Predicting Persistence of Hispanic Students in Their 1st Year of College
Students from minority and economically disadvantaged populations have traditionally
had low levels of graduation rates (Keup, 2005) and it has been observed
that the largest proportion of student attrition in college occurs in their 1st year, particularly
for Mexican American students (Hawley & Harris, 2005; Tinto, 1993).
Therefore, being able to identify 1st-year students with characteristics that increase
their probability of dropping out will facilitate the design and implementation of
early intervention strategies.
The purpose of this research was to identify which demographic and academic variables are associated with 1st-year student attrition and persistence and to determine the extent to which these factors affect the students’ decision to remain or leave the university. The methodology allowed the us to determine the extent to which each of these variables affects students’ permanence after their 1st year. After presenting and discussing our findings, we summarize the findings and offer recommendations to institutions of higher education on how to identify students who are more likely to withdraw from the institution.
Hispanic representation in colleges has increased over the past two decades; however, the proportion of Hispanic students finishing college has not improved (Hernandez & Lopez, 2004; Llagas & Snyder, 2003). Hispanic high-school graduates are enrolling in college at higher rates than other major ethnic populations but are behind other minorities in graduation rates (Hernandez & Lopez, 2004; Castillo et al., 2006). For this reason, efforts in student recruitment have to be matched with efforts to retain students until graduation. Recruitment means little if students cannot finish their degrees. Student retention is of utmost importance in an institution serving a largely Hispanic community because college attrition rates have been high for these students for some time (Llagas & Snyder, 2003; Tinto, 1993; Zurita, 2005). Among Hispanics, 1st-year college dropout rates seem to be highest for Mexican American students (Hawley and Harris, 2005).
Studies on college retention have shown that certain demographic, personal, and academic variables are related to student attrition. Traditionally accepted variables in the literature include student characteristics such as income and ethnicity, precollege academic preparation, motivation and involvement, type of institution and its image, student services offered, and the degree of student-institution interaction (Astin, 2005; Bordes & Arredondo, 2005; Keup, 2005; Umoh & Eddy, 1994; Zurita, 2005).
Perhaps the most cited model of student persistence in the literature is that of Vincent Tinto, which looks at the importance of academic integration and social integration and their impact on college retention and completion. Tinto (1975) noted that family background issues (income, values, and parents’ education), individual attributes (race, sex, and ability), and precollege success (high school GPA) are among the most important factors affecting students’ success. In a later study, Tinto (1993) also stated that in addition to precollege enrollment characteristics, students’ commitment to the institution, commitment to goals, and integration with the campus environment would be the best predictors of student retention. According to Tinto, the postenrollment variables are more important in the student’s decision to withdraw than what happens before enrollment in college.
Campus integration can be broken down into academic and social dimensions. Academic integration, as measured by such variables as grades and contacts with faculty, has been demonstrated to have the strongest positive relationship with student retention and degree attainment (Lee, 1999; Tinto, 1975). In particular, student-faculty interaction has been shown to have the greatest effect on student satisfaction and retention. A survey taken at 944 institutions (2- and 4-year public and private colleges and universities) showed that a caring attitude toward students by faculty and staff was the most important retention factor at those institutions (Roueche, 1993). Other studies have confirmed the importance of student-faculty or -counselor interaction (Bordes & Arredondo, 2005; Leon, Dougherty, & Maitland, 1997). In a study of Latino students, Bordes and Arredondo (2005) found that interaction in the form of mentoring by faculty or counselors improved the students’ perception of the university environment; according to Tinto (1993), this integration increases the students’ commitment to the institution and degree attainment. Minority students have more difficulty adjusting to the demands of public research universities because the demands for research limit the time faculty have to interact with students. Community colleges have an advantage in this regard. It is evident in the literature that student interaction with faculty is essential to student retention, in particular during their 1st year. Formal classroom student-faculty interaction is important but also nonclassroom activities such as intellectual discussions are reported to be positively correlated with student persistence (Lee, 1999).
Pascarella and Terenzini (1991) found that the importance of academic integration becomes more significant in student retention and degree attainment as the level of the students’ social integration decreases. Conversely, social integration becomes more important as the level of academic integration decreases. Furthermore, the importance of both the academic and social integration increases as the level of the students’ family education and commitment to graduation decreases. This could have implications for Hispanic-serving institutions because they tend to have large percentages of 1st-generation college students.
Other variables such as gender, placement grades, GPA, and financial aid had mixed results in the literature. In some cases they seemed to be directly related to student retention and in other studies they did not show statistically significant results. For example, Bean and Metzner (1987) found that college GPA had a direct relationship with student retention, but Rickinson and Rutherford (1995) did not find a statistically significant relationship between these two variables. Placement grades and retention did not show a relationship (Bean & Metzner, 1987; Rickinson & Rutherford, 1995). Making the institution the number one choice for their education made no difference to students in terms of retention (Laman, 1989). Availability of financial assistance did show a direct effect. The probability that the student would persist increases with the amount of financial support available to the student (Cofer & Somers, 2000). Salinas and Llanes (2003) found two early indicators linked with student persistence. In their study, conducted at a university with predominantly Hispanic students, they found that most students who did not return to continue their studies had reduced the number of credit hours taken during their previous semester or were placed on probation because of low performance.
This study was conducted in a predominantly Hispanic university located in the Southwestern United States using a cohort of freshmen students enrolled in a program designed for at-risk students. First-year students failing one of the sections of the state-mandated college placement test are required to participate in the program. There were 311 students initially enrolled in the program at the beginning of the Spring 2001 semester. At the end of the semester, 134 students volunteered to take the survey for this study.
The ethnic background of the students was 98.5% Hispanic, which mirrors closely the ethnic background of the total university student body at 91.7%. More than half of the participants, 62.6%, came from families whose total family income was less than $25,000. The survey consists of 130 questions and is divided into three major sections: background information, attitudes and behavior related to college, and attitudes and perception of mentoring. Of the 134 surveys administered, 28 were eliminated because they contained missing or invalid data, leaving 106 valid surveys.
The 106 students with valid surveys were then followed up in the Spring 2002 semester. We created a variable, enrolled, to list 1 if the student was still enrolled in the university after 1 year and 0 if the student was no longer enrolled (see Table 1). Of the 106 usable observations, 73 had a 1 coded for enrollment and 33 had a 0. Therefore, without an econometric model one can correctly predict about 69% (73 of 106) of the students’ choice by simply assuming that all students taking the survey will remain in college after 1 year. Unfortunately, this method would not tell us anything about those that will drop out of the university or what variables are associated with such a decision. A probit or logit model would report the extent to which variables affect the decision to remain in or leave college.
This enrolled variable was used as the dependent variable. The explanatory variables used in this study were based on the literature of previous research on this topic and the researchers’ judgments, taking into account the data available in the survey. Table 1 lists the variables used in the study.
The concepts attributed in the literature related with student attrition, such as academic integration, are measured by a factor scale that represents several of the dimensions of the concept. This scale is usually the combined score of the responses to different questions in each scale but often the correlation (Cronbach’s alpha) between the different dimensions these questions were trying to measure is very small. Therefore, the validity of such a scale as representative of a concept was questionable. For that reason and for the sake of simplicity, a single question was selected to represent a concept when a given question was perceived to be representative of the concept. The only exception was the mentor variable for which no single variable combined most of the functions of the mentor-student relationship. For this reason that particular variable was created by summing the answers of two questions
related to mentoring so as to better represent the concept (see Table 1). Certain variables identified in the literature to be related to student retention, such as high school or college GPA, were not present in the survey. A proxy was used where possible.
A logit statistical model was used to find which factors were likely to contribute to student retention and which variables were likely to affect the decision to leave the institution. The logit coefficients or odds ratios capture the impact of a unit change of the independent variable on the natural log of the likelihood of choosing to stay in school over the probability of choosing to leave.
Because the coefficients of the logit model are not directly interpretable, it is convenient to calculate the partial derivatives to interpret the effects of the independent variables on the decision to remain in school.
It is important to remember that the results given here apply to low-performance students and not to the general student population. It is essential to keep this in perspective because, according to the literature, the behavior of students will vary with the level of academic integration.
Table 2 reports the binomial logit results. The chi-square statistic, which tests the null hypothesis that all the coefficients in the model are 0, was significant at the 5% level. Therefore, the model can be used as a predictor of student retention.
The estimated model was able to predict correctly whether the student would remain in the university the following year in 84 of the 106 (79%) cases used. This represents a success rate (hit ratio) of approximately 4 of 5 students being correctly forecasted as dropping or continuing in the university after their 1st year. The partial derivatives of the dependent variable with respect to the explanatory variables evaluated at the sample means of the explanatory variables are listed in the last column of Table 2. The partial derivatives tell the probability that a particular variable adds to the correct forecast of the student remaining or dropping from the university during the 1st year.
Contrary to the literature, the age of the student does not contribute to the student’s permanence in college because the age variable is not statistically significant. The positive and statistically significant (p < .05) coefficient for the female variable suggests that at-risk female students are more likely than males are to remain in the university after 1 year. The partial derivative tells us that being female carries a 25% higher chance than being male for continuing studies.
Contrary to what one might expect because of the additional responsibilities brought by having a family, in general being married and having children are not related to the student’s decision to continue in college. For some students, having children or being married may be a reason for leaving school to attend to those responsibilities but for just as many these are reasons to remain in school and improve their future.
The negative and statistically significant (p < .05) coefficient of the income variable indicates that students with higher household incomes tend to leave the university. For every thousand dollars of additional income beyond $22,000, the likelihood of leaving college in the 1st year increases by about 1%. The effect on the probability is the largest at the mean of the explanatory variable, and it becomes smaller further away from the mean in either direction. That is, for every additional thousand dollars of income beyond, for example, $50,000, the probability of continuing in college continues to increase but at less than 1%.
Interestingly, the negative and significant coefficient (p < .10) for the fatheredu (father’s years of education) variable implies that the father’s education is inversely related to the student’s decision to remain in college. Apparently, fathers with more years of formal education influence their children to leave the university. According to the partial derivatives, for every additional year beyond 9.5 years of the father’s education, the likelihood of staying in college declines by about 2%.
On the other hand, the mother’s education seems to influence the student in exactly the opposite way. For every year of the mother’s education above 9.7 years, the probability of continuing in college increases by about 2%. Although this variable is not statistically significant, it is very close to being so (p = .1124).
The number of hours worked per week does not influence the student’s decision to continue or drop out of the university.
Neither considering this university as the first choice for their higher education nor planning to transfer to another institution seems to be related to the retention rate, but planning to graduate from this university does contribute greatly to the student’s permanence at the institution.
The grad variable (intention to graduate from UTB or TSC) is statistically significant (p < .05) and positively related to the choice of enduring in the university. In fact, this variable is the one that contributes the most toward the prediction of whether the student will indeed continue his or her education at the institution. Students who manifest their intention to graduate from this institution have a 42% higher chance of continuing their education at this institution after 1 year when compared to those students who do not plan to graduate from this university.
Informal interaction with faculty and staff do not contribute to the decision of whether to remain at this university according to the variables faculty-int (satisfaction with faculty interaction), staff-int (satisfaction with staff interaction), and faculty (faculty willingness to interact outside of class).
Contrary to part of the academic literature, academic integration as measured by student-faculty interaction does not affect the retention rate among these at-risk participants. According to the literature, student interaction with faculty is an essential factor in student retention. It would be interesting to see if the findings with at-risk participants apply to the general student population. The general population of students may indeed consider informal interaction with faculty as a factor affecting their decision to continue at this university.
However, these results are consistent with another part of the literature, which states that the importance of academic integration becomes more relevant in student retention and degree attainment as the level of the students’ social integration decreases. In turn, for students who consider social integration important, the level of academic integration decreases in importance. This seems to be happening with these at-risk students because, as reflected by the social integration variable, the decision to remain in this university is more related to social activities.
Social integration in the campus was measured with a student-center variable. This statistically significant (p < .05) variable reports whether “spending time socializing with friends in the Student Center or other campus areas is important” to the student. Student who agree with this premise show an 11% higher likelihood of continuing their education in this university over those students who neither agree nor disagree. Students who strongly agree that socializing with friends on campus is important have an even higher probability of staying in college.
The number of times the student has taken the state mandated entrance test does not affect the attrition rate. Academic integration is sometimes measured with the student’s college GPA but because this information is not available in the survey the statement “Getting good grades is important to me” was used as a proxy to measure academic performance. The grades variable is statistically significant (p < .10) and positively related to student retention. Students who strongly agree that getting good grades is important have a higher probability of continuing in the university.
Surprisingly, the students’ level of satisfaction with their academic experience in this university (acad-sat) does not influence their decision to continue or leave the institution in their 1st year.
Receiving financial aid (fina-aid) or using the Learning Assistance Center’s (LAC) tutoring services did not have an impact on retention of at-risk students. Similarly, having a mentor does not contribute to the choice of staying in college for their sophomore year. Because the mentor variable was created by subjectively, selecting two questions of many to represent this activity, it may be possible that a different selection of questions than the ones selected in this study may better represent this activity and may then reflect a relationship to retention. Also, some of these factors may be related to academic commitment (good grades), which in turn is positively related to student retention.
