admin
Thu, 02/12/2026 - 19:21
Edited Text
March 2014
In the following report, Hanover Research examines the impact of financial aid packages on
undergraduate student enrollment decisions at Clarion University at Pennsylvania. More
specifically, we examine how the various components of a financial aid package impact the
probability of students accepting their offer of admission. Separate analyses are carried out
for first time undergraduate and transfer applicants.
Executive Summary and Key Findings ............................................................................... 3
Introduction ...........................................................................................................................3
Key Findings ...........................................................................................................................3
Freshman Students ............................................................................................................3
Transfer Students (PA residents only) ...............................................................................4
Section I: Methodology .................................................................................................... 6
Regression Analysis................................................................................................................6
Data........................................................................................................................................7
Dependent (Outcome) Variable ........................................................................................7
Independent Variables.......................................................................................................9
Section II: Results and Analysis ....................................................................................... 12
Effect of Financial Aid Package on Student Enrollment ......................................................12
Freshman Students ..........................................................................................................12
Transfer Students (PA Residents Only) ............................................................................15
2
INTRODUCTION
In this report, Hanover Research examines the impact of financial aid packages on student
enrollment decisions at Clarion University of Pennsylvania. More specifically, we investigate
how financial aid offered to accepted applicants influences the probability of the student
accepting an offer of admission. For this analysis, we only considered accepted applicants in
2012-131 who intend to earn a degree in an undergraduate program at Clarion. Since the
effects of financial aid packages on freshman students (first time undergraduates) are likely
different from the effect of financial aid packages on transfer students, we analyzed the two
groups of applicants separately. Furthermore, for each of these types of applicants
(freshman applicants and transfer applicants), we examined the effect of financial aid on
students who are residents of Pennsylvania (PA) separately from those residing outside the
state.
KEY FINDINGS
FRESHMAN STUDENTS
1
Among the various components of financial aid packages, the amount of
subsidized loans has the strongest impact on the probability of PA residents
enrolling as freshman students at Clarion. For PA residents, a one percentage point
increase in the proportion of tuition and mandatory fees covered by subsidized
loans leads to an increase in the probability of student enrollment of 0.007.
Subsidized loans have no statistically significant impact on the enrollment decisions
of non-PA residents.
Unsubsidized loans have a positive impact on students’ enrollment decisions, for
applicants residing in PA as well as for those residing outside PA. For students
residing in PA with a reported Expected Family Contribution (EFC), the amount of
unsubsidized loans has a non-linear relationship with the probability of enrollment.
Unsubsidized loans may cover up to 82 percent of the tuition and mandatory fees
without negatively impacting students’ decision to enroll as a freshman
undergraduate. Any increase in unsubsidized loans beyond that is likely to lead to a
fall in the probability of the student enrolling at Clarion. For PA residents with and
without reported EFC, and non-resident PA applicants, unsubsidized loan coverage
has a positive but linear effect on the probability of freshman enrollment.
The amount of grant aid (Pell grants, SEOG grants, and PHEAA grants) has a small
but significant effect on PA residents’ decision to enroll at Clarion as a freshman.
However, for non-PA resident applicants, the effect is only true for those students
who have a reported EFC.
Students in 2013-14 were excluded due to insufficient data. Further details on this are provided in Section I.
3
An increase in EFC is associated with an increase in the probability of a student
deciding to accept Clarion’s offer of admission. For PA residents, an increase in the
EFC of $1,000 is expected to increase the probability of the student enrolling by
0.008. The impact is similar for non-resident applicants (an increase in probability of
enrolling by 0.01 for a $1,000 increase in EFC).
Table 1 below summarizes the impact of the financial aid components on freshman
enrollment.
Table 1: Impact of Various Financial Aid Components on Freshman Enrollment
FINANCIAL AID
COMPONENTS
ALL PA RESIDENTS
Subsidized
Loans
Unsubsidized
Loans
Positive and linear
impact
Positive and linear
impact
Positive and linear
impact
EFC
-
Grant Aid
PA RESIDENTS WITH
REPORTED EFC
Positive and linear
impact
Positive and linear
impact
Positive and non-linear
impact
Positive and linear
impact
ALL NON-PA RESIDENTS
NON-PA RESIDENTS
WITH REPORTED EFC
No effect
Positive and linear
impact
No effect
No effect
Positive and linear
impact
Positive and linear
impact
Positive and linear
impact
-
TRANSFER STUDENTS (PA RESIDENTS ONLY)
For PA residents, the amount of grant aid (Pell Grants, SEOG grants, and PHEAA
grants) has a positive impact on the decision to transfer to Clarion. A one
percentage point increase in the proportion of tuition and mandatory fees covered
by grant aid is associated with an increase in the probability of transfer enrollment
by 0.002.
Subsidized loans have a positive impact on accepted students’ decision to transfer
to Clarion. For all PA residents, a one percentage point increase in the proportion of
tuition and mandatory fees covered by subsidized loans leads to an increase in the
probability of transfer enrollment by 0.006. The effect is approximately half as
strong for PA transfer applicants who have a reported EFC.
The amount of unsubsidized loans has a non-linear relationship with the
probability of transfer enrollment for PA residents who have a reported EFC. The
probability of enrollment increases until 89 percent of the tuition and mandatory
fees are covered by unsubsidized loans. Further increases in unsubsidized loans
reduce the probability of transfer enrollment. For all applicants in general (with and
without reported EFC), unsubsidized loans have a linear and positive impact on the
probability of transfer enrollment.
There is no measurable impact of EFC on students’ decision to transfer to Clarion.
Table 2 on the following page summarizes the impact of financial aid components on
transfer enrollment.
4
Table 2: Impact of Various Financial Aid Components on Transfer Enrollment
FINANCIAL AID
COMPONENTS
Grant Aid
ALL PA RESIDENTS
PA RESIDENTS WITH REPORTED EFC
Positive and linear impact
Positive and linear impact
Subsidized Loans
Positive and linear impact
Positive and linear impact
Unsubsidized Loans
Positive and linear impact
Positive and non-linear impact
EFC
-
No effect
5
In this section, Hanover Research provides an overview of the data and methodology used
to examine the impact of financial aid packages on student enrollment decisions. The
datasets provided to Hanover by Clarion include demographic, academic, and financial aid
information for 8,611 students who applied to a program at Clarion between 2012-13 and
2013-14. Demographic data include applicants’ gender, ethnicity, state of residency,
intended living arrangement, and athlete status, while academic information includes
Clarion’s decision on the application as well as the applicants’ SAT and ACT scores and high
school GPA and rank. In terms of financial aid data, the dataset includes the amount of
grants, scholarships, and loans offered to students. It further includes the tuition and
mandatory fees that the applicant is expected to pay. For some students, the Expected
Family Contribution from the Student Aid Report is included.
REGRESSION ANALYSIS
To examine the impact of financial aid packages on the probability of enrollment, the report
uses multivariate linear regression analysis. In our models, the dependent variable indicates
whether or not an accepted applicant enrolled at Clarion, while demographic, academic,
and financial characteristics of these applicants are included as controls. In addition to
showing the impact of financial aid packages on accepted students’ decision to enroll at
Clarion, these models also indicate other factors that strongly predict student enrollment.
Our models only take into account accepted freshman and transfer undergraduate
applicants who intend to earn a degree2 at Clarion.
We estimated separate models based on the type of students: freshman or transfer
students. For each of these students, we also estimated separate models for students
identified as PA residents and students who are not PA residents.3 The table below shows
the six different types of models that we included in this analysis. Our models can be
differentiated by the type of students they examine and the residency status of the student.
For each of these types of models, we included two variations: one with Expected Family
Contribution (EFC) and one without the EFC variable. Students who did not apply for aid do
not have a reported EFC, which results in a significant number of observations being
dropped when this variable is included in the analysis. Since the applicants being excluded
are likely systematically different from the non-excluded students (since applicants who do
not apply for financial aid are likely to be financially better off than those who do apply for
aid), these exclusions have the potential to bias our estimates of the effect of financial aid
on students’ probability of enrollment. Thus, we decided to include both variations in our
models.
2
3
Excludes non-degree seeking applicants.
We could not estimate models for transfer applicants who are not PA residents due to the small number of
applicants who met these criteria.
6
Figure 1.1: Different Types of Models
MODELS
TYPE OF STUDENTS
RESIDENCY
Model 1
Freshman
PA
EXPECTED FAMILY
CONTRIBUTION
Yes
Model 2
Freshman
PA
No
Model 3
Freshman
Non-PA
Yes
Model 4
Freshman
Non-PA
No
Model 5
Transfer
PA
Yes
Model 6
Transfer
PA
No
For each of our regression models, we provide coefficient estimates, as well as an indication
of which coefficients proved statistically significant. The coefficients of our models reveal
how much we expect the probability of the enrollment dependent variable to change when
the independent variable increases by one unit, holding all of the other independent
variables in the model constant. In this context, the coefficients capture the change in the
likelihood of enrolling at Clarion, depending on the model estimated.4
Our linear probability models account for the dichotomous nature of the dependent
variable. Dichotomous variables assume one of two values; in the context of the present
analysis, we assign a value of 1 to our dependent variable in cases where the student enrolls
at Clarion and a value of 0 if they do not enroll. Robust standard errors (included in the
results within parenthesis) were used in order to correct for heteroskedasticity. Lastly, when
presenting our results, we also include the R-squared value, which reveals the percentage of
variation in the dependent variable accounted for by the model.
DATA
In addition to providing us with the dataset that contains all the applications in the years
2012-13 and 2013-14, Clarion also provided us with a list of enrolled students in each of the
two years. Accepted applicants who were matched in the enrolled student datasets were
identified as students who chose to enroll at Clarion; others are considered as not choosing
to enroll at Clarion. For each of these enrolled and not enrolled students, our dataset
included demographic, academic, and financial aid information.
DEPENDENT (OUTCOME) VARIABLE
As discussed previously, our dependent variable indicates whether an accepted applicant
enrolled at Clarion. In the tables below, we provide descriptive statistics on our dependent
variables. Figure 1.2 shows that Clarion had 1,357 and 1,393 incoming new undergraduate
students (including only freshman and transfer students) in 2012 and 2013 respectively. In
total, 2,750 students were identified as enrolled students, and 1,600 accepted applicants
chose not to enroll at Clarion. Irregularities identified within these students were dropped
4
For SAT scores, our model shows the effect of a 10 point change, while for Expected Family Contribution, the model
indicates the effect of a $1,000 change.
7
(irregularities included students whose admission date was before the application date).
Furthermore, for enrolled students, we only considered those students whose starting term
matched their intended starting term,5 resulting in a total of 91 students being dropped.
This left our final dataset with 2,659 enrolled students and 1,600 students who did not
enroll.
Figure 1.2: Number of Freshman and Transfer Students by Year
APPLICATION YEAR
2012 NEW STUDENTS
2013 NEW STUDENTS
TOTAL ENROLLED
DID NOT ENROLL
2012-13
1,349
14
1,363
771
2013-14
8
1,379
1,387
829
Total
1,357
1,393
2,750
1,600
Drop
70
21
91
0
Total
1,287
1,372
2,659
1,600
For our analysis, we only examine the impact of financial aid packages on degree seeking
undergraduate students. Thus, we dropped any students who did not intend to earn a
bachelor’s degree or who were identified as non-degree seeking students. Figure 1.3
indicates that, after these restrictions, a total of 2,386 students enrolled, while 1,255
students did not enroll.
Figure 1.3: Enrollment by Year
APPLICATION YEAR
TOTAL ENROLLED
DID NOT ENROLL
2012-13
1,188
551
2013-14
1,198
704
Total
2,386
1,255
Figures 1.4 and 1.5 below show the breakdown of the enrolled and not enrolled students by
freshman and transfer status. We also include the breakdown by students’ Pennsylvania
(PA) residency status. A majority of the students are freshmen and PA residents.
Figure 1.4: Freshman and Transfer Enrollment by Year
APPLICATION YEAR
5
FRESHMAN
TRANSFER
ENROLLED
NOT ENROLLED
ENROLLED
NOT ENROLLED
2012-13
951
436
237
115
2013-14
975
590
223
114
Total
1,926
1,026
460
229
Some students had almost a year gap between their application and admission date. These students missed their
intended starting term and deferred to a later semester. In order to maintain consistency within our dataset, we
only considered students whose intended starting term in the application matched their actual starting term at
Clarion.
8
Figure 1.5: PA and Non-PA Enrollment by Year
APPLICATION
YEAR
2012-13
PA STUDENTS
NOT
ENROLLED
ENROLLED
1,104
469
NON-PA STUDENTS
NOT
ENROLLED
ENROLLED
84
91
2013-14
1,073
580
125
124
Total
2,177
1,049
209
215
INDEPENDENT VARIABLES
In Figures 1.6 through 1.9, we provide descriptive statistics for each of the variables used in
our models. Before estimating our final models, we examined several models with various
combinations of independent variables. In our preliminary models, we included squared
terms for the academic and financial aid variables to examine the non-linear effect of these
on the probability of student enrollment. The list of variables that we included in our final
models is presented in Figure 1.6 below.
Figure 1.6: Independent Variables
INDEPENDENT VARIABLES
SUMMARY
VARIABLE TYPE
Gender
Gender of students: (1) Male and (0) Female
Ethnicity of students selected from four ethnic categories:
Black, Multiracial, White, and Others
High School GPA of students
Combined SAT scores in Math and Critical Reading. This also
includes converted ACT scores.
Whether applicant intend to live on campus or off campus:
(1) On campus and (0) off campus
Whether students is an athlete: (1) Athlete and (0) Not an
athlete
Percentage of tuition and mandatory fees covered by grant
aid. Grant aid includes Pell Grants, SEOG Grants and PHEAA
Grants.
Percentage of tuition and mandatory fees covered by
subsidized loans.
Percentage of tuition and mandatory fees covered by
unsubsidized loans.
The square of the percentage of tuition and mandatory fees
covered by unsubsidized loans.
Expected Family Contribution of the students in thousands
of dollars.
Categorical
Ethnicity
HS GPA
SAT
On campus/off campus
Athlete
Grant Aid Coverage
Subsidized loan
coverage
Unsubsidized loan
coverage
Unsubsidized loan
coverage squared
Expected Family
Contribution
Categorical
Continuous
Continuous
Categorical
Categorical
Continuous
Continuous
Continuous
Continuous
Continuous
9
Figure 1.7 shows the descriptive statistics of the demographic variables that were used in
our models. The gender ratio of students remained more or less similar between the two
years. In terms of ethnicity,6 a majority of the students are White, while approximately a
fifth of the accepted students who decided not to enroll at Clarion are Black. The table
further shows that a vast majority of the students intend to live off campus and are not
athletes.
Figure 1.7: Independent Variables - Demographics
GENDER
2012-13
2013-14
ENROLLED
NOT ENROLLED
ENROLLED
NOT ENROLLED
Female
59.7% (n=709)
58.8% (n=324)
60.2% (n=721)
61.8% (n=435)
Male
40.3% (n=479)
41.2% (n=227)
39.8% (n=477)
38.2% (n=269)
Black
6.6% (n=79)
23.2% (n=128)
8.8% (n=106)
20.9% (n=147)
Multiracial
3.2% (n=38)
5.6% (n=31)
6.3% (n=75)
4.4% (n=31)
Other
2.9% (n=34)
6.4% (n=35)
3.9% (n=47)
5.1% (n=36)
White
87.3% (n=1037)
64.8% (n=357)
81% (n=970)
69.6% (n=490)
On campus
15.4% (n=183)
7.8% (n=43)
16.1% (n=193)
5.4% (n=38)
Off campus
84.6% (n=1005)
92.2% (n=508)
83.9% (n=1005)
94.6% (n=666)
Yes
3.7% (n=44)
0% (n=0)
2.6% (n=31)
0% (n=0)
No
96.3% (n=1144)
100% (n=551)
97.4% (n=1167)
100% (n=704)
ETHNICITY
ON CAMPUS/ OFF CAMPUS
ATHLETE
Figure 1.8 shows the applicants’ high school GPA and SAT scores. The SAT scores show the
combined scores in Math and Critical Reading. Several students reported ACT scores instead
of SAT scores. For these students, we converted the ACT scores to equivalent SAT scores
based on the concordance table released by ACT and College Board.7 In terms of academic
performance, enrolled students have better academic record compared to students who did
not enroll.
Figure 1.8: Independent Variables – Academic
HS GPA
ENROLLED
NOT ENROLLED
2012-13
2.58 (n=1188)
1.98 (n=551)
2013-14
2.63 (n=1198)
1.95 (n=704)
2012-13
963.17 (n=1052)
871.88 (n=463)
2013-14
948.86 (n=1198)
915.17 (n=704)
SAT
6
Others include American Indians, Asians, Hispanic, Pacific Islanders, and other minority groups.
Compare ACT & SAT Scores. http://www.act.org/solutions/college-career-readiness/compare-act-sat/
7
10
Figure 1.9 summarizes students’ expected tuition, financial aid offers, and expected family
contribution. We present the tuition and mandatory fees together as these are fixed costs
for all the students. Of all the students who did not enroll in 2013-14, expected tuition and
mandatory fees are available for only four students. This excludes a considerable number of
students who chose not to enroll at Clarion in 2013-14. As a result, our models exclude all
2013-14 students. Since tuition and mandatory fees are missing for a considerable number
of students, including them may bias our sample, since we would have approximately 1,198
enrolled students from 2013-14, but only four non-enrolled students. Nevertheless, we
include the descriptive statistics for the 2013-14 dataset in this section.
The grant aid includes Pell grants, SEOG grants, and PHEAA grants (only offered to PA
residents) received by students. In our models, we include the percentage of tuition and
mandatory fees covered by the grants and loans. Please note that since we excluded other
fees such as room and board and supplies, it is possible for the total aid and loans to be
higher than the tuition and mandatory fees. The EFC is included in our models in thousands
of dollars.
Figure 1.9: Independent Variables – Financial (Annual)
TUITION AND MANDATORY FEES
ENROLLED
NOT ENROLLED
2012-13
$8774.03 (n=1188)
$9110.29 (n=551)
2013-14
$9078.13 (n=1198)
$2006.2 (n=4)
2012-13
$2741.06 (n=1188)
$288.62 (n=551)
2013-14
$2605.55 (n=1198)
8
9
GRANT AID
$0 (n=704)
10
SUBSIDIZED LOANS
2012-13
$2199.66 (n=1188)
$144.68 (n=551)
2013-14
$1608.76 (n=1198)
$9.84 (n=704)
2012-13
$2380.69 (n=1188)
$156.4 (n=551)
2013-14
$1774.01 (n=1198)
$5.63 (n=704)
10
UNSUBSIDIZED LOANS
10
EXPECTED FAMILY CONTRIBUTION
2012-13
$12860.32 (n=902)
$10056.29 (n=188)
2013-14
$13084.56 (n=866)
$13675.64 (n=236)
8
Tuition and mandatory fees are available for only four students in 2013-14.
Grant aid includes Pell grants, SEOG grants, and PHEAA grants.
10
Limited amounts of grant aid, subsidized, and unsubsidized loan are available for non-enrolled students in 2013-14.
9
11
This section presents the results of multivariate analyses investigating the impact of
financial aid packages on undergraduate enrollment. For each dependent variable and
model, the results presented in the figures below display regression coefficients for each
independent variable and, where applicable, asterisks indicating the level of statistical
significance. Regression coefficients in the figures below can be interpreted as the change in
the probability of enrolling at Clarion due to a one unit change in the independent variable.
Please note that for grant aid coverage, subsidized loan coverage, and unsubsidized loan
coverage, the one unit increase refers to a percentage point increase in these variables (for
instance, the effect on the probability of enrolling at Clarion of a one percentage point
increase in the grant aid coverage). For SAT scores, the coefficients indicate the effect of a
10 point change in the score, while the coefficient for the EFC shows the effect of a $1,000
change. Please note that our models include only 2012-13 data. Students in the 2013-14
dataset were excluded from the analysis due to missing tuition and mandatory fees for nonenrolled students.
EFFECT OF FINANCIAL AID PACKAGE ON STUDENT ENROLLMENT
In this study, we concentrate on the impact of financial aid on freshman and transfer
students separately. We also exclude non-degree seeking students, as their responses to
financial aid packages are likely different from those of students entering as freshmen. For
both the freshman and transfer students, we examined the effect of financial aid packages
on PA resident and non-resident students separately.
FRESHMAN STUDENTS
We estimated four models with freshman enrollment as our dependent variable. Our
primary independent variables of interest are grant aid coverage, subsidized loan coverage,
and unsubsidized loan coverage. The aforementioned components of financial aid packages
are included as a percentage of the expected tuition and mandatory fees that the student is
expected to pay, had he/she been enrolled. Other financial data include students’ EFC,
which is included in thousands of dollars. Results from our regression models for freshman
enrollment are presented in Figure 2.1. The following is a summary of the impact of the
different components of financial aid packages on freshman enrollment:
Grant Aid Coverage: In general, the percentage of tuition and mandatory fees
covered by grants and scholarships has a positive and significant impact on the
probability that freshman applicants will enroll. However, it must be noted that no
impact is observed when all non-resident PA students (both those with and without
EFC data available) are considered. For PA residents (regardless of whether the
model is restricted to applicants with EFC data) and for non-PA residents with EFC
data, the probability of enrolling at Clarion as a freshman increases with higher
coverage of tuition and mandatory fees by grant aid.
12
Subsidized Loan Coverage: The percentage of tuition and mandatory fees covered
by subsidized loans has a positive and significant impact on the probability of PA
residents enrolling as freshman students at Clarion. An increase in subsidized loan
coverage by one percentage point is likely to increase the probability of a PA
resident enrolling at Clarion by 0.007. In fact, among all the components of the
financial aid package, subsidized loans have the strongest impact on the probability
of freshman enrollment for PA residents.
However, subsidized loan coverage has no significant effect on the probability of
accepted students who are not PA residents enrolling at Clarion as freshman
undergraduates. No effect is observed either for students who have a reported EFC
in their Student Aid Report or for those who do not.
Unsubsidized Loan Coverage: There is a non-linear relationship between the
probability of a PA resident with a reported EFC enrolling at Clarion and the amount
of tuition and mandatory fees covered by unsubsidized loans. The probability of
enrollment increases up to a certain point beyond which an increase in the
unsubsidized loans reduces the probability of students enrolling at Clarion. For
students who are PA residents and have a reported EFC, the probability of
enrollment increases until 82 percent of their tuition and mandatory fees are
covered by unsubsidized loans. However, for all PA residents in general,
unsubsidized loans have a linear and positive relationship with the probability of
enrolling at Clarion.
In terms of accepted students who are not residents of PA, the amount of
unsubsidized loans has a linear, positive, and significant impact on the probability of
enrolling as a freshman. For all non-PA resident students, a one percentage point
increase in the amount of unsubsidized loan coverage is associated with an increase
of 0.009 in the probability of enrolling at Clarion. The effect is slightly weaker when
only students with a reported EFC are considered (increase in the probability by
0.005).
Expected Family Contribution: There is a positive and significant impact of EFC on
the probability of students enrolling at Clarion. For PA applicants, an increase in the
EFC by $1,000 is expected to increase the probability of enrolling by 0.008. The
effect is slightly stronger for non-PA residents, with an increase in the EFC by $1,000
increasing the probability of enrollment by 0.01.
Demographic and Academic Characteristics: In addition to the aforementioned
components of the financial aid package, accepted students’ demographics and
academic backgrounds also have some impact on the probability of enrolling at
Clarion. For instance, PA residents with high SAT scores, any applicants with strong
high school GPAs, or applicants who are athletes have a higher probability of
enrolling as freshman undergraduates at Clarion, compared to their respective
counterparts. In terms of ethnicity, White students who are significantly more likely
13
to enroll compared to any other ethnic groups among PA residents. However, PA
residents who plan to live on-campus are less likely to eventually enroll at Clarion
than those who plan to live off-campus.
Figure 2.1: Regression Results – Freshman Enrollment
VARIABLES
Male [Reference Group = Female]
PA RESIDENTS
MODEL 1 WITH EFC
MODEL 2
-0.0218
-0.0027
NON-PA RESIDENTS
MODEL 3 WITH EFC
MODEL 4
-0.0758
-0.027
(0.0218)
(0.0201)
(0.1187)
(0.1034)
-0.1001*
-0.2519***
-0.1372
-0.177*
(0.0531)
(0.0422)
(0.1206)
(0.0934)
-0.0675
-0.1771***
0.1231
0.0909
(0.0421)
(0.0441)
(0.1155)
(0.231)
-0.1297
-0.1508**
0.3029**
-0.0865
(0.085)
(0.065)
(0.1429)
(0.1861)
0.0038***
0.007***
0.0091*
0.0061
(0.0008)
(0.0007)
(0.0051)
(0.0038)
0.012
0.0198**
-0.0003
0.0635**
(0.0084)
(0.0076)
(0.0303)
(0.0263)
-0.0981**
-0.112***
(0.0321)
(0.0303)
0.0943**
0.2161***
0.2513***
0.3951***
(0.0358)
(0.0471)
(0.0757)
(0.0934)
Grant Aid Coverage
0.0023***
0.0015***
0.0098**
0.004
(0.0004)
(0.0003)
(0.0032)
(0.005)
Subsidized Loan Coverage
0.0073***
0.0069***
0.0065
0.0045
(0.0007)
(0.0007)
(0.0045)
(0.0048)
0.0052***
0.003**
0.005**
0.0091**
(0.0005)
(0.0013)
(0.0022)
(0.0031)
Black [Reference Group = White]
Multiracial [Reference Group = White]
Other [Reference Group = White]
SAT Score (per 10 points)
High School GPA
On Campus [Reference Group = off campus]
Athlete [Reference Group = Not an athlete]
Unsubsidized Loan Coverage
Unsubsidized Loan Coverage squared
-0.00003***
(0.000002)
Expected Family Contribution (per $1,000)
Intercept
R-squared
Number of observations
0.0082***
0.0101**
(0.0015)
(0.0035)
0.1495
-0.0882
-0.4728
-0.3107
(0.0932)
(0.0768)
(0.4757)
(0.3695)
0.3635
833
0.4188
1,253
0.4407
56
0.3493
95
Coefficients estimated using Ordinary Least Squares with a linear regression model, with robust standard errors in
parenetheses. Statistical significance indicator using asterisk next to the coefficients, with * = significant at 10%,
** = significant at 5%, and *** = significant at 1%.
14
TRANSFER STUDENTS (PA RESIDENTS ONLY)
For transfer applicants, we only estimated models with PA residents. We did not have
sufficient data to estimate models for non-PA residents. The models include the same
independent variables that were included in Models 1 through 4. Results from our
regression models for transfer enrollment are presented in Figure 2.2. The following is a
summary of the impact of the different components of financial aid packages:
Grant Aid Coverage: The amount of grant aid awarded to PA resident transfer
applicants has a positive and significant impact on their likelihood of enrolling at
Clarion. For all PA residents, a one percentage point increase in the percentage of
tuition and mandatory fees covered by grant aid is associated with an increase in the
probability of enrolling at Clarion by 0.0016.
Subsidized Loan Coverage: For all transfer applicants residing in PA, a one
percentage point increase in the proportion of tuition and mandatory fees covered
by subsidized loans is expected to lead to an increase in the probability of enrolling
by 0.006. For PA residents with a reported EFC, the effect is slightly weaker
(increases probability of enrollment by 0.003).
Unsubsidized Loan Coverage: There is a non-linear relationship between the
probability of a PA resident with a reported EFC enrolling as a transfer student at
Clarion and the amount of tuition and mandatory fees covered by unsubsidized
loans. The probability of transfer enrollment increases up to a certain point, beyond
which an increase in the unsubsidized loans reduces the probability of students
enrolling at Clarion. For students who are PA residents and have a reported EFC, the
probability of transfer enrollment increases until 89 percent of their tuition and
mandatory fees are covered by unsubsidized loans.
However, for all PA residents in general, unsubsidized loans have a linear and
positive relationship with the probability of a transfer enrollment at Clarion. A one
percentage point increase in the amount of unsubsidized loan coverage is expected
to lead to the probability of these students enrolling as a transfer student at Clarion
increasing by 0.004.
Expected Family Contribution: The EFC reported on a transfer student’s Student Aid
Report does not have any significant impact on their probability of enrolling at
Clarion.
Demographic and Academic Characteristics: Further analysis of Models 5 and 6
indicates that some demographic and academic characteristics of PA resident
transfer applicants influence their probability of enrolling at Clarion. Female
applicants (regardless of whether we limit the analysis to students with a reported
EFC) are less likely to enroll as transfer students compared to male applicants
residing in PA. There is also a slightly higher probability of applicants who are
15
categorized as ‘multiracial’ or ‘other’ in terms of ethnicity to enroll as transfer
students compared to White students, when limiting the analysis to those students
with a reported EFC. Applicants (with or without reported EFC) with higher SAT
scores and applicants who are athletes (only with reported EFC) also have a
significantly higher probability of enrolling at Clarion.
Figure 2.2: Regression Results – Transfer Enrollment
VARIABLES
Male [Reference Group = Female]
Black [Reference Group = White]
Multiracial [Reference Group = White]
Other [Reference Group = White]
PA RESIDENTS
MODEL 5 WITH EFC
MODEL 6
-0.0212
0.1239**
(0.06)
(0.0561)
-0.0639
-0.0942
(0.0941)
(0.0703)
0.2164*
0.0246
(0.1253)
(0.1604)
0.1425*
0.0465
(0.0794)
(0.0912)
0.0009
0.0045**
(0.0013)
(0.0017)
High School GPA
0.0122
0.0189
(0.025)
(0.0232)
On Campus [Reference Group = off campus]
-0.0672
-0.0277
(0.0879)
(0.0837)
0.18*
0.0049
(0.0976)
(0.1933)
SAT Score (per 10 points)
Athlete [Reference Group = Not an athlete]
Grant Aid Coverage (per 1%)
0.0018*
0.0016**
(0.0011)
(0.0005)
0.003**
0.006***
(0.0013)
(0.0011)
Unsubsidized loan Coverage
0.005**
0.0044***
(0.0019)
(0.0007)
Unsubsidized Loan Coverage squared
-0.00003*
Subsidized Loan Coverage
(0.00002)
Expected Family Contribution (per $1,000)
0.0029
(0.0051)
Intercept
0.5949**
-0.0162
(0.1969)
(0.1855)
R-squared
0.1438
0.4066
Number of observations
101
152
Coefficients estimated using Ordinary Least Squares with a linear regression model, with robust standard errors in
parentheses. Statistical significance indicator using asterisk next to the coefficients, with * = significant at 10%, ** =
significant at 5%, and *** = significant at 1%.
16
Hanover Research is committed to providing a work product that meets or exceeds partner
expectations. In keeping with that goal, we would like to hear your opinions regarding our
reports. Feedback is critically important and serves as the strongest mechanism by which we
tailor our research to your organization. When you have had a chance to evaluate this
report, please take a moment to fill out the following questionnaire.
http://www.hanoverresearch.com/evaluation/index.php
The publisher and authors have used their best efforts in preparing this brief. The publisher
and authors make no representations or warranties with respect to the accuracy or
completeness of the contents of this brief and specifically disclaim any implied warranties of
fitness for a particular purpose. There are no warranties which extend beyond the
descriptions contained in this paragraph. No warranty may be created or extended by
representatives of Hanover Research or its marketing materials. The accuracy and
completeness of the information provided herein and the opinions stated herein are not
guaranteed or warranted to produce any particular results, and the advice and strategies
contained herein may not be suitable for every partner. Neither the publisher nor the
authors shall be liable for any loss of profit or any other commercial damages, including but
not limited to special, incidental, consequential, or other damages. Moreover, Hanover
Research is not engaged in rendering legal, accounting, or other professional services.
Partners requiring such services are advised to consult an appropriate professional.
17
1700 K Street, NW, 8th Floor
Washington, DC 20006
P 202.559.0500 F 866.808.6585
www.hanoverresearch.com
18
In the following report, Hanover Research examines the impact of financial aid packages on
undergraduate student enrollment decisions at Clarion University at Pennsylvania. More
specifically, we examine how the various components of a financial aid package impact the
probability of students accepting their offer of admission. Separate analyses are carried out
for first time undergraduate and transfer applicants.
Executive Summary and Key Findings ............................................................................... 3
Introduction ...........................................................................................................................3
Key Findings ...........................................................................................................................3
Freshman Students ............................................................................................................3
Transfer Students (PA residents only) ...............................................................................4
Section I: Methodology .................................................................................................... 6
Regression Analysis................................................................................................................6
Data........................................................................................................................................7
Dependent (Outcome) Variable ........................................................................................7
Independent Variables.......................................................................................................9
Section II: Results and Analysis ....................................................................................... 12
Effect of Financial Aid Package on Student Enrollment ......................................................12
Freshman Students ..........................................................................................................12
Transfer Students (PA Residents Only) ............................................................................15
2
INTRODUCTION
In this report, Hanover Research examines the impact of financial aid packages on student
enrollment decisions at Clarion University of Pennsylvania. More specifically, we investigate
how financial aid offered to accepted applicants influences the probability of the student
accepting an offer of admission. For this analysis, we only considered accepted applicants in
2012-131 who intend to earn a degree in an undergraduate program at Clarion. Since the
effects of financial aid packages on freshman students (first time undergraduates) are likely
different from the effect of financial aid packages on transfer students, we analyzed the two
groups of applicants separately. Furthermore, for each of these types of applicants
(freshman applicants and transfer applicants), we examined the effect of financial aid on
students who are residents of Pennsylvania (PA) separately from those residing outside the
state.
KEY FINDINGS
FRESHMAN STUDENTS
1
Among the various components of financial aid packages, the amount of
subsidized loans has the strongest impact on the probability of PA residents
enrolling as freshman students at Clarion. For PA residents, a one percentage point
increase in the proportion of tuition and mandatory fees covered by subsidized
loans leads to an increase in the probability of student enrollment of 0.007.
Subsidized loans have no statistically significant impact on the enrollment decisions
of non-PA residents.
Unsubsidized loans have a positive impact on students’ enrollment decisions, for
applicants residing in PA as well as for those residing outside PA. For students
residing in PA with a reported Expected Family Contribution (EFC), the amount of
unsubsidized loans has a non-linear relationship with the probability of enrollment.
Unsubsidized loans may cover up to 82 percent of the tuition and mandatory fees
without negatively impacting students’ decision to enroll as a freshman
undergraduate. Any increase in unsubsidized loans beyond that is likely to lead to a
fall in the probability of the student enrolling at Clarion. For PA residents with and
without reported EFC, and non-resident PA applicants, unsubsidized loan coverage
has a positive but linear effect on the probability of freshman enrollment.
The amount of grant aid (Pell grants, SEOG grants, and PHEAA grants) has a small
but significant effect on PA residents’ decision to enroll at Clarion as a freshman.
However, for non-PA resident applicants, the effect is only true for those students
who have a reported EFC.
Students in 2013-14 were excluded due to insufficient data. Further details on this are provided in Section I.
3
An increase in EFC is associated with an increase in the probability of a student
deciding to accept Clarion’s offer of admission. For PA residents, an increase in the
EFC of $1,000 is expected to increase the probability of the student enrolling by
0.008. The impact is similar for non-resident applicants (an increase in probability of
enrolling by 0.01 for a $1,000 increase in EFC).
Table 1 below summarizes the impact of the financial aid components on freshman
enrollment.
Table 1: Impact of Various Financial Aid Components on Freshman Enrollment
FINANCIAL AID
COMPONENTS
ALL PA RESIDENTS
Subsidized
Loans
Unsubsidized
Loans
Positive and linear
impact
Positive and linear
impact
Positive and linear
impact
EFC
-
Grant Aid
PA RESIDENTS WITH
REPORTED EFC
Positive and linear
impact
Positive and linear
impact
Positive and non-linear
impact
Positive and linear
impact
ALL NON-PA RESIDENTS
NON-PA RESIDENTS
WITH REPORTED EFC
No effect
Positive and linear
impact
No effect
No effect
Positive and linear
impact
Positive and linear
impact
Positive and linear
impact
-
TRANSFER STUDENTS (PA RESIDENTS ONLY)
For PA residents, the amount of grant aid (Pell Grants, SEOG grants, and PHEAA
grants) has a positive impact on the decision to transfer to Clarion. A one
percentage point increase in the proportion of tuition and mandatory fees covered
by grant aid is associated with an increase in the probability of transfer enrollment
by 0.002.
Subsidized loans have a positive impact on accepted students’ decision to transfer
to Clarion. For all PA residents, a one percentage point increase in the proportion of
tuition and mandatory fees covered by subsidized loans leads to an increase in the
probability of transfer enrollment by 0.006. The effect is approximately half as
strong for PA transfer applicants who have a reported EFC.
The amount of unsubsidized loans has a non-linear relationship with the
probability of transfer enrollment for PA residents who have a reported EFC. The
probability of enrollment increases until 89 percent of the tuition and mandatory
fees are covered by unsubsidized loans. Further increases in unsubsidized loans
reduce the probability of transfer enrollment. For all applicants in general (with and
without reported EFC), unsubsidized loans have a linear and positive impact on the
probability of transfer enrollment.
There is no measurable impact of EFC on students’ decision to transfer to Clarion.
Table 2 on the following page summarizes the impact of financial aid components on
transfer enrollment.
4
Table 2: Impact of Various Financial Aid Components on Transfer Enrollment
FINANCIAL AID
COMPONENTS
Grant Aid
ALL PA RESIDENTS
PA RESIDENTS WITH REPORTED EFC
Positive and linear impact
Positive and linear impact
Subsidized Loans
Positive and linear impact
Positive and linear impact
Unsubsidized Loans
Positive and linear impact
Positive and non-linear impact
EFC
-
No effect
5
In this section, Hanover Research provides an overview of the data and methodology used
to examine the impact of financial aid packages on student enrollment decisions. The
datasets provided to Hanover by Clarion include demographic, academic, and financial aid
information for 8,611 students who applied to a program at Clarion between 2012-13 and
2013-14. Demographic data include applicants’ gender, ethnicity, state of residency,
intended living arrangement, and athlete status, while academic information includes
Clarion’s decision on the application as well as the applicants’ SAT and ACT scores and high
school GPA and rank. In terms of financial aid data, the dataset includes the amount of
grants, scholarships, and loans offered to students. It further includes the tuition and
mandatory fees that the applicant is expected to pay. For some students, the Expected
Family Contribution from the Student Aid Report is included.
REGRESSION ANALYSIS
To examine the impact of financial aid packages on the probability of enrollment, the report
uses multivariate linear regression analysis. In our models, the dependent variable indicates
whether or not an accepted applicant enrolled at Clarion, while demographic, academic,
and financial characteristics of these applicants are included as controls. In addition to
showing the impact of financial aid packages on accepted students’ decision to enroll at
Clarion, these models also indicate other factors that strongly predict student enrollment.
Our models only take into account accepted freshman and transfer undergraduate
applicants who intend to earn a degree2 at Clarion.
We estimated separate models based on the type of students: freshman or transfer
students. For each of these students, we also estimated separate models for students
identified as PA residents and students who are not PA residents.3 The table below shows
the six different types of models that we included in this analysis. Our models can be
differentiated by the type of students they examine and the residency status of the student.
For each of these types of models, we included two variations: one with Expected Family
Contribution (EFC) and one without the EFC variable. Students who did not apply for aid do
not have a reported EFC, which results in a significant number of observations being
dropped when this variable is included in the analysis. Since the applicants being excluded
are likely systematically different from the non-excluded students (since applicants who do
not apply for financial aid are likely to be financially better off than those who do apply for
aid), these exclusions have the potential to bias our estimates of the effect of financial aid
on students’ probability of enrollment. Thus, we decided to include both variations in our
models.
2
3
Excludes non-degree seeking applicants.
We could not estimate models for transfer applicants who are not PA residents due to the small number of
applicants who met these criteria.
6
Figure 1.1: Different Types of Models
MODELS
TYPE OF STUDENTS
RESIDENCY
Model 1
Freshman
PA
EXPECTED FAMILY
CONTRIBUTION
Yes
Model 2
Freshman
PA
No
Model 3
Freshman
Non-PA
Yes
Model 4
Freshman
Non-PA
No
Model 5
Transfer
PA
Yes
Model 6
Transfer
PA
No
For each of our regression models, we provide coefficient estimates, as well as an indication
of which coefficients proved statistically significant. The coefficients of our models reveal
how much we expect the probability of the enrollment dependent variable to change when
the independent variable increases by one unit, holding all of the other independent
variables in the model constant. In this context, the coefficients capture the change in the
likelihood of enrolling at Clarion, depending on the model estimated.4
Our linear probability models account for the dichotomous nature of the dependent
variable. Dichotomous variables assume one of two values; in the context of the present
analysis, we assign a value of 1 to our dependent variable in cases where the student enrolls
at Clarion and a value of 0 if they do not enroll. Robust standard errors (included in the
results within parenthesis) were used in order to correct for heteroskedasticity. Lastly, when
presenting our results, we also include the R-squared value, which reveals the percentage of
variation in the dependent variable accounted for by the model.
DATA
In addition to providing us with the dataset that contains all the applications in the years
2012-13 and 2013-14, Clarion also provided us with a list of enrolled students in each of the
two years. Accepted applicants who were matched in the enrolled student datasets were
identified as students who chose to enroll at Clarion; others are considered as not choosing
to enroll at Clarion. For each of these enrolled and not enrolled students, our dataset
included demographic, academic, and financial aid information.
DEPENDENT (OUTCOME) VARIABLE
As discussed previously, our dependent variable indicates whether an accepted applicant
enrolled at Clarion. In the tables below, we provide descriptive statistics on our dependent
variables. Figure 1.2 shows that Clarion had 1,357 and 1,393 incoming new undergraduate
students (including only freshman and transfer students) in 2012 and 2013 respectively. In
total, 2,750 students were identified as enrolled students, and 1,600 accepted applicants
chose not to enroll at Clarion. Irregularities identified within these students were dropped
4
For SAT scores, our model shows the effect of a 10 point change, while for Expected Family Contribution, the model
indicates the effect of a $1,000 change.
7
(irregularities included students whose admission date was before the application date).
Furthermore, for enrolled students, we only considered those students whose starting term
matched their intended starting term,5 resulting in a total of 91 students being dropped.
This left our final dataset with 2,659 enrolled students and 1,600 students who did not
enroll.
Figure 1.2: Number of Freshman and Transfer Students by Year
APPLICATION YEAR
2012 NEW STUDENTS
2013 NEW STUDENTS
TOTAL ENROLLED
DID NOT ENROLL
2012-13
1,349
14
1,363
771
2013-14
8
1,379
1,387
829
Total
1,357
1,393
2,750
1,600
Drop
70
21
91
0
Total
1,287
1,372
2,659
1,600
For our analysis, we only examine the impact of financial aid packages on degree seeking
undergraduate students. Thus, we dropped any students who did not intend to earn a
bachelor’s degree or who were identified as non-degree seeking students. Figure 1.3
indicates that, after these restrictions, a total of 2,386 students enrolled, while 1,255
students did not enroll.
Figure 1.3: Enrollment by Year
APPLICATION YEAR
TOTAL ENROLLED
DID NOT ENROLL
2012-13
1,188
551
2013-14
1,198
704
Total
2,386
1,255
Figures 1.4 and 1.5 below show the breakdown of the enrolled and not enrolled students by
freshman and transfer status. We also include the breakdown by students’ Pennsylvania
(PA) residency status. A majority of the students are freshmen and PA residents.
Figure 1.4: Freshman and Transfer Enrollment by Year
APPLICATION YEAR
5
FRESHMAN
TRANSFER
ENROLLED
NOT ENROLLED
ENROLLED
NOT ENROLLED
2012-13
951
436
237
115
2013-14
975
590
223
114
Total
1,926
1,026
460
229
Some students had almost a year gap between their application and admission date. These students missed their
intended starting term and deferred to a later semester. In order to maintain consistency within our dataset, we
only considered students whose intended starting term in the application matched their actual starting term at
Clarion.
8
Figure 1.5: PA and Non-PA Enrollment by Year
APPLICATION
YEAR
2012-13
PA STUDENTS
NOT
ENROLLED
ENROLLED
1,104
469
NON-PA STUDENTS
NOT
ENROLLED
ENROLLED
84
91
2013-14
1,073
580
125
124
Total
2,177
1,049
209
215
INDEPENDENT VARIABLES
In Figures 1.6 through 1.9, we provide descriptive statistics for each of the variables used in
our models. Before estimating our final models, we examined several models with various
combinations of independent variables. In our preliminary models, we included squared
terms for the academic and financial aid variables to examine the non-linear effect of these
on the probability of student enrollment. The list of variables that we included in our final
models is presented in Figure 1.6 below.
Figure 1.6: Independent Variables
INDEPENDENT VARIABLES
SUMMARY
VARIABLE TYPE
Gender
Gender of students: (1) Male and (0) Female
Ethnicity of students selected from four ethnic categories:
Black, Multiracial, White, and Others
High School GPA of students
Combined SAT scores in Math and Critical Reading. This also
includes converted ACT scores.
Whether applicant intend to live on campus or off campus:
(1) On campus and (0) off campus
Whether students is an athlete: (1) Athlete and (0) Not an
athlete
Percentage of tuition and mandatory fees covered by grant
aid. Grant aid includes Pell Grants, SEOG Grants and PHEAA
Grants.
Percentage of tuition and mandatory fees covered by
subsidized loans.
Percentage of tuition and mandatory fees covered by
unsubsidized loans.
The square of the percentage of tuition and mandatory fees
covered by unsubsidized loans.
Expected Family Contribution of the students in thousands
of dollars.
Categorical
Ethnicity
HS GPA
SAT
On campus/off campus
Athlete
Grant Aid Coverage
Subsidized loan
coverage
Unsubsidized loan
coverage
Unsubsidized loan
coverage squared
Expected Family
Contribution
Categorical
Continuous
Continuous
Categorical
Categorical
Continuous
Continuous
Continuous
Continuous
Continuous
9
Figure 1.7 shows the descriptive statistics of the demographic variables that were used in
our models. The gender ratio of students remained more or less similar between the two
years. In terms of ethnicity,6 a majority of the students are White, while approximately a
fifth of the accepted students who decided not to enroll at Clarion are Black. The table
further shows that a vast majority of the students intend to live off campus and are not
athletes.
Figure 1.7: Independent Variables - Demographics
GENDER
2012-13
2013-14
ENROLLED
NOT ENROLLED
ENROLLED
NOT ENROLLED
Female
59.7% (n=709)
58.8% (n=324)
60.2% (n=721)
61.8% (n=435)
Male
40.3% (n=479)
41.2% (n=227)
39.8% (n=477)
38.2% (n=269)
Black
6.6% (n=79)
23.2% (n=128)
8.8% (n=106)
20.9% (n=147)
Multiracial
3.2% (n=38)
5.6% (n=31)
6.3% (n=75)
4.4% (n=31)
Other
2.9% (n=34)
6.4% (n=35)
3.9% (n=47)
5.1% (n=36)
White
87.3% (n=1037)
64.8% (n=357)
81% (n=970)
69.6% (n=490)
On campus
15.4% (n=183)
7.8% (n=43)
16.1% (n=193)
5.4% (n=38)
Off campus
84.6% (n=1005)
92.2% (n=508)
83.9% (n=1005)
94.6% (n=666)
Yes
3.7% (n=44)
0% (n=0)
2.6% (n=31)
0% (n=0)
No
96.3% (n=1144)
100% (n=551)
97.4% (n=1167)
100% (n=704)
ETHNICITY
ON CAMPUS/ OFF CAMPUS
ATHLETE
Figure 1.8 shows the applicants’ high school GPA and SAT scores. The SAT scores show the
combined scores in Math and Critical Reading. Several students reported ACT scores instead
of SAT scores. For these students, we converted the ACT scores to equivalent SAT scores
based on the concordance table released by ACT and College Board.7 In terms of academic
performance, enrolled students have better academic record compared to students who did
not enroll.
Figure 1.8: Independent Variables – Academic
HS GPA
ENROLLED
NOT ENROLLED
2012-13
2.58 (n=1188)
1.98 (n=551)
2013-14
2.63 (n=1198)
1.95 (n=704)
2012-13
963.17 (n=1052)
871.88 (n=463)
2013-14
948.86 (n=1198)
915.17 (n=704)
SAT
6
Others include American Indians, Asians, Hispanic, Pacific Islanders, and other minority groups.
Compare ACT & SAT Scores. http://www.act.org/solutions/college-career-readiness/compare-act-sat/
7
10
Figure 1.9 summarizes students’ expected tuition, financial aid offers, and expected family
contribution. We present the tuition and mandatory fees together as these are fixed costs
for all the students. Of all the students who did not enroll in 2013-14, expected tuition and
mandatory fees are available for only four students. This excludes a considerable number of
students who chose not to enroll at Clarion in 2013-14. As a result, our models exclude all
2013-14 students. Since tuition and mandatory fees are missing for a considerable number
of students, including them may bias our sample, since we would have approximately 1,198
enrolled students from 2013-14, but only four non-enrolled students. Nevertheless, we
include the descriptive statistics for the 2013-14 dataset in this section.
The grant aid includes Pell grants, SEOG grants, and PHEAA grants (only offered to PA
residents) received by students. In our models, we include the percentage of tuition and
mandatory fees covered by the grants and loans. Please note that since we excluded other
fees such as room and board and supplies, it is possible for the total aid and loans to be
higher than the tuition and mandatory fees. The EFC is included in our models in thousands
of dollars.
Figure 1.9: Independent Variables – Financial (Annual)
TUITION AND MANDATORY FEES
ENROLLED
NOT ENROLLED
2012-13
$8774.03 (n=1188)
$9110.29 (n=551)
2013-14
$9078.13 (n=1198)
$2006.2 (n=4)
2012-13
$2741.06 (n=1188)
$288.62 (n=551)
2013-14
$2605.55 (n=1198)
8
9
GRANT AID
$0 (n=704)
10
SUBSIDIZED LOANS
2012-13
$2199.66 (n=1188)
$144.68 (n=551)
2013-14
$1608.76 (n=1198)
$9.84 (n=704)
2012-13
$2380.69 (n=1188)
$156.4 (n=551)
2013-14
$1774.01 (n=1198)
$5.63 (n=704)
10
UNSUBSIDIZED LOANS
10
EXPECTED FAMILY CONTRIBUTION
2012-13
$12860.32 (n=902)
$10056.29 (n=188)
2013-14
$13084.56 (n=866)
$13675.64 (n=236)
8
Tuition and mandatory fees are available for only four students in 2013-14.
Grant aid includes Pell grants, SEOG grants, and PHEAA grants.
10
Limited amounts of grant aid, subsidized, and unsubsidized loan are available for non-enrolled students in 2013-14.
9
11
This section presents the results of multivariate analyses investigating the impact of
financial aid packages on undergraduate enrollment. For each dependent variable and
model, the results presented in the figures below display regression coefficients for each
independent variable and, where applicable, asterisks indicating the level of statistical
significance. Regression coefficients in the figures below can be interpreted as the change in
the probability of enrolling at Clarion due to a one unit change in the independent variable.
Please note that for grant aid coverage, subsidized loan coverage, and unsubsidized loan
coverage, the one unit increase refers to a percentage point increase in these variables (for
instance, the effect on the probability of enrolling at Clarion of a one percentage point
increase in the grant aid coverage). For SAT scores, the coefficients indicate the effect of a
10 point change in the score, while the coefficient for the EFC shows the effect of a $1,000
change. Please note that our models include only 2012-13 data. Students in the 2013-14
dataset were excluded from the analysis due to missing tuition and mandatory fees for nonenrolled students.
EFFECT OF FINANCIAL AID PACKAGE ON STUDENT ENROLLMENT
In this study, we concentrate on the impact of financial aid on freshman and transfer
students separately. We also exclude non-degree seeking students, as their responses to
financial aid packages are likely different from those of students entering as freshmen. For
both the freshman and transfer students, we examined the effect of financial aid packages
on PA resident and non-resident students separately.
FRESHMAN STUDENTS
We estimated four models with freshman enrollment as our dependent variable. Our
primary independent variables of interest are grant aid coverage, subsidized loan coverage,
and unsubsidized loan coverage. The aforementioned components of financial aid packages
are included as a percentage of the expected tuition and mandatory fees that the student is
expected to pay, had he/she been enrolled. Other financial data include students’ EFC,
which is included in thousands of dollars. Results from our regression models for freshman
enrollment are presented in Figure 2.1. The following is a summary of the impact of the
different components of financial aid packages on freshman enrollment:
Grant Aid Coverage: In general, the percentage of tuition and mandatory fees
covered by grants and scholarships has a positive and significant impact on the
probability that freshman applicants will enroll. However, it must be noted that no
impact is observed when all non-resident PA students (both those with and without
EFC data available) are considered. For PA residents (regardless of whether the
model is restricted to applicants with EFC data) and for non-PA residents with EFC
data, the probability of enrolling at Clarion as a freshman increases with higher
coverage of tuition and mandatory fees by grant aid.
12
Subsidized Loan Coverage: The percentage of tuition and mandatory fees covered
by subsidized loans has a positive and significant impact on the probability of PA
residents enrolling as freshman students at Clarion. An increase in subsidized loan
coverage by one percentage point is likely to increase the probability of a PA
resident enrolling at Clarion by 0.007. In fact, among all the components of the
financial aid package, subsidized loans have the strongest impact on the probability
of freshman enrollment for PA residents.
However, subsidized loan coverage has no significant effect on the probability of
accepted students who are not PA residents enrolling at Clarion as freshman
undergraduates. No effect is observed either for students who have a reported EFC
in their Student Aid Report or for those who do not.
Unsubsidized Loan Coverage: There is a non-linear relationship between the
probability of a PA resident with a reported EFC enrolling at Clarion and the amount
of tuition and mandatory fees covered by unsubsidized loans. The probability of
enrollment increases up to a certain point beyond which an increase in the
unsubsidized loans reduces the probability of students enrolling at Clarion. For
students who are PA residents and have a reported EFC, the probability of
enrollment increases until 82 percent of their tuition and mandatory fees are
covered by unsubsidized loans. However, for all PA residents in general,
unsubsidized loans have a linear and positive relationship with the probability of
enrolling at Clarion.
In terms of accepted students who are not residents of PA, the amount of
unsubsidized loans has a linear, positive, and significant impact on the probability of
enrolling as a freshman. For all non-PA resident students, a one percentage point
increase in the amount of unsubsidized loan coverage is associated with an increase
of 0.009 in the probability of enrolling at Clarion. The effect is slightly weaker when
only students with a reported EFC are considered (increase in the probability by
0.005).
Expected Family Contribution: There is a positive and significant impact of EFC on
the probability of students enrolling at Clarion. For PA applicants, an increase in the
EFC by $1,000 is expected to increase the probability of enrolling by 0.008. The
effect is slightly stronger for non-PA residents, with an increase in the EFC by $1,000
increasing the probability of enrollment by 0.01.
Demographic and Academic Characteristics: In addition to the aforementioned
components of the financial aid package, accepted students’ demographics and
academic backgrounds also have some impact on the probability of enrolling at
Clarion. For instance, PA residents with high SAT scores, any applicants with strong
high school GPAs, or applicants who are athletes have a higher probability of
enrolling as freshman undergraduates at Clarion, compared to their respective
counterparts. In terms of ethnicity, White students who are significantly more likely
13
to enroll compared to any other ethnic groups among PA residents. However, PA
residents who plan to live on-campus are less likely to eventually enroll at Clarion
than those who plan to live off-campus.
Figure 2.1: Regression Results – Freshman Enrollment
VARIABLES
Male [Reference Group = Female]
PA RESIDENTS
MODEL 1 WITH EFC
MODEL 2
-0.0218
-0.0027
NON-PA RESIDENTS
MODEL 3 WITH EFC
MODEL 4
-0.0758
-0.027
(0.0218)
(0.0201)
(0.1187)
(0.1034)
-0.1001*
-0.2519***
-0.1372
-0.177*
(0.0531)
(0.0422)
(0.1206)
(0.0934)
-0.0675
-0.1771***
0.1231
0.0909
(0.0421)
(0.0441)
(0.1155)
(0.231)
-0.1297
-0.1508**
0.3029**
-0.0865
(0.085)
(0.065)
(0.1429)
(0.1861)
0.0038***
0.007***
0.0091*
0.0061
(0.0008)
(0.0007)
(0.0051)
(0.0038)
0.012
0.0198**
-0.0003
0.0635**
(0.0084)
(0.0076)
(0.0303)
(0.0263)
-0.0981**
-0.112***
(0.0321)
(0.0303)
0.0943**
0.2161***
0.2513***
0.3951***
(0.0358)
(0.0471)
(0.0757)
(0.0934)
Grant Aid Coverage
0.0023***
0.0015***
0.0098**
0.004
(0.0004)
(0.0003)
(0.0032)
(0.005)
Subsidized Loan Coverage
0.0073***
0.0069***
0.0065
0.0045
(0.0007)
(0.0007)
(0.0045)
(0.0048)
0.0052***
0.003**
0.005**
0.0091**
(0.0005)
(0.0013)
(0.0022)
(0.0031)
Black [Reference Group = White]
Multiracial [Reference Group = White]
Other [Reference Group = White]
SAT Score (per 10 points)
High School GPA
On Campus [Reference Group = off campus]
Athlete [Reference Group = Not an athlete]
Unsubsidized Loan Coverage
Unsubsidized Loan Coverage squared
-0.00003***
(0.000002)
Expected Family Contribution (per $1,000)
Intercept
R-squared
Number of observations
0.0082***
0.0101**
(0.0015)
(0.0035)
0.1495
-0.0882
-0.4728
-0.3107
(0.0932)
(0.0768)
(0.4757)
(0.3695)
0.3635
833
0.4188
1,253
0.4407
56
0.3493
95
Coefficients estimated using Ordinary Least Squares with a linear regression model, with robust standard errors in
parenetheses. Statistical significance indicator using asterisk next to the coefficients, with * = significant at 10%,
** = significant at 5%, and *** = significant at 1%.
14
TRANSFER STUDENTS (PA RESIDENTS ONLY)
For transfer applicants, we only estimated models with PA residents. We did not have
sufficient data to estimate models for non-PA residents. The models include the same
independent variables that were included in Models 1 through 4. Results from our
regression models for transfer enrollment are presented in Figure 2.2. The following is a
summary of the impact of the different components of financial aid packages:
Grant Aid Coverage: The amount of grant aid awarded to PA resident transfer
applicants has a positive and significant impact on their likelihood of enrolling at
Clarion. For all PA residents, a one percentage point increase in the percentage of
tuition and mandatory fees covered by grant aid is associated with an increase in the
probability of enrolling at Clarion by 0.0016.
Subsidized Loan Coverage: For all transfer applicants residing in PA, a one
percentage point increase in the proportion of tuition and mandatory fees covered
by subsidized loans is expected to lead to an increase in the probability of enrolling
by 0.006. For PA residents with a reported EFC, the effect is slightly weaker
(increases probability of enrollment by 0.003).
Unsubsidized Loan Coverage: There is a non-linear relationship between the
probability of a PA resident with a reported EFC enrolling as a transfer student at
Clarion and the amount of tuition and mandatory fees covered by unsubsidized
loans. The probability of transfer enrollment increases up to a certain point, beyond
which an increase in the unsubsidized loans reduces the probability of students
enrolling at Clarion. For students who are PA residents and have a reported EFC, the
probability of transfer enrollment increases until 89 percent of their tuition and
mandatory fees are covered by unsubsidized loans.
However, for all PA residents in general, unsubsidized loans have a linear and
positive relationship with the probability of a transfer enrollment at Clarion. A one
percentage point increase in the amount of unsubsidized loan coverage is expected
to lead to the probability of these students enrolling as a transfer student at Clarion
increasing by 0.004.
Expected Family Contribution: The EFC reported on a transfer student’s Student Aid
Report does not have any significant impact on their probability of enrolling at
Clarion.
Demographic and Academic Characteristics: Further analysis of Models 5 and 6
indicates that some demographic and academic characteristics of PA resident
transfer applicants influence their probability of enrolling at Clarion. Female
applicants (regardless of whether we limit the analysis to students with a reported
EFC) are less likely to enroll as transfer students compared to male applicants
residing in PA. There is also a slightly higher probability of applicants who are
15
categorized as ‘multiracial’ or ‘other’ in terms of ethnicity to enroll as transfer
students compared to White students, when limiting the analysis to those students
with a reported EFC. Applicants (with or without reported EFC) with higher SAT
scores and applicants who are athletes (only with reported EFC) also have a
significantly higher probability of enrolling at Clarion.
Figure 2.2: Regression Results – Transfer Enrollment
VARIABLES
Male [Reference Group = Female]
Black [Reference Group = White]
Multiracial [Reference Group = White]
Other [Reference Group = White]
PA RESIDENTS
MODEL 5 WITH EFC
MODEL 6
-0.0212
0.1239**
(0.06)
(0.0561)
-0.0639
-0.0942
(0.0941)
(0.0703)
0.2164*
0.0246
(0.1253)
(0.1604)
0.1425*
0.0465
(0.0794)
(0.0912)
0.0009
0.0045**
(0.0013)
(0.0017)
High School GPA
0.0122
0.0189
(0.025)
(0.0232)
On Campus [Reference Group = off campus]
-0.0672
-0.0277
(0.0879)
(0.0837)
0.18*
0.0049
(0.0976)
(0.1933)
SAT Score (per 10 points)
Athlete [Reference Group = Not an athlete]
Grant Aid Coverage (per 1%)
0.0018*
0.0016**
(0.0011)
(0.0005)
0.003**
0.006***
(0.0013)
(0.0011)
Unsubsidized loan Coverage
0.005**
0.0044***
(0.0019)
(0.0007)
Unsubsidized Loan Coverage squared
-0.00003*
Subsidized Loan Coverage
(0.00002)
Expected Family Contribution (per $1,000)
0.0029
(0.0051)
Intercept
0.5949**
-0.0162
(0.1969)
(0.1855)
R-squared
0.1438
0.4066
Number of observations
101
152
Coefficients estimated using Ordinary Least Squares with a linear regression model, with robust standard errors in
parentheses. Statistical significance indicator using asterisk next to the coefficients, with * = significant at 10%, ** =
significant at 5%, and *** = significant at 1%.
16
Hanover Research is committed to providing a work product that meets or exceeds partner
expectations. In keeping with that goal, we would like to hear your opinions regarding our
reports. Feedback is critically important and serves as the strongest mechanism by which we
tailor our research to your organization. When you have had a chance to evaluate this
report, please take a moment to fill out the following questionnaire.
http://www.hanoverresearch.com/evaluation/index.php
The publisher and authors have used their best efforts in preparing this brief. The publisher
and authors make no representations or warranties with respect to the accuracy or
completeness of the contents of this brief and specifically disclaim any implied warranties of
fitness for a particular purpose. There are no warranties which extend beyond the
descriptions contained in this paragraph. No warranty may be created or extended by
representatives of Hanover Research or its marketing materials. The accuracy and
completeness of the information provided herein and the opinions stated herein are not
guaranteed or warranted to produce any particular results, and the advice and strategies
contained herein may not be suitable for every partner. Neither the publisher nor the
authors shall be liable for any loss of profit or any other commercial damages, including but
not limited to special, incidental, consequential, or other damages. Moreover, Hanover
Research is not engaged in rendering legal, accounting, or other professional services.
Partners requiring such services are advised to consult an appropriate professional.
17
1700 K Street, NW, 8th Floor
Washington, DC 20006
P 202.559.0500 F 866.808.6585
www.hanoverresearch.com
18