Quantitative data are essential for the development of criteria to guide deans and faculties in the identificati on"of students who will successfully complete master's degree programs in nursing. The diversity of candidates, limited number of qualified nurses in leadership positions in nursing, and the expense of graduate education necessitated the formulation and testing of a predictive formula for determining the probability of successful completion of applicants. Further, a prediction equation may encourage nursing faculties to test theoretical explanations for the use of criteria in relation to completion status of students in master's programs.
The purpose of this investigation was to determine the significance and strength of relationships between selected academic and demographic variables and successful completion of master's degree programs in nursing. This relationship between variables was used to formulate a predictive equation. The equation was then tested with a second sample of subjects from the same programs. This investigation was designed to answer the following questions:
1. Do relationships exist between selected academic variables and completion of master's degree programs in nursing? The academic variables were: (a) undergraduate grade point average (UGPA) and (b) Graduate Record Examination scores (GRE).
2. Do relationships exist between selected demographic variables and completion of master's degree programs in nursing? The demographic variables tested were: (a) age, (b) gender, (c) race, (d) state residency status, (e) urban/rural domicile, (f) number of years of professional work experience, and (g) time elapsed between awarding the baccalaureate degree and admission to a master's degree program in nursing.
3. What specific variables or combination of selected academic and demographic variables demonstrate a significant relationship for completion status?
Students who have been granted a master's degree from a school of nursing in a university were operationally defined as completers. Students who withdrew or terminated from the program prior to the completion of the requirements for the degree were operationally defined as noncompleters.
Graduate Record Examination
Scores achieved by students admitted to graduate schools of nursing as reported by the Educational Testing Service.
Students who were residents of the same state as determined by the permanent address at the time of admission to a master's degree program in nursing. All other students were considered nonresidents.
Students who provided a permanent address from a community of 50,000 or more on the application for admission to a master's degree program were designated urban. All other students were designated rural.
The academic variables selected for study were cited in the literature. Undergraduate grade point average (UGPA) and the total and subtests of the Graduate Record Examination (GRE) score(s) are variables commonly used by schools of nursing in the selection process. Hence, these variables were utilized in this study.
Undergraduate Grade Point Average
Of 197 graduate schools of nursing, 153 (78%) have established an UGPA for selection of applicants (Moore, Ready, & Sacchetti, 1988). Burgess, Duffey, and Temple (1972), Grant (1986), and Pontious (1980) concurred that a future GPA could be predicted by utilizing a former UGPA.
Studies by Barkan (1987), Burton and Turner (1982), Mayo (1988), Rauckhorst (1986), Stein (1978), Stein and Green (1970), and Thomas (1977) had positive correlations for UGPA and success in schools of nursing.
Graduate Record Examination (GRE)
Tripp and Duffey (1981) studied completers and non-completers in master's programs. They divided 270 subjects into three categories identified as graduated (n = 91), dropped out (n = 86), and unaccepted (n = 51). The predictor variables were UGPA, GRE-V, and GRE-Q which were utilized in discriminate analysis between the groups. Their findings indicated the inability to differentiate between those subjects who graduated and those who dropped out of the program.
Studies by Ainslie, et al. (1976), Barkan (1987), Burton & Turner (1982), Lust (1981), Mayo (1988), Rauckhorst (1986), Stein (1978), Thomas (1974, 1977), Turner (1983), and Yaksich and Cox (1985) reported significant correlations for UGPA and success in schools of nursing. Studies by Ainslie, et al. (1976), Lust (1981), Stein (1978), Thomas (1974, 1977), and Yàksich and Cox (1985) reported significant correlations for various GRE subtests and success in schools of nursing.
Findings related to demographic variables and academic proficiency were limited, inconsistent, and inconclusive. Meer, Stein, and Geertsma (1955) and Krekeler (1987) found subjects to lose mental rapidity according to age; whereas, Geisler ( 1983 ) Wild, Durso, and Rubin (1982), Chase, Ludlow, Pugh, and Pomeroy (1964), and Stein (1978) found no difference. Geisler (1983) and Seashore (1962) reported no difference according to gender. Marston (1971) indicated that the GRE was particularly difficult for African-American students who were otherwise qualified for graduate school. Findings relative to domicile were not comparable (Lavin, 1965). Denman (1976), Creager (1965), and Chase, Ludlow, Pugh, & Pomeroy (1964) found significant differences between elapsed time and GRE scores; however, King (1953) found these scores to stabilize after 5 years.
Several research studies focused on completion status (Mayo, 1988; Tripp & Duffey, 1981; Stein, 1978). The majority of these studies evidenced inconsistent results. Most investigators recommended an increase in the number of variables and additional studies to differentiate completers from noncompleters.
These variables had been investigated at several universities with confusing findings. It was projected that including them in this study would further clarify some of the inconsistencies of the preceding studies.
Combination of Variables
The review of literature demonstrated contradictions in the findings. It provided limited support for any combination of variables as a criterion for the selection of students.
Although the request for participation of all graduate schools in the nation was made, only four schools elected to participate thereby limiting the generalizability of the findings. The sample consisted of the records of 289 students in four master's degree programs in nursing from 1980 through 1984. These years were selected as the universities permitted 5 years to complete the requirements for a master's degree in nursing.
The categorical variable of completion status required the use of discriminant analysis, specifically, the stepwise Mahalanobis method (Hair, Anderson, & Tatham, 1987) which is similar to a stepwise regression analysis. Discriminant analysis determines the significance and strength of the relationship between completion status (dependent variable) and academic and demographic characteristics (independent variables). The Mahalanobis method is based on distances that (a) adjust for unequal variances, (b) utilize original data, and (c) perform better with a larger number of variables. It maximizes the use of available information. This method determined those specific academic and demographic variables most likely to predict completion status of students in master's degree programs in nursing. The discriminant analysis with the application of the stepwise Mahalanobis method randomly assigned subjects to Group A, Analysis Group (n = 215, 74%) and Group B, Test Group (n = 74, 26%). The size of the groups were proportional to the percentage of completers (75%) and noncompleters (25%) in the total group. The Mahalanobis method develops a completion prediction equation based on the findings from Group A and then tests that equation on Group B. The Mahalanobis method of discriminant analysis is available with SPSS/PC + V3.0 (1988).
The number of subjects who completed programs for the total group was 212 (73%). The number who completed in Group A was 159 (74%) and Group B was 53 (72%). All academic variables were significant at the p = .05 level between schools (n = 289) except for UGPA which was p = .09. Demographic variables were significant at the ? <. 05 level for all variables except gender (p = . 198). These groups were essentially comparable. In both groups, 50% or more of the male students did not complete a program. Minority subjects in Group A who completed a program were 9 (56%); whereas, in Group B the minority subjects were 7 (70%). There were more subjects in Group A with an urban status (n = 82, 76%) as compared to Group B (n = 25, 34%) who completed master's programs in schools of nursing.
Relationship Between Academic Variables and Completion
Comparison of academic variables by t test in Group A (n = 215) resulted in the following: UGPA of completers and noncompleters was -1.99, GRE-V was -2.4, GRE-Q was 2.03, and GRE-A was -2.32. These findings were only suggestive relationships between success in a program and the selected academic variables, and were not significant at the p = .05 level, hence, they were rejected.
Relationship Between Demographic Variables and Completion
Age, gender, race, state resident status, and time elapse between awarding the baccalaureate degree and admission to a master's degree program in nursing differentiated students who completed a master's degree program in nursing from those who did not.
Several significant relationships at p = .05 were found for subjects who completed programs. These relationships were age (n = 159) at r = .0084, female gender (n = 152) of r = .0124, race (n = 150 Caucasians) of r=. 0068, and in-state resident status was significant for completers at r = .0111. The minimal time that elapsed between awarding of the baccalaureate degree and admission to a master's degree program in nursing was also significant at r = .0O91.
Relationship Between Combination of Selected Academic and Demographic Variables and Completion
The hypothesis was rejected. The combination of variables of UGPA, age, gender, race, and years of professional experience were significant (p<.013) for subjects who completed programs.
Younger completers (22 to 27 years of age) scored higher in UGPA, GRE scores, and graduate GPA. The majority of the male students (n = 8, 53%) were noncompleters although minority status (n = l) was equivalent for both completers and noncompleters among the male students. Even when students classed in the minority group scored lowest in the GRE-A (220-410), a majority of those students completed the programs (n = 9, 56%). The variables of race and GRE scores were significant at the p<.05 level (n = 159i The combined variables of race and UGPA were not significant (n = 159, p = .0619). Years of professional work experience were negatively associated with GPA upon completion of the programs.
The results of the discriminant analysis of Group A, the analysis sample, formed the weighted coefficients for all variables in the prediction equation. The weighted coefficients of Group A, Analysis Group, were applied in the prediction equation to be tested with Group B.
Completion = gender × .648 + race × .431 + Age × .427 + YrsWk × .223 + UGPA × .152 + GRE-Q × -.099 + StRes × -.098 + GRE-V × .094 + Urban × .083 + GRE-A × -.022 + Time × .002
The 11 variables were entered simultaneously in analysis and provided significant correlation coefficients (n = 215, r = . 128, p< .05) as determining differences between completers an non-completers in Group A. This formula classified correctly 75% of the subjects according to completion status.
This equation was tested with Group B. The equation was formulated and confirmed by stepwise discriminant analysis. The value of this function was significant (df= 10, r = .189,p<.013). Fifty-three (72%) of the subjects were identified correctly by the application of the predictive equation. The proportional chance criterion for correct classification of the sample is 62%. Therefore, the prediction equation provided a method of selection which differentiated between students who did and did not complete master's degree programs in nursing.
The number of minority and male students in this study limits generalizations related to these variables; however, these variables do reflect the populations in the majority of graduate schools of nursing. Variables, such as motivation and maturity, which may be associated with the completion of a master's degree program in nursing, were not assessed. As the events that precipitated student withdrawal from programs could not be verified, there may have been other inducements than academic proficiency involved. The demographic characteristics of the graduate students were obtained by selfreport and recorded in the student's records.
The findings of this study reinforced the importance of considering demographic variables as well as criteria representing academic proficiency in the selection process. Further, these results concur with Madaus (1985) that academic variables should be used for student advisement rather than as a criterion for applicant selection. For example, the majority of minority students who did not score well on the GRE completed the master's programs. Support services of reading, writing, and mathematics for all students with marginal scores would result in a higher rate of completion. Frierson (1989) found test scores were significantly increased by participation in learning teams and by developing skills in test taking.
Although male students obtained higher GRE scores, the majority did not complete the program. Career guidance would be of value for all applicants who achieve high scores. The combination of academic and demographic variables in the selection process may increase the number of nurses prepared at the advanced level as well as correctly reflect the demographic characteristics of the general population in the nursing profession. This congruence would provide a higher quality of patient care which is the ultimate goal of all nursing programs.
Criteria for admission and completion of master's programs should reflect the behavior needed in the nursing profession such as a wide knowledge base, leadership capabilities, openness to diverse cultures, critical thinking, and skills in problem solving and decision making. Suggestions for further research include the development of valid instruments which diverge perceptibly from the Caucasian culture. Additionally, research may identify multiple indicators to measure capabilities of non-traditional students such as hours of continuing education and honors received (Gieske, 1987).
In summary, academic variables singly did not differentiate between master's degree nursing students who did or did not complete programs. The combination of academic and the demographic variables of age, gender, ethnic origin, and years of professional experience were discriminatory regarding completion status.
- Ainslie, B. S., Andersen, L.E., Colby, B.K., Hoffman, M.A., Meserve, KP., O'Connor, C, & Ouimet, A. (1976). Predictive value of selected admission criteria for graduate nursing education. Nursing Research, 25(4), 296299.
- Barkan, H. (1987). Prediction of student performance: The strengths of predictions based on preadmission data for nursing master's students. Sonoma, CA: Sonoma State University.
- Burgess, M.M., Duffey, M., & Temple, RG. (1972). Two studies of prediction of success in a collegiate program of nursing. Nursing Research, 21(4), 357-366.
- Burton, N. W., & Turner, N.J. (1982). University of Utah Department of Nursing validity study report. Unpublished report, Educational Testing Service, Princeton, NJ
- Chase, CL, Ludlow, HG., Pugh, RC, & Pomeroy, M.C. (1964). Predicting success for advanced graduate students in education. Bloomington, IN: Indiana University Press.
- Creager, JA. (1965). Predicting doctorate attainment with GRE and other variables. (Report No. 25). Washington, DC: Office of Scientific Personnel, National Academy of Sciences, National Research Council.
- Denman, L. M. (1976). "No shows* in a master of science in nursing program: They helped us improve our selection process! Journal of Nursing Education, 15, 3-6.
- Frierson, H.T., Jr. (1989). The impact of testing skills intervention upon Black nursing students' licensure examination performance. The Journal of Negro Education, 58(1), 8291.
- Geisler, M.P. (1983). The older graduate student: A descriptive study. Madison, WI: University of Wisconsin.
- Gieske, R.M. (1987). Influence of selected social status criteria and their correlation with the MAT as a selector of candidates for a master's program in nursing. Unpublished manuscript, Valdosta, GA: Valdosta State College School of Nursing.
- Grant, R.E. (1986). Predicting academic success. In WL. Hölzerner (Ed.), Review of research in nursing education, 93-105. New York: National League for Nursing.
- Hair, J.E, Jr., Anderson, R.E., & Tatham, RL. (1987). Multivariate data analysis: With readings. New York: Macmillan Company.
- King, D.W. (1953). Graduate record examination scores and grade-point averages of graduate students at the College of the Pacific. Unpublished master's thesis. College of the Pacific, Stockton,
- Krekeler, Sr. K. (1987). GRE scores as predictors of academic and career success. Unpublished report, St. Louis University Medical Center, St. Louis, Missouri.
- Lavin, D.E. (1965). The prediction of academic performance: A theoretical analysis and review of research. New York: Russell Sage.
- Lust, B.L. (1981). A study of the predictors of achievement of nurses in graduate school. Dissertation Abstracts International, 42(3), 986-B.
- Madaus, G.F. (1985). Test scores as administrative mechanisms in educational policy. PAi Delta Kappan, 66, 611-617.
- Marston, A.R. (1971). It is time to reconsider the GRE. American Psychologist, 26, 653-655.
- Mayo, P.S. (1988). Predictors of success in the MSN program. Unpublished study. University of North Carolina School of Nursing, Graduate Nursing Program, Charlotte, North Carolina.
- Meer, B., Stein, M.I., & Geertsma, R. (1955). An analysis of the Miller Analogies Test for a scientific population. American Psychologist, 10, 33-34.
- Moore, T.C., Ready, B.C., & Sacchetti, R.D. (Eds.). (1988). Graduate programs in the biological, agricultural and health sciences. Princeton, NJ: Peterson's Guides.
- Pontious. S.L. ( 1980). Multivariate prediction of academic achievement in a collegiate program of nursing. Unpublished doctoral dissertation, New Mexico State University, Las Cruces.
- Rauckhorst, L. ( 1986). Predictive validity of GRE scores. Unpublished statistical report, University of Wisconsin School of Nursing, Oshkosh, Wisconsin.
- Seashore, H.G. (1962). Women are more predictable than men. Journal of Counseling Psychology, 9(3), 261-270.
- Statistical package for the social sciences. ( 1988). V3.0. New York: Microsoft.
- Stein, R.F. (1978). The graduate recored examination: Does it predict performance in nursing programs? Nurse Educator, 16-19.
- Stein, RF., & Green, E. J. (1970). The graduate record examination as a predictive potential in the nursing major. Nursing Research, 29(1), 42-47.
- Thomas, B. (1974). Prediction of success in a graduate nursing service administration program. Nursing Research, 23(2), 156-159.
- Thomas, B. (1977). Differential utility of predictors in graduate nursing education. Nursing Research, 26(3), 100-102.
- Tripp, A., & Duffey, M. (1981). Discriminant analysis to predict graduation-nongraduation in a master's degree program in nursing.
- Research in Nursing and Health, 4, 345-353. Turner, N.J. ( 1983). University of Utah Department of Nursing validity study report. Unpublished report. Educational Testing Service, Princeton, NJ.
- Wild, CL., Dureo, R., & Rubin, D.B. (1982X Effect of increased test-taking time on test scores by ethnic group, years out of school, and sex. Journal of Educational Measurement, 29(1), 19-28.
- Yaksich, S., & Cox, R. (1985). Predictors of success in a graduate nursing program. Unpublished study. University of Pittsburgh, Pittsburgh, PA.