Access versus Success: policy insights from a study funded

Download Report

Transcript Access versus Success: policy insights from a study funded

A Quantitative
Approach to the
Identification of Factors
that Contribute to
Graduation & Attrition
at an Hispanic Serving
Institution
The University of Texas at El Paso
Center for Institutional Evaluation, Research
and Planning
CIERP gratefully acknowledges the support of
Lumina Foundation for Education
Presentation Outline
UTEP’s Efforts to Ensure Student Success
• UTEP’s context
• Efforts over last 20 years and their impact
• Current challenges
 Student Success Project: methodology
Findings, implications and next steps
UTEP’s Context
El Paso, Texas
• El Paso County-3rd
poorest large county in the
US1
• Population: 724,000
• 81% Hispanic
• Border community, very
dynamic flow of residents
and students across the
border
• Limited educational
opportunities
1
2005 American Community Survey of the US Census
UTEP’s Efforts over last
20 years
President Diana Natalicio has
rededicated UTEP’s mission to
ensure the widest possible
access to all students from the
region, and to focus on serving
the El Paso area
Institution has made efforts to
ensure success at all levels of the
pipeline, with impressive results
• K-12
• Admissibility/ Affordability
• Student engagement
UTEP Demographics
Enrollment by Race/Ethnicity
N
%
White Non-Hispanic
2277
11.8%
Black Non-Hispanic
480
2.5%
Hispanic
13947 72.4%
Asian/Pacific Islander 240
1.3%
Am. Indian or Alaskan 46
0.2%
International*
2132
11.1%
Total Enrollment by Residence
N
%
El Paso County 16349 84.9%
New Mexico
244
1.3%
Mexico
1798
9.3%
Other Int’l
430
2.2%
*includes Mexican Nat’l. students
Percent of financial aid awardees with family income of $20,000 or less:
Percent of UTEP students with reported family income of $20,000 or less:
43%
33%
Nationally:
% of students with family income of less than $20,000 at large public
1
research (doctoral) universities:
10%
% of students with family income of less than $20,000 at small &mid-sized
1
private colleges and universities:
12%
2
% of students with family income less than $20,000 at community colleges: 29%
1Council of Independent Colleges: http://www.cic.edu/makingthecase/data/access/income/index.asp
2Lumina Foundation Focus, Fall 2005, p.5
El Paso Collaborative for Academic Excellence
Founded in 1991; partners include the National Science Foundation,
the U.S. Department of Education and The Pew Charitable Trusts, in
addition to businesses and organizations in the community and
throughout Texas
Goals of the Collaborative
 Ensure academic success for all
students, K-16
 Ensure that all students
graduate from high school
prepared to succeed in a fouryear college or university
 Close achievement gaps among
different groups of students
Completion of Recommended High School Program
or Higher, El Paso Districts & Statewide
Class of 2004
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
91%
72%
All Students
90%
72%
Hispanic
El Paso Urban Districts
93%
91%
73%
65%
White
African
American
Statewide
Source: Texas Education Agency, http://www.tea.state.tx.us/adhocrpt/GraduateReports
Affordability
UT System Full-time Undergraduate Students
with Need-Based Grant Aid, AY 2005-06
Average In-State Total
Academic Cost,
Fall 2005 - Spring 2006
combined w/ fees
% Students
Receiving NeedBased Grant Aid
Average % Discount
Arlington
$5910
37.0%
71.6%
Austin
$7288
46.8%
80.8%
Brownsville
$3709
57.9%
65.1%
Dallas
$6838
30.3%
61.5%
El Paso
$4984
47.4%
100.00%
Pan American
$3605
65.5%
100.00%
Permian Basin
$4282
36.3%
54.3%
San Antonio
$6016
47.0%
64.3%
Tyler
$4671
42.0%
89.1%
$5093
46.7%
76.9%
UT System
Institution
Average
Source: UT System Fast Facts, 2007
UTEP’s Institutional Successes
•
The National Survey of Student Engagement and
the American Association for Higher Education
identified UTEP as one of 20 colleges and
universities that was “unusually effective in
promoting student success”.1
UTEP is identified as only one of six Model
Institutions for Excellence in the nation by the
National Science Foundation for its success in
creating educational opportunities for non-traditional
students.
UTEP’s College of Engineering was identified as
the top engineering school for Hispanics by
Hispanic Business Magazine. The magazine says
UTEP “is changing the face of engineering and
producing highly trained graduates heavily recruited
by the industry’s leading companies”.2
•
•
1
1989 UTEP Alumnus
Danny Olivas, NASA
astronaut scheduled for
flight on the shuttle
Atlantis in June 2007
NSSE Institute for Effective Educational Practice, Project DEEP Final Report, p. 4
Hispanic Business, September 2006
2
Same Institution Gradation Rate
Despite All the Success,
UTEP has More Work to Do
35%
29.40%
30%
25.10%
25%
20%
14.80%
15%
16.70%
10%
5%
2.50%
3.90%
0%
4yr Fall 4yr Fall 5yr Fall 5yr Fall 6yr Fall 6yr Fall
1997
2001
1996
2000
1995
1999
Entering Cohort
Goals of the UTEP Student
Success Project Funded by
Lumina Foundation
 Identify factors that affect
students’ success—timely
progress toward a degree.
 Identify & implement
strategies to improve the
success of students at UTEP.
 At this preliminary stage, there
were two study questions and
quantitative approaches.
Research Questions
1. Predictors of Success
What factors explain graduation within
6 years at the University of Texas at El
Paso?
2. Predictors of Risk
What factors explain why students fail
to graduate within 6 years?
Factors that Affect Student Success
ENROLLMENT (FT/PT)
FINANCIAL
AID
SOCIO-ECONOMIC STATUS
NUMBER OF DROPS,
WITHDRAWLS &
INCOMPLETES PER
SEM.
ADVISING QUALITY
EXTERNAL
RESPONSIBILITIES
OTHER
THAN WORK
(e.g., CHILDREN)
FACULTY INTEGRATION/HAPPINESS
QUALITY OF HIGH
SCHOOL PREPARATION
LOW GPA
READING/WRITING PLACEMENT
INTENTIONS (NOT ASPIRATIONS)
MATH PLACEMENT
WORK HOURS (FT/PT)
NUMBER OF HIGHER ED
INSTITUTIONS
ATTENDED (TRANSFERS)
External Factors
MOTIVATION
CLARITY OF DEGREE
REQUIREMENTS
NUMBER OF
CREDIT HRS
EARNED
PER SEM.
INTEGRATION INTO UNIVERSITY
ENVIRONMENT (EXTRA-CURRICULARS)
G
R
A
D
U
A
T
I
O
N
R
A
T
E
INTEGRATION INTO UNIVERSITY
ENVIRONMENT (FACULTY/STAFF)
Shared External/Institutional Factors
Institutional Factors
Student-Based Factors
Data and Sample
Cohort Entering Students from Fall 1999 and
Fall 2000 (Sample size = 2,065)
Data provided clean, reliable
set to examine the 6 yr.
graduation rate
Undergraduates only
1st time (non-transfer)
students
Full-time students only
Research Approach:
Development of a
Model to Predict Graduation in 6 Years
Logistic Regression
• Conducted with the binary prediction of
Graduation-Yes (1) or Graduation-No (0)* within
six years
• Conducted in five steps
• Variables entered sequentially, from demographics
to college academic performance
*Includes “not yet” graduated
23 Variables Considered as Predictors of
Graduation:
• Age (standardized)
• Educational level of parents
• Hrs. per week spent working
(anticipated for 1st term, self-report) 1
• Number of dependents (children &
other family members) 1
• ACT Scores (standardized)
• Math Placement Level (BANM)
• Reading Placement Level (BANR)
• Writing Placement Level (BANW)
• English Placement Level (BANE)
• High School Percentile Rank
• 1st Term GPA standardized (GPA for
which we have all students’ data;
“past behavior is a predictor of future
behavior”)
• Number of classes failed in 1st term
1
• Personal perception that one will
drop out before graduation1
• Personal perception that one will
need to study harder at UTEP than in
high school to get good grades1
• Personal perception that it is
important to prepare for class1
• Personal perception that one will
change majors at least once1
• Personal perception that it is
important to prepare for class1
• Average number of credit hours
attempted (standardized)
• Need (based on Household Income)
• Grant- Amount Paid
• Scholarship- Amount Paid
• Student Loan- Amount Paid
• Work-Study- Amount Paid
New Student Survey administered to all new students annually by UTEP-CIERP
Understanding Logistic
Regression Outputs
Statistically Significant Predictors
Odds Ratio
P<
• Gender (Female)
1.723
.001
• Educational level of parents*
 Both parents graduated from college
1.781
.05
• Dependent (do not have dependent)
1.858
.05
________________________________________________________
* Reference group is Neither parent ever attended college.
Note: Nagelkerke R Square is .040.
Percentage correctly classified is 66.3.
Income and age did not qualify into the regression equation.
Modeling Factors that Explain
At-Risk Group Membership
Multinomial Logistic Regression is ideal for
examining factors that help predict membership in
a defined category when the outcome is more than
binary (i.e. graduating or not).
• 7 Categories versus 2 Categories considered in the
model.
The model produced offers information about
institutional interventions for particular groups.
Objective: Resources can be more effectively
targeted toward these at-risk groups.
Identifying At-Risk Students
1.
1st Semester Leavers: Students who left UTEP in the first
semester and never returned. These individuals are arguably a
very high-risk group.
2.
1st Year Leavers: Students who left UTEP in the first year and
never returned; arguably also a high-risk group.
3.
2nd Year Leavers: Students who left at some point in the
second year and never returned to UTEP.
4.
3rd Year Leavers: Students who left at some point in the third
year and never returned to UTEP.
5.
Sporadic Leavers: Students who left at some point in their
college career and returned, yet did not graduate.
6.
Persisters: Students who were continually enrolled at UTEP
between the Fall of 1999 and the Fall of 2005, yet did not
graduate.
Understanding Multinomial Logistic
Regression Outputs
Significant Predictors
Odds Ratio
P<
Failing one or more classes
Working more hours
Lower High school percentile
2.738
1.381
1.364
.001
.001
.01
Higher Likelihood of Sporadic Leaving
---------------------------------------------------------------- 1.00 ---------------------------Lower Likelihood of Sporadic Leaving
Higher Math Placement Score
Average Credits Attempted
Increasing 1st Term GPA
.759
.543
.363
.05
.001
.001
_
Descriptive Stats
College Graduation within 6 Years
by High School Rank Percentile
Percent of Students who Graduated
from UTEP within Six Years
60
Percentage graduated
50
49.1
40
30
20.8
20
9.7
10
0
top 25%
75% to 50%
High School Rank Percentile
less than 50%
Descriptive Stats
Percent of Students who Graduated from UTEP
within Six Years
College Graduation within 6 Years
by Math Placement Level
80
75
Percentage graduated
70
60
52.5
50
37.1
40
30
23.3
20
10
0
Level 1-- Two
Level 2-- One
Level 3--College Level 4--College
Semesters Below Semester Below
College
College
Math Placement Level
Descriptive Stats
Percent of Students who Graduated from UTEP within
Six Years
College Graduation within 6 Years
by Hours Worked per Week
45
40
39.5
percentage graduated
38.7
35
30
28.1
25
20
14.9
16.7
15
10
5
0
Not
1-19 hours
20-29
Employed per week hours per
week
30-39
40+ hours
hours per per week
week
Anticipated Hours of Employment per Week indicated on
the UTEP New Student Survey
Descriptive Stats
Pecent of Students who Graduated from UTEP
within Six Years
College Graduation within 6 Years
by Household Income
45
percentage graduated
37.4
38.2
50 to 65k
65k or
more
40
35
34.5
35.4
20 to 35k
35 to 50k
29.9
30
25
20
15
10
5
0
20k or less
Houshold Income
Descriptive Stats
Percentage of Students who Graduated within
Six Years
College Graduation within 6 Years
by Classes Failed-1st Semester
50
45.4
percentage graduated
45
40
35
30
25
20
14.8
15
10
5
1
0
No failed classes
failed one class
failed two or more classes
Number of classes failed
Distribution of GPA
250
200
Frequency
100
50
0
250
Not Graduated
200
150
100
50
0
0.00
2.00
First term GPA
4.00
Graduation Indication
Graduated
150
Distribution of ACT
200
ACT 18
100
Frequency
50
0
200
Not Graduated
150
100
50
0
5.00
10.00
15.00
20.00
ACT
25.00
30.00
35.00
Graduation Indication
Graduated
150
Distribution of ACT for different
Income groups
ACT 18
Descriptive Statistics of ACT
Scores by Income Group
Income group
N
Mean
Median
Std.
Deviation
Less than 20k
553
18.06
18.00
3.26
20k to 35k
453
18.40
18.00
3.20
35k to 50k
246
19.37
19.00
3.41
50k to 65k
122
19.76
19.00
3.60
65k or more
95
20.51
20.00
3.73
1,469
18.68
18.00
3.41
Total
Step 1: Demographics
In this step, only demographic
variables were considered:
•
•
•
•
•
Gender
Income
Educational level of parents
Having a dependent
Age
Step 1: Logistic Regression
Statistically Significant Predictors
Odds Ratio
P<
• Gender (Female)
1.723
.001
• Educational level of parents*
 Both parents graduated from college
1.781
.05
• Dependent (do not have dependent)
1.858
.05
________________________________________________________
* Reference group is Neither parent ever attended college.
Note: Nagelkerke R Square is .040.
Percentage correctly classified is 66.3.
Income and age did not qualify into the regression equation.
Step 2: High School Preparation
Variables considered in this step:
•
•
•
•
•
•
High school class rank
Math Placement score
ACT /SAT score
Writing Placement score
Reading Placement score
English Placement score
Step 2: Logistic Regression
Statistically Significant Predictors
•
•
Math Placement Score*
 Level 2
 Level 3
 Level 4
Gender (Female)
Odds Ratio
P<
1.401
2.195
5.564
1.672
.05
.001
.05
.001
-------------------------------------------------------------------------------------------------•
High school percentile**
 Between 75 and 50
 Less than 50
0.355(2.82)
0.159(6.29)
.001
.001
________________________________________________________
* Reference group is level 1.
** Reference group is top 25.
Note: Nagelkerke R Square is .216.
Percentage correctly classified is 72.2.
Educational level of parents and dependent did not qualify for inclusion in the
regression equation.
Step 3: Student Commitment
 Variables considered in this step are
personal perceptions1 that:
• One will drop out before graduation
• One will need to study harder at UTEP than in high
school to get good grades
• It is important to prepare for class
• One will change majors at least once
• One will need more than four years to graduate
• Confident one will graduate
1
Data were collected from the UTEP New Student Survey
Step 3: Logistic Regression
Statistically Significant Predictors
• Math Placement Score*
 Level 2
 Level 3
 Level 4
Odds Ratio
P<
1.408
2.227
5.384
.05
.001
.05
• Gender (Female)
1.537
.001
• Do not need more than 4 years to graduate
1.212
.05
--------------------------------------------------------------------------------------------------------• Disagree it is important to prepare for class
0.729(1.37)
.05
• High school percentile**
 Between 75 and 50
 Less than 50
0.361(2.77)
0.144(6.94)
.001
.001
______________________________________________________________
* Reference group is level 1.
** Reference group is top 25.
Note: Nagelkerke R Square is .229.
Percentage correctly classified is 71.6.
The rest of the variables considered in step 3 did not qualify to the regression equation.
Step 4: First Semester
Performance
Variables considered in this step:
•
•
•
•
GPA
Credits attempted
Hours worked per week
Number of classes failed
Step 4: Logistic Regression
Statistically Significant Predictors
•
•
Odds Ratio
P<
GPA
2.733
.001
Math Placement Score*
 Level 2
1.237
.05
 Level 3
1.512
.001
 Level 4
2.954
.10
•
Gender (Female)
1.443
.05
•
Do not need more than 4 years to graduate
1.267
.05
•
Credits attempted
1.165
.05
---------------------------------------------------------------------------------------------------------------------------------•
Disagree it is important to prepare for class
0.698(1.43)
.05
•
High school percentile**
 Between 50 and 75
0.643(1.56)
.05
 Less than 50
0.379(2.64)
.001
•
Hours worked per week***
 Working 30-39 hours per week
.448(2.23)
.05
•
Number of classes failed****
 Failing one class
.530(1.89)
.001
 Failing two or more classes
.105(9.52)
.001
______________________________________________________________________________
* Reference group is level 1. ** Reference group is top 25. *** Reference group is not working.
**** Reference group is not failing a class.
Note: Nagelkerke R Square is .40.
Percentage correctly classified is 76.6.
Significant Predictors (in Bold)





















Gender
Income
Educational level of parents
Having a dependent
Age
High school class rank
Math Placement score
ACT /SAT score
Writing Placement score
Reading Placement score
English Placement score
Will drop out before graduation
Will need to study harder at UTEP than in high school to get good grades
It is important to prepare for class
Will change majors at least once
Do not need more than four years to graduate
Confident that one will graduate
GPA
Credits attempted
Hours worked per week
Number of classes failed
Identifying At-Risk Students (cont.)
Fall 1999 & Fall 2000 Cohort
Graduated in </= 6 Years
Left UTEP in the 1st Semester
Left UTEP in the 1st Year
Left UTEP in the 2nd Year
Left UTEP in the 3rd Year
Sporadically Left UTEP
Persisters
Total
N
%
419 33.0
75
5.9
171 13.5
133 10.5
45
3.5
347 27.3
81
6.4
1271 100.0%
Summary of the Multinomial
Model
Variable
First
Term
First Yr.
Second
Yr.
Third Yr.
Sporadic
Persisters
GPA
7.69
3.37
2.30
----
2.89
---
SCH
----
1.79
---
----
1.31
---
Failing class
3.289
3.897
2.717
2.106
2.623
---
Working
1.78
1.398
1.332
----
1.405
----
Math
---
---
1.50
---
1.29
---
Need
1.625
---
---
----
-----
----
No dependent
----
2.35
----
----
-----
---
Disagree it is important
to prepare for class
2.154
---
1.528
----
---
1.651
High School rank
1.629
1.76
1.411
1.626
1.486
1.629
Not needing more than
four years
----
---
---
----
-----
1.67
Loan Paid
1.39
Grant Paid
2.34
1.49
---
----
-----
---
Institutional Responsibility Vs
Student Responsibility
“Student responsibility (the intersection of choice with
opportunity) is a major theme. . . some recommendations
for students, who are partners in their own education
fate, who shouldn’t wait around for someone else to do
something for them, and who are rarely addressed in
studies of attainment,” Toolbox Revisited, p. xxvi.
Students are explicit, rather than implicit. . . They are
respected adults playing large roles in their own
destinies. . . While we trust that school and college
actions will not leave them behind, they have equal
responsibilities.
Source: Adelman, C. The Toolbox Revisited: Paths to Degree Completion from High School Through
College. Washington, D.C.: U.S. Department of Education, February 2006
Implications from the UTEP Student
Success Project
All students can be
successful—
• Family/ household Income is
not a predictor of student
success
• Ethnicity is not a predictor of
success
• ACT (SAT) score is not a
predictor of success
But, student’s choices are
important in ensuring
The challenge is to create institutional
success…
structures to shape student choices
that will ensure success.
Next Phase of Study
Refine “at-risk” and explore alternative
models
Latent growth models
Markov models
Explore major issues more closely
(advising, role of faculty, working, financial
aid, course failure)
Develop interventions
Contact Information
Roy Mathew, Ph.D.
Director
Center for Institutional Evaluation,
Research and Planning
The University of Texas at El Paso
Administration Building, Room 318
(915) 747-5117
[email protected]
Denise Carrejo, Ph.D.
Assistant Director
Center for Institutional Evaluation,
Research and Planning
The University of Texas at El Paso
Administration Building, Room 318
(915) 747-5117
[email protected]
Thomas J. Taylor
Doctoral Student
Center for Institutional Evaluation,
Research and Planning
The University of Texas at El Paso
Administration Building, Room 318
(915) 747-5784
[email protected]
This AIR 2007 Presentation is available at:
http://cierp.utep.edu