1. Closing equity gaps and improving student completion of transfer-level math and English California Acceleration Project CCCCO Webinar December 5, 2016
Myra Snell
CAP Co-Founder
Math Professor
Los Medanos College
Contact us at
I will be reiterating and expanding upon points highlighted in the PPIC research and then touching on promising reforms both in CA and in other states

2. Students labeled “underprepared” much less likely to achieve their goals
This reinforces PPIC findings that 80% of students are placed into dev. ed. and these students are less likely to transfer

3. Students labeled “underprepared” much less likely to achieve their goals
Statewide, more than three-quarters of degree, certificate or transfer-seeking students are classified “unprepared”
This reinforces PPIC findings that 80% of students are placed into dev. ed. and these students are less likely to transfer

4. Not prepared?
Why are students who are labeled “underprepared” less likely to transfer or complete the Associate’s Degree?
Is it because they are not prepared for college? Or are there other reasons?
How do we determine who is, and who is not, ready for college level course work?
If someone is not prepared, what interventions will improve his/her chances of completing transfer requirements in English and Math?

5. How do we determine who is “college ready”?
Sample Item: Accuplacer “Sentence Skills” Test
Writing a best seller had earned the author a sum of money and had freed him from the necessity of selling his pen for the political purposes of others.
Rewrite, beginning with The author was not obliged
The new sentence will include
A) consequently he earned
B) because he had earned
C) by earning
D) as a means of earning

6. Are you college ready in math?

7. Placement tests do a poor job of determining who is college ready.
In most CA CCs “college readiness” is determined by a placement test. But these tests do a poor job identifying who will – and will not – do well in college.
Placement tests have low predictive validity, e.g. Accuplacer scores in English explain about 1% of the variation in course grades; in math less than 4% (Cal-Pass data).
PPIC study documents that testing is the primary means for determining college-readiness.

8. Placement tests do a poor job of determining who is college ready.
In most CA CCs “college readiness” is determined by a placement test. But these tests do a poor job identifying who will – and will not – do well in college.
Placement tests have low predictive validity, e.g. Accuplacer scores in English explain about 1% of the variation in course grades; in math less than 4% (Cal-Pass data).
Placement tests have high error rates: Severe under-placement error is three times more prevalent than over-placement error (those placed into remediation who could have earned a B or better in a college course vs. those placed into college course who fail) (Scott-Clayton, 2012).
PPIC study documents that testing is the primary means for determining college-readiness.

9. Placement tests do a poor job of determining who is college ready.
In most CA CCs “college readiness” is determined by a placement test. But these tests do a poor job identifying who will – and will not – do well in college.
Placement tests have low predictive validity, e.g. Accuplacer scores in English explain about 1% of the variation in course grades; in math less than 4% (Cal-Pass data).
Placement tests have high error rates: Severe under-placement error is three times more prevalent than over-placement error (those placed into remediation who could have earned a B or better in a college course vs. those placed into college course who fail) (Scott-Clayton, 2012).
In math tests test the wrong stuff: Fewer than 10% of the topics in Elementary and Intermediate Algebra are needed for the study of Statistics, yet tests of these skills block students’ access to Statistics courses.
PPIC study documents that testing is the primary means for determining college-readiness.

10. Are you college ready? Depends on where you want to go to college
At recent CAP events CA community colleges examined their placement data:
In English (35 colleges) direct placement into transfer-level English varied from 8% to 66%.
In Math (20 colleges) direct placement into transfer-level Math varied from 1% to 48%.

11. Placement policies have a disproportionate impact
The lower a student places in the developmental sequence, the less likely he or she will complete a transfer-level math or English course. Students of color are much more likely to be placed into lower levels of developmental education.

12. The Problem with Multi-Level Remedial Sequences: English-Writing in California
Students’ Starting Placement
English-Writing
% Completing Transfer-Level English in 3 Years
One Level Below
48%
Two Levels Below
34%
Three or more Levels Below
19%
Data on racial demographics from: Perry, M.; Bahr, P.R.; Rosin, M.; & Woodward, K.M. (2010). Course-taking patterns, policies, and practices in developmental education in the California Community Colleges. Mountain View, CA: EdSource.
Statewide data, Basic Skills Cohort Tracker, Fall 2009-Spring 2012

13. The Problem with Multi-Level Remedial Sequences: English-Writing in California
Students’ Starting Placement
English-Writing
% Completing Transfer-Level English in 3 Years
One Level Below
48%
Two Levels Below
34%
Three or more Levels Below
19%
Data on racial demographics from: Perry, M.; Bahr, P.R.; Rosin, M.; & Woodward, K.M. (2010). Course-taking patterns, policies, and practices in developmental education in the California Community Colleges. Mountain View, CA: EdSource.
Across CA, students of color 2-3 times more likely to begin in lowest levels than white students
Statewide data, Basic Skills Cohort Tracker, Fall 2009-Spring 2012

14. The Problem with Multi-level Remedial Sequences: Math in California
Students’ Starting Placement
Mathematics
% Completing Transfer-Level Math in 3 Years
One Level Below
35%
Two Levels Below
15%
Three or more Levels Below 6%
Statewide data, Basic Skills Cohort Tracker, Fall 2009-Spring 2012

15. The Problem with Multi-level Remedial Sequences: Math in California
Students’ Starting Placement
Mathematics
% Completing Transfer-Level Math in 3 Years
One Level Below
35%
Two Levels Below
15%
Three or more Levels Below 6%
Across CA, more than half of Black and Hispanic students in remedial math begin here
Statewide data, Basic Skills Cohort Tracker, Fall 2009-Spring 2012

16. Inequity in placement drives inequity in longer term college completion
According to a study by Greg Stoup (Senior Dean Research and Planning, CCCCD), initial placement explains 50-60% of the inequities in degree or transfer completion in the Contra Costa Community College District.

17. What can we do to improve completion and reduce equity gaps?
High-Leverage Strategy #1: Let them in! Broaden access to transfer-level course work by changing placement criteria. In other words, address the severe underplacement problem with more accurate placement criteria.

18. Broadening Access to Transfer-level English Improves Outcomes: Butte College
While setting cut scores for a new placement test, Butte College discovered that twice as many students were now assessing as “college ready” in English. Placement into college English: Increased from 23% to 48% of incoming students

19. Broadening Access to Transfer-level English Improves Outcomes: Butte College
While setting cut scores for a new placement test, Butte College discovered that twice as many students were now assessing as “college ready” in English.
Placement into college English:
Increased from 23% to 48% of incoming students
One-Year Completion of College English
Tripled for African American students
Doubled for Hispanic and Asian students
1.6 times higher for white students
Achievement Gaps:
The gap between White and Black students’ completion of college English was cut nearly in half.
Under-Estimation of Students:
40% of the students who previously would have been placed into remediation earned As and Bs in college English

20. Multiple Measures = More accurate placement
MMAP (Multiple Measures Assessment Project): data driven use of multiple measures. Placement rules describe students who have a 70% chance of passing the targeted course (based on an analysis of tens of thousands CA community college students) High school GPA is the best indicator of student performance in college. Example: Students are predicted to pass college English at a rate of at least 70% if they have a high school GPA of 2.6 or higher Statewide, 61% of students met this criterion.
Only 3 colleges out of 35 at CAP events were allowing 61% or more students to place directly into college English. This suggests that our current placement policies are severely underplacing students. This is a big problem when we consider the high attrition in developmental course sequences.

21. Multiple Measures = More accurate placement
MMAP (Multiple Measures Assessment Project): data driven use of multiple measures. Placement rules describe students who have a 70% chance of passing the targeted course (based on an analysis of tens of thousands CA community college students) High school GPA is the best indicator of student performance in college. Example: Students are predicted to pass college English at a rate of at least 70% if they have a high school GPA of 2.6 or higher Statewide, 61% of students met this criterion. Underplacement in CA: Only 3 out of 35 colleges at CAP events allowed 61% or more of students to place directly into college English.
Only 3 colleges out of 35 at CAP events were allowing 61% or more students to place directly into college English. This suggests that our current placement policies are severely underplacing students. This is a big problem when we consider the high attrition in developmental course sequences.

22. Broadening Access to Statistics: College of the Canyons
Old Model:
Direct placement into Statistics if a student meets any of the following criteria:
Accuplacer (weighted 90%) AND
Multiple Measure Questions (weighted 10%)
Results: 14% placed into college level math.
Use of disjunctive “or” and self-reported HS information.
There are numerous studies on the accuracy of self-reporting.

23. Broadening Access to Statistics: College of the Canyons
Old Model:
Direct placement into Statistics if a student meets any of the following criteria:
Accuplacer (weighted 90%) AND
Multiple Measure Questions (weighted 10%)
Results: 14% placed into college level math.
New Model:
Direct placement into Statistics if a student meets any of the following criteria: 
HS GPA of 3.0 or higher   OR
 HS GPA of 2.7 and Algebra 2 or higher with C or higher  OR
 HS GPA of 2.7 and Algebra 1 with B- or higher  OR
 Algebra 2 with B- or higher
Use of disjunctive “or” and self-reported HS information.
There are numerous studies on the accuracy of self-reporting.

24. Broadening Access to Statistics: College of the Canyons
Old Model:
Direct placement into Statistics if a student meets any of the following criteria:
Accuplacer (weighted 90%) AND
Multiple Measure Questions (weighted 10%)
Results: 14% placed into college level math.
New Model:
Direct placement into Statistics if a student meets any of the following criteria: 
HS GPA of 3.0 or higher   OR
 HS GPA of 2.7 and Algebra 2 or higher with C or higher  OR
 HS GPA of 2.7 and Algebra 1 with B- or higher  OR
 Algebra 2 with B- or higher
Results: 72% are eligible to take Statistics. The average placement has increased two levels.
Use of disjunctive “or” and self-reported HS information.
There are numerous studies on the accuracy of self-reporting.

25. Broadening Access to Statistics: College of the Canyons
Old Model:
Direct placement into Statistics if a student meets any of the following criteria:
Accuplacer (weighted 90%) AND
Multiple Measure Questions (weighted 10%)
Results: 14% placed into college level math.
New Model:
Direct placement into Statistics if a student meets any of the following criteria: 
HS GPA of 3.0 or higher   OR
 HS GPA of 2.7 and Algebra 2 or higher with C or higher  OR
 HS GPA of 2.7 and Algebra 1 with B- or higher  OR
 Algebra 2 with B- or higher
Results: 72% are eligible to take Statistics. The average placement has increased two levels.
COC uses self-reported high school information.
Use of disjunctive “or” and self-reported HS information.
There are numerous studies on the accuracy of self-reporting.

26. Lessons Learned: Los Medanos College
Implementation of multiple measures: Assessment Center staff with help from A&R checked all transcripts to determine placement. A glitch: students who did not bring a transcript to their ed. planning session were not able to complete ed. plans or to register. For expediency, they were instructed to test and test score was used in placement and ed. planning. The result: approximately 70% of students placed into lowest levels of remediation were eligible to take transfer-level courses or an accelerated algebra option one level below IF they had showed transcripts. The solution: We plan to use self-reported information in the future with random spot checks to assess accuracy of self-reported information.
Growing body of evidence that self-reported HS information is highly accurate.

27. Summary: The Power Moves for Reforming Placement
High School GPA – the strongest indicator of how students will perform in college is how they performed in all their classes during four years of prior schooling
“Disjunctive” placement – students qualify for college-level courses based on high school measures OR test score, whichever is higher
Self-reported information – high school GPA and coursework provided in response to questions during assessment process so that all students have access to multiple measures placement

28. What about students who really are not ready for college courses in Math and English?
What happens to them now?
Students scoring below “college-level” are required to enroll in 1-4 remedial courses in math and/or English
In lower-level remedial classes, tasks are more basic, less challenging than in higher levels
In math, remediation is a repeat of K-12 math content through Algebra II, students need to make it through all of this content before taking any college-level math course

29. Attrition among Students Placed into Remediation: A Structural Problem
Students placed 2 levels below transfer-level English/Math face 5 “exit points” (if they choose to enroll):
If they enroll, do they pass the first dev. ed. course?
If they pass, do they enroll in the next dev. ed. course?
If they enroll, do they pass the second dev. ed. course?
If they pass, do they enroll in the transfer-level course?
If they enroll, do they pass the transfer-level course?
Students placed 3 levels down face 7 exit points.

30. Illustration: Los Medanos College
Students beginning two levels below Transfer-level Math:
Do they enroll in the first course?
?? %
If they enroll, do they pass the first dev. ed. course (Elem. Alg.)?
69%
If they pass, do they enroll in the next course (Int. Alg.)?
90%
If they enroll, do they pass the second dev. ed. course (Int. Alg.)?
85%
If they pass, do they enroll in the transfer-level course?
71%
If they enroll, do they pass the transfer-level course?
76%
(0.69)(0.90)(0.85)(0.71)(0.76)= 28%
Fall 2012 Cohort. Students tracked from their first developmental math enrollment and followed for all subsequent math enrollments for 3 years. Pass rates includes students passing on first or repeated attempts within timeframe. Basic Skills Cohort Tracker, DataMart.

31. Illustration: Los Medanos College
What if we could dramatically improve course success rates and persistence rates?
(0.69)(0.90)(0.85)(0.71)(0.76)= 28%

32. Illustration: Los Medanos College
What if we could dramatically improve course success rates and persistence rates?
(0.69)(0.90)(0.85)(0.71)(0.76)= 28%
(0.80)(0.90)(0.85)(0.71)(0.76)= 33%

33. Illustration: Los Medanos College
What if we could dramatically improve course success rates and persistence rates?
(0.69)(0.90)(0.85)(0.71)(0.76)= 28%
(0.80)(0.90)(0.85)(0.71)(0.76)= 33%
(0.85)(0.90)(0.85)(0.90)(0.85)= 50%
Phenomenal (and unrealistic) improvements in course success rates and persistence rates do not move the completion needle enough.

34. What about students who really are not ready for college courses in Math and English?
What does the research suggest is a better approach? High Leverage Strategy #2: Concurrent Support Students previously classified as “below transfer level” enroll in transfer-level course with support: Mandatory co-requisite support course, mandatory SI-instruction, embedded tutoring, etc. High Leverage Strategy #3: Redesigned Remedial Courses Multi-level sequences in English and math are replaced with a single accelerated course that is well aligned with the transfer-level requirements in students’ chosen pathway.
In math, both of these strategies align remediation with the math course or courses students will take to meet lower division requirements for transfer. Algebra-based remediation for math intensive majors; quantitative reasoning and skills for statistics for everyone else.

35. Co-requisite models dramatically increase completion of college-level courses
Only co-requisite models of remediation are offered at Tennessee public colleges and universities
Fall 2015 Statewide Data: Tennessee Board of Regents (TBR)

36. How do the low-scoring students do in co-requisite models?

37. How do the low-scoring students do in co-requisite models?

38. Cuyamaca College: Co-requisites in math
No student places lower than intermediate algebra with a concurrent support course. (no more arithmetic, prealgebra or elementary algebra)
All transfer-level math courses have a linked concurrent support course option. If a student does not test into transfer-level, multiple measures can place him/her into transfer level with concurrent enrollment in the support course.
All students are eligible for Statistics with support.

39. High-Leverage Strategy #3
Redesigned Remediation: Multi-level sequences in English and math are replaced with accelerated courses that are well aligned with the transfer-level requirements in students’ chosen pathway.
Add Solano slide?

40. Redesigned remedial courses dramatically improve student completion of transfer-level courses
RP Group’s study of 16 CAP colleges. Used statistical methods to control for pre-existing differences between students (2,489 accelerated students: 1,836 in English, 653 in math)
In English, students’ odds of completing a transfer-level course were 2.3 times greater in high-impact models of acceleration than students in traditional remediation
In math, students’ odds of completing a transfer-level math course were 4.5 times greater than students in traditional remediation
Add summary slide #2 strategy

41. CAP Math Pathways: Eliminating the Achievement Gap for African-American Students
Completion of Transfer-Level Math (within years)
The first 8 colleges piloting accelerated statistics pathways in CAP
(Descriptive data, no statistical controls)

42. CAP Math Pathways: What about the lowest students?
Completion of Transfer-Level Math (within years)
Students at 3-4 Levels Below Transfer in Math
(Descriptive data, no statistical controls)

43. RP Group Evaluation of CAP: Benefit to All Students
Significant completion gains among all student subgroups studied, including:
all ethnic groups
low-income students
students who had taken ESL courses
students who had not graduated from high school
students with low GPAs
students with disabilities
Talking point:
Acknowledge the common faculty concern that SOME students need more time, worry about leaving people behind…
But MULTIPLE studies have confirmed what the RP Group found – that there is no group of students we can identify who have better outcomes in a non-accelerated curriculum – the Tennessee statewide co-requisite data, the CCRC study of Chabot College’s single-semester accelerated English course, internal data from Chabot and Las Positas Colleges...any potential gains we might see in course success rates are offset by the impact of attrition across exit points. Fewer students complete the college level course.

44. These strategies change lives Capacity Unleashed: The Faces of Community College Math Pathways Accelerationproject.org
Add photos and names: Chris, prison guy, Terrell or Kevin, Donna Rose

45. Three High-Leverage Strategies
Changing Placement Policies: Colleges broaden access to transfer-level courses, and make access more equitable, by adjusting cut scores, using robust multiple measures, and requiring algebra-based testing and remediation only for access to courses that require substantial algebra. Implementing Co-requisite Models: Students classified as “below transfer level” are allowed to enroll in a transfer-level course with extra concurrent support, saving them at least a semester of stand-alone remediation and reducing their chances of dropping out (e.g., “1A-plus” models: students co-enroll in English 1A and 2 additional units with the same instructor). Redesigning Remedial Courses: Multi-level sequences in English and math are replaced with accelerated courses that are well aligned with the transfer-level requirements in students’ chosen pathway.
Talking points – these strategies have two things in common:
they shorten students path through transfer-level courses and eliminate exit points, and
all are grounded in the belief that students are capable of doing higher level work than we have traditionally assumed