RETRACTED ARTICLE: Summer Research Training Provides Effective Tools for Underrepresented Minorities to Obtain Doctoral Level Degrees

The ethnic, racial, and cultural diversity of the USA is not reflected in its healthcare and biomedical workforce. Undergraduate research programs are used to encourage underrepresented minorities to pursue training for biomedical careers, but there is limited published data on doctoral degree completion rates by participants in these programs. This study evaluates the attainment of doctoral degrees by a NIH-funded cohort of undergraduates from the 1998–2002 Baylor College of Medicine’s Summer Medical Research Training (SMART) program. Variables collected included race, ethnicity, gender, tier status of undergraduate institution, and terminal degree status. The majority of cohort participants were underrepresented minorities, while the rest were socio-economically underserved, or first generation college students. Of the 163 participants, 65 (56%) were confirmed to have completed doctoral level biomedical or healthcare degrees: 48 MDs (41%), 22 PhDs (19%), 4 MD/PhDs (3%), and 1 PharmD (1%). There was no statistical significant difference in doctoral degree attainment based on gender, ethnicity, or tier status of participants. The completion rates of doctoral degrees by underrepresented minority participants (65%) in this program exceeds that of underrepresented minority graduate students (38%) and that of non-minority graduate students (51%) nationally based on National Science Foundation data for US citizens and permanent residents, suggesting that this 9-week summer research experience provides a pathway for underrepresented minorities to successfully attain the doctoral degrees required for careers in the health sciences.


Introduction
African Americans, Latinos, and indigenous people make up over 30% of the US population but less than 9% of health and biomedical professions, and this low percentage is problematic since a culturally competent and diverse workforce is an integral part of eradicating health disparities [1]. Additionally, the percentage of underrepresented minorities graduating with doctoral degrees in science, technology, engineering, and mathematics (STEM) is likewise low, increasing slightly from 9.5 to~10.9% from 2003 to 2008 [2]. Interventions to increase the diversity of doctoral level scientists include undergraduate research experiences [3][4][5][6][7][8][9][10][11][12][13]. These research experiences are either part-time during the school year or full-time during the summer and offer the opportunity to learn new skills and develop an appreciation for applications of scientific concepts that were introduced in the classroom while learning the scientific skills required to think as a scientist [3][4][5][14][15][16].

R E T R A C T E D A R T I C L E
Despite the integration of undergraduate research experiences into many STEM undergraduate programs and extensive investments from federal, state, and private organizations into these programs, there is only limited published data on doctoral degree completion by participants of these programs, due to an absence of funding to carry out extensive longitudinal evaluation studies [9,17,18]. Some published evaluations use admission to graduate or professional school as yardsticks of success despite data from the National Science Foundation that reveals a persistent and significant attrition rate for attaining doctoral level degrees in STEM fields by underrepresented minorities. Specifically, only 38.1% of enrolled underrepresented minorities attained their PhDs as compared to 51.3% of non-minorities [19,27]. Others including the highly successful Meyerhoff scholarship program at the University of Maryland Baltimore County, which significantly increased the likelihood of pursuing doctoral degrees in STEM, focus on academic year long STEM Undergraduate Research Programs [20][21][22][23].
In an effort to close the missing gap in the literature, a goalbased Tylerian model evaluation of the 1998-2002 cohort of the Summer Medical and Research Training (SMART) Program at the Baylor College of Medicine (BCM) was performed, with the objective to determine doctoral level advanced degree completion by underrepresented minorities as well as to conduct a sub-analysis on the impact of choice or self-efficacy on long-term outcomes. This evaluation was completed after sufficient time had elapsed to allow participants to complete doctoral level degrees.

Analysis Framework
A goal-based Tylerian model was used as a framework to investigate the long-term doctoral degree completion of the 1998-2002 cohort of the SMART program. This cohort allowed sufficient time for attaining the principal goal of this summer research program, which was training students to successfully achieve doctoral-level degrees in biomedical sciences. Furthermore, this time framed allowed a sub-analysis on choice and doctoral degree completion.

Setting and Participants
Participant description and demographics are listed in Table 1. Baylor College of Medicine (BCM) is one of the top ranked medical schools in the USA and is located within the Texas Medical Center, the world's largest medical metropolis. BCM has four schools: the medical school, graduate school, school of allied health, and the national school of tropical medicine, which all offer professional and post-baccalaureate degrees and diplomas. BCM does not offer any undergraduate degrees. With over a century of biomedical research, BCM has been a top 20-funded NIH-funded medical school since 2006 with $205,461,582 in NIH funding in 2015.
The participants received funding from the National Institute of General Medi cal Sci ences (NIGMS R25 GM56929) or the National Heart, Lung, and Blood Institute (NHLBI HL07480). Sub-analysis on choice as a measure of self-efficacy was facilitated because participants funded by the NIGMS could choose a research program of interest (from >22 research programs at Baylor College of Medicine), while those funded by the NHLBI were limited to specific research mentors involved in heart, lung, or blood research. To minimize the effects of other variables such as career development activities, the participants were chosen from overlapping years because these students were involved in identical activities through the SMART program.

Program Description
The SMART program was developed to provide frontier-level, biomedical summer research projects for undergraduates in a supportive environment with supplemental educational activities (Table 2).

Results
Terminal degrees completed by participants are listed in Table 3 while Tables 4, 5, 6, and 7 list the percentages of students that attained degrees based on the independent variables.

Success Regardless of Choice of Research Area
A sub-analysis was performed to determine if the ability to choose the discipline of their research projects impacted the attainment of doctoral degrees by participant. The challenges to completing doctoral degrees are well known and studied [24]; however, on a fundamental level, it is believed that aspects of the goal (doctoral degree) that make achievement of the goal important to the individual and the individual's selfefficacy (for example, choice) play a role in attaining goals [25]. The confirmed advanced degree outcomes for participants from both programs are summarized in Table 3. Similar percentages of students in both cohorts completed doctoral level biomedical degrees (Table 4), whereas a greater percentage of those who had no choice did not complete any advanced degree (Fig. 1). A statistical analysis of completion of doctoral level or advanced degrees by participants with and without choice in research revealed no statistical differences between the two groups (Tables 8 and 9).

Undergraduate Institution Type and Doctoral Degree Completion
Access to undergraduate research opportunities varies across different types of tertiary institutions. Tier 1 and research-intensive universities typically have the ability to provide students the exposure and opportunities to interact with faculty conducting state-of-the-art research. Undergraduate students at tier 1-type research-intensive universities expect to have access to research opportunities, and these colleges are ranked by US News and World reports based on this access. Conversely, such opportunities are not as readily available at schools that are not research intensive [26]. Since the students participating in the summer research programs were from diverse types of undergraduate institutions, the data was probed to determine if attending a tier 1 vs. a non-tier 1 undergraduate school correlated with increased advanced degree completion rates. A larger percentage of participants from both cohorts attended non-tier 1 schools, and for some of these students, this summer program was their first exposure to research outside their classes. Despite this difference in educational background, similar percentages of participants from tier 1 and non-tier 1 schools successfully completed doctoral level degrees (Fig. 1, Table 5). There are no statistically significant differences in doctoral or advanced level degree completion between participants from tier 1 and other schools (Tables 8 and 9).

Gender and Doctoral Degree Completion
Gender imbalances in STEM fields are well known, and overall, more American men complete doctorate and masters level degrees than women. Interestingly, the situation differs for underrepresented minority women who attain doctorate degrees at higher rates than underrepresented minority men [27,28]. Male and female participants had high biomedical sciences doctoral rates (Table 6), and there was no statistical difference in post-graduate degree completion by gender (Fig. 1, Tables 8 and 9). The gender analysis reveals one of the major limitations in long-term follow-up studies-the difficulty tracking subjects, especially after more than a decade since participation in a program, and females were harder to track than men.

Race, Ethnicity, and Doctoral Degree Completion
A major objective of this summer program is to increase the numbers of underrepresented minorities with doctoral level degrees in the biomedical sciences. This cohort of participants reveals high numbers and percentages of doctoral degrees by participants that self-identified as African American or Latino/ a (Tables 3 and 7). Another limitation to the more detailed analysis of the ethnicity and race was the small numbers of white students in the cohort, since the program targeted mostly underrepresented minorities. Instead, members of underrepresented groups were compared to participants that were not underrepresented minorities, and there was no statistical difference in completion of advanced degrees (Fig. 1, Tables 8 Fig. 1 a Doctoral level degree completion percentage. b All advanced degree completion percentage

Comparison with the Meyerhoff Program
As mentioned previously, one of the limitations of this analysis is the large numbers of lost participants. This analysis was conservative and treated lost participants as non-degreed. If less conservative analysis were performed and missing participants were discarded, 84% of the SMART participants completed doctoral degrees while 94% completed some advanced Masters level degree. Regardless of the methodology, the outcomes support previous observations that undergraduate research experiences enhance the likelihood of attaining advanced biomedical degrees [6,8,11,13]. Furthermore, doctoral degree completion percentages for the SMART program are comparable to the enrollment percentages of the most successful year-long research programs on record for the Meyerhoff scholarship program at the University of Maryland Baltimore county [20][21][22]. The Meyerhoff program is considered the national model of a successful program and 68.6% of the 1996-2005 cohort entered either a doctoral level degree in the STEM or MD [20]. The Meyerhoff Scholars program takes a holistic approach to ensuring that scholars achieve a research-based PhD degree and includes comprehensive training program that offers summer research, tutoring, mentorship, and other support over the 4 years of undergraduate studies [20][21][22]. This work goes further by tracking the completion of advanced degrees and not just admission into graduate programs. The SMART program is similar to the Meyerhoff program that it has mentorship support and creates a community environment over a single summer. The SMART community is integrated within the BCM community of over 100 underrepresented minority postbaccalaureate, PhD, MD/PhD, and post-doctoral scholars.

Conclusion
A majority of members of this cohort of the SMART program (65%) completed doctoral level degrees in the biomedical sciences. This analysis was undertaken at least 12 years after participating in the SMART program, allowing sufficient time to complete doctoral level degrees. A conservative assessment was undertaken, and only counted as degreed are those participants for whom conclusive evidence was obtained regarding degree completion. The analysis presented here is just a small portion of the extensive data available from the more than two decades of undergraduate research training at the Baylor College of Medicine. Funding is required to probe the entire SMART database to identify specific factors and variables that lead to the success of this 9-week summer research program.

Supplementary Methods
The SMART database is stored using Filemaker Pro database software, and it contains information on all participants in SMART since its inception. Relevant data was exported into a text file after removing all personal identifying information.
The data text file was edited to ensure consistent textual definitions for variables and exported into SQLite database tables for further analysis (https://www.sqlite.org). Tier 1 versus non-tier 1 colleges were identified with the Carnegie Classification of Institutions of Higher Education's listing of very high research activity (RU/VH) universities http://carnegieclassifications.iu.edu/downloads/cc2010_ classification_data_file.xls. The list of RU/VH universities was downloaded as a CSV file, and the classification was added into the SQLite database table. SQL queries were made for final analysis. Chi-Square test was used to determine the significance of the differences between variables. Data from the SQLite was queried through RSQLite R package to get the data into R data frame for statistical analysis. Participants that we could not verify their completion of advanced degrees were treated as status unknown.