Secondary Logo

Successful Modeling of Factors Related to Recruitment and Retention of Prenursing Students

Stankus, Jo-Ann, PhD, RN; Hamner, Mark, PhD; Stankey, Michael, PhD; Mancuso, Peggy, PhD, RN, CNM, CNE

doi: 10.1097/NNE.0000000000000579
Feature Articles

Background The traditional admission methodology used by many schools of nursing is the second-tier admission based on specified criteria. This method has been associated with a student body that is predominately female, white, and English speaking.

Problem Rational judgment modeling (points assigned to criteria) used to evaluate applicants for admission into prenursing programs potentially overlooks students who do not fit into the traditional model.

Approach This study compared predictive logistic regression with traditional rational judgment models to classify potential nursing school applicants.

Outcomes A higher number of Hispanic and black prenursing students were identified for potential upper-level nursing program admission.

Conclusions The use of logistic regression modeling can identify a more diverse student population. Advisors at both high school and university/college levels can use results of this model to help students determine their progress, identify academic weaknesses, and develop individual plans of action to help students successfully complete the prenursing curriculum requirements.

Author Affiliations: Coordinator, RN BS/MS Program, and Assistant Professor (Dr Stankus), College of Nursing; Associate Provost of Institutional Research & Improvement (Dr Hamner); Director of Analytics (Dr Stankey); and Associate Dean for Research and Clinical Scholarship/Professor (Dr Mancuso), College of Nursing, Texas Woman's University, Denton.

The authors declare no conflicts of interest.

Funding was provided by Health Resources and Services Administration Scholarships for Health Professions Students from Disadvantaged Backgrounds (award CFDA #93.925).

Correspondence: Dr Stankus, College of Nursing, Texas Woman's University, 1216 Oakland St, ASB 124, PO Box 425498, Denton, TX 76204 (

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (

Accepted for publication: June 1, 2018

Published ahead of print: July 26, 2018

Cite this article as: Stankus J-A, Hamner M, Stankey M, Mancuso P. Successful modeling of factors related to recruitment and retention of prenursing students. Nurse Educ. 2019;44(3):147-150. doi: 10.1097/NNE.0000000000000579

In the United States, nursing education is a high-stakes endeavor. The student expends money, time, and intellectual effort in the hopes of acquiring a rewarding career.1-3 The educational institution deploys ever decreasing faculty and clinical resources in the hope of producing graduates who will be excellent practitioners.4-6 Society expends taxpayer dollars to support universities and colleges in the hope that graduates will be nurses who provide care to our most vulnerable populations.7 Current admission criteria tend to generate nursing student populations that are predominantly white and female, even in those states with diverse ethnic populations.8,9 To maximize the benefits of nursing education, we must ensure that the students who are accepted into nursing programs represent the populations they will serve and can successfully address the rigors of the required academic work needed for successful graduation. To meet these objectives, this research project created an innovative modeling process using logistic regression to predict the future success of prenursing students who apply to nursing school through a second-tier admissions process.

Back to Top | Article Outline


Admission to nursing programs today is no easy feat. Today's prenursing students grapple with the reality of a larger number of applicants for each available opening.10 To compete for these openings, prenursing students must be academically equipped to meet the challenges of higher-education courses. Students who have successfully passed high school courses may not be adequately prepared for rigorous college/university science courses.1,2,6,11 Students who come from lower socioeconomic status households are competing with students from a higher socioeconomic status, who may have experienced academic advantages and increased financial support, as well as additional emotional support and guidance, fostering the successful transition to university course work.12,13

To provide the nursing care needed for predicted US demographic changes, educational institutions are urged to respond with increasing numbers of well-prepared, diverse nursing graduates.7,11,14 At the same time, schools of nursing are grappling with decreasing numbers of prepared faculty members, clinical sites, and available preceptors.10 These decreasing numbers often limit the numbers of students who can be accepted into clinical course work. Currently, advances in technology, an increased emphasis on community-based health care, and shifting demographics from homogenous to diverse populations are spurring rapid changes in health care. As a result, many states do not have a nursing workforce that reflects current demographic trends of increasing ethnic and minority groups,5,6,15 currently at 37% of the US population.10,16

These challenges, combined with the personal and societal expense of a university education, serve to create the high-stakes environment surrounding entry into upper-level nursing course work.12,17,18 Educational institutions therefore need to invest in prenursing students who will not only achieve admission to prenursing programs but also successfully attain licensure, perform as competent clinicians, and represent the diverse populations that the nursing profession serves. Despite the efforts of the American Association of Colleges of Nursing and other nursing organizations to increase the numbers of 21st-century nursing graduates, in 2016, there was only a 3.6% increase in the numbers of students admitted to prenursing programs. This small increase is insufficient to meet future nursing workforce demands.1,19 To increase the number of students recruited from all demographics and who will also successfully matriculate to upper-level nursing courses, a new and innovative admissions methodology is needed.

The traditional admission methodology used by many schools of nursing for prenursing programs has been a second-tier type of admission, with admission following completion of college-level nursing prerequisites based on a system of metrics that includes prenursing course work. This methodology assigns “points” or “weights” to institutionally specific, previously identified variables. These variables comprise the criteria leading to a summed admission score.20-22 Cunningham et al23 proposed that this traditional method reflects a rational or judgmental admission model, with the chosen factors and the weighting assigned each factor varying according to the institution's current admission policies and procedures.

Our investigation of dynamic predictive modeling, on the other hand, specifies those factors that have the greatest influence on a specific, successful end measure (eg, admission to prenursing programs, time to graduation, or successfully passing the NCLEX) for a certain point in time. As such, variables and weights associated with end-point success could change over time, depending on the data. These changes can be influenced by factors such as high school grade inflation, which inhibits predictable values of previous high school grade point average (GPA), and demographic changes in high school populations and districts, which change the predictability of that institution's class ranking related to future academic success in the university setting. Our model was created for those students who are first applying to the university or college setting, based on information at that point in time (high school ranking, high school GPA, Scholastic Aptitude Test [SAT]/American College Test [ACT] individual scores), and precluded variables that do not yet exist (grades in college-level nursing prerequisite courses, HESI/Assessment Technologies Institute [ATI] scores). The end point for success of the model was admission into upper-division, nursing course work.

Back to Top | Article Outline


The purpose of this study was to create an innovative modeling process (logistic regression) to predict future success for prenursing students entering the second-tier admission processes. For this study, success was defined as admission into upper-level nursing. The accuracy of the model predictions was compared with the institution's traditional method of summation of points from predesignated criteria using historical data.

Back to Top | Article Outline


The initial goal of this study was to identify a sufficiently large pool of diverse high school applicants from which 10 students would be selected to receive special assistance: (a) guaranteed admission into the upper-division nursing program (provided they met minimum cumulative GPA requirements), (b) substantial financial assistance for 4 years, and (c) academic coaching during the first 2 years. Ultimately, this involved testing the efficacy of 2 models used to predict the success of ethnic minorities, primarily Hispanic and black high school graduates, which would have broader implications for nursing admissions practices.

Back to Top | Article Outline

Method 1: Rational Use of Conventional Wisdom

The first approach was to use the traditional “conventional wisdom” process to identify high school students who would be offered guaranteed admission into the nursing program and then select the 10 grant recipients from the resulting pool. This approach used a student's high school GPA, high school science grade point average (SGPA), and SAT (or ACT) composite scores (verbal and mathematics scores combined into a single score). As part of the traditional application ranking, each of these 3 variables was weighted equally, and students who met minimum levels on all criteria (eg, GPA ≥3.50, SGPA ≥3.50, and SAT ≥1000) were offered guaranteed admission to upper-level nursing course work.

Back to Top | Article Outline

Method 2: Empirical Use of Logistic Regression

The second and new empirical approach was to model historical student success (acceptance into upper-level nursing course work) based on high school application data. Logistic regression was used to identify relevant variables and appropriate weights for predicting the binary outcome of getting accepted into nursing (yes or no). Preliminary analyses of applicant data from 2008 to 2013 narrowed the list of possible explanatory variables to high school GPA (weight = 2.550), SAT verbal (weight = 0.007), and SAT math (weight = 0.012). This list was attractive because it was similar to the criteria used in method 1, with 2 notable exceptions. First, the combined SAT score used in method 1 was split into its base components for method 2. Second, the variable weights were determined empirically instead of through conventional wisdom, with SAT math subsequently being weighted more heavily than SAT verbal scores. The resulting regression equation was used to calculate a success probability score for each high school applicant.

Back to Top | Article Outline


Method 1 (rational method) was used to identify 37 students, from an applicant pool of 1102 students, who met at least the minimum values for GPA, SGPA, and SAT (or ACT) and were subsequently offered guaranteed admission into the nursing program. In this group of students, only 10 of the 37 students were Hispanic (n = 8) or black (n = 2). Unfortunately, for the initial purposes of the study, none of the 10 Hispanic or black students met the financial need requirements of the grant.

Method 2 (logistic regression) was used to assign a success probability score for all 1102 applicants. These students then were ranked from high score (0.965) to low score (0.000) and subdivided into deciles (0.00-0.90) for further analysis.

Figure 1, Supplemental Digital Content,, illustrates the distribution of applicants by model score decile (bars) and by the historical success rates (dots) associated with those deciles for getting accepted into upper-division nursing course work. Nearly three-fourths (799 of 1102) of the applicant pool fell into the lowest model score decile (shown here as 0.02 and 0.00 subcategories), with a very low chance of being accepted into upper-division nursing course work. For applicants in these 2 subcategories, the average model score was 0.006, which indicated that there was less than a 1% chance of being accepted into upper-division nursing course work. For this study, these students were eliminated from further analysis. More than one-fourth (303 of 1102) of the applicants fell into deciles 0.10 to 0.90, which were associated with 38% to 100% historical success rates. Forty-eight of those students fell into deciles 0.60 to 0.90. These students would be low-risk candidates for guaranteed admission because their model scores were associated historically with 100% success.

Figure 2, Supplemental Digital Content,, illustrates the distribution of the guaranteed admission offers generated by method 1 (rational) against the success probability model score deciles generated by the method 2 (empirical). Eighteen of the 37 method 1 guaranteed admission offers (49%) went to students in deciles 0.60 to 0.90, which are associated with 100% success in getting admitted into nursing. However, another 18 of the 37 guaranteed offers (49%) went to students in deciles 0.20 to 0.50, which are associated with historical admission rates of 44% to 75%. One guaranteed offer (2%) went to a student in decile 0.10, associated with a historical success rate of 36%. In this sense, model 1 did a poor job identifying students whose predicted success in nursing was “guaranteed” and resulted in guaranteed offers being made to students who may not have made it on their own.

Figure 3, Supplemental Digital Content,, illustrates a sector of high-potential Hispanic and black students identified by method 2 (empirical) that was largely overlooked by method 1(rational). Most of the guaranteed admission offers made to Hispanic and black students generated by method 1 (7 of 10) went to students in model score deciles 0.50, 0.60, and 0.70. This represented an offer rate of 35% (7 offers made from the “pool” of 20 Hispanic and black applicants in those deciles). Deciles 0.20, 0.30, and 0.40 include more than twice the number of Hispanic and black applicants (n = 43), but method 1 identified only one of these students as meeting the criteria to receive a guaranteed offer of admission. This represents an offer rate of only 2% among Hispanic and black applicants in these deciles, despite the finding that 9 of the non-Hispanic/black applicants (n = 68) in the same deciles received a method 1 guaranteed offer, representing an offer rate of 13.2%. Importantly, Hispanic and black applicants with method 2 model scores in deciles 0.20, 0.30, and 0.40 represent a large group of high-potential students who have been essentially overlooked by previous method 1 approaches. Students with these decile scores have been associated with nursing program acceptance rates of 44% to 63%, well above the overall acceptance rate of 15%. For the initial purposes of the study, 18 of the 43 Hispanic and black applicants met the financial need requirements of the grant, and the target of 10 was awarded.

Back to Top | Article Outline


Results from method 2 (logistic regression) were used to assign an overall predictive model score, which ranged from 0.00 to 1.00, to determine those ethnically diverse students who could benefit most from scholarship support to be successful, as defined by admission to upper-division nursing course work. A finding of this analysis was that students with a model score of 0.60 or greater had a 100% probability of being accepted into upper-division nursing, with or without scholarship support. Also, students in this category had multiple sources of financial aid in place. Students with a model score of less than 0.10 were unlikely to be accepted into any nursing program. The target population for financial support would be ethnically diverse students with model scores of 0.20 to 0.49 (deciles 0.20, 0.30, and 0.40) who would be more likely to attain successful admission with financial and academic support.

The use of this dynamic modeling tool can help nursing programs identify a diverse student population for admittance into the prenursing program. Advisors at both the high school and university/college level can use the results of this model to help students determine their progress, identify academic weaknesses, and develop individual plans of action to help the students initiate and successfully complete the prenursing curriculum requirements. For those not able to complete the prenursing classes, these data would help the advisor work with the student to choose another major that would optimize previously completed courses and meet the student's career aspirations.

Dynamic modeling enabled advisors and faculty members to ascertain boundaries regarding student success and identify those students who would receive the most benefit from academic and financial support. In consideration of the allocation of limited resources, this approach is similar to the classic triage method used in disaster situations. One group of applicants possessed attributes that were associated with a high probability of success. This group of applicants tended to be those who had already received scholarships and financial aid. Another group was composed of applicants who were highly unlikely to be successful in attaining admission to upper-division nursing. Logically, the allocation of limited resources would generate the most benefit-to-cost ratio if these resources were given to the middle group of applicants who would be more likely to be successful if they received support.

Back to Top | Article Outline


The use of dynamic modeling through logistic regression identified students for upper-level nursing program admission. Also, dynamic modeling simplified ranking of students using different weights for previously defined criteria. This process can be adapted to other educational contexts that have a defined end-point criterion for success and access to past performance of students.

Back to Top | Article Outline


1. Murray TA. Pedagogy and academic success in prelicensure nursing education. J Prof Nurs. 2016;32(5S):S24–S29.
2. Newton SE, Smith LH, Moore G, Magnan M. Predicting early academic achievement in a baccalaureate nursing program. J Prof Nurs. 2007;23(3):144–149.
3. Norris TL, Wicks MN, Cowan PA, Davison ES. A summer prematriculation program: bridging the gap for disadvantaged underrepresented minority students interested in a nursing career. J Nurs Educ. 2016;55(8):471–475.
4. Abele C, Penprase B, Ternes R. A closer look at academic probation and attrition: what courses are predictive of nursing student success? Nurse Educ Today. 2013;33(3):258–261.
5. McLain RM, Fifolt M, Dawson MA, et al. Student success survey: supporting academic success for at-risk nursing students through early intervention. Nurse Educ. 2017;42(1):33–37.
6. Murray TA, Pole DC, Ciarlo EM, Holmes S. A nursing workforce diversity project: strategies for recruitment, retention, graduation, and NCLEX-RN success. Nurs Educ Perspect. 2016;37(3):138–143.
7. American Association of Colleges of Nursing Support of IOM Future of Nursing Report. Available at Published October 2012. Accessed July 1, 2017.
8. National League for Nursing. Achieving diversity and meaningful inclusion in nursing education. Available at Published 2016. Accessed July 1, 2017.
9. US Department of Health and Human Services Health Resources and Services Administration Bureau of Health Workforce National Center for Health Workforce Analysis. Supply and demand projections of the nursing workforce: 2014-2030. Available at Published 2017. Accessed January 20, 2018.
10. American Association of Colleges of Nursing. AACN Enhancing Diversity in the Workforce, AACN Fact Sheet Containing the Latest Data. Available at Published 2013. Updated March 2015. Accessed July 1, 2017.
11. Jeffreys MR. Nursing Student Retention: Understanding the Process and Making a Difference. New York, NY: Springer Publishing Co; 2012.
12. Jury M, Smeding A, Stephens NM, Nelson JE, Aelenei C, Darnon C. The experience of low-SES students in higher education: psychological barriers to success and interventions to reduce social-class inequality. J Soc Issues. 2017;73(1):16–34.
13. American Psychological Association. APA education and socioeconomic status factsheet. Available at Accessed December 1, 2017.
14. Wros P, May BA. Creating and sustaining a diverse student community in a school of nursing. Hisp Health Care Int. 2013;11(3):131–141.
15. Mulholland J, Anionwu EN, Atkins R, Tappern M, Franks PJ. Diversity, attrition, and transition into nursing. J Adv Nurs. 2008;64(1):49–59.
16. US Census Bureau. State, County, City and Town Quickfacts. Available at Published 2012. Updated 2017. Accessed January 20, 2018.
17. Latham J, Morris-Thompson T, Plata R. Role of careers advisers in nurse recruitment. Nurs Manag. 2013;19(10):20–25.
18. Metcalfe SE, Neubrander J. Social determinants and educational barriers to successful admission to nursing programs for minority and rural students. J Prof Nurs. 2016;32(5):377–382.
19. American Association of Colleges of Nursing. AACN Nursing Shortage Fact Sheet. Available at Published 2013. Updated May 2017. Accessed July 1, 2017.
20. Byrd G, Garza C, Nieswiadomy R. Predictors of successful completion of a baccalaureate nursing program. Nurse Educ. 1999;24(6):33–36.
21. Timer JE, Clauson MI. The use of selective admissions tools to predict students' success in an advanced standing baccalaureate nursing program. Nurse Educ Today. 2011;31(6):601–606.
22. Wolkowitz AA, Kelley JA. Academic predictors of success in a nursing program. J Nurs Educ. 2010;49(9):498–503.
23. Cunningham JL, Manier A, Anderson A, Sarnosky K. Rational versus empirical prediction of nursing student success. J Prof Nurs. 2014;30(6):486–492.

admissions; diversity; nursing students; predictive modeling; retention

Supplemental Digital Content

Back to Top | Article Outline
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved