22
Dropouts Ii'om Higher Education: An Interdisciplinary Review and Synthesis 1 William G. Spady The dropout process from higher education is examined from a variety of operational definitions and ~ntel- lectual perspectives. A methodological analysis, critique, and synthesis of the empirical literature suggest that a more rigorous interdisciplinary approach must be attempted. A model based on Durkheim's theory of suicide provides a fruitful vehicle for sum- marizing a large proportion of current research, and focusing future attention on the interaction between student attributes (i.e., dispositions, interests, attitudes, and skills) and the influences, expectations, and demands imposed by various sources in the university environment. Both the academic and social systems of the university are regarded as important frameworks from which the dropout process must be examined. An empirical analysis operationalizing the variables in the model will be presented in the sequel to this paper. Le processus d'abandon au niveau des 6tudes suprrieures est examin6 h la lumirre d'une varirt6 de drfinitions oprrationnelles et de perspectives intellectu- elles. Une analyse mrthodologique, une critique, et une synth~se de la littrrature exprrimentale sur le sujet suggrrent qu'une approche interdisciplinaire plus rigoureuse devrait ~tre appliqure. Un modrle bass sur la throrie du suicide de Durkheim est proposr, servant avantageusement a rrsumer une grande partie des recherches rrcentes et permettant, dans l'avenir, d'attirer l'attention sur l'interaction entre les attributs de l'rtudiant (i.e., dispositions, intrr~ts, attitudes, et habiletrs) et les influences, les attentes, et les demandes imposres par diverses sources dans le milieu universi- taire. Les systrmes acadrmique et social de l'universit6 constituent d'importantes structures darts lesquelles le processus d'abandon doit &re 6tudi6. Suit une analyse empirique visant h rendre op6rationelles les variables du modrle suggrrr. Since at least six reviews of the literature on college dropouts have been published within the past decade (Knoell, 1960, 1966; Marsh, 1966; Sexton, 1965; Summerskill, 1962; Waller, 1964), yet another treat- ment of this extensively researched topic would appear to be superfluous. 2 However, beyond a few comfort- able and familiar generalizations about the relation- ship between attrition and family background, ability, or academic performance, this literature lacks both theoretical and empirical coherence. Although each of these reviewers laments the lack of conceptual clarity, methodological rigor, complexity of design, breadth, and analytic sophistication that characterizes most existing work, their recommendations regarding new and more thorough research approaches also lack a definite theoretical basis. Specifically, however, we feel it necessary to emphasize their belief that future work must be more eclectic in its intellectual approach and focus more explicitly on the relationship between the attributes of students and those of the institutional environment as they pertain to both its academic and social subsystems? The task before us, then, is to move beyond a mere summary of available studies of "college success" toward a more interdisciplinary-based, theoretical synthesis of the most methodologically satisfactory find- ings and conceptually fruitful approaches to this prob- lem. Although no one theoretical model can hope to account for most (let alone all) of the variance in dropout rates either within or across institutions, we suggest how a variety of currently distinct approaches may be combined within the framework of a single design in order to treat several clusters of relevant variables simultaneously. In the sequel to the present paper, we shall aim for further empirical clarification of these relationships by applying the conceptual ap- proaches suggested here to an existing body of longi- tudinal data. Before we attempt to deal explicitly with the vast literature on college dropouts, however, it is necessary to acknowledge its inseparable relationship with the equally prodigious and troublesome body of empirical work on academic performance. This acknowledge- ment does not imply, of course, that grade performance is either a necessary or sufficient condition for college a The compilation and synthesis of materials for this paper would have been impossible without the able assistance of Mrs. Joan Duchastel and Miss Patricia Erickson. Valuable suggestions were also provided by Miss Rondo Wood and Miss Martha Schmidt. 2 The reader should note that an extensive search for recent empiri- cal studies on attrition from higher education in Canada failed to yield any material. As a result, this discussion relates essentially to four-year undergraduate colleges and universities in the United States. Although data from junior college studies are mentioned from time to time, the primary assumptions and focus of this paper are based on institutions that offer the bachelor's degree. a This point is also stressed in three reviews of the literature on college admissions criteria and academic performance by Lavin (1965), O'Shea (1969), and Stein (1963).

Spady 1977

  • Upload
    nrgag

  • View
    251

  • Download
    1

Embed Size (px)

DESCRIPTION

desercion

Citation preview

  • Dropouts Ii'om Higher Education: An Interdisciplinary Review and Synthesis 1 William G. Spady

    The dropout process from higher education is examined from a variety of operational definitions and ~ntel- lectual perspectives. A methodological analysis, critique, and synthesis of the empirical literature suggest that a more rigorous interdisciplinary approach must be attempted. A model based on Durkheim's theory of suicide provides a fruitful vehicle for sum- marizing a large proportion of current research, and focusing future attention on the interaction between student attributes (i.e., dispositions, interests, attitudes, and skills) and the influences, expectations, and demands imposed by various sources in the university environment. Both the academic and social systems of the university are regarded as important frameworks from which the dropout process must be examined. An empirical analysis operationalizing the variables in the model will be presented in the sequel to this paper.

    Le processus d'abandon au niveau des 6tudes suprrieures est examin6 h la lumirre d'une varirt6 de drfinitions oprrationnelles et de perspectives intellectu- elles. Une analyse mrthodologique, une critique, et une synth~se de la littrrature exprrimentale sur le sujet suggrrent qu'une approche interdisciplinaire plus rigoureuse devrait ~tre appliqure. Un modrle bass sur la throrie du suicide de Durkheim est proposr, servant avantageusement a rrsumer une grande partie des recherches rrcentes et permettant, dans l'avenir, d'attirer l'attention sur l'interaction entre les attributs de l'rtudiant (i.e., dispositions, intrr~ts, attitudes, et habiletrs) et les influences, les attentes, et les demandes imposres par diverses sources dans le milieu universi- taire. Les systrmes acadrmique et social de l'universit6 constituent d'importantes structures darts lesquelles le processus d'abandon doit &re 6tudi6. Suit une analyse empirique visant h rendre op6rationelles les variables du modrle suggrrr.

    Since at least six reviews of the literature on college dropouts have been published within the past decade (Knoell, 1960, 1966; Marsh, 1966; Sexton, 1965; Summerskill, 1962; Waller, 1964), yet another treat- ment of this extensively researched topic would appear to be superfluous. 2 However, beyond a few comfort- able and familiar generalizations about the relation- ship between attrition and family background, ability, or academic performance, this literature lacks both

    theoretical and empirical coherence. Although each of these reviewers laments the lack of conceptual clarity, methodological rigor, complexity of design, breadth, and analytic sophistication that characterizes most existing work, their recommendations regarding new and more thorough research approaches also lack a definite theoretical basis. Specifically, however, we feel it necessary to emphasize their belief that future work must be more eclectic in its intellectual approach and focus more explicitly on the relationship between the attributes of students and those of the institutional environment as they pertain to both its academic and social subsystems?

    The task before us, then, is to move beyond a mere summary of available studies of "college success" toward a more interdisciplinary-based, theoretical synthesis of the most methodologically satisfactory find- ings and conceptually fruitful approaches to this prob- lem. Although no one theoretical model can hope to account for most (let alone all) of the variance in dropout rates either within or across institutions, we suggest how a variety of currently distinct approaches may be combined within the framework of a single design in order to treat several clusters of relevant variables simultaneously. In the sequel to the present paper, we shall aim for further empirical clarification of these relationships by applying the conceptual ap- proaches suggested here to an existing body of longi- tudinal data.

    Before we attempt to deal explicitly with the vast literature on college dropouts, however, it is necessary to acknowledge its inseparable relationship with the equally prodigious and troublesome body of empirical work on academic performance. This acknowledge- ment does not imply, of course, that grade performance is either a necessary or sufficient condition for college

    a The compilation and synthesis of materials for this paper would have been impossible without the able assistance of Mrs. Joan Duchastel and Miss Patricia Erickson. Valuable suggestions were also provided by Miss Rondo Wood and Miss Martha Schmidt.

    2 The reader should note that an extensive search for recent empiri- cal studies on attrition from higher education in Canada failed to yield any material. As a result, this discussion relates essentially to four-year undergraduate colleges and universities in the United States. Although data from junior college studies are mentioned from time to time, the primary assumptions and focus of this paper are based on institutions that offer the bachelor's degree.

    a This point is also stressed in three reviews of the literature on college admissions criteria and academic performance by Lavin (1965), O'Shea (1969), and Stein (1963).

  • Dropouts from Higher Education: 65 An Interdisciplinary Review and Synthesis

    graduation. Rather, it merely helps to bring a well- known fact into perspective: a large proportion of low performers become dropouts as the result of institu- tional grading standards and policies. Hence, students whose grade-point average falls below a specified level may be either advised to leave or dismissed from col- lege; but there is assurance neither that such marginal performers will all withdraw, nor that high performers will remain. 4 Even though this relationship is statisti- cally imperfect, many of the factors that distinguish low from high performers will also differentiate between dropouts and those who remain. The question that most researchers fail to ask, however, is whether the vari- ables that account for large differences in performance also influence attrition with performance levels held constant. In other words, does a variable such as verbal aptitude influence persistence in college only because of its well-documented relationship with academic achievement, or do dropouts have lower verbal aptitude regardless of the grades they receive? Unfortunately the majority of studies cited here do not provide suffi- cient data to document this crucial distinction. Instead they tend merely to explore a number of zero-order "correlates" (independent variables) without either examining the relationships among them or testing for spuriousness.

    In her first review, Knoell (1960) separated this type of research into four categories: census studies, which attempt to document the magnitude of attrition, transfer, and retention rates either within or across in- stitutions; autopsy studies, which seek self-reported reasons for students leaving school; case studies, which generally involve long-term follow-ups of students in- itially regarded as potential risks at the time of admis-

    4 In a national sample of students who attended college during the early fifties, Iffert (1958) fround that 61% left their first school. However, if only students who finished in the top fifth of their high school class had been admitted to college in the first place, the dropout rate would have decreased only to 44%. Albers (1965) found, for example, that after two years, 43% of the 6400 students at six Colorado state-supported colleges failed to return to their original school. Of these nearly 3000 "dropouts," 44% had earned at least a C average or better while in college. According to Summerskill (1962), academic failure in the initial year is usually disastrous, yet less than half of those who leave actually "flunk out." Dressel (1943) showed, however, that not all poor performers leave either. Of 181 Michigan State students advised by a special committee to withdraw because of poor grades, 84% ignored the advice. Over a period of time 144 of these students received a total of 714 warnings or notices from the committee concerning their performance. In addition, less than a third would accept a reduced course load as recommended, and only 21 eventually graduated.

    sion; and prediction studies, which utilize a range of admissions variables to generate prediction equations for a variety of college "success" measures. Marsh (1966) described the literature in different terms: philosophical and theoretical studies, which usually include recommendations for action on the assumption that dropouts should be prevented; descriptive studies, which describe the characteristics of dropouts, how they live as students, and the reasons they give for leav- ing; and predictive studies (similar to the fourth type described by Knoell).

    The absence of a category that might be called analytical-explanatory (with a focus on isolating un- derlying explanatory mechanisms) is, of course, un- fortunate and glaring. But before we can turn our attention to the development of such an approach, we must deal with the important and interrelated issues of definition and measurement of attrition and summarize the major findings currently available.

    Definition and Measurement o[ College Attrition: Their Methodological Symptoms Two operational definitions of the college dropout are generally accepted, but the assumptions and research approaches appropriate to each are definitely distinct: Definition 1 includes anyone leaving a college at which he is registered, and Definition 2 refers only to those who never receive a degree from any college. The former and most traditional of these two definitions is appropriately geared primarily to the concerns and policies of specific institutions, although scholars with a system-wide orientation find it narrow and mislead- ing. Essentially this definition regards any student who leaves his institution of initial matriculation as a drop- out. It is certainly an important criterion for admissions officers, institutional planners, guidance and counsel- ing personnel, social scientists, and others concerned with student morale, institutional commitment, and with the prediction, explanation, or prevention of stu- dent turnover. Because college registration information is updated on a regular basis and usually available on relatively short notice, Definition 1 is methodologically much easier to handle and more reliable than Definition 2. It is also the operational definition that we emphasize in developing our theoretical model later in the paper. However, since research on social stratification, educa- tional mobility, and human resources requires informa- tion on entire educational careers, this definition is inadequate and inappropriate for those purposes.

  • 66

    For example, Iffert (1958) drew a national sample of 12,667 students from 149 institutions in an attempt to document the magnitude of attrition from higher education in the United States and to specify the back- ground, residential, and academic factors that differ- entiated dropouts from persisters. His criterion was Definition 1 modified to include all students who failed to receive a bachelor's degree within eight consecutive semesters of first matriculation. Although only 40% of his sample managed to graduate from their original college within four years, he estimated that 60% do eventually receive degrees somewhere.

    Despite its several virtues, the limitations of this modified operational definition are fairly obvious. First of all, it fails to account for students who transfer to other schools but complete a degree program "on time"; and second, it overlooks students who may even- tuaUy receive a degree from either a transfer college or their home institution after an absence or delay of one semester or more. Although KnoelI (1960, 1966) and Ecldand (1964a) have been among the most out- spoken critics of this purely institutional approach, work by Albers (1965), Bayer (1968), Bemis (1962), Du Bois (1965), Faunce (1966), Fox (1967), Panes and Astin (1968), Pervin (1966), Sewell and Shah (1967, 1968), and Trent and Medsker (1968) also takes at least one of these two factors explicitly into account. Essentially, then, there must be logical con- sistency between the use of a given operational defini- tion and the inferences and conclusions that can be drawn from its use. In addition, Bemis (1962), Ex- tence (1965), Kamens (1968), Kubie (1966), Little (1959), McCammon (1965), Pervin (1966), Robin- son (1967), Rose (1965), Rose and Elton (1966), and Vaughan (1968) all argue that distinctions must also be made between students who are forcibly dismissed from college for academic or disciplinary reasons and those who voluntarily withdraw, s Hence, predic- tion equations for college grades can be applied directly to retention only as long as the issue at hand is the likelihood of failure and dismissal. Even though Bloom

    5 When these distinctions are made, the utility of what KnoeU and Marsh term prediction studies is, of course, immediately subject to question. We have already noted, for example, that a sizable proportion of students who leave their original college have both high admi~iOll8 credentials and adequate college grades. Whether they leave in order to upgrade their education at a better institu- tion (DonneUy, 1966; Kamens, 1968) or to escape the artificiality and constraints of college llfe for the freedom and responsibilities of the "real world" (Kubie, 1966) is not altogether clear.

    and Peters (1961) point out that the prediction of grade performance can be exceptionally accurate when the necessary data are available, the analysis of with- drawal rather than dismissal certainly requires a more complex set of variables than those generally used for strict prediction purposes.

    Given the diversity and range of institutions of higher education in the United States (see Jencks and Riesman, 1968; Riesman, 1958), mobility between institutions and availability of alternative colleges cer- tainly play a role in the attrition process. Not only can the failing student at most colleges do remedial work or make up necessary credits at some less selective school, but outstanding performers also may leave their current schools for the presumably greener pastures of more highly selective institutions. ~

    This mobility, argues Eckland (1964a), plus leaves of absence are major sources of error in dropout studies that use Definition 1 to make inferences and draw conclusions appropriate only to Definition 2. Using a sample of University of Illinois freshmen that entered in 1952, he shows that only 29% remained in school and received a degree within four years. By Iffert's criteria, the remaining 71% would be regarded as dropouts. In reality, however, over a period of 10 years 24% of this sample had returned to Illinois and com- pleted a degree, and an additional 5% were still regis- tered and likely to graduate. Sixteen percent graduated from some other institution, and 11% were registered but not expected to finish. In short, Definition 1 yields a dropout rate of 71%, which is over twice as large as either estimate using Definition 2.

    Knoell (1960) also cites evidence from several long-term studies at other large state universities that generally supports Eckland's findings: apparently no more than 35% of the entering freshmen at large state

    e These factors are the main thesis of Kamens' (1968) investigation of the students at 99 colleges and universities: i.e., controlling for college performance, the dropout rates at smaller and less selective schools should be higher because such schools typically lack the "resources" necessary to retain their student bodies. A goodly proportion of these "dropouts" will apparently become transfers at "better" institutions.

    Although Nelson (1966) was unable to apply the controls necessary for duplicating Kamens' study, he found that a whole series of characteristics differentiated colleges with high freshman attrition rates from those with fewer losses. On the whole, the institutions who lose the most students are (in order of impor- tance): public, inexpensive, low on estheticism and artistic orienta- tions, less selective, predominately female, nonselective, noncompetitive, lacking in special areas of strength, and high on student-faculty ratios.

  • Dropouts from Higher Education: 67 An Interdisciplinary Review and Synthesis

    institutions manage to graduate within a four-year period, although the eventual completion rate is much higher. Findings by Trent and Medsker (1968) based on a quasi-national sample of 10,000 high school stu- dents graduating in 19597 suggest that these rates might have increased slightly in the intervening decade since Knoell's first review. Although only 27% of the students who entered four-year public colleges in the fall of 1959 managed to graduate in four years, the completion rate for entrants of public universities was 36%, with an additional 30% still registered and working on a degree. In addition, 47% of the students at private liberal arts colleges finished "on time," com- pared with 55% of those who entered private uni- versities, s

    The Panos and Astin (1968) data suggest instead that a considerable upswing in completion rates may have taken place during the post-Sputnik years. They show that within an American Council on Education national sample of 127,212 freshmen at 248 four-year colleges in 1961, 65% of the respondents to a follow- up questionnaire had already received at least four years of schooling by 1965. Since their criterion does not differentiate between graduates and others still in attendance, it is impossible to estimate with much ac- curacy the actual proportion of degree holders in this group. However, it may be over 50% since 56% re- mained in continual attendance at their original schools for four straight years. Nonetheless, this rate is at least 13% higher than the Trent and Medsker completion rates for students in four-year colleges during roughly the same period. However, it is 10% less than figures presented by Bayer (1968). This anomaly may be due in part to the fact that Bayer's data were obtained five rather than four years after initial college entrance. Based on responses from a follow-up study of 100,000 high school graduates in 1960 who participated in Pro- ject Talent, he concluded that 60.5% of all college entrants had received a degree, 17.9% had not finished but were still enrolled, and 21.6% had withdrawn without graduating.

    "t Their sample is drawn from 16 communities of different kinds across the country. The communities were not, however, selected on the basis of strict probability sampling procedures.

    8 Presumably the more selective the institution, the greater the chance its students have of completing a degree program within a four-year period.

    Sewell and Shah (1967), on the other hand, find that only 48% of the college entrants in their one-third sample of all 1957 Wisconsin high school graduates had actually received a bachelor's degree within a seven-year period. Yet Robinson (1967) found that about 50% of the University of Illinois freshman class of 1962 had received degrees after four consecutive years.

    Reconciling the differences in these studies is diffi- cult, however, even when time span and chronology are taken into account. For example, even if all of Trent and Medsker's 24% still working on degrees after four years finished during the fifth year after high school graduation, their five-year graduation total would be only 52%, i.e., over 8% less than Bayer's. In addition, about half of their 48% dropouts would have had to return to college and continue working on a degree during that fifth year in order to resemble Bayer's para- meters. When their parameters for students who entered four-year colleges are compared with Panos and Astin's figures, similar discrepancies result: only 37% of the Trent and Medsker entrants graduated within four years, although another 25% were still registered and working on a degree. A conservative estimate of the Panos and Astin four-year graduation rate is 50%.

    One explanation for these discrepancies is that the Trent and Medsker sample contains a less selective cohort than either the Project Talent or American Council on Education sample; therefore, because of their poorer college qualifications the former are less able to survive four years of college. Since population parameters presented by Trent and Medsker indicate that, if anything, their sample is biased toward the high end of the SES and intelligence distributions, this explanation does not seem adequate. However, the other logical alternative--selective response rates--is a definite possibility since data extrapolated from these as well as other longitudinal studies all suggest that either more successful, or higher status, or higher ability respondents are more likely to cooperate in follow-up inquiries than are individuals who in some way might be regarded as educational or social failures (see Kubie, 1966).9

    Essentially both Bayer's and Panos and Astin's parameters are estimated on the basis of returns re- ceived, even though their longitudinal respondents comprise only a fraction of the original samples. The latter's respondents are acknowledged to have had

  • 68

    either higher educational aspirations, better high school grade averages, or better educated fathers than their nonrespondents, while 55% of Bayer's returns came from students who had been to college even though less than 45 % of the nationwide cohort during that period had attended (see Cooper, 1960). Even Trent and Medsker found that not only were college students overrepresented in their longitudinal sample, but the proportion of respondents also increased consistently with the amount of college that they had earned. Their attendance figures are more likely to be accurate than those just reported, however, because of an elaborate series of follow-ups and checks against both high school and college records. The results of these follow- ups, rather than the estimates from questionnaire returns, were used in establishing their attendance parameters. Sewell and Shah's thorough follow-up pro- cedures may also be associated with lower reported completion rates for the same reason.

    A final suggestion that selective response rates may account for an inadvertent overestimation of success rates comes from Pervin (1966). In a longitudinal study of former Princeton students, he received follow- up responses from 77% of those who had graduated but from only 57% of those who had dropped out. Had he estimated his completion rates on the basis of these returns, he would have underestimated the drop- out rate by at least 10%.

    The conclusion to be drawn from this exercise in extrapolation and inference may not in the long run seem worth the effort expended, but several important points now seem clearer than before. Essentially, each of the various definitions of dropping out has both its benefits and costs. Definition 1 is methodologically con- venient, but it fails to provide a broad enough perspec- tive on the actual rates of retention and attrition pertinent to the system of higher education as a whole. Definition 2, however, requires the mobilization of immense data gathering and follow-up resources, which are further complicated by the elements of time, inter-

    0 In a longitudinal study by this author (Spady, 1970), for example, careful searches through school guidance records indicated that none of the 15% nonrespondents to a follow-up questionnaire (mailed to theft latest known address four years after high school graduation) had ever requested that his high school transcript be sent to a college. Under normal procedures, admission is impossible without a grade transcript. Hence, none of these nourespondents had ever gone to college, even though 61% of the entire sample had.

    institutional mobility, and the social stigma associated with the failure to persevere academically, t0 Elaborate safeguards must be taken before data obtained from partial samples can be used to estimate population parameters.It

    Background Variables: Parents, Potential, and Past Performance Although the socioeconomic and academic background of college students are known to influence their overall chances of graduating, L~ the independent influences of these factors on leaving a particular institution are less well documented. Nor are the results of the apparent democratization of American higher education really clear; i.e., will the increased admission of more students from non-middle-class backgrounds mitigate or rein-

    10 This stigma is precisely one of the main factors that make what Knoell calls "autopsy studies" methodologically suspect. Slocum (1956) questions their validity because students are apt to explain failure to remain in college by means of socially "acceptable" reasons or mere rationalizations, rather than revealing their true feelings. Knoell (1966) cites evidence from Lins and Pitt (1953) to show that students downplay their true academic difficulties and inflate their financial problems. She also questions the merits of using these techniques with students whose basis for leaving is simply academic failure. Panes and Astin (1968), Robinson (1967), Trent and Medsker (1968), and Yoshino (1958) all suggest that finances are at best a contributory rather than deter- mining factor in attrition, and Fox (1967) reports that a large proportion of dropouts from 21 liberal arts colleges cited lack of finances as a major reason for leaving, yet a considerable propor- tion hoped to transfer to higher prestige (and presumably more expensive) schools. In comparing reasons for dropping out derived from self-administered questionnaire responses with those obtained in lengthy interviews with counselors, Demos (1968) found that both men and women exaggerated financially based reasons and underplayed reasons dealing with the difficulty of academic work, their lack of motivation, and their indecision about a major field.

    i t One reliable although imperfect device for measuring system- wide attrition is the use of official census reports. While having the obvious disadvantage of large time lags between measurements, cen- sus reports do provide estimates of college attrition over several decades. Data extrapolated from a report made in 1962 (Spady, 1967a) do corufirm an inference made earlier, however; i.e., graduation rates for men at least are increasing slowly over time. Men aged 55 to 64 in 1962 would typically have attended college during the twenties. Fifty-nine percent of this age cohort who reached college actually finished, compared with only 50% of the men who attended during the thirties. The Depression years rep- resented the apparent high point in attrition rates over the last 40 years. Roughly 56% of the men who entered college in the forties managed to graduate, compared with nearly 59% for men who attended during the fifties. Since some of these men were as young as 25 when sampled, however, the final graduation rate for this youngest age group may eventually exceed 60%.

  • Dropouts from Higher Education: 69 An Interdisciplinary Review and Synthesis

    force the selection mechanisms that have operated in the past? xa It is possible, for example, that the flow of non-middle-class and lower aptitude students primarily into junior colleges may actually strengthen existing relationships because of the disproportionately low graduation rates among junior college entrants.

    Large-scale documentation for this trend is pre- sented by this author in an earlier study (Spady, 1967a). These census figures show that family origins (as meaured by father's education) have had an in- creasing influence on the completion rates of college entrants over time, even though graduation rates are generally improving for men from all social strata. The greatest exceptions to this rule are the sons of men who failed to complete grade school. Their chances of finishing college have remained close to 47% over the past 30 years. Those particularly likely to complete their studies during the same period have been the sons of college-educated men: 67 %.

    However, on the basis of their data, Trent and Medsker (1968) argue that family background prob- ably has its greatest influence on the decision to go to college rather than on the chance of finishing. Accord- ing to their interpretation, intelligence rather than socioeconomic status is the more important factor in college completion. 14 Although we were unable to con- trol for intelligence, our findings are not inconsistent with their interpretation: during the past 30 years, variations in father's education account for 45% dif- ferences in the likelihood of high school graduates actually reaching college, but only 20% differences in their chances of finishing.

    Perhaps the most direct confirmation of this inter-

    12 Skeptics who may not be convinced of this conclusion after reading the evidence presented in the next few pages may wish to consult other relatively recent work. In addition to the studies cited in the text, the relationship between various measures of family socioeconomic status and attrition has also been documented by Astin (1964), Eckland (I964b), Lembesis (1965), McCammon (1965), and Wolford (1964). The relationship between academic factors and attrition is documented further by Bayer (1968), Goble (1957), Iffert (1958), Kamens (1958), Lembesis (1965), Lins (1954), Little (1959), Pants and Astin (1968), and Spady (1967b). For a lengthy review of the literature on academic factors and college performance, see Lavin (1965).

    18 For an extended discussion of this and related questions, see Jencks and Riesman (1968).

    14 This fact remains true even though they (Iffert, 1958; Little, 1959; Sewell & Shah, 1967) all document the relatively high attrition rates among high-ability students.

    pretation can be found in the work of Sewell and Shah (1967, 1968). In their earlier paper, they show that family status has a major independent influence on graduation rates controlling for student IQ (about 18 % ). The influence of IQ with status controlled, how- ever, is over twice as large (about 40% ).1~ Men who fall in the lowest quartile on both variables have only a 4.4% chance of completing a degree program within seven years, compared to a 70.6% chance for their counterparts in the upper quartile of both variables.

    Somewhat similar inferences can also be drawn from the work of Gurin, Newcomb, and Cope (1968). Using data from a sample of nearly 4400 from the 1962 and 1963 arts and sciences freshmen at the Uni- versity of Michigan, they compared the characteristics of dropouts and students still registered in the fall of 1965. A large number of significant zero-order dif- ferences for each sex separately were found among the battery of demographic, attitude, aspiration, activity, and self-concept measures. From these findings, two indices were built: (1) Cosmopolitanism (containing parents' religious affiliation 16 and level of education, rural-urban background, and size of high school)17 and (2) Academic Preparation (containing high school class rank and Scholastic Aptitude Test scores). With Cosmopolitanism alone controlled, almost all of the relationships between the remaining independent vari- ables and attrition became nonsignificant or disappear- ed. Virtually the same process occurred controlling for Academic Preparation alone. When these two indices were controlled simultaneously, however, Cosmopoli- tanism had a slightly greater independent influence on attrition than did Academic Preparation, particularly among the men.

    The basic implication of their findings is somewhat contrary to what some researchers might expect to emerge from a set of variables such as this, for they suggest that when socioeconomic and attitudinal factors

    15These figures were derived using the standardized partialing technique known as Weighted Net Percentage Differences, develop- ed and first used by Davis (1964).

    16 For an analysis of the relationship between religious background and graduate school plans using national sample of college grad- uates, see Davis (1964).

    17 For a more extensive discussion of the relationship between high school size and college success, see Sexton (1965). Bayer's (1968) national sample, however, showed only slight differences between students from small and large high schools.

  • 70

    are considered simultaneously, the advantages thought to accrue to individuals with particular kinds of atti- tudes do not exist independently of their family back- ground. It is background experiences, then, that both lead to and account for the attitudinal differences often associated with attrition. These experiences are cer- tainly influenced by some of the basic elements that comprise what is called socioeconomic status, but work by several authors suggests that other important ele- ments may be involved as well. is

    Residential differentiation appears to be such a variable, since Gurin et al., and Iffert (1958) both show that students with rural or small-town backgrounds are less likely to remain at their first school. 19 Eckland (1964b) also found that students from large central cities were less likely than others to drop out at any point during their college career; but if they did leave, they were also less likely to return and complete a degree.

    A more important factor, however, appears to be the quality of relationships and central values expressed within the family. Congdon (1964), for example, shows that students who were succeeding enjoyed more casual, accepting, and open relationships with their parents, while the parents of failing students were dis- proportionately more demanding and overprotective. Findings by Jones (1955) and Weigand (1957) also suggest that poor achievers come from households characterized by greater tension, more disturbance, and more stringent parental discipline. In addition, Hood (1957) reports that dropouts were either less certain of parents' attitudes about college or more ambivalent regarding the wisdom of disagreeing openly with their parents on controversial matters. Graduates, according to Trent and Ruyle (1965), were more likely than dropouts to turn to their parents for advice, receive praise from them, and have parents with an interest in their college success.

    18 Eckland (1964b), Iffert (1958), and Trent and Medsker (1968) all note, for example, that dropouts come from less affluent families, but others such as Demos (1968), Panos and Astin (1968), Slocum (1956), and Yoshino (1958), clearly suggest that a clear lack of financial resources plays only a minor part in most students' decisions to withdraw.

    19 For a summary of the literature on the relationship between community of residence and college aspirations, see SeweU and Armer (1966). The influence of hometown on the graduate school aspirations of a national sample of college graduates is discussed at some length by Davis (1964).

    The pattern suggested by these findings, then, complements the research pertaining to specific socio- economic attributes of the family. Apparently better educated, more urbane, and aflfluent parents enjoy more open, democratic, supportive, and less conflicted re- lationships with their children. These same qualities, of course, may also be associated with both high apti- tude and outstanding high school performance since most researchers agree that the latter characteristics are more typical of students from higher status back- grounds as well.

    Since we have already acknowledged the dramatic impact of academic aptitude on college performance and attrition while controlling for socioeconomic status, there is little need to belabor the obvious. Two less conspicuous sets of findings deserve attention, how- ever. First, Hedley (1968) found that high school grades and scholastic aptitude did not differentiate be- tween dropouts and persisters at a small liberal arts college. This finding may be an important example of the distinction between students who are dismissed on academic grounds (those with low college potential) and those who withdraw voluntarily in order to en- hance their opportunities at a better school (these more likely to have high potential). The students who re- main, then, would be characterized by moderate but adequate potential, and the statistical relationship be- tween potential and attrition would be close to zero. -~~

    Second, the term "inadequate preparation" is rele- vant to this discussion because of its various implica- tions. One of its two most common meanings refers to the poor quality of instruction, facilities, curriculum, and grading standards at the student's secondary school; -~1 the other refers to his own lack of exposure and mastery of the fundamental skills that most college curricula assume rather than provide. Although the two are no doubt related, Jellison (1965), McLaughlin (1966), Summerskill (1962), and Wolford (1964) all suggest that dropouts have a tendency to attribute

    20 A related but actual reversal between background variables and attrition is discussed by Du Bois (1965). In his study of education majors at a relatively unselective state school, those whose fathers had high occupational status were more likely to leave (including transfer) than were students with blue collar fathers.

    21 Paradoxically, Sexton (1965) notes that, all things being equal (such as social class and ability), students from public secondary schools usually outperform private school graduates in college, even though private schools are generally regarded as offering superior college preparation.

  • Dropouts from Higher Education: 71 An Interdisciplinary Review and Synthesis

    their academic deficiencies to the quality of preparation provided by their secondary schools, rather than to weaknesses in their own ability or motivation to learn. This tendency for dropouts to find fault first with the system and only secondarily with themselves is a re- curring element in this discussion, z2

    The Sex-linked Role of Educational Goals and Interests Before pursuing the issue of personal motivation to any extent, however, we first consider the relationship that educational aims and interests have with attrition. In the process we take into account variables such as the level and intensity of student aspirations and intel- lectual interests, their clarity and realism, how and when they were formulated, their substantive differ- ences, their relationship to programs of study, the vari- ability in these patterns by sex, and their translation into actual study behavior.

    We begin by assuming that students who aspire to finish college will be more likely to graduate than those with more modest goals. Although this assumption is documented by several authors, three studies stand out in terms of their methodological adequacy and un- ambiguous results. Sewell and Shah (1967) show, for example, that with family socioeconomic status and student intelligence simultaneously controlled, educa- tional plans measured during the senior year of high school had by far the strongest independent influence on college graduation. Using similar controls, Trent and Medsker (1968) found that graduation rates were strongly influenced by the importance that students attached to being in college. In a study of National Merit Scholars, Astin (1964) found that dropouts were less likely than eventual graduates to have plans for graduate or professional school. In addition, Leathers (1966) shows that, as freshmen, high-ability students are fairly accurate in predicting their own persistence or attrition.

    Several other writers have suggested, however, that lofty goals facilitate graduation only when they are clear and realistic. Pervin (1966) shows, for example, that a large proportion of dropouts from Princeton in the forties left school in order to pursue clear voca-

    20 We have already noted, for example, that in giving reasons for leaving college, dropouts tend to avoid the stigma of being a failure. In the face of more objective criteria, however, many of these explanations seem inadequate.

    tional goals, but dropouts a generation later were more likely to be uncertain or ambivalent about their plans. Abel (1966), on the other hand, found that dropout rates were highest among students with clear vocational goals but poor academic records. Their lack of realism, he argued, made them particularly susceptible to the consequences of low performance. Knoell (1966) also argues that in some cases unrealistic goals may have a more negative influence on persistence than having no specific goals at all, particularly for those without strong intrinsic interests.

    In addition, findings by Slater (1957) and Weigand (1957) both suggest that dropouts are more likely to allow others to make important educational or voca- tional decisions for them, rather than making their own decisions based on personal interests and aptitudes. However, survival for students in a Western teachers' college, according to Whitmer (1963), was facilitated rather than impeded by having others make decisions about college for them. Gurin et al. (1968) also found that dropouts were more certain of their major field than others, but the relationship disappeared when cosmopolitanism was controlled. Unfortunately, the contradiction in these findings cannot be resolved on the basis of the data at hand. If we allow, however, that students who remain certain in their choice of field while doing only marginal academic work may be less flexible and open to change, then perhaps dropping out is characteristic of either excessively docile and de- pendent (Slater; and Weigand) or unbending and in- flexible (Abel; Gurin et al.; and Whitmer) students.

    The substance of one's educational and curricular interests and orientations is also related to attrition, but the findings available are not consistent. Sarnoff and Raphael (1955) found, for example, that failing students typically failed to see their education as a process involving intellectual growth and self-realiza- tion. Instead they were extremely utilitarian. Faunce (1966) also showed that students with strong practical and applied orientations were more likely to drop out, even with ability controlled. Her findings complement those of Gurin et al. (1968), who show that female dropouts have lower intellectual and aesthetic interests, lower introspective and theoretical orientations, and weaker intellectual reasons for choosing their college. Bayer (1968) too found that graduates had higher scores of interest, creativity, and abstract reasoning measures, but Iffert (1958) found that retention was highest among students in occupationally oriented

  • 72

    fields, possibly because their own vocationalism was reinforced and supported in these practical academic contexts, za Summerskitl (1962) noted that having high vocational goals usually facilitates persistence, but the reverse may occur in a highly intellectual climate. In support of the notion of congruity between personal orientations and curricular emphases, Trent and Meds- ker (1968) found that persisters were more likely to value education as a process of gaining knowledge and appreciating ideas, to be more introspective in their thinking, and to be in academic rather than applied fields. 24 Hood (1957) also found that persisters had a stronger orientation toward improving their ability, while dropouts were typically concerned with develop- ing their personalities.

    There is also a growing body of data that suggests that the nature and strength of college goals and orien- tations are differentially linked to certain outcomes, depending on the sex of the student. It is fairly clear, for example, despite the recent upsurge of feminist rhetoric, that men face the necessity of establishing a position in the occupational structure on which their future income and status will depend. For women, on the other hand, the decision to pursue a career is less often dictated by social or economic necessity. As a result, women are both freer to deal with college as an intrinsically rewarding experience and face less pres- sure to finish.

    Out of pure "necessity," then, we might expect a higher proportion of men to finish their degrees and a higher proportion of women to drop out, even though women may feel less constrained to attend college in the first place. 25 The corollary to this hypothesis is that men are less likely to be voluntary dropouts than are wom- en. The evidence available suggests that these hypo- theses are generally correct. The major anomaly is that women who do graduate are more likely to finish "on time." Data from Bayer's (1968) national example

    23 For a more detailed and well-documented discussion of the differences in educational and instructional goals across university departments and their potential consequences for students, see Gamson (1966, 1967) and Vreeland and Bidwell (1966).

    24 Smith (1967) found, however, that congruence between student interests and curricular choice was associated with satisfaction, but not with either achievement or attrition.

    25 Sewell and Shah (1968) show, for example, that as high school seniors 37% of the men in their sample planned on college, com- pared to 30% of the women.

    show that after five years 65% of the women have graduated, 25% are no longer in school, and 10% are still registered and working on a degree. The men have somewhat fewer graduates and dropouts (58% and 19% respectively) but considerably more who are still registered (23%). The Trent and Medsker (1968) data reflect similar patterns. After four years, 31% of the men were still working toward a degree compared to only 16% of the women (a difference significant at the .001 level), but 51% of the women were classified as dropouts compared to 46% of the men. After seven years the Sewell and Shah (1967) graduation rates were nearly identical; 50% of the men were finished compared with 47% of the women. They also show, however, that aspirations were more closely tied to actual attainments for the women than for the men. -~6 Apparently women who want to finish are more likely to do so than are men with similar aspirations.

    The more frequent thwarting of male aspirations is confirmed by several other studies as well. Lembesis (1965) shows, for example, that among the second, third, and fourth year dropouts at a Midwestern state university, a greater proportion of women left volun- tarily. Robinson (1967) shows that 68% of the male dropouts from a large Midwestern university were dis- missed compared with only 44% of the women. In addition, Gurin et al. (1968) show that female drop- outs have lower educational aspirations than their counterparts who remain in school, while the aspiration levels among the men are virtually the same.

    The major inference to be drawn from this entire set of findings would appear to be that survival in college is dependent largely on a clear and realistic set of goals and having interests that are compatible with the influences and expectations of departmental faculty and curricula. Men in particular, however, appear to maintain high expectations despite the academic reali- ties of college life. According to Jellison (1965) and Sarnoff and Raphael (1955), in fact, dropouts are typically unable to translate their goals into effective patterns of study. Although Malloy (1954) finds that female underachievers apply themselves only in cur- ricular areas of particular interest, Trent and Medsker (1968) show that time spent studying is more highly associated with persistence for men than for women.

    26 The beta weights in their path models are .55 and .67 for men and women respectively.

  • Dropouts f rom Higher Educat ion: 73 An Interdiscipl inary Review and Synthesis

    Personality Dispositions: In Search of the "Mature" Student Since we have been able to isolate and integrate only a few general trends in the recent empirical literature, much of our discussion contains reaffirmations of established findings and interpretations. If the relation- ship between grade performance and attrition comes as no surprise, neither will the ubiquitous conclusion that dropouts are less "mature" than persisters? r The diffi- culty, of course, is in isolating the various components of maturity and examining the literature for more specific confirmation of this general hypothesis.

    From our reading of this immense body of litera- ture, maturity is generally equated with the following attributes: motivation, independence, flexibility, in- volvement, impulse control, self-confidence, responsi- bility, and rationality. For our present purposes, each of these terms should be viewed in its broadest frame of reference, allowing for overlap between categories.

    In his review of the literature, Summerskill (1962) isolated motivation as a critical variable in the dropout process. The difficulty, he argued, was that it had rarely been operationalized well or partialed out in most analyses. Since then a number of other studies have re- affirmed his conclusions. -~s Heilbrun (1964), for ex- ample, showed that dropouts expressed a lower need for achievement ( as measured by a revised form of the Gough Adjective Check List). Daniel's (1963) South- ern freshman dropouts had rated themselves as less persevering and curious at the beginning of the year, and Stone (1965) found that both persistence and need for achievement were positively related to remaining in a Northern state college beyond the first year. Pervin (1966) also found that poor motivation was a con- tributing factor in attrition at Princeton, and Demos (1968) used intensive interviews to find that poor motivation in the face of difficult academic demands

    27 It is important, however, that maturity not be confused with ability or achievement, even though the latter are usually regarded as important attributes of the mature individual. Even Astin (1964) notes, for example, that dropouts in a national sample of highly talented National Merit Scholars were more immature than those who remained in college. Pervin (1966) too concludes that immaturity contributes to the attrition process among highly se- lected Princeton students.

    -~ her review of the literature, Sexton (1965) also regards motivation as the crucial explanatory variable. Trent and Medsker (1968) conclude that high interest and motivation are more important conditions for success than are high ability and finan- cial support.

    was a major underlying factor behind male attrition, despite the subjects' tendency to attribute the problem to other causes.

    Robinson (1967), however, combines poor study habits with other factors in suggesting that dropouts are more dependent and less able to manage things on their own than are persisters. In a study done a decade earlier, Grace (1957) drew the same conclusion. But the work of Rose and Elton (1966) suggests that the problem is not so simple. They differentiated a sample of freshmen into four groups on the basis of two criteria: how well the student performed academically during the freshman year, and whether or not he re- turned for his second year. "Defaulters" are those who left during the freshman year, and "dropouts" are those in good academic standing who left after the first year. Essentially students with low grade perfor- mance were lower on the Omnibus Personality Inven- tory Dependency Scale than those who performed well, regardless of whether or not they dropped out. In other words, those who "defaulted" by leaving early were more independent than those who dropped out at the end of the year, despite their better grade performance. The successful persister, then, according to Rose and Elton, would be characterized as more immature than the defaulter by virtue of his greater conformity and dependence. Brown (1960) too finds dropouts to be less conforming than persisters. 29

    Whether conformity, in turn, implies a less rigid stance toward life is not altogether clear, but in her earlier work Rose (1965) found that defaulters had less tolerance for conformity than other students. To the extent that this intolerance suggests a lack of adaptability, these findings, in fact, support Summer- skill's (1962) conclusion that dropouts are usually less adaptable.

    According to Jones (1955), inflexibility and an incapacity for variety are typical of students on aca- demic probation, and the work of Stern, Stein, and Bloom (1956) at the University of Chicago further bolsters the argument that "stereopaths" (students with more rigid and inflexible orientations) are more likely to become dropouts. More recent work by Gurin et al. (1968) also shows that dropouts are more au- thoritarian, intolerant, and inflexible.

    2a Although some of this apparent contradiction is resolved in the next section of the text, Lavin (1965) does note that an element of conformity to achievement demands is definitely a part of the academic success process.

  • 74

    Another aspect of the "maturity-persistence syn- drome" that appears in the literature involves the individual's capacity for decisiveness and involvement. This phenomenon is, of course, related to motivation, but it also involves elements of ambivalence, detach- ment, passivity, and submissiveness. Slater (1957), for example, argues that indifference toward aspects of the curriculum is associated with attrition. Brown, Abeles, and Iscoe (1954) also suggest that indecision and procrastination are characteristic of dropouts. Us- ing a more psychoanalytic approach, Blaine and Me- Arthur (1961) showed that dropouts reacted to Rorschach pictures without feeling and emotional in- volvement, rather than with either anxiety or unusual perception. Congdon's (1964) findings also support their results. He found that failing students were typi- cally reluctant to commit themselves to anything, and dropouts had strong dispositions toward avoidance. Patterns of overcompliance or overrebellion toward parents, he argued, impair the normal functioning of the student in task-related activities.

    Since noninvolvement and impulse control have a good deal in common, it is perhaps not surprising that Summerskill (1962) found dropouts at the other ex- treme of these continua, i.e., more rebellious and non- conformist. Heilbrun (1964) suggests, in fact, that passivity is not only part of the stereotypic female role, but is also linked to lower frustration in the classroom setting. On the whole, the dropouts in his study were more assertive and had a higher need for change, con- trolling for both sex and ability. In addition, Astin's (1964) finding that the dropouts among his National Merit Scholars were more assertive is paralleled by Faunce TM (1966) ; her high-ability dropouts also had more problems with impulse control. Stone (1965) too found self-control to be a significant factor in retention, while Gurin et al. (1968) showed that dropouts were definitely higher on impulse expression. Beyond this, Hood (1957) and Rose and Elton (1966) both found that dropouts clearly showed more hostility than per- sisters.

    The necessary empirical synthesis of the findings related to these two personality dispositions is pre- sented by Vaughan (1968). In an attempt to differenti- ate between dismissals and voluntary withdrawals, he found that the former both lacked deep emotional com- mitments and were more impulsive, overly active, and restless. In terms of these two factors, then, maturity implies moderation.

    To the clinical psychologist, however, these ex- tremes in behavior might appear to be the symptoms of a more fundamental problem related to a lack of self-esteem and self-confidence. Barker (1968), for example, found that among the men in a sample of West Virginia high school graduates, college dropouts who had been academically successful had lower self- concepts than persisters, particularly those with low grades. Among the women, dropouts also had lower self-concepts, irrespective of their grade performance. Gurin et al. (1968) found that several self-concept measures were associated with attrition among the men but not among the women. Early in the freshman year, subsequent dropouts showed more concern over their academic and life success, presumed adequacy in mari- tal, heterosexual, and parental roles, and were more preoccupied with identity seeking. In other words, these men had conspicuous doubts about their own capaci- ties to perform adequately in the very roles that symbo- lize masculine maturity.

    Alfert (1966) also uses the notion of self-concept to explain her findings that commuters and students living in boarding houses or private rooms are more likely to drop out. Because they have fewer inter- personal ties within the university milieu, they feel like outsiders and have difficulty clarifying their self-con- cepts. Since she does not control directly for friendship ties in her analysis, her interpretation does not rest on firm empirical support. 30

    She argues further, however, that dropout rates in cooperatives and fraternities are lower partly because the constituents in these living units develop a sense of mastery that they derive from exercising responsibility in the management and maintenance of their houses. Other research also suggests that persisters may be more responsible and mature (Brown, 1960; Grace, 1957; and Summerskill, 1962), but the concept of responsibility is as difficult to define operationally as maturity is.

    Perhaps the most problematic aspect of maturity, however, is the broad constellation of disposition sub- sumed here under the term rationality. Chief among its components, perhaps, are clarity of insight and critical thinking. Findings by Daniel (1963), Faunce (1966), and Rose and Elton (1966) all suggest that dropouts either lack insight and capacities for self-analytic, criti-

    ao Her basic findings, however, do provide support for the theoreti- cal model discussed later in this paper.

    nreyesResaltadoEMOCION Y DESERCION

    nreyesResaltado

    nreyesResaltado

    nreyesResaltadolos procesos de adaptacion buscan generar un grado de maduracion en el estudiante e implican mayor control en la toma de deciciones o lo que en el texto refierene a moderacin. Por tanto lo programas de orientacion debe facilitar la generacion de madurez en el estudiante y y buen manejo de su atonmia

  • Dropouts f rom Higher Educat ion: 75 An Interdiscipl inary Review and Synthesis

    cal thinking, or reject these processes as important parts of their personalities.

    Another phenomenon that is closely related to rationality is anxiety, the irrational preoccupation with imagined fears and threats. According to Sexton (1965), maturity is a critical aspect of the dropout process because it represents the ability to control the irrational anxieties that impede motivation and achieve- ment. Grace (1957), for example, found that dropouts were more anxious than persisters, while Brown (1960) showed that among females, dropouts were generally more depressed than those who continued. Rose (1965) also finds that "defaulters" (early drop- outs) were both more anxious and maladjusted than persisters on probation, but persisters with high grades also had higher scores on these two scales. Part of the problem, suggests Summerskill (1962), is not only that dropouts are more anxious than others, but failure may be a way of irrationally acting out other kinds of problems. This suggestion is extended by Anderson (1954) who maintains that underachievers have un- consciously negative attitudes that predispose them to failure? 1 Vaughan (1968) concludes, in fact, that dropouts tend to deny (refuse to recognize) the pos- sibility that they have any psychological or intellectual limitations. Their defensiveness takes the form of com- pensation and projection, a tendency to paint a rosy picture of themselves despite their shortcomings.

    More often than not, then, empirical research de- picts dropouts as individuals who lack at least some of the attributes commonly associated with psychological maturity. Although these findings suggest that failing dropouts in particular are low on motivation, inde- pendence, flexibility, involvement, impulse control, self-confidence, responsibility, and rationality, we should note that falling at either extreme on these di- mensions is probably inconsistent with a truly mature (healthy) personality, a-~ Nonetheless, our results closely parallel those of Lavin (1965) who, in a com- prehensive review of the literature on academic per- formance, synthesized the findings on 26 different attributes and behaviors into six underlying dimen- sions. He found that higher levels of performance were generally associated with: (1) Social Maturity in the

    ~1 Pervin, Reik, & Dalrymple (1966), however,suggest a comple- mentary theory: i.e., that some students use academic achievement as a way of either compensating for a covering up more funda- mental insecurities.

    Student Role (greater social presence, responsibility, and restraint in social behavior); (2) Emotional Sta- bility (higher morale, stability, and freedom from neu- rotic orientations toward work) ; (3) Achievement Motivation Syndrome (higher motivation, activity level, and endurance); (4) Cognitive Style (greater curiosity, flexibility, originality, and liking for think- ing) ; (5) Achievement via Conformance (greater need for order, femininity, and conformance); and (6) Achievement via Independence (lower need for affilia- tion and peer conformity, greater independence, and lack of constrictedness).

    Later in his discussion, however, Lavin argues that it is misleading to deal with these attributes in isolation from the social environment in which the student finds himself. Standards of competition, the content of reference group value, and the student's need for achievement and affiliation all interact to influence the relationship between his personality dispositions and his level of performance. We would argue that this set of factors, plus academic performance must be consid- ered in the same analysis before an adequate, inter- disciplinary understanding of the dropout process is possible. Since interpersonal relationships would play an integral part in such an explanatory model, we briefly examine the related literature for examples of existing trends.

    Interpersonal Relationships: Profmae and Pro|ouad Although the literature on academic failures and the work on dropouts also have certain parallels with re- spect to interpersonal contacts, a careful comparison of the two suggests that they lead to opposite conclusions.

    The association between interpersonal relationships and college grade performance has recently been re- viewed by O'Shea (1969). His findings are primarily one-sided and apparently predicated on a simple fact: because of the finite amount of study time available during any one semester, students who wish to establish

    a2 In addition, we need to emphasize that these findings represent trends rather than categorical imperatives. Suczek and Alfert (1966) show, for example, that withdrawals from Berkeley in good academ- ic standing were generally more intellectually oriented, autono- mous, complex, open to ambiguity, and innovative in their thinking than were persisters. Failing withdrawals, however, were the least autonomous and intellectually oriented. The former students, then, are undoubtedly the ones Kubie (1966) had in mind when he discussed the act of dropping out as representing for some a refusal to sacrifice their creativity and spirit to the demands of the system.

    nreyesResaltado

    nreyesResaltado

  • 76

    extensive bonds within the social system of the college must dip into this time reserve in order to do so. Simply stated, time spent socializing is usually time lost to studying, as The result, according to O'Shea's review, is: 14 studies reporting that good social relationships and good grades are negatively related, 11 studies suggest- ing that they are essentially unrelated, and only 1 study showing a positive association. By implication, it would appear that under most circumstances extensive social- izing and exceptional academic performance are sel- dom combined in the same individual. Instead, as Lavin (1965), Phillips (1966), Sussmann (1960), and Wallace (1966) point out, more often than not a strain is created between the academic and social systems of the college that leaves the serious student only a small margin for compromise.

    According to theoretical ideas proposed by Lavin (1965), Nasatir (1969), and Sussmann (1960), how- ever, some of this strain may be reduced if friendship ties are established with people having strong aca- demic orientations. In this way, academic and social system influences may coalesce, providing opportuni- ties for both social interaction and mutual assistance. ~4 The major drawback, according to Wallace (1966), is that in viewing upperclassmen as significant role models (as is often done), freshmen expose themselves to the nonacademic group norms that are sometimes prevalent among older students. The result is a lower- ing of their own short-run achievement goals, a5

    The literature on dropouts, however, reveals that various measures of interpersonal orientations, friend-

    33 Findings by Andrew (1956) suggest, however, that the problem may involve more than just time. He found that the number of hours per week spent working at a job or in classes and labora- tories was basically unrelated to grade performance. Students with fewer hours available for study did just as well as those with more time to spare. The critical factor, of course, may be how efficiently those available hours are used.

    34 At least indirect evidence for this point is provided by Malloy (1954). He found that friendship and performance patterns were mutually reinforcing: underachievers had friends who were under- achievers, and overachievers befriended other overachievers. Whether performance influences choice of friends or friends in- fluence one's performance is not firmly established, however.

    35 Although upperclassmen had a dramatic impact on the orienta- tions of the freshman in Wallace's sample, work by Dertzin (1966) suggests that faculty occupy the top "campus-situational" reference group position for both male and female undergraduates. Friends, however, were regarded with nearly the same importance as faculty within the campus frame of reference, and were also the most significant others within their overall frame of reference.

    ship support, and extracurricular involvement are gen- erally associated with staying in college. There are exceptions, of course, but in most cases they distinguish what we have called the "profane" from the "pro- found." The profane refer primarily to what may be regarded as patterns of excessive and largely superficial socializing with casual acquaintances. Either this pat- tern or its attitudinal counterpart (a narcissistic or hedonistic orientation toward self and social relation- ships) is associated with dropping out in studies by Astin (1964), Trent and Medsker (1968), and Your- glich (1966). The dropouts in Hood's (1957) study had closer ties with upperclassmen rather than their own classmates, but the qualitative differences between the two, if any, are unknown (Wallace, 1966, not- withstanding).

    Otherwise, findings by Alfert (1966), Brown (1960), Hood (1957), Iffert (1958), Newcomb and Flacks (1964), Pervin (1966), and Spady (1967b) all suggest that an absence of "profound" relationships is definitely associated with attrition. By profound we mean relationships that imply significant meaning or closeness to the student. In three of these studies (Brown; Newcomb and Flacks; and Spady) the nega- tive impact on women of having no close friends is particularly apparent. In three others (Gurin et al., 1968; Rose, 1965; and Summerskill, 1962) dropouts were characterized as having either less social adequacy or lower social orientations. Since neither of these three studies included an actual friendship measure, we can only presume that their dropouts had fewer friendships (by virtue of their lower social skills or interests).

    The evidence relating to extracurricular participa- tion is less equivocal, however. Studies by Bemis (1962), Goble (1957), Koelsche (1956), Stone (1965), and Wolford (1964) all show that dropouts were less likely to participate in some form of campus activity than were persisters. 36 Goble in particular in- terprets his findings to suggest that participation is a major link to the social system of the college. Non- participants lack the social supports provided by pur- poseful contact with other students and the rewards that accrue from group accomplishments.

    86In addition, Reed (1965), Sexton (1965), and Spady (1970) all note that both the amount and kind of involvement in high school activities have a bearing on college success. On the whole, persisters are more likely to have participated in "important" activi- ties that demand a good deal of personal endeavor and responsi- bility.

  • Dropouts from Higher Education: 77 An Interdisciplinary Review and Synthesis

    In the main, then, the findings discussed here sug- gest that interpersonal relationships facilitate greater integration of the student into the social system of the college. To the extent that peer group norms either emphasize or denigrate academic endeavor, they may also influence achievement within the academic system, but this influence is more often implied in theory than verified empirically. 37 It is toward a synthesis of the variables relevant to both the academic and social sys- tems of the college that we now turn, in the hope of generating a more adequate theoretical and opera- tional model of the dropout process than we have seen in the literature to date.

    Toward a Sociological Model of the Dropout Process Since no one theoretical model or research design could possibly systematize or operationalize the specific re- lationships among all of the variables mentioned so far in this discussion, we do not attempt the absurd. We recommend, however, that with the more advanced multivariate statistical techniques and standardized computer programs now available, further a-theoreti- cal, bivariate research on the "correlates" oJ dropping out should be abandoned. Now! :~s We also incorporate into our model several variables shown earlier to be unambiguously associated with dropping out, and a few others that, in combination, may help to explain attri- tion as a conditional phenomenon.

    Our basic starting point is the assumption that the dropout process is best explained by an interdisciplin- ary approach involving an interaction between the individual student and his particular college environ- ment in which his attributes (i.e., dispositions, interests, attitudes, and skills) are exposed to influences, ex- pectations, and demands from a variety of sources (including courses, faculty members, administrators, and peers)Y 'q The interaction that results provides the student with the opportunity of assimilating success-

    a7 Perhaps, to date, Wallace (1966) has presented the best empiri- cal data of this phenomenon on the undergraduate level.

    as Researchers incapable of carrying out some form of multivariate analysis on their data (in order to at least test for spuriousness among their key variables) should contact their alma mater for a refund of their tuition.

    9 ~'J Brown (1967) describes this process of initial interaction as resembling a virtual assault on the student's self-image and esteem. The result can be a frantic search for meaningful sources of sup- port.

    fully into both the academic and social systems of the college. 4~ To the extent that the rewards available with- in either system appear insufficient, however, the stu- dent may decide to withdraw. 41 This decision, we have argued earlier (Spady, 1967b), may be influenced by at least two factors relevant in each system.

    Within the academic system, grades represent the most conspicuous form of reward. They are basically extrinsic and are used as tangible resources in the quasi- occupational role-playing of the career-oriented student in his negotiations for improved opportunities for success. 42 Intellectual development, however, may be more important to those oriented toward education as an integral part of their personal development. Though more subjective, it represents the intrinsic rewards of the system. Within the social system, "success" is de- fined first by having attitudes, interests, and personality dispositions that are basically compatible with the attributes and influences of the environment. This con- dition we call normative congruence. The second im- portant factor is the establishment of close relationships with others in the system, a condition we call/riend- ship support. Together these two conditions resemble the major social components of what Durkheim (1951), in his treatise on the social nature of suicide, called social integration.

    According to Durkheim, breaking one's ties with a social system stems from a lack of integration into the common life of that society. The likelihood of suicide increases when two kinds of integration are

    40 The term "assimilating" is used here in defence to Olsen's (1965) distinction between "integration" and "assimilation." He argues that social integration is a property of groups, and social assimila- tion is a relational property pertaining to individuals within those groups. Hence, according to his interpretation, a person is assimi- lated rather than integrated into a group's structure.

    41 Munger (1954) points out that early success during the fresh- man year is very important in preventing attrition, particularly among low-ability students. Williams (1967) argues that there are four major dimensions of the college environment that influence this interaction process: its degree of structure, its emphasis on questioning and inquiry, its non-nurturance and complexity, and the ability levels of others (competitive standards). When student behavior and orientations are not reinforced, dropping out results.

    42 Evidence presented by Davis (1966) and Munger and Goecker- man (1955) suggests, however, that the meaning and perceived utility of grades may vary considerably with the ability of the student and the quality of the institution. High ability students are less likely than others to view a passing grade with favor (Munger and Goeckerman), whereas grading standards across institutions force students in selective schools to work harder for the same marks than students with the same ability in less selective colleges.

  • 78

    absent: insufficient moral consciousness (viz. low normative congruence) and insufficient collective af- filiation (viz. low friendship support). Although drop- ping out is clearly a less drastic form of rejecting social life than is suicide, we assume that the social conditions that affect the former parallel those that produce the latter: a lack of consistent, intimate interaction with others, holding values and orientations that are dis- similar from those of the general social collectivity, and lacking a sense of compatibility with the immediate social system. However, since the student's academic role has many parallels with his future occupational role, it would not be inappropriate to extend this analogy a step farther. Poor performance in one's occupational role (viz. low grades) and inadequate identification with the norms of the occupational group (viz. low intellectual development) are also plausible additions to this system, since Durkheim views the occupational role as a critical component of the integration process.

    The elementary Durkheimian model that we pro- pose, then, consists of five independent variables, four of which influence the fifth, social integration, which in turn interacts with the other four to influence attrition (see Spady, 1967b). We would like to suggest further, however, that the link between social integration and dropping out is actually indirect. Intervening are at least two critical variables that flow from the integra- tion process: satisfaction with one's college experi- ences, and commitment to the social system (i.e., college). The addition of these two variables is based on two assumptions: first, that one's satisfaction with the college experience will depend on the available social as well as academic rewards; and second, that sustaining one's commitment to the college first re- quires both a sense of integration in the system and a sufficient number of positive rewards. 43

    What this model fails to account for, however, are the two variables that Gurin et al. (1968) found to be so important in their research: a comprehensive mea-

    43 This discussion also raises an issue that has not yet emerged with sufficient clarity: commitment to finishing a degree (viz. to the system of higher education itself) is not synonymous with commit- ment to a particular institution. Panos and Astin (1968) show that nearly half of the students in their national sample left their original college for reasons associated with the quality of its environment. Yet a considerable proportion of these students apparently found the environment elsewhere satisfactory enough to remain and finish a degree.

    sure of family and cultural background variables and a measure of academic potential. Since both potential and the elements that comprise normative congruence are influenced by family background, the latter pro- vides the foundation on which the remainder of the system rests (see Figure 1).

    The model as represented both implies a definite time sequence and depicts the assumed direct causal connections between pairs of variables. 44 Its most problematic aspects involve the meaning and opera- tionalization of normative congruence, since so much is implied in this one component. It represents not only all of the student goals, orientations, interests, and personality dispositions discussed earlier, but the con- sequences of the interaction between these attributes and various subsystems of the college environment as well. For example, a student may enter College A with strong utilitarian achievement orientations, but the col- lege itself may emphasize a humanistic, developmental undergraduate program. In the most general sense, then, this student could be considered normatively in- congruent. 4~ But he might, as Newcomb and Flacks (1964) emphasize, establish close relationships with other "deviants" who share his orientations, in which case subgroup support might be sufficient to override the more diffuse influences of the general system. 46 Or, as Gamson (1966) and Vreeland and Bidwell (1966) suggest, he might major in a department whose faculty members share and reinforce his own orientations. In any event, we are suggesting that the broad range of attributes subsumed in this component will have a

    44 The solid arrows imply that either previous research or the theoretical ideas generated from Durkheim's work suggest the existence of a direct causal link between that pair of variables. When this model is tested empirically, other direct links may also emerge from the data analysis, of course, but they are not assumed at this time.

    45 Although one might expect incongruent students or potential dropouts to view the environment differently from other students, work by Connor (1966) suggests that they do not. The scores of dropouts and persisters on the College and University Environment Scales (see Pace, 1964) did not differ appreciably in any systematic direction.

    46 Two things are implied in their results. First, friendships arc more likely to emerge between students with similar rather than complementary values and attitudes (balance theory); and second, students whose values are in a distinct minority are less likely to drop out if they have close friendships with students similar to themselves (solidary "deviant" subgroups). For a more explicit theoretical and empirical treatment of the balance theory model, see Newcomb (1961).

  • Dropouts from Higher Education: 79 An Interdisciplinary Review and Synthesis

    Figure 1/An Explanatory Sociological Model of the Dropout Process

    r Family Background

    ~ Academic Potential

    ~Normative Congruence ~

    I

    ~Grade [l'erformance ~_

    4Intellectual __ } "lDevel~

    ~ Friendship ]Support

    JSocial "[Integration

    TT ~Satisfaction , I

    ioccisio f v[ Dropout

    ;lIo i, ,,o.ol I l"mm"m ~ I

    I

    I

    I I I

    direct influence not only on the student's friendship support and social integration, but on his grade per- formance and intellectual development as well. The latter two variables will also be directly influenced by his patterns of interpersonal relations (friendship sup- port) and by his academic potential.

    The role of grade performance is also complex and, therefore, deserves a word of explanation. Unlike the other connections in the model, the arrow moving directly from grade performance to dropout decision should be interpreted as implying an absolute condition rather than a normal relationship between variables. It is reserved for those students whose performance is so low that institutional policy (dismissal) overrides the theoretical pattern of the model. In other words, a student doing failing work may be forced to leave de- spite his having a high degree of integration, satisfac- tion, or commitment to the institution.

    The broken arrow leading from institutional com- mitment back to normative congruence is particularly

    important as well, since it implies that the model is cyclical and flexible rather than immutable. We are suggesting here that the result of this whole process may lead to changes in attitude, interest, goals, or motivation that will in turn have repercussions at later stages of the college career. By definition, these changes in personal attributes will alter the conditions subsumed under normative congruence and will affect the remainder of the process as a result.

    Minimally, then, we hope that the ideas suggested here will serve both as a reasonable synthesis of some of the more consistent findings on college attrition cur- rently available, and as a worthwhile conceptual frame- work for guiding further research. An analysis of the relationships implied in this model based on longitudi- nal data from an entire entering class will be presented in the sequel to this paper.

  • 80

    About the Author William G. Spady is Associate Professor of Sociology at The Ontario Institute for Studies in Education. He re- ceived his PhD in the sociology of education from the University of Chicago in 1967 and served as Assistant Professor of Education and Social Relations at Harvard University for two years. His research has focused on the effects of family, institutional, and peer group influences on educational aspirations and attainment. His work has appeared in the American Journal o/ Sociology, Adult Education, and Educational Testing Service Research Memoranda.

    References

    ABEL, W. H.

    Attrition and the student who is certain. Personnel and Guidance Journal, 1966, 44, 1042-1045.

    ALBERS, T. E.

    Student attrition in six Colorado State-supported institu- tions o/higher education. ( Doctoral dissertation, University of Colorado) Ann Arbor. Mich. : University Microfilms, 1965. No. 66-2757.

    ALFERT, E.

    Housing and selection, need satisfaction, and dropout from college. Pxychological Reports, 1966, 19, 183-186.

    ANDERSON, J . R.

    Do college students lack motivation? Personnel and Guidance Journal, 1954.33,209-210.

    ANDREW, D. C.

    Relationship between academic load and scholastic success of deficient students. Personnel and Guidance Journal, 1956, 34, 268-270.

    AST1N, A. W.

    Personal and environmental factors associated with college dropouts among high aptitude students. Journal o/Educational Psychology, 1964, 55. 219-227.

    BARKER, L. W. A n analysis o/achievement, motivational, and perceptual variables between students classified on the basis o/success and persistence in college. (Doctoral dissertation. West Virginia University) Ann Arbor, Mich.: University Microfilms, 1968. No. 68-13,718.

    BAYER, A.

    The college drop-out: factors affecting senior college completion. Sociology o/ Education, 1968, 41, 305-316.

    BEMIS, 3". F.

    A study o/undergrad,tate students who voluntarily withdrew from the University o/Washington during 1959-60. (Doctoral dissertation, University of Washitagton) Ann Arbor, Mich.: University Microfilms, 1962. No. 62-6581.

    BLAINE, G. B., JR . , & MCARTHUR, C.

    Problems connected with studying. In G. B. Blaine, Jr. & C. McArthur (Eds.), Emotional problems o/the student. Garden City, N.Y.: Doubleday, 1961. Pp.83-109.

    BLOOM, B. S. & PETERS, F. R.

    The use o/ academic prediction scales/or counselling and selecting college entrants. New York: Free Press of Glencoe, 1961.

    BROWN, D. R.

    Student stress and the institutional environment. Journal o/Social Issues, 1967, 23, 92-107.

    BROWN, F. G.

    Identifying college drop-outs with Minnesota Counselling Inventory. Personnel and Guidance Journal, 1960, 39, 280-282.

    BROWN, W. F. , ABELES, N., & ISCOE, I.

    Motivational differences between high and low scholar- ship students. Journal o/Educational Psychology, 1954, 45, 215-223.

    CONGDON, W. L.

    Personality factors and capacity to meet curriculum demands. Personnel and Guidance Journal, 1964, 42, 17-3 I.

    CONNER, 3". D.

    The relationship between college environmental press and freshman attrition at Southern Methodist University. (Doctoral dissertation, North Texas State University) Ann Arbor, Mich.: University Microfilms, 1966. No. 66-10,949.

    COOPER, S.

    Employment of June 1959 high school graduates, October 1959. Monthly Labor Review, 1960, 83, 501.

  • Dropouts from Higher Education: 81 An Interdisciplinary Review and Synthesis

    DANIEL, K. L. B. A study of dropouts at the University of Alabama with respect to certain academic and personality variables. (Doctoral dissertation, University of Alabama) Ann Arbor, Mich.: University Microfilms, 1963. No. 64-9118.

    DAVIS, 3". A. Great aspirations. Chicago: Aldine, 1964.

    DAVIS, 3.. A. The campus as a frog pond. American Journal of Sociology, 1966, 72, 17-31.

    DEMOS, G. D. Analysis of college dropouts--some manifest and covert reasons. Personnel and Guidance Journal, 1968, 46, 681-684.

    DENZIN, N. K. The significant others of a college population. Sociological Quarterly, 1966, 7, 298-310.

    DONNELLY, M. 3.. M. A study of the factors which influence women college students to withdraw before completing their degree requirements. (Doctoral dissertation, University of Virginia) Ann Arbor, Mich.: University Microfilms, 1966. No. 66-15,219.

    DRESSEL, P. L. Liberal arts students advised to withdraw. Journal of Higher Education, 1943, 14, 43-45.

    DU BOIS, L. V. A study of ]actors related to persistence and withdrawal among sophomore students in the college of education at Oklahoma State University. (Doctoral dissertation, Oklahoma State University) Ann Arbor, Mich. : University Microfilms, 1965. No. 66-3999.

    DURKHEI M, E.

    Suicide. Translated by John A. Spaulding and George Simpson. Glencoe: The Free Press, 1951.

    ECKLAND, B. K. A source of error in college attrition studies. Sociology o/Education, 1964, 38, 60-72. (a)

    ECKLAND, B. K. College dropouts who came back. Harvard Educational Review, 1964, 34, 402-420. (b)

    EXTENCE, D. Non-intellective ]actors which may be related to voluntary withdrawal of college freshmen. (Doctoral dissertation, University of Southern California) Ann Arbor, Mich. : University Microfilms, 1965. No. 65-6904.

    FAUNCE, P. S. Personality characteristics and vocational interests related to the college persistence of academically gifted women. (Doctoral dissertation, University of Minnesota) Ann Arbor, Mich.: University Microfilms, 1966. No. 67-7722.

    FOX, D. E. Voluntary withdrawal in twenty-one liberal arts colleges for the period