparison group, rates are even higher: 82% like to learn new things to
a high extent and 80% to look for additional information. Nonparticipants find it more
problematic to determine how different ideas fit together (49% could figure it out to a
high extent) than does the comparison group (64% to a high extent). More problematic
for nonparticipants is relating new ideas into real life (24% to a high extent).

Computer Use
Computer use is high; 73% of nonparticipants report ever having used a computer
before. The rate of computer use climbs to 95% for adults younger than 25 years, and
is more than 65% for adults 55 years and older. In comparison group, 91% has ever
54 Adult Education Quarterly 68(1)

Nonparcipants

Figure 9.  Learning strategies, used to a high extent, of nonparticipants and comparison
group, PIAAC (2012/2014).
Note. PIAAC = Program for the International Assessment of Adult Competencies.

used a computer, including more than three fourths of those 55 years and older; both
rates are significantly higher.
Of nonparticipants using the Internet outside of work, 90% use e-mail at least
sometimes and 54% e-mail daily (respective rates for e-mail use and daily e-mail in
comparison group are 95% and 66%). Hence 89% of nonparticipants go online at least
sometimes to better understand health or financial issues (vs. 94% of comparison
group), and 73% to conduct financial transactions (less than 83% of comparison group
members doing so). Only 31% of nonparticipants get online at least sometimes to meet
virtually, compared with half (47%) of comparison group members. All rates for non-
participants are significantly lower than for comparison group.

Community Involvement and Social Trust


PIAAC survey data also offer insights into nonparticipants’ infrequent community
involvement and low levels of social trust. Most nonparticipants (67%) do not volun-
teer in nonprofit organizations, charities, or political parties. Three fourths (78%) trust
only a few people completely. Four in five (83%) agree if they are not careful, other
people take advantage of them. Nearly half (48%) agree that “people like me don’t
have a say” in what the government does. Interestingly, 34% disagree with the same
Patterson 55

statement. Older nonparticipants tend to be more polarized as to whether they have a


say, and younger ones more neutral, neither agreeing nor disagreeing.
A greater proportion of comparison group, on the other hand, tends to volunteer
(27% regularly and 27% less than monthly). They are significantly more trusting than
nonparticipants—71% of adults in the comparison group trust only a few people com-
pletely. Still the comparison group believes others take advantage (80%) at approxi-
mately the same rate as for nonparticipants. Two fifths (42%) disagree that they have
no say in government, but like nonparticipants, younger adults in the comparison
group tend to be more neutral.
When asked where they get information about current events, three fifths of nonpar-
ticipants receive it “a lot” from television (60%) and one third online (34%); for the com-
parison group, television rates are lower (52%) and online rates are much higher (58%).
A fourth of nonparticipants receive information a lot from family, friends, or coworkers
(22%) or radio (22%); rates for the comparison group are higher (28% and 26%, respec-
tively). Overall, family, friends, or coworkers, along with television, are the most frequent
sources of current events information (“a lot” or “some”), with 94% of nonparticipants
and adults in comparison group getting at least some information from these sources.

Assessed Skills in Literacy, Numeracy, and PSTRE


The second research question considers differences in assessed skill levels for both
groups. Adults were assessed in three skill domains: literacy, numeracy, and PSTRE.
As displayed in Figure 10, on average adults score at Level 2, except for nonpartici-
pants who were assessed in numeracy. At Literacy Level 2, respondents were tasked
with matching text and information, paraphrasing, and making low-level inferences.
At Numeracy Level 2, PIAAC respondents could respond to mathematical content in
common contexts and apply two or more steps to solving math problems (Patterson &
Paulson, 2016). Nonparticipants score at Level 1 in numeracy, on average. For both
groups, average numeracy scores are lower than those in literacy and PSTRE.
Significant gaps occur in mean scores between groups, as shown in Figure 10.
Although literacy and PSTRE skill levels average at Level 2, these average scores are
near the bottom of the level, indicating that many nonparticipants score lower. The
magnitude of score differences is moderate for all three domains.

Logistic Regression Findings


Propensity scores were developed as described in the methodology section, with all
but 75 cases (n = 4,033) successfully matched on HS completion, assessed numeracy
skills, and several demographic and background variables. Propensity scores were
then entered into a logistic regression model to predict likelihood of nonparticipation.
Table 2 presents the most important predictors of nonparticipation, along with odds
ratios to estimate likelihood of nonparticipation. Predictors with negative beta weights
and odds ratios below 1.0 exhibit reduced likelihood of nonparticipation; predictors
with positive beta weights and odds ratios above 1.0 exhibit increased likelihood of
56 Adult Education Quarterly 68(1)

Nonparcipants

Figure 10.  Assessed skills in literacy, numeracy, and PSTRE of nonparticipants and
comparison group, PIAAC (2012/2014).
Note. PSTRE = problem-solving via technology rich environments; PIAAC = Program for the International
Assessment of Adult Competencies.

nonparticipation. The Nagelkerke pseudo R2 is low, indicating that additional predic-


tors are needed to more fully explain the likelihood of nonparticipation.
As presented in Table 2, the strongest predictor is the propensity to complete HS, as
based on assessed numeracy levels and demographic/background variables.
Nonparticipation in education of adults with high propensity to graduate is approxi-
mately six times less likely than for adults with low propensity. Other important negative
predictors are health status and ability to relate new ideas to real life. Being able to relate
new ideas to real life reduces nonparticipation in education; with greater extent of this
learning strategy, the likelihood of nonparticipation decreases by 34%. Incrementally for
decreasing health (e.g., good to fair, fair to poor), the likelihood of nonparticipation
decreases by 31%. A lesser predictor is a measure of social trust: The more adults dis-
agree that others take advantage of them, the more nonparticipation likelihood is reduced.
Strong positive predictors of nonparticipation include work reasons and gaining
information online. Adults who report being too busy at work to take nonformal edu-
cation have a 45% greater likelihood of nonparticipation. The likelihood of nonpartici-
pation increases by approximately 31% for those who tend to primarily get information
online in contrast with other sources.
Two lesser positive predictors are age group and work schedule flexibility. The
likelihood of nonparticipation increases 13% for each increment of age (16-19 years to
20-24 years, 20-24 years to 25-29 years, etc.). Those with more work schedule flexi-
bility have an 11% greater likelihood of nonparticipation per increment.
Patterson 57

Table 2.  Important Predictors From Logistic Regression of Nonparticipation, PIAAC


(2012/2014).

Standard Wald
Predictor b error statistic p Odds ratio
High school completion −1.8 0.8 4.55 <.05 0.17
propensity
Health status (excellent to poor) −0.3 0.1 5.23 <.05 0.77
Incomplete education level −0.1 0.1 3.51 <.10 0.88
Learning strategy: can relate new −0.3 0.1 6.63 <.01 0.75
ideas to real life
Monthly income (deciles) −0.1 0.2 2.35 >.10 0.92
Social trust: others take −0.1 0.1 1.51 >.10 0.87
advantage if not careful
Age group 0.1 0.0 7.51 <.01 1.13
Primarily gains information 0.3 0.1 5.75 <.05 1.31
about current events from the
Internet
Work schedule flexibility 0.1 0.1 1.50 >.10 1.11
Being too busy at work was 0.4 0.3 1.44 >.10 1.45
reason for no nonformal
education last year
Constant 2.1 1.7 1.49 >.10 7.67
Pseudo R2 0.20  

Note. PIAAC = Program for the International Assessment of Adult Competencies.

Discussion
To summarize major findings, nonparticipation is associated with both deterrents and
supports. Most deterrents identified in PIAAC are situational: increasing age, variable
role models from parental education, low income, work responsibilities, and family
responsibilities all appear to contribute to nonparticipation. Dispositional deterrents
from PIAAC include lack of interest in education, health and disability challenges,
low social trust, and difficulties in relating new ideas to real life. Two institutional
deterrents are cost of education and little work schedule flexibility. Nonparticipants
also experience lower assessed skills, particularly in numeracy. Supports nonpartici-
pants report are liking to learn new things, use of computers, and getting information
from television and trusted people in their lives.
A first situational deterrent is age. Moreover, 70% of nonparticipants are younger than
60 years and could be recruited to education if interested or in the workforce. Many edu-
cators may assume that middle-aged adults do not want or need further education, but as
the number of adults 25 and older entering PSE rises (Ryu, 2010), that assumption needs
challenged. A sizable percentage of adults do not have family role models for pursuing
further education and may not know how to do so or what advantages may result. Adult
and PSE programs may need to supply this information (Patterson, 2014; Patterson &
Paulson, 2016) and to offer examples where participation has been successful.
58 Adult Education Quarterly 68(1)

Deterrents affect least educated adults in middle age. Where nonparticipant men in
their early 50s have fewer HS diplomas and higher rates of unemployment and dis-
ability, they could be recruited through health care professionals, vocational rehabili-
tation agencies, or disabilities service providers. For the unemployed, recruiting
methods offering career pathways could be fruitful and perhaps less intimidating than
starting PSE directly.
Other situational deterrents are being “too busy at work” and, to a lesser extent,
family caretaking. How can adult education programs and nonformal education pro-
grams accommodate the needs of adults who work full-time, have family responsibili-
ties, and have limited means? How can educators help adults and employers to see
relevant connections between learning and work—and ways learning can support their
careers? Communication among employers and educators is key to reducing deter-
rents. As Bergson-Shilcock (2017) notes, employers need to recognize implications of
supporting employees’ skill building, either through support of nonformal education
opportunities or through release time to gain external instruction.
The incidence of fair and poor health is a strong concern for nonparticipants. How
can educators reach adults who may struggle with long-term illness and cannot work?
One approach might be to offer further learning to prevent skills from eroding until
they return to work. Education programs could support needs of prospective partici-
pants with health challenges in several ways. Recruitment might need to occur through
doctors, dentists, and health clinics. This recruitment involves more than simply leav-
ing brochures in a doctor’s waiting room, rather actively building relationships with
local medical professionals who recognize what the program offers and can quickly
refer adults who struggle with health literacy or numeracy. Flexible scheduling may be
required across time, and flexibility with deadlines for assignments could also be nec-
essary. Participants with health care or disability issues—as well as family caregiv-
ers—may also require individualized support and instruction. Flexibility may be even
more beneficial to middle-aged adults dealing with long-term illness.
The top occupations of nonparticipants—personal service workers, building trades,
metal and machinery trades, and drivers and mobile plant operators—offer clues as to
where to recruit them. Adult and PSE programs not already doing so could foster rela-
tionships with chambers of commerce, small businesses, business development cen-
ters, and local trade associations or unions in these occupational areas. Industry
partnerships, including multiple firms and education partners, could collaborate to
identify needs, determine available training resources, and develop career pathways
for adults (Bergson-Shilcock, 2017). As with other growing relationships, adult educa-
tors will need to cultivate patience in developing new connections and willingness to
learn more about occupations and potential participants from them.
Institutional deterrents are costs of education and inflexible work schedules. Do
nonparticipants realize adult education programs have minimal or no cost? Even PSE
programs may cost less than expected, and understanding the difference between
grants and loans is necessary as they consider costs and learn to navigate financial aid
processes (Fike & Fike, 2008).
As well as recruiting prospective learners, adult or postsecondary educators could
also convince businesses of the advantages of flexible scheduling for their employees
Patterson 59

(Bergson-Shilcock, 2017). Flexible work hours can promote learning, especially


“learning by doing” that nonparticipants are already somewhat comfortable with and
that would directly benefit the company. The high rate of self-employment among
nonparticipants also offers insights for recruitment, in that programs could point out
benefits of additional skills for leading and growing businesses.
Since most adults report liking to learn new things and to look for new information,
programs could offer occasional open house events on topics of community interest or
to adults in certain occupations. An added bonus would be in appealing to adults’ curi-
osity and explaining how they could use new information to their advantage in real life
applications. In recruitment materials, adult educators can also capitalize on adults
liking to learn new things. Programs may believe their materials already do so, but the
best way to reevaluate materials would be to ask nonparticipating adults with low
skills from the community for their reactions—these could be friends, relatives, or
neighbors of current adult learners.
In the current decade, educators can no longer argue that adults are not online.
Nearly all adults are getting online, whether on computers or smartphones, for more
information. Are programs taking full advantage of those interactions to recruit new
learners? Also, programs need to investigate how adult education can be delivered
through online learning—what are the most effective methods and how can potential
adult learners be encouraged to engage in these opportunities? Another medium that
nonparticipants prefer is television. Do adult and postsecondary educators take full
advantage of television services as a medium for recruiting and serving adult learners?
Another area of concern is lack of community involvement and of social trust.
Since lack of trust is high, educators should consider expanding word of mouth to
recruit adults to education. Even though many nonparticipants believe others take
advantage of them, they tend to place faith in family, friends, and trusted coworkers to
gain information. To the extent that adult and postsecondary educators can ally them-
selves with community and family members who live near or work closely with non-
participants, the chances of successful recruitment would rise.
Most important predictors of nonparticipation are actionable. While poor health,
advancing age, or low income may not be directly actionable, they offer valuable
insights into opportunities for recruitment, individualized supports, flexible program-
ming, and financial aid. Educators’ efforts will go a long way to closing wide skills
gap, reducing literacy classism (Quigley, 2017), and preparing adults for the future
workforce (Autor, 2014).

Limitations and Conclusion


A first limitation of the article is associated with the measure employed to identify
nonparticipation. As a cross-sectional survey, PIAAC addresses participation in for-
mal or nonformal education within the past 12 months rather than measuring it as time
invariant or longitudinally. Some adults who pursued learning 2 or more years earlier
perhaps took a break from participation. Still the article’s results provide evidence,
such as the much lower skill levels of nonparticipants, which suggests that nonpartici-
pation for many may be long-standing.
60 Adult Education Quarterly 68(1)

This article leaves many questions unanswered, in part because certain background
questions were not included in PIAAC and in part because findings raise new ques-
tions which are beyond the scope of the article to address. With the exception of PSA,
all analyses in this article were descriptive, and causality should not be inferred.
Future research could consider geographic differences, age and parental back-
ground, and the “reasons behind the reasons” of other deterrents. Nonparticipants need
to be an integral part of this future research and no longer “forgotten” in future studies
of deterrents (Quigley, 2006). While residence density is evenly distributed across the
United States, the overrepresentation of nonparticipants in the South and in rural areas
of the Midwest and South for the least educated adults is a concern. Future researchers
could investigate qualitatively how nonparticipant experiences differ and what reasons
residents in these areas give for not participating. How could adult and postsecondary
educators from different regions of the country collaborate to change this balance?
Further qualitative investigation is also needed on the role of age, interest in educa-
tion, and parental education. What are the interest levels of middle-aged adults in
further education, and how do those levels fluctuate with advancing age (Smith,
2009)? How long are they planning to stay in the workforce, and what types of learn-
ing might appeal to them? What family role models, if any, might encourage them to
pursue education? A sizable proportion of nonparticipants are simply not interested in
nonformal education. Why might nonparticipants be uninterested, especially at work?
Do they anticipate nonformal education as “going to school again”—as stigmatizing,
irrelevant, or boring (Quigley, 1997, 2017)? Whatever is gleaned in answer to this
question can be used in recruitment materials to show prospective participants how
adult or PSE may differ from previous “school” experiences and how further educa-
tion connects to the workplace.
Deterrents such as costs of education, work and family responsibilities, and even health
challenges are a reality. Research needs to determine what is behind these deterrents—
what feeds into challenges—and how some adults can identify solutions where others can-
not. Greater understanding of the “reasons behind the reasons” will help adult and
postsecondary educators identify and implement solutions to the deterrents adults face.
In-depth PIAAC-USA analysis of the backgrounds and deterrents of least educated
adults adds to the knowledge base of nonparticipation in formal or nonformal education.
This article also identifies solutions that might trigger them to reengage. Knowing even
more about what happened educationally to the forgotten 90% is informative to educa-
tors and policy makers—and most important, to prospective adult learners themselves.

Acknowledgments
The author wishes to thank B. Allan Quigley, Jaleh Soroui, and two anonymous reviewers for
insightful comments on earlier drafts of the article.

Author’s Note
Paper presented on April 2017 at American Educational Research Association conference, San
Antonio, TX. Content responsibility is solely the author’s.
Patterson 61

Declaration of Conflicting Interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of
this article.

References
Autor, D. H. (2014). Skills, education, and the rise of earnings inequality among the “other 99
percent.” Science, 344, 843-851.
Bai, H. (2015). Methodological considerations in implementing propensity score matching.
In W. Pan & H. Bai (Eds.), Propensity score analysis: Fundamentals and developments
(pp. 74-88). New York, NY: Guilford Press.
Bergson-Shilcock, A. (2017). Foundational skills in the service sector: Understanding and
addressing the impact of limited math, reading, and technology proficiency on workers and
employers. Washington, DC: National Skills Coalition.
Carnevale, A., Smith, N., & Strohl, J. (2013). Recovery: Projections of job and education
requirements through 2020. Washington, DC: Georgetown Public Policy Institute, Center
on Education and the Workforce.
Diplomas count 2013. (2013, June). Education Week. Retrieved from http://www.edweek.org/
ew/articles/2013/06/06/34execsum.h32.html?intc=EW-DC13-LNAV
Fike, D. S., & Fike, R. (2008). Predictors of first-year student retention in the community col-
lege. Community College Review, 36, 68-88.
Hahs-Vaughn, D. (2015). Propensity score analysis with complex survey samples. In W. Pan &
H. Bai (Eds.), Propensity score analysis: Fundamentals and developments (pp. 236-266).
New York, NY: Guilford Press.
Hahs-Vaughn, D. (2017). Applied multivariate statistical concepts. New York, NY: Routledge.
Hogan, J., Thornton, N., Diaz-Hoffmann, L., Mohadjer, L., Krenzke, T., Li, J., . . . Khorramdel,
L. (2016). U.S. Program for the International Assessment of Adult Competencies (PIAAC)
2012/2014: Main study and national supplement technical report (NCES 2016-036).
Retrieved from https://nces.ed.gov/pubs2016/2016036.pdf
Kis, V., & Field, S. (2013). Time for the U.S. to reskill? What the Survey of Adult Skills
says. Paris, France: OECD. Retrieved from http://www.oecd-ilibrary.org/docserver/
download/8713041e.pdf?expires=1503991935&id=id&accname=guest&checksum=1061
0AC52E9E620358AC143665C451C7
McAnnaney, S. (2009). Assessing barriers to education among GED graduates entering col-
lege. College Park: University of Maryland, Department of Social Work.
Organization for Economic Cooperation and Development. (2013). OECD skills outlook 2013:
First results from the Survey of Adult Skills. Paris, France: OECD. Retrieved from http://
www.oecd-ilibrary.org/docserver/download/8713011e.pdf?expires=1503991925&id=id&
accname=guest&checksum=0570A5C3E9083C88D028144DAC0DE612
Patterson, M. B. (2013). Post-GED®-credential employment experiences of adults with spe-
cial needs. Journal of Research and Practice for Adult Literacy, Secondary, and Basic
Education, 2(3), 156-169.
62 Adult Education Quarterly 68(1)

Patterson, M. B. (2014). Post-GED®-credential college prospects for adults with special needs.
Journal of Research and Practice for Adult Literacy, Secondary, and Basic Education,
3(3), 22-35.
Patterson, M. B., & Paulson, U. (2014, December). Adult transitions to learning in the USA:
What do PIAAC survey results tell us? Paper presented at American Institutes for Research’
PIAAC Invitational Research Conference, Washington, DC.
Patterson, M. B., & Paulson, U. (2016). Adult transitions to learning in the USA: What do
PIAAC survey results tell us? Journal of Research and Practice for Adult Literacy,
Secondary, and Basic Education, 5(1), 5-27.
Quigley, B. A. (1997). Rethinking adult literacy education: The critical need for practice-based
change. San Francisco, CA: Jossey-Bass.
Quigley, B. A. (2006). Building professional pride in literacy. Malabar, FL: Krieger.
Quigley, B. A. (2017). Will anything be different in the 21st century? How 107 million
adults and the field of adult literacy became so marginalized. PAACE Journal of Lifelong
Learning, 26, 39-54.
Randolph, J., Falbe, K., Manuel, A., & Balloun, J. (2014). A step-by-step guide to propensity
score matching in R. Practical Assessment, Research, & Evaluation, 19(18), 1-6.
Reder, S. (1999). Adult literacy and postsecondary education students: Overlapping populations
and learning trajectories. Retrieved from http://files.eric.ed.gov/fulltext/ED508706.pdf
Reder, S. (2010). Adult literacy development and economic growth. Retrieved from https://
lincs.ed.gov/publications/pdf/AdultLiteracyDevEcoGrowth.pdf
Ross-Gordon, J. M. (2011). Research on adult learners: Supporting the needs of a student popu-
lation that is no longer nontraditional. Peer Review, 13(1), 26-29.
Ryu, M. (2010). Minorities in higher education: Twenty-fourth status report. Washington, DC:
American Council on Education.
Schleicher, A. (2013). Skilled for life? Key findings from the Survey of Adult Skills. Brussels,
Belgium: OECD.
Schneider, B., Carnoy, M., Kilpatrick, J., Schmidt, W. H., & Shavelson, R. J. (2007). Estimating
causal effects using experimental and observational designs (Report from the Governing
Board of the American Educational Research Association Grants Program). Washington,
DC: American Educational Research Association.
Smith, M. C. (2009). Literacy in adulthood. In M. C. Smith & N. DeFrates-Densch (Eds.),
Handbook of research on adult learning and development (pp. 601-635). New York, NY:
Routledge.
Spellman, N. (2007). Enrollment and retention barriers adult students encounter. Community
College Enterprise, 13(1), 63-79.
Zhang, J., Guison-Dowdy, A., Patterson, M. B., & Song, W. (2011). Crossing the bridge: GED
credentials and postsecondary educational outcomes: Year two report. Washington, DC:
American Council on Education.

Author Biography
Margaret Becker Patterson is currently Senior Researcher with Research Allies for Lifelong
Learning in the Washington, DC, metro area (www.researchallies.org). She led the award-­
winning Adult Learner Leadership in Education Services (ALLIES) evaluation for VALUEUSA,
the national organization of adult learners. Previously, she served as Research Director at GED
Testing Service and administered and taught in adult education programs in Nebraska, Nevada,
and Kansas.
Copyright of Adult Education Quarterly is the property of Sage Publications Inc. and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

Vous aimerez peut-être aussi