AEA Presentation explores US college access education system as a complex system with a special focus on inequality. Simple models are included using vensim of achieving Obama goals of US being first in college attainment by 2020
Us college access programs complex adaptives system nov2011
1. AEA November 2011
Anaheim California
Treating the US Educational System as a Complex Adaptive System
and Investigating Computational Simulation of Federally-Funded
Access Program Impacts
Prepared by:
Margaret Cahalan
Margaret.Cahalan@pellinstitute.org
Pell Institute for the Study of Opportunity in
Higher Education
Council for Opportunity in Education (COE)
2. Topics----Invitation to:
Explore systems concepts
Use what for me is a new tool to help in understanding
policy choices---dynamic modeling software—Vensim
Tool
Walk through examples in college access—specifically
related to President Obama’s goals to make US once
again be first in college attainment
3. Definitions of key terms
System—Greek word –susiatani— “cause to stand
together”
Complex system is a system that consists of a large
number of agents that interact with each other in
various ways
System is adaptive if these agents change their actions
as a result of events in the process of interaction
(Vriend, 1994)
Social systems are complex dynamic non-linear with
multiple feedback loops
4. Relevant Complex Systems
Concepts
Nested or fractal quality— Entrainment (competitors become
patterns replicate at levels more alike)
Self similarity—parts & whole Ultrastability—change internal to
similar adapt
Attractors and basins what settles Sensitivity to initial conditions-
down to over time and attracted to small differences large impacts
Feedback—negative dampens- Chaos-deterministic –patterns
positive amplifies emerge-- sensitive to initial conditions
Disconnect between proximity in
Saddle points—unstable-areas -- time and cause and effect—small
direction may be more important can bring massive change and large
than strength-tipping seemingly less—unintended and
Bifurcation-new systems—sudden unforeseen
jumps Importance of timing and
Self-organization—self iteration direction
Emergent behavior of system- Synchronicity-” temporally
patterns emerge—different rule sets coincident occurrences of acausal
events”-Jung
5. Steps in Modeling (Andrew Ford—
Modeling the Environment 2010)
1. Get to know system 5. Estimate the
2. Get specific about parameter values
dynamic problem 6. Run reference model
3. Construct—stock 7. Check units and
(level) and flow Sensitivity
diagram 8. Experiment or test
4. Draw causal loops potential impact of
policies
6. Know systems and get specific about
problem of Obama’s 2020 Goals
Lift US to be first in college attainment—once again
US ranked 12th of 36 in percent of 25 to 34 with Tertiary
(BA and associate) degrees---Korea first, Canada
second
Increase 25 to 34 year olds with college from 15 million
to 26 million
Increase work force 25 to 64:
College 46 million to about 63 million--college
BA and above from about 32 million to about 45 million
7. Obama’s goals are adaptive to wider global systems
but to achieve nested systems must change
Global
US
State
City -
Neighborhood
School
Home
Stud
ent
8. Five major features of system to
capture in models
1. Dynamic large increases in attainment high school and
postsecondary over last 60 yrs.
2. High levels of inequality characterize system
3. Role of education sorting in keeping equilibrium of
work value inequality and legitimizing through the
meritocracy and income stratification
4. Role of government in mediating the inequality
effects—closing the gaps –mission of US Department of
education—equal access---Pell and College access
support programs (TRIO, Gear UP)
5. Education driver of innovation role for jobs
9. Inequality
Meritocracy
Stability
Education and Class Innovation
and Workforce System and Change
equilibrium
Opportunity
and mobililty
10. Dynamic change ---Note Large Increase Over
Time in High School Diploma—Saturation—
leveling—S curve
Figure 1. Percent of population 25 years of age and older who have a high school
diploma or equivalent by race/ethnicity: Decennial Census Data 1940-
2000
90
84
80
78
72
70 69
63
60
55
51 52
50
50
43 44
40
36
30 31
26
20 22
14
10
8
0
1930 1940 1950 1960 1970 1980 1990 2000 2010
Black Hispanic White
Note: Based on Decennial census. White category does not exclude those of Hispanic Origin. Hispanic
Origin can be of any race.
SOURCE: U.S. Census Population Division Census 2000 PHC-T-41. A Half-Century of Learning: Historical
Statistics on Educational Attainment in the United States, 1940 to 2000
11. Note Some Convergence in State Variation
Figure 3. Percent of total population 25 and older with high school diploma or
equivalent by state: 1940-2000
100
90
80
70
60
50
40
30
20
10
0
1930 1940 1950 1960 1970 1980 1990 2000 2010
NOTE: This distribution ranged from 15 percent in Arkansas to 41 percent in the District of Columbia and 37
percent in California in 1940; and ranged from 73 percent in Mississippi to 88 percent in 4 states, Utah,
Wyoming, Minnesota, and Alaska in the year 2000.
SOURCE: U.S. Census Population Division Census 2000 PHC-T-41. A Half-Century of Learning: Historical
Statistics on Educational Attainment in the United States, 1940 to 2000
12. Exit exams and high school
graduation: Cohort Survival Rate
90
80
78
76 76 75 75
74 74 74 73 73 74
70 71 70 70
69 69 68 67 68 67 67 67 68
67 66 66 66
64 65 64 64 64 65
60
50
40
30
20
10
0
1988 1990 1992 1994 1996 1998 2000 2002 2004 2006
US Mean of states without exit exams Mean of states with exit exams
13. Note Large Increase in BA, but larger gaps
Figure 2. Percent of population 25 years of age and older who have a BA degree:
Decennial Census Data 1940-2000
30
27
26
25
22
22
20
17
15
14
11 11
10 10
9
8
8 8
7
5 5 4
4
2
1
0
1930 1940 1950 1960 1970 1980 1990 2000 2010
Hispanic Black White White non-Hispanic
Note: Based on Decennial census. White category does not exclude those of Hispanic Origin. Hispanic
Origin can be of any race.
SOURCE: U.S. Census Population Division Census 2000 PHC-T-41. A Half-Century of Learning: Historical
Statistics on Educational Attainment in the United States, 1940 to 2000
14. Rising BA and above Expectations in all SES
groups NCES high school longitudinal studies
(HS&B, NELS, ELS) BA and above expectations
100
92.9
90
86.6
80
78.7
70.3
70
58.8 66.2
60
50
40
36.6
36.6
30
22.0
20
10
0
1975 1980 1985 1990 1995 2000 2005
Lowest SES Middle SES Highest SES
15. Composite achievement test score
and educational expectations of BA
100
95.5
90 88.3
87.7
79.4
80
71.8 68.3 75.7
70
60 55.8
59.5
50
47.5
43.7
40
40.6
32.3
30
27.2
20
19.4
10
0
1975 1980 1985 1990 1995 2000 2005
Lowest quartile Second quartile Third quartile Highest quartile All
16. Progress---One of TRIO Eligibility Criteria—First Generation
NCES Data on Percent of High School Students with at Least one
Parent with a BA Degree (Data from NLSY and ELS:2002)
1972 2002 % change
All 21% 38% 81%
AM. Indian or AK. Native 11% 29% 164%
Asian 23% 52% 126%
Black 8% 31% 288%
Hispanic or Latino 7% 21% 200%
White 22% 43% 95%
17. Inequality...
How much richer are the richest 20% in
each country than the poorest 20%?
Income gaps
How many times richer 9.7
are the richest fifth than
the poorest fifth? 8.5
8.0
7.2
6.7 6.8 6.8 7.0
6.1 6.2
5.6 5.6 5.6 5.7
5.2 5.3
4.8
4.3 4.6
3.4 3.7 3.9 4.0
ly
a
ia
l
e
um
d
ce
ar k
y
nd
UK
d
A
ai n
l
d
n
en
s
re
ay
da
ae
ga
an
ali
nc
nd
lan
pa
lan
an
US
st r
It a
po
ee
rw
ala
ed
na
Isr
r tu
l gi
Sp
nm
st r
rm
Fra
Ja
erl
r la
Au
Ir e
Fin
ga
Gr
No
Sw
Ca
Ze
Be
Po
Au
Ge
De
t he
izt
Sin
w
Sw
Ne
Ne
Wilkinson & Pickett, The Spirit Level www.equalitytrust.org.uk
18. Bachelor’s degree by age 24 by family income
quartiles:1970-2009-nationally about 30 percent
Source: US Census Bureau,
1970-2006-P20 table 14,
Report on School
Enrollment;
2007-2009 unpublished
Census tabulations.
Analysis by
Tom Mortenson;
Graph prepared by
the Pell Institute
of the Council for Opportunity
in Education (COE)
(October 2011)
19. Schools highly segregated by SES &--strong
inverse relationship of school poverty rates
and school proficiency scores
40.0 .
Figure reads: 1.6 percent of
35.0 the highest poverty schools
were in the 90 -100th
30.0 percentile in reading compared
with 36.1 percent
Highest of lowest poverty schools. Conversely, 0.5 percent of
25.0 Poverty--80% the lowest
and over free poverty schools (under 20% free
20.0 lunch) were at or below
lunch (17.1 %
the 10th percentile on reading proficiency ;
of schools) while 33.8 percent of highest
15.0 Lowest Poverty- poverty schools (80% or above free lunch)
-Under 20% free were at or below the 10th percentile .
10.0 lunch (17.5 % of
schools)
5.0
0.0
20. Health and social problems are worse
in more unequal countries
Index of health and social problems
Index of:
• Life expectancy
• Math & Literacy
• Infant mortality
• Homicides
• Imprisonment
• Teenage births
• Trust
• Obesity
• Mental illness
– incl. drug &
alcohol
addiction
• Social mobility
Wilkinson & Pickett, The Spirit Level www.equalitytrust.org.uk
21. More children drop out of High School
in more unequal US states
Wilkinson & Pickett, The Spirit Level www.equalitytrust.org.uk
22. Uncomfortable in the Inequality
Federal College Access Programs Designed
to Mitigate
Improved
college
Increased going and
student success
engagement, rates
Changes
(e.g. supplemental self
programs, increased esteem, caree
collaboration and
professional r goals, and
development among college
staff, use of
Federal data, rigorous course preparation
college enrollment and
success, strength
access finding and asset
program building among Changes in
students) school culture
23. Evidence from evaluations that these programs
can be effective (Random Assignment National
Evaluation of Upward Bound—6 year BA
attainment)
Difference 7.0 ****
14.1% 50% increase
UB/UBMS participation (TOT/CACE)
21.1%
15.2%
Difference 5.8***
UB/UBMS compared with other non-UB/UBMS 39 %increase
service only
21.0%
Difference 14.4***
Any pre-college with academic component 6.5% 223% increase
compared with no pre-college service
reported 20.9%
0.0% 5.0% 10.0% 15.0% 20.0% 25.0%
Comparison Treatment
24. However serve only small
minority of eligible population
100%
90%
80%
70%
60%
50%
98% 96% 91%
40% 96%
30%
20%
10%
2%
0% 4%
9%
Upward Bound 4%
% of free lunch Talent Search %
grades 9-12 of free lunch GEAR UP % of
grades 7 to 12 free lunch Student Support
grades 7 to 12 Services % of
PELL
undergraduates
Percent served Percent not served
25. Estimated tax impact of UB compared to estimated cost
based on average length of participation—4.9 times
more taxes based on earnings
$40,000
$35,000
$30,000
$25,000
$20,000
$15,000 $36,453
$10,000
$5,000 $7,350
$-
Cost of UB per average
length
Estimated additional
average lifetime taxes
from UB participation
26. Rate of Return to College Education
Increased over Last Two Decades
Source: Hill et al. The value of higher education , 2003, p. 15
27. Limiting factor—equilibrium of job
placement with educational system-BLS
new jobs by 2018
BA and Associates –27 percent BA and
Associates
23 % Percent BA or above
4.7 % Associates
5.7 % Postsecondary voc certificate
66.6 % On the job training --
Longer/medium/shorter (38.8 %)
29. Projections of Job Requirements Through
2018 (Carneval, Smith, Strohl 2010)
30. Developing 20/20 Vision on the 2020
Degree Attainment Goal, (A. Nichols
Pell Institute, 2011)
If could raise the bottom half to top
half—would easily meet goal
Top—already has almost 60 percent in BA
alone (59 percent)
Bottom half of country has 12 percent BA
attainment
If got bottom up to top then would have
2.9 million per year or 29 million more in
10 years
31. Dynamic Models
Levels----stocks—Population of college—set initial
level
Flows---number coming into system over time
Rates of flows---(observe what happens when change)
Auxiliary variables related to flows—
Can be exogenous or
Result of defined feedbacks from previous times
32. Rate of Flows—Increases and
Decreases
Increases Diversions away
High expectations of students Inequality measurements of achievement
themselves (direct and indirect)
Potential of college access programs to Raising standards to get out of high school
work if funded and implemented Reductions in school college access
more widely counselors
Economic downturn of families
Lack of high paying jobs for high
school grads Rising costs and decline in public share of
costs—increase family contributions
Social pressure to go to college— Fear of debt
presidential call—school climate-- Increase in percent of k-12 students who
peers have disabilities
Careers specification of Increase in percent of students who are in
requirements—certifications poverty---50 percent
Federal, state, local and non-profit Difficulty of peers who went to college
sector programs and encouragement finding jobs
Availability of credit and grants Low self efficacy to meet challenges
Remediation requirements
Focus on college readiness
33. Simple model of BA increase
Rate of
conferral
BA population
Conferrals--
births
Amount of BAjobs
--carrying capacity
34. Simple Model—BA growth
Population-----INTEG
(conferrals-births)
Initial value ---32 million
Population
Conferrals-births = Rate of 40 M
conferral*Population * (1- 35 M
Population * Amount of 30 M
jobs—carrying capacity) 25 M
Rate of conferral is a 20 M
constant in simulation (.3) 2008 2010 2012 2014
Time (Year)
2016 2018 2020
Population : Current
Amount of jobs—carrying
capacity = 38 million
35. Simple BA increase model
Changed carrying to 45 Change rate to .58—keep
million limits the same
Population Population
60 M 60 M
50 M 50 M
40 M 40 M
30 M 30 M
20 M 20 M
2008 2010 2012 2014 2016 2018 2020 2008 2010 2012 2014 2016 2018 2020
Time (Year) Time (Year)
Population : Current Population : Current
36. Increase rates to show system
dynamics
Overshoot and crash
Chaos—increase to 3.0
(bifurcating)---2.0
Population
Population
60 M
60 M
45 M
50 M
30 M
40 M
15 M
30 M
0
20 M 2008 2010 2012 2014 2016 2018 2020
2008 2010 2012 2014 2016 2018 2020 Time (Year)
Time (Year) Population : Current
Population : Current
37. A bit more complex
Rate of
conferral
Population
Rate of conferral Conferrals--
for top half births
Rate of conferral Amount of jobs
for bottom half --carrying capacity
Increase from
innovation
38. Models with innovation and
inequality in rates as auxiliaries
Estimating 25-64 year old See innovation allows job
college rate capacity to increase
Population base = 46.7
million 80 M
Population
Amount of jobs—carrying 70 M
capacity = 63 million + 60 M
Increase from innovation
50 M
Increase from innovation =
40 M
Population * .10 2008 2012 2016 2020
Time (Year)
2024 2028
Population : Current
Rate of top half = .588
Rate of bottom half = .120
39. Model 50 percent increase for the
bottom half
Features From .12 to .18
Innovation = Popuation * .10
Allows the carrying to move
from 63 million up to 70 80 M
Population
million
70 M
See leveling off due to 60 M
carrying constraints even
50 M
with innovation
40 M
Kept top half at .588 2008 2012 2016 2020
Time (Year)
2024 2028
Population : Current
40. More complex model
Rate of
conferral
College Population
Conferrals--
Rate of conferral births
for top half
Amount of jobs
Rate of conferral
--carrying capacity
for bottom half
Increase from Increase from
Federal program innovation reduction in
Increase in parents
inequality
impact who have college
41. Reduction in inequality due to impact of
increased reach of college access programs and
parent education
Increased ED levels More jobs
College Population Amount of jobs --carrying capacity
80 M 80 M
70 M 77.5 M
60 M 75 M
50 M 72.5 M
40 M 70 M
2008 2012 2016 2020 2024 2028 2008 2012 2016 2020 2024 2028
Time (Year) Time (Year)
College Population : Current "Amount of jobs --carrying capacity" : Current
Editor's Notes
Worked as evaluator of federal programs—seeped in experimental and quasi-experimental linear modeling of program effectivenessInterest for a long time is systems work---but qualitative thinking ---no attempt to specify these relationships mathematicallySocial systems---Multiple loop non-linear feedback systems---dynamic mathematical modelsExplore using tools of dynamic systems modeling in evaluation of policy decisions potential outcomes.Using Vensim Software Tool—much like Stella
High inequality—schools very segregated by SES and growing more soLow and decreasing proportion of public financing—roughly –53 percent family/student; 8 percent feds; 38 percent state and localRising costs—higher than inflation—reduction in Pell buying power---$12,000 in 1973 when started in todays dollars compared to $5000 Debt burden increasing in amountLink to employment-major determinant of income differentialsGrowth in poor (50 percent eligible for free or reduced in 2009)Growth in children with diagnosed disabilities (11 percent---20 percent in high poverty schools)Growth in at-risk minorities
Levels are also known as stocks, accumulations, or state variables. Levels change their values by accumulating or integrating rates. This means that the values of Levels change continuously over time even when the rates are changing discontinuously. Rates, also known as flows, change the value of levels. The value of a rate is not dependent on previous values of that rate; instead the levels in a system, along with exogenous influences, determine the values of rates. Intermediate concepts or calculations are known as auxiliaries and, like rates, can change immediately in response to changes in levels or exogenous influences.