Journal of Nursing Education

The articles prior to January 2012 are part of the back file collection and are not available with a current paid subscription. To access the article, you may purchase it or purchase the complete back file collection here

Prediction of Academic Success in a Baccalaureate Nursing Education Program

Evelyn R Hayes, RN, PhD

Abstract

HYPOTHESIS THREE: There is no significant difference in the predictive accuracy of the regression equation derived from the Class of 1977, when applied to the Class of ?976.

The regression equation derived as a result of the identification of the significant cognitive variables was used. Cross validation of the regression equation was accomplished by use of the Fisher Exact Test which yielded a p- .0001. Thus, Hypothesis Three was accepted.

Discussion

The findings suggest that cognitive variables are the most powerful predictors of academic success.

A methodological problem in the form of a probable type II error existed with the many scales and the low sample size for the California Psychological Inventory and Survey of Interpersonal Values. It was not conceptually possible to combine the scale scores for each inventory in order to reach overall scores. Therefore, it is possible that under more representative conditions, the addition of the noncognitive variables might have been significant.

Through cross validation it has been demonstrated that the same cognitive variables remain statistically significant when applied to another sample. The regression equation derived for the Class of 1977 retained its predictive accuracy when applied to the Class of 1976.

Based on these findings, it seems reasonable to use the information based on the cognitive variables to facilitate students' progress toward successful completion of the program. The findings demonstra te thatitis possible to identify students who are not likely to graduate because of academic problems early in the college program (by the end of the freshman year). It is recommended that other schools consider adoption of this approach to prediction and develop their own regression equation.

Implications

Prevention and early detection have direct application to the academic advisement programs in our institutions of higher education. Complementary with identification of students at risk for not completing the program is the establishment of an advisory system. Advisement should begin with an orientation of both faculty and students to the system and continue with planned conferences to evaluate student progress throughout the educational program.

Summary

The multiple regression equation approach based on valid predictor variables coupled with an advisory system would facilitate a student's educational experience. The ultimate goal is to reduce attrition, thereby maximizing human and educational resources, while maintaining academic and professional standards.

TABLE

REDUCTION OF THE INITIAL BATTERY OF COGNITIVE PREDICTOR VARIABLES…

Although predicting student success concerns the total educational community, it is a crucial issue in nursing. The emerging role of nurses in the delivery of primary care, the change and increase in the number of health care settings and the need for expert clinicians all require effective and efficient use of limited nursing education resources.

The interest in and search for valid predictors of academic success is motivated by concern about attrition. There is an urgent need to alleviate currently high attrition rates without sacrificing academic standards. Traditionally, admissions and guidance officers have relied heavily on measures of intellectual achievement for predicting outcomes. The strength of cognitive predictors (those predictor variables that indicate knowledge or achievement), has been well documented; in fact, no other set of predictor variables accounts for as much criterion variance as do ability and achievement. However, there is reason to believe a peak has been reached in the usefulness of these predictors. Cattell and Butcher note "considerable improvement in the prediction of scholastic achievement can, in principle and sometimes in practice, be obtained by adding motivation and personality measures; in later achievement, these probably become equally important with abilities."1 Indeed, several studies indicate that noncognitive variables have a marked influence on academic performance.

The purpose of the study to be reported here, therefore, was to determine the validity of selected cognitive variables and the additive value of certain noncognitive variables in predicting academic success, as measured by completion of the curriculum in a baccalaureate nursing education program. This study included development of a predictive regression equation to identify students who are likely to succeed in the program and cross validation of the results.

Studies of nursing students have found ability and achievement measures to be good predictors of success. The Scholastic Aptitude Test (SAT) has been identified as a reliable predictor variable at the college level.2 Furthermore, since learnings in advanced courses are built on foundations from earlier courses it follows that achievement at an earlier stage should be considered in predicting achievement at a later stage. Indeed, the predictive ability of previous grade point average (GPA) is documented by several studies.3'4 For example. Burgess, Duffy and Temple5 found that overall achievement (cumulative GPA) was predicted by several variables, but prenursing GPA emerged as the most significant predictor. Moreover, Cattell and Butcher's research indicates success in nursing school is related to proficiency in the natural sciences, mathematics and English. This finding suggests that first year grades in these courses provide extremely relevant criteria.

The available evidence indicates achievement indices are powerful predictors of academic performance. Therefore, it seemed reasonable to include in the present study the cognitive variables of SAT verbal and quantitative scores, the grade point average for the two freshman semesters and grades in freshman courses (two chemistry courses, two psychology courses, and one course in each of the following: English, history, mathematics, philosophy, and speech).

Available evidence also suggests that variables in the noncognitive domain need to be taken more seriously and need to be investigated as potential predictors since learning is a multisensory experience involving the total being of the learner.6 The present study used as the noncognitive variables the personality factors measured by the California Psychological inventory (CPI) and the Survey of interpersonal Values (SIV). The CPI assesses the areas of dominance, capacity for status, sociability, social presence, self-acceptance, sense of well-being, responsibility, socialization, self-control, tolerance, good impression, communality, achievement via conformance, achievement via independent, intellectual efficiency, psychological-mindedness, flexibility and femininity. The SIV yields scores in the area of support, conformity, recognition, independence, benevolence and leadership.

Hypotheses

The study tested three null hypotheses:

1. There will be no significant difference between the graduates and nongraduates based on the regression equation utilizing cognitive variables.

2. There will be no significant difference in the predictive accuracy of the regression equation based on cognitive variables with the addition of noncognitive variables.

3. There will be no difference in the predictive accuracy of the regression equation derived from the Class of 1977 when applied to the Class of 1976.

Sample

Students from the Classes of 1976 and 1977, at a state university school of nursing in New England, were the subjects in this study. The sample was limited to those students enrolled at the university in June 1973 and June 1974 who had completed the freshman year and who planned on entering the sophomore year. Students who transferred to the university after June 1974 were excluded from the sample. These limits were imposed because the data analysis relied heavily on grades attained in courses taken at that university during the freshman year.

Eligible students from the Class of 1977 were assigned to one of twogroups: Group O consisted of 34 students who did not graduate due to academic reasons, and Group 1 consisted of 122 students who did graduate. The students of the Class of 1976 were similarly assigned to groups for purposes of cross validation: Group O had 28 nongraduates and Group 1 had 106 graduates.

Procedure

A packaged computer program, Statistical Package for the Social Sciences (SPSS) was used. The .05 level was required for evidence of statistical significance.

The data were subjected to multiple regression analysis techniques. F ratios were used to determine if the two groups could be significantly distinguished, based on a regression equation using cognitive variables. Another F test of significance was conducted to determine the amount of change in the predictive accuracy of an equation composed of both cognitive and noncognitive variables. The incremental value of the noncognitive variables was of particular interest. Cross validation of the findings was accomplished by applying the regression equation to another class and testing for a significant difference in predictability.

Results

HYPOTHESIS ONE; There is no significant difference between graduates and nongraduates based on the regression equation utilizing cognitive variables.

The ability of cognitive variables to predict academic success, as measured by graduation, was demonstrated. The cognitive variables collectively accounted for 62% of the criterion variance. The initial battery of 13 cognitive variables was reduced to a set of 8 significant variables including: grade point averages for the two freshman semesters, grades in 2 chemistry courses, mathematics, psychology, philosophy and SAT verbal score. The predictor accounting for the greatest variance (37%) was the grade point average for the second freshman semester. This combined with grades received in Mathematics 104 and Psychology 132 accounted for 50% of the variance (Table).

The F ratios indicate that 8 of the cognitive variables contribute significantly to the prediction of academic success as measured by the dependent variable, graduation (graduates and nongraduates). Therefore, the first null hypothesis was rejected.

HYPOTHESJS TWO; There is no significant difference in the predictive accuracy of the regression equation based on cognitive variables with the addition of noncognitive variables.

The CPI variables collectively accounted for 27% of the variance with the following 9 scales accounting for 25% of the variance: responsibility, achievement via independence, self -acceptance, sociability, good impression, self-control, communality, dominance, and flexibility. The support, leadership, and conformity scales from the SIV were the significant contributors to the 7% of the variance accounted for collectively by the SIV scales. The resultant F ratios with the addition of the noncognitive variables indicate that noncognitive variables do not contribute to a significant degree beyond that already established by the use of the cognitive variables alone. Therefore, Hypothesis Two was accepted.

Table

TABLEREDUCTION OF THE INITIAL BATTERY OF COGNITIVE PREDICTOR VARIABLES

TABLE

REDUCTION OF THE INITIAL BATTERY OF COGNITIVE PREDICTOR VARIABLES

HYPOTHESIS THREE: There is no significant difference in the predictive accuracy of the regression equation derived from the Class of 1977, when applied to the Class of ?976.

The regression equation derived as a result of the identification of the significant cognitive variables was used. Cross validation of the regression equation was accomplished by use of the Fisher Exact Test which yielded a p- .0001. Thus, Hypothesis Three was accepted.

Discussion

The findings suggest that cognitive variables are the most powerful predictors of academic success.

A methodological problem in the form of a probable type II error existed with the many scales and the low sample size for the California Psychological Inventory and Survey of Interpersonal Values. It was not conceptually possible to combine the scale scores for each inventory in order to reach overall scores. Therefore, it is possible that under more representative conditions, the addition of the noncognitive variables might have been significant.

Through cross validation it has been demonstrated that the same cognitive variables remain statistically significant when applied to another sample. The regression equation derived for the Class of 1977 retained its predictive accuracy when applied to the Class of 1976.

Based on these findings, it seems reasonable to use the information based on the cognitive variables to facilitate students' progress toward successful completion of the program. The findings demonstra te thatitis possible to identify students who are not likely to graduate because of academic problems early in the college program (by the end of the freshman year). It is recommended that other schools consider adoption of this approach to prediction and develop their own regression equation.

Implications

Prevention and early detection have direct application to the academic advisement programs in our institutions of higher education. Complementary with identification of students at risk for not completing the program is the establishment of an advisory system. Advisement should begin with an orientation of both faculty and students to the system and continue with planned conferences to evaluate student progress throughout the educational program.

Summary

The multiple regression equation approach based on valid predictor variables coupled with an advisory system would facilitate a student's educational experience. The ultimate goal is to reduce attrition, thereby maximizing human and educational resources, while maintaining academic and professional standards.

References

  • 1. Cattell R, Butcher H: The Prediction of Achievement anil Creativity. New York, The Bobbs Merrill Co, Ine, 1968.
  • 2. Munday L, Hoyt DP: Predicting academic success for nursing students. Nurs Res 13:222-229, 1964.
  • 3. Plapp JM, Psathos G, Caputo DV: Intellective predictors of success in nursing school. Educationa! and Psychocagical Measurement 25:565-577, Summer 1965.
  • 4. Owen S, Feldhusen J: Effectiveness of three models of multivariate prediction of academic success in nursing education. Nurs Res 19(&):517525, 1970.
  • 5. Burgess M, Duffy M, Temple F; Two studies of prediction of success in a collegiate program in nursing. Nurs Res 2l(4):357-366, 1972.
  • 6. Lavin DE: The Prediction of Academic Performance. New York, Russell Sage Foundation, 1965.

TABLE

REDUCTION OF THE INITIAL BATTERY OF COGNITIVE PREDICTOR VARIABLES

10.3928/0148-4834-19810601-03

Sign up to receive

Journal E-contents