The Myth of Total Retention

by Gregory Bovasso

Introduction

When faced with tasks like increasing student retention or performance, one often has a sense that “nothing works.” Like Sisyphus, we feel we labor in vain. However, as Camus observes, the existential task is to derive meaning from the futility of such efforts. This sense of meaning does not require that one accurately calibrate the degree of futility in their efforts, but it nonetheless helps one to set a realistic standard without being overly optimistic or pessimistic. Thus, establishing the parameters for the increase in student retention that one can be reasonably expected to produce provides an important standard for evaluating our efficacy. The great effort required to produce a modest impact is neither a counsel to despair nor a signature of grandeur. The virtue of the endeavor is its own reward apart from its outcome if that endeavor is realistic and in good faith. Efforts to increase retention that are simply a pretense or which mask financial interests or self-aggrandizement are likely neither to succeed nor to give meaning to an almost futile task.

The realism of our efforts informs us of our responsibility. As I sit here and peruse my grade book, I find that none of my students who have obviously stopped attending my classes had a failing grade when they stopped attending. Was it something I said? Did the students simply achieve their short term financial objectives? Perhaps the pedagogical system that I implemented needs some fine tuning, or a complete overhaul, or perhaps I am accepting responsibility for a lightning strike? Perhaps it’s none of the above, but the students’ exercise of a free choice reflects the value of other matters more important than education. Although we may wish to minimize needless attrition, education is no more than a self-serving indoctrination business if students are not free to opt out and do not have a sense that they are freely participating in the first place. The exercise of this choice to opt out may reflect a personal moral victory as much as a failure if we consider this choice in the context of the students’ world rather than the world we wish the student to inhabit. We can no more force or persuade students to continue their education, live in our world, or actually learn something than we can make a plant grow by pulling on it.

One does not provide a more engaging pedagogical process because it increases student numbers, but because this engagement reflects a meaningful educative process which may not be fully measurable in grades or retention. The measurable outcome of such a process is likely to be modest, but the value of the process must be justified on its own terms. Educational institutions that set modest goals for retention efforts may recognize that total retention, however gradually approached, may result in a kingdom of imaginary students who are present in body, but not mind. Modest and realistic goals also protect an engaging educative process from degenerating into busy work for student and teacher alike. Only the freedom to opt out will protect the integrity of the educative process by assuring that it is freely chosen. The empirical research on self determination theory demonstrates the motivational losses and performance decrements resulting from the excessive use of extrinsic incentives, particularly in the field of education.

Given the need for an engaging educative process in it own right efforts to increase retention are best directed at supporting students by providing resources needed to overcome obstacles to their persistence. These resources cannot be provided in the context of the classroom without altering its fundamental character, and the role of the larger institution in meeting these needs is arguable. Ultimately, the taxpayers and their representatives will decide if educational institutions should have a broader social function of creating communities whose members are willing and able to participate in an educative process. If not forthcoming, the student who leaves school to care for his/her family because he/she cannot find childcare, or for other similar reasons, has made the right choice.

The Relationship of Student Persistence to Academic Ability and Other Factors

A previous study (Doshi & Grosset, 2004) examined Community College of Philadelphia students’ stated reasons for not meeting academic goals, but the data on which this study is based may suffer from retrospective and selection biases. The current study examined objective risk factors that predict student attrition in an effort to direct interventions that target at-risk students.

The data were obtained through the Community College of Philadelphia Office of Institutional Research for all Community College of Philadelphia students registered for Psychology 101, 2002-2003 (N= 2,976). The sample was limited to Community College of Philadelphia students who took placement tests in the 2002-2003 academic year to better track the progress of a cohort of entering students over the course of a year. This sample is not necessarily representative of the entire Community College of Philadelphia student population or other Community College of Philadelphia sub-populations, which makes replication of the study necessary using other samples representing populations of interest.

Analysis 1

In the first analysis, attrition was measured as the number of credits for which students registered that they failed to earn. The only risk factor correlated with attrition in the Fall 2002 semester was financial aid status, but this correlation was low (r = .11, p < .05). Several of the risk factors examined significantly correlated with attrition in the Spring 2003 semester (see Table 1), but the magnitude of the correlations of the risk factors with attrition was generally low.

Regression analysis indicates that the single best predictor of Spring semester attrition was Fall semester GPA, in the sense that GPA was the strongest predictor of attrition after accounting for the inter-relationships among the predictors. The loss of one credit is associated with one additional grade point (B = - 1.0, SE = .09). Fall GPA explained 13% of the variance in Spring semester attrition. Fall semester attrition and reading scores modestly explained an additional 5% of the variance in Spring semester attrition.

Analysis 2

In the second analysis, attrition was measured by an indicator of students who failed to earn any number of credits for which they registered. Regression analysis indicated that an increase in 1 grade point in the Fall semester was associated with a 10% decrease in students failing to earn credit in the Spring semester. An ROC analysis indicated that GPA was an inefficient indicator of future attrition (See Table 2). The students identified as having GPAs of 1.0 or less in the Fall semester predicted 27% of all students who failed to earn credits for which they registered in the following semester. However, 10% of the students with GPAs of 1 or less in the Fall earned all credits for which they registered in the Spring. Thus, a modest 27% true positive rate was achieved with a low 10% false positive rate, but required the identification of 24% of the entire sample. Raising the threshold for GPA increased the true positive rate, but also increased the false positive rate, and identified increasingly larger proportions of the student body.

A better rate of prediction was found using multiple indicators based on a measure that indicated students with Fall GPAs below 2, reading placement scores below 26, and a failure to earn 3 or more credits in the Fall. These cut-off scores were based on the thresholds that correctly identified 50% of at-risk students for each respective indicator (other thresholds could produce better results). Using the multiple indicator, 67% of the students who failed to earn credits in the Spring were identified, but with a 38% false positive rate. This indicator flagged 36% of the sample. (See Table 2, above.)

Conclusions

Measures of academic ability are significant, but weak predictors of attrition, which is not consistent over semesters. Thus, interventions aimed at improving academic performance are likely to require great improvements that may only produce modest reductions in attrition. Given that correlations are not indicative of causation, increases in scholastic aptitude may have even less influence on attrition than indicated by the already modest correlations.

The use of indicators of students at risk for attrition provides an effective, but inefficient means of targeting at-risk students for interventions aimed at reducing attrition. The majority of students who will fail to earn credits can be identified prior to their failure to earn credits, but doing so involves a high false positive rate and requires targeting a large proportion of the student body. The risk of falsely labeling students as potential “drop-outs” and the resources needed to intervene in such as large proportion of the student body needs to be weighed against benefits of intervention.

Recommendations

Interventions aimed at reducing attrition may need to focus on factors other than students’ academic ability. Students at risk for attrition may require intervention by social workers who can refer students to the services required to address the problems that impede their persistence. These problems are likely to be unrelated to academic ability, and may require social, financial, family, legal and health services that are beyond the resources of Community College of Philadelphia. For example, a student who is not able to get consistent day care for her children may miss classes and eventually drop classes. Such a student may be more likely to persist if she is referred to the appropriate professionals, such as a social worker who can help her find day care.

Reference

Doshi & Grosset. Reasons for student attrition at Community College of Philadelphia, Office of Institutional Research report #137. March 2004. Available Online

©Copyright 2006. Contact author for permission

Maintained by Jay Howard,Jan 2006