Click for next page ( 26


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 25
2 MODELS OF THE FACTORS THAT AFFECT STUDENT CHOICE AND TIME TO THE DOCTORATE: A LITERATURE SURVEY Over the past 30 years, an extensive literature has developed addressing 11D and the factors affecting student decisions to pursue postgraduate education. The literature has focused on five lines of inquiry: (1) persistence and attrition, or factors that cause students either to complete their education or to terminate it before a degree is received; (2) educational aspirations, or students' plans for pursuing additional education and training; (3) enrollment in college, which is similar in focus to the literature on aspirations but often uses different assumptions and statistical approaches to study the problem of student choice; (4) expected or perceived value of investing in education; and (5) TTD. This review is selective in nature, focusing mainly on findings that aid in an understanding of student choice. Literature on Persistence and Attrition The focus of much of the early research on attrition identified factors that caused students to quit school at the undergraduate and graduate levels (Berelson, 1960; Summerskill, 1962), not the processes that caused individuals to drop out or the quantitative impact of the factors involved. Descriptive approaches of this type can still be found in the current literature (e.g., Teague- Rice, 1981; Dolph, 1983), but more recent studies, beginning with the work of Spady (1970) and the model proposed by Tinto (1975), focus on causality. The model by Vincent Tinto (1975) is important because it explains how the interaction of many factors affects decisions to remain in school or to drop out. Longitudinal and theoretical in nature, the model assumes individuals enter institutions with specific attributes, background characteristics, prior experiences, and commitments that are integrated into their academic and social lives. The institution itself may have important effects on grade performance, intellectual development, peer group interaction, and faculty interaction with students. 25

OCR for page 25
In Tinto's model, grade performance and intellectual development contribute to academic integration and thus to goal commitment, and peer-group and faculty-student interactions contribute to social integration and to the student's commitment to the institution. And the interplay between the individual's commitment to completing college and his/her commitment to the institution affects the decision whether to drop out and for how long. Other researchers have used the Tinto model as a basis for regression and path analyses, and their findings tend to support Tinto's theory. For example, Pascarella and Terenzini (1983) found that social and academic integration have about equal effects on persistence and that students who are better integrated into an institution are more likely to complete the undergraduate degree than those who are not. Several other causal models are discussed in Bean (19801. For the most part, the factors identified by Tinto as having an impact on students are the same across studies, although some researchers differ as to which factors have direct and which have indirect effects. For example, Smart and Pascarella (1986) argue that schooling plays a direct role in determining social mobility. Differential levels of educational attainment yield different levels of achievement among persons with equivalent social backgrounds. Education also indirectly affects social mobility by serving as a "mediator" through which individual resources such as ability and background are converted into earnings and occupational status. Two aspects of the Tinto model warrant further comment. First, the role assigned to quality of college is ambiguous in the theory. Some researchers have found that better colleges produce a "higher yield" of graduates from the entering class (Knapp and Goodrich, 1952; Knapp and Greenbaum, 1953) while others have suggested the opposite (Davis, 1966~. Many studies have looked at the role of college characteristics and college environment in affecting persistence and educational aspirations (see, for instance, Pascarella, Terenzini, and Hibel, 1978~. Recent research shows that student interaction with faculty has a very small, albeit positive, effect on academic performance. In the Tinto model, faculty-student interaction affects persistence directly through its effect on social interaction and indirectly through its effect on grades. These two, in turn, affect 5 Specifically, Tinto questioned whether students at higher-quality schools have lower expectations. Davis had posited a "frog pond" effect, wherein the higher the average ability of the student body, the lower the grades of individuals of given ability, as compared to the grades they would have received at institutions populated by students of lower ability. Since grades affect expectations and expectations affect dropouts, a person of given ability level may be more likely to drop out at a higher-quality Han at a lower-quality institution. 26

OCR for page 25
academic integration and the decision to drop out (see Pascarella and Terenzini, 1979). Second, the Tinto model relegates changes in economic factors, such as unemployment and expected future earnings, to a category called "external impacts upon dropout." Changes in these variables are assumed to affect persistence indirectly by operating on student commitment to finish school and to the educational institution itself but do not directly enter the model as observables. Interestingly, Tinto assumes that an individual goes through a benefit-cost calculation to determine if it is worthwhile to stay in college, but he ignores the role that opportunity-cost considerations might play. Some who have relied on the Tinto model have considered economic factors as of secondary importance, although most studies assign a role to financial aid (e.g., Ethington and Smart, 1986~. The relegation of economic factors to a secondly role makes it difficult to study the impact of economic factors other than availability of funds on the decision to drop out and also precludes researchers from using the Tinto model to explore the effects of market forces on student choice. Both descriptive and causal studies point to parents' education, student grade-point average (GPA), race, and educational characteristics as affecting student persistence at the undergraduate level. These studies also tend to validate the importance of the interaction between students and faculty in keeping students in school. Other student variables associated with high attrition rates are upbringing in a rural area, father with less than a high school education, religion, and separation from one's spouse. Attrition at the doctoral level has been less carefully studied, is less well understood, and is most often expressed in descriptive rather than model form. For example, Tucker, Gottlieb, and Pease (1964) present data based primarily on student responses to questionnaires, indicating that the largest single reason for dropouts is student finances. Students without money to meet expenses or not having a teaching assistantship, research assistantship, or other financial aid were more likely to drop out than those with adequate financial support. Teague-Rice's (1981) study of female doctorates at Auburn from September 1971 to 1977 and Dolph's (1983) study of Georgia State students from 1970 to 1980 confirm the importance of scholarship, assistantship, or fellowship support. Students who are full-time, have a positive relationship with their dissertation chairperson, and score high on comprehensive exams also tended to remain on the doctoral track, according to these studies. A recent causal analysis by Girves and Wemmerus (1988) used the Tinto model to explore "degree progress" at the graduate level. For doctoral-level students, academic involvement appeared to have a direct impact on degree progress, while for master's-level students, such involvement appeared not to be important. Moreover, social integration did not seem to plan an important role in students' persistence, suggesting Tinto's conceptualization may not be entirely 27

OCR for page 25
valid at the graduate level. Grades were important determinants of persistence for master' e-level students, while the effect of grades on degree progress disappeared at the doctoral level.6 Girves and Wemmerus argue that involvement in the academic program, the role of the advisor, the number of faculty members a student gets to know, the faculty/student relationship, and the type of financial support are all important in affecting degree progress, but the effect of some of these variables is indirect. Differences among fields, identified by Biglan (1973), also are important in influencing a student's decision to complete a doctoral program, Girves and Wemmerus say. Literature on Educational Aspirations There is a substantial body of research that attempts to identify the reasons students decide to attend graduate school (e.g., Baird, 1976; Gropper and Fitzpatrick, 1959~. More recent studies are less interested in "why" than in the process by which key variables interact to shape student educational aspirations. One major line of inquiry looks at how the structural and environmental characteristics of colleges influence students to seek graduate training [see, for example, Astin and Panos (1968~. Pascarella's 1984 study, which used a causal model of educational aspirations based on Tinto's dropout model, finds that the direct effects of any single aspect of the college environment are "quite modest" and the best predictor of educational aspirations at the end of the second year of college is the level of educational aspirations at entrance to college. The only other factors directly affecting the decision to continue to a higher level of training are a student's cumulative GPA and a cumulative measure of college environment, according to Pascarella. Other studies use somewhat different causal models and include different variables but, nonetheless, reach similar conclusions. Alwin (1974), for example, found that a small amount of the variation in student aspirations can be attributed to differences in the college environment after student inputs are controlled, and Heyns (1974) found that verbal achievement and curriculum placement affect the relationship between student inputs and student aspirations. More recently, Ethington and Smart (1986) modified the Tinto model to test how the decision to enter graduate school is made. The model assumes the . ^^ . ,. . , . . . ~ 6 The variability in grades is probably small at the doctoral level. Thus, the finding of no effect does not necessarily imply that academic performance doesn't matter. 28

OCR for page 25
decision is the culmination of a series of choices made as students progress through the educational system, and it differs from those used in earlier studies by giving less influence to certain variables (for example, factors that affect decisions early in the choice process exert subsequent influences only indirectly). The Ethington and Smart model assumes that decisions regarding graduate education are based on "blocks" of independent variables that interact with each other. Student background characteristics and high school experiences comprise one block, which affects the choice of undergraduate institution. A second block measures student social and academic integration within the undergraduate institution, which in turn is influenced by the background "block" of variables. At a certain point, the effects of background characteristics wane as undergraduate experiences, financial aid, and receipt of the undergraduate degree replace them in importance. Enrollment in graduate school is dependent on all the measured variables, but results of the study indicate degree completion and receipt of financial aid have, by far, the greatest impact on graduate school enrollment. Student background characteristics have, at best, a marginal impact on the decision; the only student background variable showing a direct effect is the educational level of the student's family. For men, selectivity of the undergraduate institution has a strong positive effect on graduate school attendance while, for women, size of the undergraduate institution is important. Ethington and Smart found students with greater social and academic involvement in their undergraduate institutions are more likely to go to graduate school than those less involved. - Spaeth's (1968) study of factors that "allocate" college graduates to graduate and professional school, more empirical than theoretical, assumed that parental socioeconomic status (SES), students' intellectual ability, undergraduate academic performance, and the quality of the college from which they graduated influenced choice of graduate school. The Spaeth study looked at the career plans of 1961 college graduates, using a path-analytic model to relate quality of graduate school attended to student input and family background characteristics. Student undergraduate grades and the "intellectual caliber" of the undergraduate college attended were found to be major determinants of the quality of graduate school attended. Literature on Enrollments Some studies have equated enrollments with demand (Heath and Tuckman, 1986), although the former variable includes elements of supply while the latter does not. For example, Campbell and Siegal (1967) looked at demand for higher education using time-series data to estimate the ratio of undergraduate degree enrollments to the number of those eligible to enter undergraduate 29

OCR for page 25
institutions. Likewise, Carroll et al. (1977) analyzed the effects of Basic Educational Opportunity Grants on enrollment decisions, and Alexander and Frey (1984) attempted to identify the determinants of enrollment in MA programs. Most studies of these types focused on the direct effects of a set of independent variables and used regression analysis to identify what factors determine enrollment. But such studies largely ignore the interactive relationships captured by a path analysis, and many relegate sociological and psychological variables to a secondary role, although controlling for student characteristics such as race, age, and ability. Researchers usually place great importance on the role of tuition, family income, and financial aid in determining enrollment. Some also assume that external economic variables, such as unemployment, affect enrollments. A number of literature reviews have explored factors that determine demand for higher education (Becker, 1986; Jackson and Weathersby, 1975; Leslie and Brinkman, 1986~. Heath and Tuckman's (1989) review found that early studies that relied on a net tuition variable (gross tuition less financial aid) were flawed because net tuition fails to recognize that changes in tuition and financial aid have different effects on student demand. It also found that type of financial aid was important, with the evidence suggesting fellowships have a larger effect on demand than teaching assistantships. The review also revealed that the price elasticity of demand is lower at high-quality undergraduate institutions than at other 4-year schools and is less for graduate education than for undergraduate education. Finally, the review showed that decreases in financial aid at the graduate and professional levels reduce matriculation, increase the dropout rate, and lengthen time to the doctorate. Heath and Tuckman developed a model of the determinants of the demand for higher education that breaks the group of potential graduate students into five subpopulations: recent college graduates; persons in the work force; homeworkers who might return to graduate school; those discharged from the armed forces interested in higher education; and non-residents of the United States who attend U.S. institutions of higher education. Demand for graduate training in any given field is based on family characteristics, individual abilities and interests, tuition and financial aid variables, the characteristics of the educational organizations, and economic and social variables. The model can be used to explain the demand for both graduate and undergraduate training and, by introducing time notation, can also be used to explain persistence. 30

OCR for page 25
Literature on Expected Returns Since the early 1960s, researchers have recognized and written about the economic returns from investment in schooling. Implicit is the assumption that fields with higher returns attract more students than fields with less lucrative returns. Some studies have shown that when salaries rise in a field (e.g., business), more students major in it (Berryman, 1982), while others have actually formulated tests designed to show a specific causal relation. For example, Koch (1972) computed internal rates of return by academic field and compared them to changes in enrollment in 17 major fields at Illinois State University. He found that a small group of students do indeed shift to fields where salaries are high. A more recent and complete study by Cebula and Lopes (1982) looked at enrollment data for 28 fields at Illinois State University from 1973 through 1976 and confirmed that future earnings are an important consideration in selecting a major. But changes in earnings differentials were more important than the absolute value of the earnings differential, and neither the outlook for a given field nor Graduate Records Examination (ORE) scores were statistically significant predictors of field choice. Freiden and Staaf (1973) introduced an opportunity-cost approach to the student-choice literature, albeit indirectly, arguing that students switch curriculum groups as they progress through college and acquire information about alternative educational opportunities. According to this approach, students prefer "bundles" of courses that fulfill specific degree requirements and tend to pursue curriculum groups in which they have a comparative advantage, as defined by their verbal and quantitative Scholastic Aptitude Test (SAT) scores. More rigorous modeling of the relationship between enrollments and earnings potential in an academic field can be found in the work of Freeman (1971), who argued that differences in relative earnings signal potential students to enter fields experiencing shortages. He formulated a set of equations based on interactions between changes in starting salaries, government research and development expenditures, and student enrollments. Freeman showed markets adjust to changes in demand gradually and the nature of this time lag varies among fields. In some fields, a cycle of periodic shortages and excesses develops, emulating the cobweb pattern found in agricultural employment. Freeman's model has been tested and modified in the last decade, and while the cobweb pattern is in dispute, most research supports the conclusion that expected earnings affect student decisions (e.g., Hansen et al., 1980~. Trusheim and Crouse (1981) examined the effect of relative earnings on student decisionmaking in a different way, focusing on the effects of college prestige and selectivity on income. They found that, for men, type of occupation depends heavily on having gone to college but not very much on the prestige or 31

OCR for page 25
selectivity of the college attended. College selectivity did have a statistically significant effect on income, however. In an interesting and provocative piece, Berger (1988) tested Freeman's assumption of student myopia by estimating conditional logit models that incorporate alternative predicted future earnings measures. Using data from the National Longitudinal Survey of Young Men for five broad fields of study, Berger used alternative earnings measures to see if students were more likely to choose a field of study based on its potential future earnings than on its sting salary. After controlling for background characteristics, he found the probability of choosing one field over another increases as the present value of its predicted earnings stream increases relative to that in other fields. Researchers agree that relative income is important, but they disagree about how it should be measured. Should starting salaries, mid-career salaries, or future earnings profiles be used as a proxy for expected future earnings? Should salaries be measured relative to a numeraire field (e.g., a common base) or in absolute terms? Should the salary average be for a field or an occupation? These and related questions are addressed in future chapters. Literature on TTD Literature on the factors determining AD is limited. Interest in AD emerged in the early 1960s, when demand for graduate education led to a temporary shortage of Ph.D.s. Early studies by Berelson (1960) and Carmichael (1961) used survey analysis and data provided by the National Research Council to explore what was happening to AD over time. Among Berelson's findings were that IT'D can be shortened if full-time support is provided to a large number of doctoral students. Shortening AD will allow more students to be educated, Berelson found, but it would do more to increase the quality of training than to increase the number of available places. He also found that the main cause of the rise in AD was time spent in nondoctorate-related pursuits, such as work as a teaching assistant or research assistant, or time spent in work-related pursuits. Berelson's work contains little-information on the background characteristics of students and how they have changed through time. Although it does not address the interactions between students and their environment in model form, it does suggest specific institutional policies that might shorten rrD. Early on, researchers realized that "the Ph.D. is an open-end degree [that] cannot be circumscribed by an exact preordained time limit" (Prior, 1962~. Prior's work, like that of Berelson and Carmichael, provides useful information on institutional policies, but it does not explain changes in 1TD nor does it show the quantitative effects of the various factors causing increased I l D. 32

OCR for page 25
A study by Wilson (1965), based on a questionnaire sent to graduates, graduate deans, and departmental representatives in a representative group of fields at 23 doctoral institutions, is more useful, since it identifies the factors that affect TTD. Graduate deans, graduate faculty, and doctorate recipients all felt discontinuity of attendance, work as a teaching assistant, and writing the dissertation off-campus contributed to increased TTD. Similarly, financial problems, inadequate preparation in a foreign language, lack of coordination between beginning and advanced stages of graduate work, family obligations, inadequate undergraduate preparation in the major, and transfers among graduate institutions were named by all three groups as factors leading to lengthened I-ID, the study found. Continuity of study and adequate time to devote to study were seen as key to rapid completion of the doctorate. Clarity of institutional and departmental expectations regarding doctoral requirements were cited by deans as critical. Respondents to the Wilson questionnaire made two recommendations of special note: (1) students need to be insured adequate amounts and appropriate forms of financial support so they minimize their reliance on nondoctorate- related employment and (2) expectations of the skills and competencies that doctoral candidates have should be better articulated. While the Wilson study is thorough and thought-provoking, it does not provide insight into the role of student input variables in TTD, nor does it provide a quantitative estimate of institutional impacts. Abedi and Benkin (1987) attempted to fill this gap by studying over 4,000 students who received doctoral degrees from UCLA between 1976 and 1985. The Abedi-Benkin study postulated two regression equations with mean TTD as one dependent variable and mean RTD as the other. Three key sets of independent variables- demographic, financial, and academic-were included in the analysis. Using stepwise regression to find the statistically significant variables, the authors found that source of support was the most important predictor of TTD (using the F-ratio as the criterion for importance), while "postdoctoral plans" was the second most important. Average AD was lower for those in the postdoctoral study/trainee category than for those who planned to enter the labor force after receiving their degree, suggesting many who plan to enter the labor force post-degree are already employed, perhaps slowing their progress toward the doctorate. Other significant variables were the number of dependents, sex, and field of study. Summary A great deal of research has been conducted on what determines student decisions regarding higher education. There is more literature assessing decisionmaking at the undergraduate rather than the graduate level. Similarly, 33

OCR for page 25
more studies have assessed decisionmaking in the fields of education and sociology than in economics. Studies have moved in the direction of causal modeling and away from pure empirical analysis. The literature review suggests that many recent studies performed by non-economists have relegated economic factors to a secondary role and the significance of market forces in student decisionmaking has been neglected. There is a dearth of studies about time to the doctorate partly because researchers seem to have lost interest in the question when the shortages of the early 1960s turned to surpluses in the 1970s, and partly because researchers seem unaware of the trend toward increasing DID. Most studies of aspirations, dropouts, enrollments, and expected returns were noncausal and largely descriptive in the early 1960s, giving way in the 1970s to more formal modeling (path analyses or deterministic demand models). The Tinto model provided the basis for much subsequent educational and sociological research, but it failed to integrate the economic variables considered important in studies of enrollment and expected returns. And most studies in these latter two areas have tended to ignore demographic and sociological variables, while others have not paid adequate attention to institutional environment. Overall, findings from several avenues of inquiry have not been integrated into comprehensive theory of what determines time to the doctorate and, as a result, studies of AD have been largely noncausal and empirical. Despite this, several variables appear to affect student choice consistently: . . . . . Financial aid (this raises the question of whether the variable is also important in determining '1-11)); Main source of support (the literature provides little insight into the quantitative importance of this variable in determining HID); Immediate, rather than past, background characteristics (for example, current grades are more likely than past ones to affect current decisions. Many socioeconomic factors that affect the decision to enter college- for example, parent's education and income-are unlikely to have a major effect on 1-11) at the doctoral level. Work is needed on personal factors that have an immediate effect on 1 11)~; Quality of the undergraduate and graduate college (at present, little is known on the quantitative effects of organizational environment on l Liz); and Differences in expected earnings and changes in market conditions (to date, such variables have not been added into models of ITD). These insights are the basis of the theory and model discussed in the following chapters. 34