September 11, 2007
Predictors of Persistence to Graduation: Extending a Model and Data on the Transition to University Model
By Wintre, Maxine Gallander Bowers, Colleen Dorothy
Abstract Persistence to graduation was investigated with 944 (272 males, 672 females) undergraduate students at a large, commuter Canadian university. Within six years, 57.9% of the students had graduated, 9% remained enrolled, and 33.1% were not enrolled or graduated. Results indicate that gender, parental support, stress, depression, and first-year GPA were direct predictors of persistence. When Baker and Siryk's (1985) adaptation scales were added to the model, social adaptation and goal and school commitment became direct predictors in addition to parental support and first- year GPA, whereas high school average and academic adaptation were indirect predictors of graduation. Results indicate that the predictors of persistence to graduation are not the same as the predictors of the transition to university.Resume
La persistance en vue de l'obtention d'un diplome a ete etudiee dans un groupe de 944 etudiants du premier cycle (272 hommes; 672 femmes) d'une grande universite canadienne sans residences. Durant une periode de six ans, 57,9 % des etudiants ont obtenu leur diplome, 9 % sont demeures inscrits et 33,1 % n'etaient pas inscrits ou n'ont pas obtenu de diplome. Les resultats indiquent que les predicteurs directs de la persistance sont le sexe, le southen des parents, le stress, la depression et la moyenne ponderee cumulative (MPC) de la premiere annee d'etudes. Apres l'ajout des echelles d'adaptation de Baker et Siryk (1985), l'adaptation sociale et l'engagement envers le but et l'etablissement d'enseignement sont devenus des predicteurs directs en plus du southen des parents et de la MPC de la premiere annee d'etudes, alors que la moyenne obtenue a l'ecole secondaire et l'adaptation a l'universite etaient des predicteurs indirects de l'obtention d'un diplome. Les resultats demontrent que les predicteurs de la persistance en vue de l'obtention d'un diplome ne sont pas les memes que les predicteurs de la transition entre l'ecole secondaire et l'universite.
In North America, approximately 60% of adolescents attend some form of post-secondary education, up from less than 15% in the 1930s (Steinberg, 1999). However, 30-40% of those entering postsecondary education may drop out, interrupt, or fail to complete their degrees or programs of study (e.g., Pantages & Creedon, 1978; Smith, 1991). These considerable attrition rates highlight the importance of investigating undergraduate persistence through to graduation.
The present study uses a theoretical framework to examine the relevance of numerous predictor variables (both past and current) to predict university graduation. For example, one can question the continuing role of parenting styles and relations with parents on academic success (cf. Wintre & Yaffe, 2000) at this later "emerging adult" developmental stage (Arnett, 2000). It seems appropriate to establish variables of potential relevance (e.g., the role of parents, social adjustment, etc.) before attempting to manipulate variables that may enhance university graduation. By longitudinally extending a model and the data in a transition to university study (Wintre & Yaffe, 2000), this paper illuminates the potential links between familial and background variables, parenting styles, current relations with parents, psychological well-being, academic achievement, and university adaptation to students' persistence to graduation.
The Wintre and Yaffe (2000) model (see Figure 1) proposed a developmental sequence determining student adjustment at university. It was theorized that students' pre-entry characteristics and experiences (with a focus on relations with parents) influence participation in the formal and informal aspects of the academic and social systems of the university, thereby shaping students' academic and social integration. The students' pre-entry characteristics include gender, demographic variables such as generational status, socio-economic status (finances and parental education levels), and parents' parenting styles, which help shape childhood experiences and achievement in school. Chronologically, this is followed by the students' current relationships with their parents (including reciprocity in relations with parents, parental social support, autonomy, and discussion with parents about university) upon starting postsecondary school. Wintre and Yaffe (2000) assumed these variables would be associated with students' psychological well- being (perceived stress, depressive symptomology, and self-esteem). The original model proposed that these variables would predict (indirectly and directly) adaptation and adjustment to university, as divided into academic and social integration into the university, as well as personal-emotional adjustment, and university goals and school commitment. The present study extends the Wintre and Yaffe model to predictors of persistence in university by empirically testing the previous variables along with whether positive adaptation and academic achievement predict commitment to graduation and degree completion.
The theoretical framework for the above was derived from two models of student development (Chickering, 1969, Chickering & Reisser, 1993; Weidman, 1989) and one model of student departure (Tinto, 1975, 1993). Chickering recognized an absence of a systematic framework for research on student development. He proposed seven "vectors of development," including a sense of competence, managing emotions, autonomy, identity, increased ability to interact with others with a degree of tolerance, developing purpose, and " the clarification of a set of beliefs that has internal consistency and that provide a tentative guide for behaviour" (Chickering, 1969, p. 17). The present model incorporates the first three vectors by examining self-esteem as a measure of perceived competence, depression and perceived stress as measures of managing emotions, and autonomy.
Weidman (1989) developed a more explicit model of undergraduate socialization. He theorized an important role for parents in a continuing socialization capacity. He believed that students' performance in university might be affected by their ability to cope with problems at home and in other community settings. The present model, by examining parenting styles, parental social support, perceived reciprocity with parents, and discussion with parents about university, integrates Weidman's emphasis on socialization forces, past and present.
Tinto (1975, 1993), in examining university attrition, believed that students' attributes and background characteristics, combined with their goals and institutional commitment, provide an initial basis for integration into the university environment. In distinguishing between the academic and social domains of university life, Tinto suggested that one might achieve integration in one domain without doing so in the other. Tinto's model includes the differential influences of the familial, academic, social, and cultural systems. Empirical support of his theory, evaluated according to testable propositions, (Braxton, Sullivan, & Johnson, 1997) indicates only modest support of Tinto's theory (Braxton, Bray, & Berger, 2000). The present model includes Tinto's conceptualization of the academic and social domains, as well as goals and institutional commitment.
The results from Wintre and Yaffe (2000) were analyzed separately by gender using multiple regression models. For the males, with regard to overall adaptation to university, the model accounted for 56% of the explained variance, with perceived stress and depression negatively related to adaptation, and identity and reciprocity with parents positively related. For the females, the model accounted for 63% of the explained variance, with depression and stress being negatively related, and self-esteem and discussion about university with parents being positively linked to adaptation. Furthermore, additional analyses found indirect positive effects for authoritative parenting and indirect negative effects for authoritarian parenting on the current relationships with parents. The models used to examine first-year grade point averages (GPA) also varied by gender. For the males, there was a three-variable model accounting for 48% of the variance, including perceived academic adaptation, graduating high school average, and paternal level of education. For the females, there was a sixvariable model accounting for 38% of the explained variance, with graduating high school average, academic adaptation, stress, lack of maternal authoritarianism, lack of discussion with parents about university, and maternal level of education predicting first-year grade point average.
The present study tests the related and extended theoretical model of students' persistence to graduation in their undergraduate careers. The first analysis employs the same participants and model from Wintre and Yaffe (2000), with the addition of first-year GPA, to predict their persistence to graduation. The second analysis, with a subsample of the above group, adds the adjustment variables (academic adaptation, social adaptation, personal emotional adaptation, and goal and institutional commitment) as predictors in the model of undergraduate persistence to graduation from the first analysis. Gender and Demographic Variables
Background characteristics encompass factors that may contribute to students' academic success and graduation (Tinto 1975, 1993; Weidman, 1989). Given that Wintre and Yaffe (2000) found different predictors by gender for adaptation to university and academic achievement in first year, it was hypothesized that predictors of undergraduate persistence would differ depending upon gender. Other demographic variables also seemed relevant. For example, paternal level of education was a predictor of first-year GPA for males, and maternal education was a predictor of first-year GPA for females (Wintre & Yaffe, 2000). Finances could also be implicated in graduation, given that if students need to work to support their education, this may interfere with their ability to attend to school- related activities. The present study included information to examine immigrant generational status (i.e., country of birth, parents' countries of birth) that, although a nonsignificant predictor in Wintre and Yaffe's model on adjustment to university, was explored in relation to persistence. The rationale was based on findings that immigrant generational status is associated with parenting and psychological well-being (Wintre, Sugar, Yaffe, & Costin, 2001).
The influence of parents is present in Weidman's model (1989), as weU as Chickering's (1969) vectors of identity development, and could be included in the individual characteristics Tinto (1975, 1993) refers to when students enter college. Several researchers have demonstrated that parent-child relationships characterized by authoritative parenting (e.g., Buri 1989; Steinberg, Lamborn, Dornbusch, & Darling, 1992), mutuality (e.g., Wintre & Yaffe 2000), parental support (e.g., Cutrona, Cole, Colangelo, Assouline, & Russell, 1994), and discussion with parents about university (e.g., Pancer, Pratt, Hunsberger, & Alisat, 1995) are associated with positive outcomes. These outcomes include adolescent school success, higher academic achievement, self-esteem, self-reliance, social competence, less depression, and better adjustment.
Wintre and Yaffe (2000) extended the findings of the beneficial associations of authoritative parenting beyond high school. They found that authoritative parenting style had a positive, indirect, relationship with adjustment to university that was mediated by current relationships with parents (perceived reciprocity and support) and psychological well-being. The present study includes parenting styles, perceived reciprocity with parents, support from parents, discussion with parents, and autonomy as potential parent socialization predictors of graduation.
Psychological 'Well-Being Variables
Wintre and Yaffe (2000) added psychological well-being constructs to the predominantly sociological theories of student transition to university and they are incorporated in the present study to examine their potential contribution to students' persistence and graduation. For example, moving or changing residence is associated with depression (Fisher & Hood, 1987), which may affect adjustment and academic success in university. Chickering's (1969) model includes a vector for managing emotions. Furthermore, psychopathology among university students is increasing (O'Malley, Wheeler, Murphey, O'Connell, & Waldo, 1990). It is not unusual for students to withdraw from university for personal, emotional, or medical reasons (Meilman, Manley, Gaylor, & Turco, 1992; Wintre, Bowers, Gordner, & Lange, 2006).
Self-esteem, stress, and depression have been individually examined in relation to university graduation. First, research has shown that self-esteem is positively related to adjustment and first- year GPA in university (Bettencourt, Charlton, Eubanks, Kernahan, & Fuller, 1999; Hickman, Bartholomae, & McKenry, 2000). Second, while stress is associated with increased symptomatology (e.g., somatic symptoms, depression, anxiety) and decreased grade point average, it can also be a motivating factor for growth and positive adjustment (Zitzow, 1984). Third, depression is related to university attrition rates (Hammen, 1980; Meilman et al., 1992). Finally, Wintre, and Yaffe (2000) found that self-esteem, stress, and depression were differentially related to achievement and adjustment according to gender. For females, stress and depression were negatively related, while self-esteem was positively related to adjustment. Stress was also a direct predictor, while depression and self-esteem were indirect predictors of first-year GPA. For males, stress and depression were negatively related to adjustment and indirectly related to first-year GPA.
First-Year Academic Achievement
First-year academic achievement is another factor expected to contribute to students' persistence to graduate. Students must pass their courses and meet minimum academic requirements, referred to as structural integration in Tinto's theory, in order to graduate. Research on university adjustment and success demonstrated that high school average is the best predictor of first-year university GPA (McCausland & Stewart, 1974; Wintre & Yaffe, 2000), and that GPA increases over consecutive semesters at college (Wilczenski, 1993). Furthermore, when compared to other predictors, researchers found high school average (McDonald & Gawkoski, 1979; Neely, 1977) and university GPA (Ryland, Riordan, & Brack, 1994) to be the best predictors of persistence. Therefore the Wintre and Yaffe model (2000) was expanded here to include first-year academic achievement as a predictor of undergraduate persistence.
First-Year Adjustment to University
While adjustment to university, as measured by the Student Adaptation to College Questionnaire (SACQ: Baker & Siryk, 1989), is widely used as a dependent measure (see Lamothe et al., 1995; Oppenheimer, 1984; Pratt et al, 2000; Wintre & Yaffe, 2000), it is seldom examined as an independent measure. Gerdes and Mallinckrodt (1994) provide an exception in one of the few longitudinal studies on attrition. In a sample of 208 students, after six years of enrollment, 70% had graduated, 2% were still enrolled, and 28% were labeled "leavers." Given their findings that academic, social, and personal-emotional adaptation were all significant positive predictors of persistence, and Tinto's focus on social and academic integration, we expanded the present model to include all four categories (academic, social, personal-emotional, and goal and institutional commitment) of first-year adaptation as predictors of graduation.
The Model and Related Hypotheses
As a result of the above literature review and research findings, the present study employed a chronological, developmental sequence to examine the potential direct and indirect effects of gender, demographic variables (generational status, fathers' and mothers' highest level of education, and perceived socio-economic status), parenting style (authoritative, authoritarian, permissive / laissezfaire), graduating high school average, current relations with parents (perceived reciprocity, social support from parents, discussion with parents, and autonomy), psychological well-being (self-esteem, perceived stress, and depressive symptomotology), first-year GPA, and adaptation to university (academic, social, personal /emotional, and goal and institutional commitment) on students' graduation. See Figure 1 for a representation of the present model.
Based on the findings of Wintre and Yaffe (2000) and the model in Figure 1, we made the following predictions. Hypothesis 1 predicted that the same-sex parental level of education would be a predictor of persistence, such that paternal level of education would be significant for the males but not the females, whereas maternal level of education was significant for the females but not the males. Hypothesis 2 expected that parenting styles would have indirect effects on persistence. More specifically, it was hypothesized that authoritative parenting would continue to be a significant positive predictor and authoritarian parenting a negative predictor of academic persistence. Hypothesis 3 predicted that past parenting styles would be mediated by current relations with parents. Hypothesis 4 predicted that current relations with parents, especially reciprocity with parents and discussion with parents about university, would be related to graduation. Hypothesis 5 predicted that the largest contribution to the outcome variables would be academic achievement, with first-year GPA as a mediator of graduating high school average on persistence. Hypothesis 6 predicted that current relations with parents would be mediated through psychological well-being as predictors of persistence. Depression and perceived stress would be negatively related to academic persistence, and self-esteem positively related. Finally, Hypothesis 7 proposed that first-year adjustment would also predict graduation. More explicitly, it was expected that all four adjustment scales would be positive predictors of graduation.
All participants were enrolled at a large, commuter Canadian university (over 40,000 students). A total of 944 (272 males and 672 females) first-year students between the ages of 17 and 27 (M = 19.14, SD = 1.21) participated in a data collection during the first week of classes in September 1995 and the data was used for the first set of analyses. A randomly selected subsample of 388 (109 males and 279 females) participated in a second data collection (March 1996) when they completed the adaptation measures (SACQ) and the data was used for the second set of analyses.
Demographic data. The majority of the students (78.8%) reported that their families were intact, 74% lived at home with their families, and 18.5% lived in residence halls. This is congruent with reports that the majority of today's university and college students are commuters (Slade & Jarmul, 1975; Stewart, Merrill, & Saluri, 1985). Participants identified 108 majors (25.5% psychology majors and 15.6% with undeclared majors). Culturally, although 77.5% of the participants were born in Canada, the remainder cited 53 countries of origin. A total of 94 countries of origin were identified for the parents. Data for "Immigrant Generational Status-Canadian" (Wintre et al., 2001) indicated that 22.2% of the sample were immigrants to Canada, 42.2% were Canadian-born children of two immigrant parents, 11.6% were Canadian-born children of one immigrant parent, and 24.2% were (at least) second-generation Canadians. Although 70.3% of the sample reported English as their primary language, the remaining students identified 65 different languages spoken at home. Furthermore, participants declared 242 different cultures, and 30.8% considered themselves members of a visible minority. With respect to parents' highest level of education, 24.4% of fathers and 23.2% of mothers attended elementary school and/or some high school, 11.1% of fathers and 21.9% of mothers completed high school, 27% of fathers and 25.3% of mothers completed technical training or community college, 6.9% of fathers and 8.1% of mothers attended university, 13.6% of fathers and 14.1% of mothers completed a bachelor's degree, and 16.4% of fathers and 6.7% of mothers completed (some) graduate work or attained a professional degree. Financially, 62.3% of the students considered their family to be of average means, and 29.6% considered their family to be above or well above average means.
In addition to the demographic measures, the following measures were collected in questionnaire packages in September of the students' incoming year: parenting style, current relations with parents, psychological well-being, and graduating high school average. Data on the adaptation to university were collected in February 1996, with the subsample. School records data were collected during the sixth academic school year, as the majority of students intending to complete a degree were expected to have graduated within this time.
The Parental Authority Questionnaire (PAQ; Buri, 1991). Three 10- item scales were created based upon Baumrind's (1971) authoritarian, authoritative and permissive parenting prototypes. The psychometric properties, reliability (Cronbach's alpha ranges from .74 to .87), criterion validity, and discriminant validity are reported to be very good (Buri, 1991). Alphas for the present data ranged from .74 to .90.
The Perception of Parental Reciprocity Scale (POPRS; Wintre et al, 1995). This 43-item measure, based upon the developmental task of late adolescence described by Youniss (1980; Youniss & Smollar, 1985), examines the degree of perceived reciprocity in the parent- child relationship from the offspring's perspective. The psychometric properties, reliability (Cronbach's alpha of .95), criterion validity, construct validity, and discriminant validity are well established (Wintre et al., 1995). From the present study, the alphas were .95 for the overall scale, .83 for the general subscale, and .92 and .91 for the mother and father subscales, respectively.
The Social Provisions Scale-Parent Version (SPS-P; Cutrona, 1989). This 12-item measure examines perceived social support from parents. The psychometric properties, reliability (Cronbach's alpha between .81 and .91), and validity are good (Cutrona, 1989). The present alpha was .81 for the SPS-P.
The Discussion with Parents Scale (Pancer et al., 1995). This 4- item scale measures the extent to which participants discuss features of university life with their parents (e.g., "what classes will be like,""what social life at university will be like"). Wintre and Yaffe (2000) found these items to be reliable and met acceptable criteria to be used as a short scale. In this study, the alpha was .90.
The Autonomy Scale of the Psychosocial Maturity Inventory (Greenberger, Josselson, Knerr, & Knerr, 1974). Three ??-item subscales reflect different dimensions of autonomy: self-reliance, a healthy sense of identity, and work orientation. The psychometric properties, reliability (Cronbach's alphas for the subscales are .72, .83, and .67, respectively), discriminant validity, concurrent validity, and construct validity, are good (Greenberger et al., 1974). Cronbach's alpha for autonomy in this study was .89.
The Self-Esteem Scale (Rosenberg, 1965). This 10-item scale focuses on the self-acceptance aspect of self-esteem. The psychometric properties, reliability (testretest correlation over two weeks of .85; Silber & Tippett, 1965), and convergent, discriminant, and predictive validity (Rosenberg, 1965; Silber & Tippett, 1965), are excellent. Alpha in this study was .87.
The Perceived Stress Scale (PSS; Cohen, Kamarch, & Mermelstein, 1983). This 14-item scale measures one's appraisal of his/her life as stressful. The psychometric properties, reliability (Cronbach's alpha ranges from .84 and .86), and validity are very good (Cohen et al, 1983). Present alpha was .85.
The Beck Depression Inventory (BDl; Beck, Ward, & Mendelson, 1961). This 21-item inventory measures the presence of depressive symptoms and their depth or severity. The psychometric properties of reliability and validity are good (Beck & Beamesderfer, 1974; Hammen, 1980) and the scale is widely used. The alpha for the BDI in the present sample study was .84.
The Student Adaptation to College Questionnaire (SACQ; Baker & Siryk, 1984). This 67-item scale yields four subscales measuring different dimensions of adaptation to university: academic adjustment, social adjustment, personal emotional adjustment, and goal and institutional commitment. The psychometric properties, reliability (Cronbach's alpha ranges from .92 to .94), and criterion validity are highly acceptable (Baker & Siryk, 1984). The alphas for the present sample were .91 for the overall scale, .86 for academic adjustment, .85 for social adjustment, .85 for personal emotional adjustment, and .81 for university identity and commitment.
School records. According to the school records, of the 944 students, 310 had not graduated, 87 were still enrolled, and 547 had graduated within six years. Upon examination of the means and related t-tests, it appeared that those students still enrolled were more similar to the graduating group. As a result, it was decided that the outcome variable persistence would be based on the following dichotomy: not graduated from the present university within six years and graduated or still enrolled.
There were two sets of analyses. The first set comprised the predictor variables available from the first week of classes. The second set of analyses, based on the subsample, allowed for the addition of the adjustment to university predictors, while retaining the larger overall sample for all other predictors. Both sets of data were analyzed using logistic regression to predict the outcome of persisting to graduation.
Descriptive statistics for each of the variables can be found in Table 1. Data were organized into eight blocks according to the hypothesized model (see Figure 1) with its developmental sequence. Block 1 (Gender) consisted of the participants' gender; Block 2 (Demographics) consisted of immigrant generational status, parents' education level, and perceived finances; Block 3 (Parenting Style) consisted of the parent ratings of authoritative, authoritarian, and permissive; Block 4 (Previous Academic Achievement) consisted of graduating high school average; Block 5 (Current Relations with Parents) consisted of perceptions of parental reciprocity, parental support, discussion with parents, and autonomy; Block 6 (Psychological Well-Being) consisted of self-esteem, perceived stress, and symptoms of depression; Block 7 (Adjustment to University) consisted of the ratings on the SACQ and the four subscales of academic adaptation, social adaptation, personal- emotional adaptation, and goal commitment and school attachment; Block 8 (Academic Achievement) consisted of first-year GPA. The outcome variable was whether or not the participant persisted toward graduation in the six-year period. Table 2 summarizes the correlations between variables.
Of the 944 participants, 67.2% had persisted and 32.8% were not graduated or enrolled. Of the persistence group, 24.8% were males and 75.2% were females. The persistence group breakdown for GS-C was: 22.4% immigrants, 44.1% first-generation Canadians, 10.1% Canadian-born children with one immigrant parent, and 23.4% second- generation Canadian.
Analysis 1: Predictors of Persistence
Given the dichotomous outcome for persistence, logistic regression was used to predict persistence for all students (n = 825; this "n" reflects a decrease from 944, due to partially missing data evident in Table 1 and the fact that the type of analyses conducted using SPSS software omits the participant entirely if one data entry is missing (e.g., "how did you get along with your roommate?" but the participant did not have a roommate). Predictors within each of the blocks (1 through 6, and 8) were examined both sequentially (when entered) and partially (final model) for all students. Father's education level, mother's education level, finances, father's parenting style, mother's parenting style, discussion with parents, autonomy, and self-esteem were nonsignificant in both cases. Thus, some blocks were revised or removed from the model (see Table 3). There were no significant interactions.
As the predictors were entered sequentially, gender (chi^sup 2^ = 9.94, p
Analysis 2: Predictors of Persistence, Adding Adjustment to University Measures
The subsample for this analysis consisted of 388 students (this "n" decreased from 388 to 337 again due to partially missing data evident in Table 1 and the type of analyses conducted using SPSS software). Data from the subsample were compared to data from the full sample using a one-way ANOVA with an alpha level of .01, as all variables were included. There were no significant differences between the sample and the subsample except for the three most recent and related predictors. Those in the subsample scored significantly higher on high school graduating average, F(1, 908) = 12.06, p
The significant blocks from the previous logistic regression were entered with the addition of the four adaptation to university predictors. In order to keep the developmental sequence, Blocks 1 to 5 remained the same, the adjustment to university variables were entered as Block 6, and first-year GPA was entered as Block 7. Predictors within Block 6 were examined both sequentially (when entered) and partially (final model).
The overall adjustment scale was significant when entered (chi^sup 2^ = 22.57, p
Follow-Up on Two Surprising Findings
Additional analyses were conducted to investigate why perceived stress was a positive predictor in the first model and why social adjustment was a negative predictor in the second model. These results were surprising given that the former was inconsistent with the lack of correlation between stress and group found in Table 2, and the latter was inconsistent with the higher social adjustment scores for the persisting group compared to the not-graduated group in Table 1. Chi-square tests of independence were conducted to examine: a) three levels of stress (low, medium, and high) by persistence (graduated or enrolled versus not graduated) and b) three levels of social adjustment by persistence. Results indicated that persistence was not dependent upon level of stress, chi^sup 2^(2) = 1.78, p > .06. The differences for stress scores were small and nonsignificant (means for those who persisted: 13.37, 25.71, and 37.24 and those who did not persist: 13.13, 25.61, and 38.85). However, persistence was dependent upon level of social adjustment, chi^sup 2^(2) = 11.12, p
The purpose of the current study was to extend a model used for the transition to university (Wintre & Yaffe, 2000) to predict students' persistence to graduation. The finding that 32.8% had not graduated is consistent with previous U.S. and Canadian research (e.g., Gilbert, 1991; Pantages & Creedon, 1978; Smith, 1991). A breakdown of the attrition rate, following telephone interviews with students "leavers," is provided elsewhere (Wintre et al, 2006).
Gender and Demographic Variables
Females were more likely to persist to graduation than males (71.0% vs. 57.7%). The importance of gender in predicting persistence is harmonious with the gender effects that Wintre and Yaffe (2000) found in predicting adjustment; however, gender did not interact with other predictors. This gender difference in graduation is supportive of some previous research (Astin, 1972; Demos, 1968; McCausland & Stewart, 1974; Nelson, 1966; Trent & Ruyle, 1965), whereas other research on gender and attrition rates has been inconsistent. Most studies found no significant differences in the overall attrition rates of males and females (e.g., Bragg, 1956; Johansson & Rossmann, 1973; Sewell & Shah, 1967; Slocum, 1956; Summerskill & Darling, 1955) and others report that females were more likely to drop out than males (Astin, 1964, 1972; Cope, 1971; Panos & Astin, 1968; Spady, 1970; Tinto, 1975). More recent research (e.g., Gerdes & Mallinckrodt, 1994) has not studied gender as a contributing factor to persistence.
No other demographic variables were significant. This contrasts with Wintre and Yaffe (2000) where same-sex parents' level of education was significant in predicting first-year GPA.
Parenting Style and Current Relations with Parents
Previous research on the effects of parenting style has revealed the benefits of authoritative parenting (e.g., Steinberg et al., 1992) prior to postsecondary school. Wintre and Yaffe (2000) extended the relationship between authoritative parenting styles to university adjustment. However, contrary to hypotheses, parenting styles did not predict persistence directly or indirectly. This finding indicates that the models of predictors of university persistence differ from the model for adjustment to university (cf. Wintre & Yaffe, 2000).
Of the current relationships with parents, perceived parental support was the positive predictor of persistence, partially supporting Hypothesis 4. This finding provides corroboration for the importance of parental support beyond adolescence into emerging adulthood (cf. Cutrona et al., 1994) and demonstrates the benefit of parental social support throughout the undergraduate years. Note, however, that in the transition to university model (Wintre & Yaffe, 2000), it was perceived reciprocity with parents, discussion with parents, and autonomy that were significant. Again, we see differences between the models for persistence to graduation and for the transition to university.
High School Average and First-Year GPA
As predicted, both high school average and first-year GPA were positive predictors of persistence, with first-year GPA not remaining significant when GPA was entered. These findings are consistent with the literature (Ryland et al., 1994; Wilczenski, 1993) and provide empirical evidence for the importance of academic integration in Tinto's (1975) theory of student departure.
The present negative relation of depression to persistence is consistent with and extends previous research. First, it is harmonious with Chickering's (1969) second vector of development: managing emotions. Second, depression not only contributes to adjustment to university (Wintre & Yaffe, 2000), but also predicts persistence to graduation, even though it is mediated when university adjustment is added to the model. Meilman et al. (1992) demonstrated that depression is a factor for many students withdrawing from university for health reasons.
The positive relationship between perceived stress and graduation was unexpected given that stress was a negative predictor of first- year adjustment (Wintre & Yaffe, 2000), and is negatively correlated to the adjustment scores (r = -.54) in the present study (see Table 2). Furthermore, in the zero-order correlations in the bottom row of Table 2, we find the correlation with persisting to be not significant (r = .03). Therefore that stress was significant and positive in the logistical regression appears to be due to trivial differences revealed in the means comparing persisters and nongraduates across the low, medium, and high stress levels.
Adjustment to University
This study was the first to use all four subscales of the adjustment to university scale (Baker & Siryk, 1984) as potential predictors of persistence to graduation. The present findings extend those of Gerdes and Malinckrodt (1994) who studied single items of academic, social, and personal-emotional adjustment as predictors of graduation. The current results indicate differential effects for the subscales. Academic adjustment was an indirect predictor in that it became non-significant when first-year GPA was added. As predicted, goal and institutional commitment was a positive predictor of persistence, but surprisingly personal-emotional adjustment was unrelated and social adjustment was a negative predictor. In the investigation of this latter finding, we uncovered a significant curvilinear relationship between social adjustment and persistence. Social adjustment was facilitative for the persisters as the lower and middle levels of adaptation, but not at the higher level. Future researchers will need to be sensitive to possible curvilinear relationships between predictor variables and persistence to graduation. The finding that academically adjusted students in first year were more likely to graduate is consistent with Gerdes and Malinckrodt (1994) and supports the widely held belief that initial adjustment to university contributes to degree completion. Of note, and perhaps more appropriately, it is the more objective measure of first-year GPA that is more important.
However, whereas Gerdes and Malinckrodt (1994) found that first- year social adjustment (using selected items from the scale) was a positive predictor of persistence, the present finding indicates that students who felt moderately socially adjusted were most likely to persevere to graduation. There appears to be a need for moderation in students' social adjustment. Although students low in social adjustment may not develop a peer support network, which helps one deal with stress or change (Stokes, 1985), students high in social adjustment may be spending too much time with friends (Scarbro, 2002), to the detriment of their academic work. This finding addresses the social integration component of Tinto's theory (Braxton et al., 2000).
The result that students with scores indicating goals and school commitment were more likely to graduate is consistent with Tinto's (1975) theory of student departure. Of particular importance is that the single item, "I expect to stay at _____ for a bachelor's degree," which differentiated those who had graduated from those who did not, t = -8.56, p
The current findings demonstrate that the adaptation subscales contributed differentially (two directly although one was a positive predictor and the other a negative predictor, one indirectly, and one not at all) to persistence, and adds to the limited empirical research examining adjustment to university as an independent measure (e.g., Gerdes & Mallinckrodt, 1994), rather than a dependent measure (see Lamothe et al., 1995: Oppenheimer, 1984; Pratt et al., 2000; Wintre & Yaffe, 2000). Furthermore when the adaptation scales were added to the model, the explained variance increased by 2.1% to 26.5%, gender and parental support were no longer significant predictors, and depression became an indirect predictor. These findings need to be replicated with a larger sample size, preferably from multiple universities (cf. Lerner, Almerigi, Theokas, & Lerner, 2005).
Limitations, Future Research, and Implications
We identified differences in the previous models of adjustment to university and first-year GPA (Wintre & Yaffe, 2000) with the present models of graduation. The reluctance to draw direct comparisons between the models is due to the basic differences in model construction. Major differences between the models include that the transition to university models: a) were sequential regression models employing continuous outcome variables, not logistic regression models with a categorical dichotomous variable; b) analyzed separately for males and females, rather than using gender as a predictor; and c) contained multiple measures of depression, stress, and self-esteem over time, as opposed to just the first-time September well-being measures.
In the present study, the predictors of persisting to graduation from university were gender, high school graduating average, parental support, depression, stress, first-year GPA, academic adaptation, social adaptation, and goal and institutional commitment. Notable in the current models, compared to the transition to university models, are the absence of parenting styles as either direct or indirect predictors, as well as self-esteem, identity, perceived reciprocity with parents, discussion with parents, or parents' levels of education. Instead, perceived parental support, social adjustment, and goal and institutional commitment are now significant, and first-year GPA has mediated the significance of high school average and academic adaptation.
While no study to date has empirically included as many variables in a model of student graduation from university, the model remains nonexhaustive, excluding potential predictors such as peer support, personality traits (cf. Wintre & Sugar, 2000), and other responsibilities such as employment. More important, future research should aspire to collect data from multiple, varying institutions to allow generalizability beyond a large, commuter institution (cf. Lerner et al., 2005). Also needed are investigations of institutional "process" factors such as class size, availability of the professors, the student-university environment fit, and the services or support provided to floundering students. And, of course, one needs to be mindful that regression models still refer to correlations, which do not imply cause and effect relationships.
The results of the current study generate several implications. There appear to be several risk (e.g., depression, experiencing too little or too much stress, and over involvement in one's social life) and protective (e.g., higher grades in high school and first- year, GPA parental support, good academic adaptation to university, and feeling attachment to the university) factors for university persistence. Given that the parenting role extends into emerging adulthood, we need to consider ways to encourage parental support of their university students, in addition to working with the students alone. Institutional programs seeking to increase student retention should focus on facilitating the adjustment to university, improving academic achievement, and promoting psychological well-being (cf. Pratt et al., 2000). Of importance to university administrators is the need to enhance goals and commitment to the university.
The current study provides empirical models for predicting students' persistence in university, both with and without university adaptation measures. Results indicate that females were more likely to graduate than males. Depression, while initially a negative predictor was mediated by first-year adaptation scales. Social adjustment was a curvilinear predictor of persistence to graduation. Parental support, goal and school commitment, and first- year GPA were positive direct predictors of graduation. Although stress was significant, further examination revealed that these differences were trivial. Finally, high school average and academic adjustment were positive indirect predictors.
Several aspects of the model enhance the literature surrounding student development. First is the contribution of perceived parental support into emerging adulthood as a significant predictor of persistence to graduation. Second, the variables that predict persistence to graduation are not simply an extension of those that predict adjustment to university. Third, the integration of first- year adaptation scales with persistence to graduation is new to the empirical literature. Fourth, the negative contribution of social adjustment identified that curvilinear relationships need to be considered in future research in the area. Finally, the study confirms the prevailing belief that first-year GPA is the most important factor in predicting graduation. First-year GPA mediates high school average, as well as academic adjustment to university.
Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55, 469-480.
Astin, A. W. (1964). Personal and environmental factors associated with college dropouts among high aptitude students. Journal of Educational Psychology, 55, 219-227.
Astin, A. W. (1972). College dropouts: A national profile. ACE Research Reports, 1, 1-71.
Baker, R. W., & Siryk, B. (1984). Measuring adjustment to college. Journal of Counseling Psychology, 31, 179-189.
Baker, R. W., & Siryk, B. (1989). Manual for Student Adaptation to College Questionnaire. Los Angeles, CA: Western Psychological Services.
Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology Monographs, 4, 1-103.
Beck, A. T, & Beamesderfer, A. (1974). Assessment of depression: The depression inventory. Psychological Measurements in Psychopharmacology, 7, 151-169.
Beck, A. T., Ward, C. H., & Mendelson, M. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561- 571. Bettencourt, B. A., Charlton, K., Eubanks, J., Kernahan, C., & Fuller, B. (1999). Development of collective self-esteem among students: Predicting adjustment to college. Basic and Applied Social Psychology, 21, 213-222.
Bragg, E. A. (1956). A study of student withdrawal at "W. U." Journal of Educational Psychology, 47, 199-202.
Braxton, J. M., Bray, N. J., & Berger, J. B. (2000). Faculty teaching skills and their influence on the college departure process. Journal of College Student Development, 41, 215-227.
Braxton, J. M., Sullivan, A.V. S., & Johnson, R. M. (1997). Appraising Tnito's theory of college student departure. In J. C. Smart (Ed.), Higher education: Handbook of theory and research, Vol. 12 (pp. 107-164). New York: Agathon.
Buri, J. R. (1989). Self-esteem and appraisals of parental behavior. Journal of Adolescent Research, 4, 33-49.
Buri, J. R. (1991). Parental authority questionnaire. Journal of Personality Assessment, 57, 110-119.
Chickering, A. (1969). Education and identity. San Francisco, CA: Jossey-Bass.
Chickering, A., & Reisser, L. (1993). Education and identity. (2nd Ed.). San Francisco, CA: Jossey-Bass.
Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385-396.
Cope, R. (1971). An investigation of entrance characteristics related to types of college dropouts. Washington, DC: Office of Education Reports (BR-0-1-068).
Cutrona, C. E. (1989). Ratings of social support by adolescents and adult informants: Degree of correspondence and prediction of depressive symptoms. Journal of Personality and Social Psychology, 57, 723-732.
Cutrona, C. E., Cole, V., Colangelo, N., Assouline, S. G., & Russell, D. W. (1994). Perceived parental social support and academic achievement: An attachment theory perspective. Journal of Personality and Social Psychology, 66, 369-378.
Demos, G. D. (1968). Analysis of college dropouts - some manifest and covert reasons. Personnel and Guidance Journal, 46, 681-684.
Fisher, S., & Hood, B. (1987). The stress of the transition to university: A longitudinal study of psychological disturbance, absent-mindedness, and vulnerability to homesickness. British Journal of Psychology, 78, 425-441.
Gerdes, H., & Mallinckrodt, B. (1994). Emotional, social, and academic adjustment of college students: A longitudinal study of retention. Journal of Counseling and Development, 72, 281-288.
Gilbert, S. (1991). Attrition in Canadian universities. Prepared for the Commission of Inquiry on Canadian University Education.
Greenberger, E., Josselson, R., Knerr, C., & Knerr, B. (1974). The measurement and structure of psychosocial maturity. Journal of Youth and Adolescence, 4, 127-143.
Hammen, C. L. (1980). Depression in college students: Beyond the Beck Depression Inventory. Journal of Consulting and Clinical Psychology, 48, 126-128.
Hickman, G. P., Bartholomae, S., & McKenry, P. C. (2000). Influence of parenting styles on the adjustment and academic achievement of traditional college freshman. Journal of College Student Development, 41, 41-54.
Johansson, C. B., & Rossman, J. E. (1973). Persistence at a liberal arts college: A replicated, five-year longitudinal study. Journal of Counseling Psychology, 20, 1-9.
Lamothe, O., Currie, E, Alisat, S., Sullivan, T., Pratt, M., Pancer, S. M., et al. (1995). Impact of a social support intervention on the transition to university. Canadian Journal of Community Mental Health, 14, 167-180.
Lerner, R. M., Almerigi, J. B., Theokas, C., & Lerner, J. V. (2005). Positive youth development: a view of the issues. Journal of Early Adolescence, 25, 10-16.
McCausland, D. F., & Stewart, N. E. (1974). Academic aptitude, study skills, and college GPA. Journal of Educational Research, 67, 354-357.
McDonald, R. T., & Gawkoski, R. S. (1979). Predictive value of SAT scores and high school achievement for success in a college honors program. Educational and Psychological Measurement, 39, 411- 414.
Meilman, P. W., Manley, C., Gaylor, M. S., & Turco, J. H. (1992). Medical withdrawals from college for mental health reasons and their relation to academic performance. College Health, 40, 217-223.
Neely, R. (1977). Discriminant analysis for prediction of college graduation. Educational and Psychological Measurement, 37, 965-970.
Nelson, A. G. (1966). College characteristics associated wit freshman attrition. Personnel and Guidance Journal, 44, 1046-1050.
O'Malley, K., Wheeler, I., Murphey J., O'Connell, J., & Waldo, M. (1990). Changes in levels of psychopathology being treated at college and university counseling centers. Journal of College Student Development, 31, 464-465.
Oppenheimer, B. (1984). Short-term, small group intervention for college freshmen. Journal of Counseling Psychology, 31, 45-53.
Pancer, S. M., Pratt, M., Hunsberger, B., & Alisat, S. (1995, March). Great expectations: Parent-child discussion and the impact of pre-university expectations on adolescent transition to university. Paper presented at the biennial meeting of the Society for Research in Child Development, Indianapolis, IN.
Panos, R., & Astin, A. (1968). Attrition among college students. American Educational Research Journal, 5, 57-72.
Pantages, T., & Creedon, C. (1978). Studies of college attrition: 1950-1975. Review of Educational Research, 48, 49-101.
Pratt, M. W, Hunsberger, B. E., Pancer, S. M., Alisat, S., Bowers, C., Mackey K., et al. (2000). Facilitating the transition to university: Evaluation of a social support intervention program. Journal of College Student Development, 41, 427-441.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.
Ryland, E. B., Riordan, R. J., & Brack, G. (1994). Selected characteristics of high-risk students and their enrollment persistence. Journal of College Student Development, 35, 54-58.
Scarbro, J. (2002). First-year students' adjustment to a university environment The role of peer support. Dissertation Master's thesis, York University, Toronto, ON.
Sewell, W., & Shah, V. (1967). Socioeconomic status, intelligence, and the attainment of higher education. Sociology of Education, 40, 1-23.
Silber, E., & Tippett, J. (1965). Self-esteem: Clinical assessment and measurement validation. Psychological Reports, 16, 1017-1071.
Slade, I., & Jarmul, L. (1975). Commuting college students: The neglected majority. College Board Review, 95, 16-21.
Slocum, W. L. (1956). Social factors involved in academic mortality. College and University, 32, 53-64.
Smith, S. L. (1991). Report of the Commission of Inquiry on Canadian University Education. Ottawa, ON: Association of Universities and Colleges of Canada.
Spady, W. (1970). Dropouts from higher education: An interdisciplinary review and synthesis. Interchange, 1, 64-85.
Steinberg, L. (1999). Adolescence. (5th Ed.). Boston, MA: McGraw- Hill.
Steinberg, L., Lamborn, S. D., Dornbusch, S. M., & Darling, N. (1992). Impact of parenting practices on adolescent achievement: Authoritative parenting, school involvement, and encouragement to succeed. CWW Development, 63, 1266-1281.
Stewart, S., Merrill, M., & Saluri, D. (1985). Students who commute. In L. Noel, R. Levitz, D. Saluri, & Associates (Eds.), Increasing student retention: Effective programs and practices for reducing the dropout rate (pp. 162-182). San Francisco, CA: Jossey- Bass.
Stokes, J. P. (1985). The relation of social network and individual difference variables to loneliness. Journal of Personality and Social Psychology, 48, 981-990.
Summerskill, J., & Darling, C. D. (1955). Sex differences in adjustment to college. Journal of Educational Psychology, 46, 355- 361.
Tinto, V. (1975). Dropout from higher education: A theoretical synthesis of recent research. Review of Educational Research, 45, 89- 125.
Tinto, V. (1993). Leaving college: Rethinking the causes and cures of student attrition. (2nd ed.). Chicago, IL: University of Chicago Press.
Trent, J., & Ruyle, J. (1965). Variations, flow and patterns of college attendance. College and University, 41, 51-76.
Weidman, C. J. (1989). Undergraduate socialization: A conceptual approach. Higher Education: Handbook of Theory and Research, 5, 289- 322.
Wilczenski, F. L. (1993). Comparison of academic perfor- manecs, graduation rates, and timing of drop out for LD and NONLD college students. College Student Journal, 27, 184-194.
Wintre, M. G., Bowers, C., Gordner, N., & Lange, L. (2006). Re- evaluating the university attrition statistic: A longitudinal follow- up study. Journal of Adolescent Research, 21, 111-132.
Wintre, M. G., & Sugar, L. A. (2000). Relationships with parents, personality, and the university transition. Journal of College Student Development, 41, 202-214.
Wintre, M. G., Sugar, L. A., Yaffe, M., &, Costin, D. (2001). Generational status: A Canadian response to the editors' consortium statement with regard to race/ethnicity. Canadian Psychology, 41, 244-256.
Wintre, M. G., & Yaffe, M. (2000). First-year students' adjustment to university life as a function of relationships with parents. Journal of Adolescent Research, 15, 9-37.
Wintre, M. G., Yaffe, M., & Crowley, J. (1995). Perception of parental reciprocity scale (POPRS): Development and validation with adolescents and young adults. Social Development, 4, 129-148.
Youniss, J. (1980). Parents and peers in social development. Chicago, IL: University of Chicago Press.
Youniss, J., & Smollar, J. (1985). Adolescent relations with mothers, fathers, and friends. Chicago, IL: University of Chicago Press.
Zitzow, D. (1984). The college adjustment rating scale. Journal of College Student Personnel, 25, 160-164.
Received April 10, 2006
Revised October 1, 2006
Accepted January 16, 2007
MAXINE GALLANDER WINTRE and COLLEEN DOROTHY BOWERS
Partial funding for this study was provided by a Faculty of Arts Research Grant from York University to the first author.
The authors are indebted to the anonymous reviewers for their astute and constructive criticisms that contributed to a greatly improved paper. We also wish to express our gratitude to Wendy Busby in the Registrar's office for her invaluable assistance. Correspondence concerning this article should be directed to the first author (E-mail: [email protected]).
Copyright Canadian Psychological Association Jul 2007
(c) 2007 Canadian Journal of Behavioural Science. Provided by ProQuest Information and Learning. All rights Reserved.