Cantor JC, Bergeisen L, Baker LC. Effect of an Intensive Educational Program for Minority College Students
and Recent Graduates on the Probability of Acceptance to Medical School. JAMA. 1998;280(9):772–776. doi:10.1001/jama.280.9.772
Context.— Increasing the number of minority physicians is a long-standing goal
of professional associations and government.
Objective.— To determine the effectiveness of an intensive summer educational program
for minority college students and recent graduates on the probability of acceptance
to medical school.
Design.— Nonconcurrent prospective cohort study based on data from medical school
applications, Medical College Admission Tests, and the Association of American
Medical Colleges Student and Applicant Information Management System.
Setting.— Eight US medical schools or consortia of medical schools.
Participants.— Underrepresented minority (black, Mexican American, mainland Puerto
Rican, and American Indian) applicants to US allopathic medical schools in
1997 (N=3830), 1996 (N=4654), and 1992 (N=3447).
Intervention.— The Minority Medical Education Program (MMEP), a 6-week, residential
summer educational program focused on training in the sciences and improvement
of writing, verbal reasoning, studying, test taking, and presentation skills.
Main Outcome Measure.— Probability of acceptance to at least 1 medical school.
Results.— In the 1997 medical school application cohort, 223 (49.3%) of 452 MMEP
participants were accepted compared with 1406 (41.6%) of 3378 minority nonparticipants
(P=.002). Positive and significant program effects
were also found in the 1996 (P=.01) and 1992 (P=.005) cohorts and in multivariate analysis after adjusting
for nonprogrammatic factors likely to influence acceptance (P<.001). Program effects were also observed in students who participated
in the MMEP early in college as well as those who participated later and among
those with relatively high as well as low grades and test scores.
Conclusions.— The MMEP enhanced the probability of medical school acceptance among
its participants. Intensive summer education is a strategy that may help improve
diversity in the physician workforce.
INCREASING the number and proportion of minority physicians is a long-standing
goal of government and professional associations in the United States.1 Four racial/ethnic groups—blacks, Mexican Americans,
mainland Puerto Ricans, and American Indians—are designated as underrepresented
in medicine by the Association of American Medical Colleges (AAMC).2 Together these groups comprise 21% of the US population,3 but only 8.6% of the US physician population1 and about 13% of entering medical school classes.4 In addition to concerns about achieving equitable
access to careers in medicine for minority groups, the lack of parity of minority
populations in medicine raises concerns about access to medical care for underserved
populations. Minority physicians are considerably more likely to serve minority
and low-income communities that traditionally have poor access to care.5- 7
In this article, we assess one strategy for improving the representation
of minorities in medicine: academic enrichment programs for college students
and recent graduates. As of 1991, nearly half of all medical schools sponsored
such preprofessional programs.8 We evaluate
the effectiveness of one large 8-site program, the Robert Wood Johnson Foundation
Minority Medical Education Program (MMEP). Like other programs of its type,
the MMEP seeks to improve the medical school acceptance rate of its participants.
Thus, we evaluate the MMEP by comparing the probability of medical school
acceptance of its participants with that of other underrepresented minority
applicants. We also explore the degree of impact of MMEP on selected subgroups
that may benefit more or less from participation.
The MMEP, established by the Robert Wood Johnson Foundation, Princeton,
NJ, in 1989, was initially administered at 6 sites around the country. In
1995, through a national competition, 4 of the original sites were renewed
and 4 new sites were selected. Today, the MMEP is administered by the AAMC
and conducted in 8 medical schools or consortia of schools (the complete listing
of schools is at the end of the article).
The 8 MMEP sites share the objective of enhancing the chances of medical
school acceptance for their participants. All are 6-week residential programs,
and all share a set of core program elements. The core elements include training
in studying, test taking, interviewing, and presentation skills. All sites
also conduct classroom training in the sciences, writing, and verbal reasoning,
and all provide clinical experiences.
The MMEP accepts participants from the AAMC-defined underrepresented
minority groups and is limited to US citizens and permanent residents. Participants
are required to have completed at least 1 year of college by the start of
the program. The focus of MMEP is on enrichment not remediation. Accordingly,
program acceptance criteria include a minimum overall grade-point average
(GPA) of 3.0 (on a 4.0 scale), including a minimum 2.75 in the sciences, and
an Scholastic Aptitude Test (SAT) score of at least 950 or an American College
Test score of at least 20. The MMEP uses a rigorous admission application
process designed to parallel that which is used by the American Medical College
Application Service (AMCAS). Applicants must submit college admission test
scores, transcripts, a personal statement, and letters of recommendation.
Data are drawn from 3 medical school application cohorts (1992, 1996,
and 1997) from sources maintained in the AAMC Student and Applicant Information
Management System.9 Freshman and cumulative
GPAs in science courses (biology, physics, chemistry, and mathematics) and
all other subjects, Medical College Admission Test (MCAT) scores, race/ethnicity,
and medical school acceptance data were drawn from medical school application
data of the AMCAS or the University of Texas System Medical and Dental Application
Center. Data on the highest level of educational attainment of either of the
applicant's parents are from the AAMC Prematriculation Questionnaire, which
is administered with the MCAT. Medical school applicants who participated
in the MMEP between 1987 and 1996 were identified by matching MMEP program
records to AMCAS data.
We begin by describing MMEP participants by program year. Then we focus
on the central objective of the MMEP, enhancing the chances of acceptance
to medical school among participants. Annual medical school acceptance rates
of MMEP participants are compared with acceptance rates of other underrepresented
minority applicants (nonparticipants). After providing descriptive comparisons
of participants to nonparticipants, we use a χ2 test to measure
whether the medical school acceptance rate of MMEP participants differs from
that of nonparticipants. However, estimating the true contribution of the
MMEP to the acceptance probability requires adjustment for nonprogrammatic
factors that could influence acceptance. Thus, we use logistic regression
to calculate odds ratios (ORs) of medical school acceptance for participants
compared with nonparticipants. In addition to estimating an unadjusted OR,
we estimate models that adjust for measures of academic achievement, demographics,
and other factors.
Two regression specifications are used to adjust for academic factors.
First, we estimate a baseline adjustment model. This model controls for the
freshman GPAs in science courses and all other subjects, and the competitiveness
of the undergraduate institution (measured by the mean SAT score of each applicant's
undergraduate institution). Second, we estimate a full adjustment model in
which we replace the freshman GPA variables with the applicants' cumulative
GPA for science and all other subjects, and we add each applicant's total
MCAT score. The baseline adjustment model controls only for academic factors
that were observable prior to MMEP participation. The full adjustment model
includes control variables that may have been influenced by MMEP participation
(ie, college major, cumulative grades, and MCAT scores) but potentially provides
a stronger self-selection adjustment.
Both the baseline and full adjustment models also control for nonacademic
factors that we expected to be associated with acceptance to medical school.
First, we control for the number of years elapsed between college graduation
and medical school application. This variable is highly correlated with (r=0.75; P <.001) and effectively
controls for applicant age. Second, we control for whether an individual is
applying to medical school for the first time or had applied in prior years.
Third, we control for whether the applicant decided to pursue a medical career
prior to high school graduation as a proxy measure of the degree of applicant
motivation to pursue a medical career. Fourth, we control for the highest
level of education of either of the applicant's parents as a measure of socioeconomic
status (SES). Because a large proportion of data on parental income is missing
from the AAMC Student and Applicant Information Management System and because
of concerns about students' ability to accurately report their parents' income,
parental education is the only SES included in the models. Although parental
education data are more complete than income data, just over one third of
applicants is missing parental education data. To avoid biasing our results,
we include a dummy variable indicating missing data in the regression models.
Finally, the models include variables that classify applicants by their race
The analyses described above are conducted on the 1997 medical school
application cohort. In that year, there were 4604 underrepresented minority
medical school applicants, 526 of whom had previously participated in the
MMEP. Data for one or more of the independent variables in the regression
models were lacking for 774 cases, leaving a final analysis sample of 3830
cases (85.6%). The probability of being excluded from the analysis due to
missing data is not significantly different between the MMEP and comparison
In addition to estimating the models on the full 1997 cohort, we examine
several subgroups and additional populations. First, we estimate program impact
in the 1997 cohort separately for those who participated in MMEP after their
freshman or sophomore year (lower division) and those who participated after
their junior year or later (upper division). In the unadjusted and fully adjusted
models, we test whether the estimated program effects for lower- and upper-division
participants are significant. In addition, using the fully adjusted model,
we test whether the program effects in the upper- and lower-division participants
Second, we estimate unadjusted and fully adjusted ORs in 1996 and 1992
application cohorts to compare the program effect in each of these years with
that measured in the 1997 cohort. We make the 1996 to 1997 comparison because
the medical school application environment for minority candidates is changing
rapidly. By 1997, affirmative action had come under intensive scrutiny,10 and the number of minority medical applicants had
declined.11 The models estimated for the 1992
application cohort provide a window on the effect of MMEP in its early years.
Using coefficients from variables that interact application year with MMEP
participation, we test whether program effects are equal between 1992 and
1997 and between 1996 and 1997 in the fully adjusted model.
Third, we examine the impact of MMEP in 3 groups stratified by their
level of academic achievement. The MMEP may be more or less effective in different
subgroups, and knowing the program impact relative to academically similar
peer groups may be helpful in shaping future program recruitment strategies.
The first of the 3 groups has both relatively low cumulative science GPAs
(<3.0) and low MCAT scores (an average score <7). The second group has
discordant science grades and MCAT scores: either the GPA or MCAT is below
the threshold, but not both. The third group has science grades and MCAT scores
both above these levels. Data from the 1996 and 1997 application years are
pooled to reach an adequate sample size for this analysis. A variable indicating
application cohort is added to the regression models used to estimate both
the unadjusted and adjusted ORs. The total available sample size for these
models was 3147 (low grades and scores), 3211 (discordant grades and scores),
and 2027 (high grades and scores). We test for equality of program effects
between the high and low and between the discordant and low academic groups
by estimating coefficients for the interaction of MMEP participation with
Table 1 shows the number
of participants and medical school application and acceptance rates for each
program year. Between the summers of 1989 and 1997, the MMEP has served 6479
participants. Of these, 2742 have applied to medical school, 1728 were accepted,
1640 matriculated, and 509 have graduated. Medical school application rates
were between about 50% and 60% through 1994 and are lower thereafter. The
lower recent application rates reflect the fact that many of the more recent
participants have not yet reached the point at which medical school application
is normally made. Among MMEP participants who subsequently applied to a US
allopathic medical school between 1990 and 1997, 63.0% have been accepted.
The cumulative acceptance rates are 59% or higher for each MMEP class through
1994. After 1994, the rates were lower, although resubmitted applications
to medical school by previously unsuccessful applicants in these MMEP classes
can be expected to lead to higher cumulative acceptance rates in the future.
The remaining analyses focus on medical school application cohorts.
In the 1997 cohort, those who participated in MMEP differed from other underrepresented
minority applicants in several important respects. Table 2 shows that there are statistically significant differences
between participants and nonparticipants in levels of academic achievement,
but the direction of selection bias is not uniform. The GPAs of participants
were about 0.1 points higher on a 4.0 scale (0.20 SD), but the mean SAT scores
of their colleges were lower by 54 of 1600 points (0.31 SD). The MMEP participants
were about 10 percentage points more likely to have set their sights on medical
school by the end of high school. No differences were observed in total MCAT
scores, the percentage of applicants who were science majors, or the percentage
who were applying to medical school for the first time. The distribution of
parental educational background was nearly identical for the 2 groups, including
whether education data were missing. Finally, compared with nonparticipants,
black women were more likely to have participated in MMEP, while nonblack
men were less likely to have participated.
The medical school acceptance rate for 1406 (49.3%) of the MMEP participants
was higher than for 223 (41.6%) of the nonparticipants in 1997 (P=.002). The ORs for acceptance among MMEP participants compared with
nonparticipants in the 1997 application cohort are 1.37 in the unadjusted
regression (P=.002), 1.60 in the baseline adjusted
model (P<.001), and 1.69 in the fully adjusted
model (P=.001) (Table 3). These results provide evidence that the MMEP increases
the chances of acceptance among its participants. (Although ORs provide a
useful analysis tool, some care should be used in interpreting their magnitude.
Because medical school acceptance is relatively common, these ORs are larger
than the corresponding relative risks.)
The ORs of most other variables in the adjusted models are consistent
with expectations. Notably, OR estimates for GPAs, average SAT scores, and
MCAT scores are all greater than 1 and significant in both multivariate models.
Contrary to our expectation, being a first-time applicant does not significantly
increase the probability of acceptance. Parental education has the expected
effect in the baseline model, but the effect reverses in the fully adjusted
model. The estimates on this variable may be unstable because of a high degree
of association between parental education and academic performance.
Table 4 shows the interaction
of the MMEP program effect with the timing of participation. We find positive
and significant program effects for both lower- and upper-division participants.
The ORs for lower-division participants are larger than those for upper-division
participants, although this difference is not statistically significant (P=.20 in the fully adjusted model). Beyond the program-effect
terms, coefficients in these models are virtually identical to those shown
in Table 3 and are not repeated
The MMEP effects in the 1992 and 1996 application cohorts were similar
to those found in 1997. In 1992, 207 (56.6%) of the 366 MMEP participants
were accepted to medical school, compared with 1505 (48.8%) of 3081 nonparticipants
(P=.005); and in 1996, 243 (45.6%) of 533 participants
were accepted, compared with 1638 (39.7%) of 4121 nonparticipants (P=.01). These acceptance rates are expressed as ORs in the top part
of Table 5. The unadjusted ORs
are about the same for all 3 years, and there is little difference between
adjusted ORs in the 1996 and 1997 cohorts (P=.22).
Although the adjusted OR in 1997 is about 40% higher than the OR in 1992,
this difference is also not significant (P=.45).
The bottom part of Table 5
shows the MMEP impact in 3 groups defined by level of academic achievement.
The MMEP participants have a significantly higher probability of medical school
acceptance within each group. The MMEP participation generates a higher OR
in the group with both high GPAs and MCAT scores compared with the other academic
groups. In the fully adjusted model, the program effect in the high-academic
group is significantly higher than the program effect in the low group (P=.02), but the discordant does not differ significantly
from the low group (P=.15).
The MMEP appears to enhance the probability that its participants will
be accepted to medical school. In the 1997 medical school application cohort,
the odds of acceptance were significantly higher for MMEP participants compared
with nonparticipants, after controlling for college grades, MCAT scores, and
other factors likely to influence acceptance.
The effect of intensive preparation programs like MMEP may be even greater
than our analysis suggests. First, it is likely that many of the medical school
applicants in our comparison group participated in activities similar to those
provided by MMEP. In 1991, nearly half of all medical schools sponsored undergraduate
preprofessional enrichment programs.8 If, like
MMEP, these programs contribute to the probability of medical school acceptance
then we have underestimated the effect of efforts of this type. Second, the
MMEP provides activities to improve academic performance, including practice
MCAT examinations and class work in the sciences. Therefore, adjusting for
cumulative grades and MCAT scores may control away part of the program effect.
Nevertheless, even after controlling for these academic performance measures
we find our highest point estimate of the program impact.
We found a consistently positive and statistically significant effect
of MMEP in each of several subgroups of participants that we examined. Estimates
of the program effects of participant subgroups provide some insights into
how the MMEP achieves its impact. First, the MMEP effect was as great for
those who participate after their freshman or sophomore years as it was for
upper-division participants. This finding is consistent with the view that
the MMEP is not simply a "crash course" for the MCAT or other aspects of preparation
for application and that it has a sustained effect. The MMEP may shape the
way its participants prepare for medical school application well after the
Second, MMEP had a positive and significant impact on groups of participants
with a broad range of academic preparation, but the degree of impact varied.
The MMEP had a significantly larger impact among applicants with relatively
high cumulative science grades and MCAT scores compared with those with lower
grades and scores. This evidence supports the view that MMEP is not remedial,
but that it augments skills even in the group with relatively strong academics.
Finally, estimates of program impact in the 1992 and 1996 application
cohorts are statistically equivalent to the 1997 estimate. These additional
observations show that the MMEP effect is robust in the face of changes in
the medical school application pool and the changing affirmative action environment,
at least through 1997.
Several important evaluation questions were beyond the scope of our
analysis. First, additional studies of participant subgroups or selected programmatic
components of the MMEP (or similar programs) would assess more fully the mechanisms
by which the program achieves its effects. Second, description of the group
of MMEP participants who never apply to medical school could help program
managers refine their recruitment strategies. Finally, long-term follow-up
comparing MMEP participants who are accepted to medical school with nonparticipants
would measure the degree to which MMEP may lead to different outcomes (eg,
attrition rates, specialty choice, and levels of service to underserved populations).
Our analyses draw on a rich source of existing data derived from the
medical school application process, but these data have some limitations.
Data available to adjust for the SES of the study population are limited.
In addition, other than controlling for the timing of the decision to seek
a career in medicine, we could not adjust for the degree of applicant motivation
to become a physician. Thus, although we were able to adjust for many factors
related to MMEP participant self-selection and the underlying probability
of acceptance, we cannot rule out the possibility that failing to include
better measures may have biased our results. But because we control for some
aspects of SES, and applicant motivation and controls for academic factors
are strong, we believe that the likelihood of substantial bias is small.
In conclusion, this study provides strong evidence that intensive premedical
preparation programs have their desired effect—boosting the chances
of medical school acceptance among minority applicants. In the near term,
the MMEP and similar programs cannot bring the minority physician population
into parity, but continuing or expanding summer programs can be an important
part of a broader strategy to diversify the physician workforce.