CHAPTER 7

How College Affects Students:

A Summary

In the preceding 9/10 chapters, we have reviewed the evidence on a wide range of specific college outcomes. In this chapter, we attempt to tie this evidence together into a comprehensive summary of what we know about the impact of college on students. In developing this global synthesis, a somewhat different organizational framework is employed than that used in Chapters 3 through 11/12. In each of those chapters, the evidence pertaining to a specific category or outcome (e.g., learning, psychosocial development) was, where appropriate, summarized across six fundamental questions: 1) Do students change during the college years, and if so, how much and in what directions? (this is the “change” question); 2) To what extent are these changes attributable to college attendance as distinct from other influences, such as normal maturation or noncollege experiences? (the “net effects” question); 3) Are these changes differentially related to the kind of institution attended? (the “between-college” effects question); 4) Are these changes related to differences in students’ experiences at any given institution? (the “within-college effects” question); 5) Are these changes differentially shaped by individual student characteristics? (the “conditional effects” question); and 6) Is the influence of college durable? (the “long-term effects” question).

In the present chapter, this organizational framework is inverted. Here we synthesize the evidence with respect to each of the six fundamental questions posed by the book across the various outcome categories. The result is a somewhat different perspective than the previous chapters have afforded in that the focus is on the various impacts of college on a broad spectrum of outcomes rather than on how a specific category of outcomes may be influenced by different elements of the college experience.

In addition to the main objective of providing a comprehensive summary of our major conclusions, the chapter also seeks to accomplish three other goals. First, where possible, it attempts to determine the degree of consistency between the conclusions of our 1991 synthesis in How College Affects Students and the evidence we uncovered from the 1990s. Second, where possible, it articulates the extent to which the evidence is supportive of major theories or models of student development and college impact. Finally, the chapter offers suggestions for future research and briefly discusses how different methodologies can be used to optimize our understanding of the impact of college.

Change During College

In our 1991 synthesis, we concluded that the evidence indicated that the college years are a time of student change on a broad front. While the changes that occurred from freshman to senior year were generally the largest “effects” we uncovered in our synthesis, it is the breadth of change that was perhaps the most striking characteristic of the evidence. Students not only made statistically significant gains in factual knowledge and in a range of general cognitive and intellectual skills, they also changed significantly on a broad spectrum of value, attitudinal, psychosocial, and moral dimensions. Thus, the change coincident with the college years did not appear to be confined to a few isolated areas. Rather, the changes occurred in an integrated way, with change in any one area appearing to be part of a mutually reinforcing network or pattern of change in other areas.

Although the evidence from the 1990s on change during college is nowhere near as extensive as the evidence uncovered in our 1991 synthesis, there is little that would lead us to change our basic conclusion. Maturation that occurs during the undergraduate years is holistic in nature and embraces many facets of individual change. (Comments here re: theories of change or development.)

There are some very clear directions to this overall pattern of change during the undergraduate years. We turn now to a brief summary of those changes under the general topics of learning and cognitive change, attitudes and values, psychosocial changes, and moral development.

Learning and Cognitive Change

In our 1991 synthesis, we concluded that students make statistically significant freshman-to-senior gains on a variety of different dimensions of learning and cognition. Our estimates of the dimensions along which gains occur and the magnitude of the gains were as follows:

Dimension	Effect Size
Verbal skills	.56 sd
Quantitative skills	.24 sd
Specific subject matter knowledge	.84 sd
Speaking skills	.60 sd
Written communication	.50 sd
Piagetion (formal) reasoning	.33 sd
Critical thinking skills	1.00 sd
Reflective judgment/thinking (use of reason and evidence to address ill-structured problems; post formal reasoning)	1.00 sd
Conceptual complexity	1.20 sd

The evidence on gains during college in the areas of learning and cognition is much less extensive in the literature of the 1990s than it was in the literature we reviewed for our 1991 synthesis. However, with two possible exceptions, we uncovered little in the research of the 1990s that would lead us to fundamentally alter our major, if unsurprising, conclusion that students not only make significant gains in subject matter knowledge during the undergraduate years, but they also become more critical, reflective, and sophisticated thinkers. Our estimates of the freshman-to-senior gains indicated by the literature of the 1990s are as follows:

Dimension	Effect Size
English (reading and literature, writing)	.77 sd
Mathematics (general mathematics proficiency, algebra, geometry)	.55 sd
Science (laboratory and field work, understanding fundamental concepts)	.62 sd
Social studies (history, social sciences)	.73 sd
“Liberal arts competencies” (e.g., using science, using art, solving problems)	.80 sd
Critical thinking skills	.50 sd
Critical thinking disposition	.50 sd
Reflective judgment/thinking	.90 sd
Epistemological sophistication or maturity	2.00 sd

The two notable differences between the conclusions from our 1991 synthesis and our conclusions from the literature of the 1990s are in the areas of quantitative mathematics competencies and critical thinking skills. The research from the 1990s suggests the possibility that our previous synthesis may have underestimated typical student growth in quantitative/mathematics competencies in college and overestimated students’ acquisition of critical thinking skills during the undergraduate years. However, such a conclusion needs to be interpreted cautiously. The different estimates from the two syntheses on these dimensions could also be attributable to use of different measurement instruments administered to different samples.

It is important to note that our estimates of the gains made during the undergraduate years on various dimensions of academic learning and intellectual sophistication reflect only relative advantages of seniors over beginning students. As such, they should be understood within the context of two additional, and important, findings from the research of the 1990s. The first suggests that college graduates as a group do not always perform particularly well in terms of absolute standards of knowledge acquisition or cognitive functioning. (For example, only about 50% of all college graduates appear to be functioning at the most proficient levels of prose, document, or quantitative literacy.) Such a finding suggests an important future alternative to research that simply documents change during college or relative senior advantages over freshmen. Documenting the proficiency of college graduates on absolute standards of subject matter knowledge and cognitive skills is of equal, if not greater, importance than simply knowing how much they change or grow during the undergraduate years.

The results of a second line of inquiry in the 1990s are grounds for greater optimism. There is evidence to suggest that undergraduate students can retain as much as 70-85% of the subject matter content typically introduced in postsecondary settings. Of particular importance, it would appear that increasing the level of original learning during college increases retention subsequent to college.

(Section on attitudes & values and psychosocial changes here)

Moral Development

In our 1991 synthesis, we concluded that there was clear and consistent evidence that undergraduates make statistically significant gains during college in the use of principled reasoning to judge moral issues. This finding held across different measurement instruments and even different cultures. The relative magnitude of these gains was difficult to determine from the evidence presented in the research. However, the major shift that occurred during college was from conventional to post-conventional reasoning.

We found little in our present synthesis that would lead us to change our previous conclusion. The major contribution of the research of the 1990s was that it permitted us to estimate an effect size. From the literature reviewed, we estimate that the average advantage of seniors over freshmen in principled moral reasoning is about .77 of a standard deviation. Furthermore, consistent with the conclusions from our previous synthesis, the research from the 1990s indicates that the major change taking place during the undergraduate years is a shift from using moral reasoning that concedes to societal authority (conventional moral reasoning) to using reasoning that is founded on the application of universal moral principles (post-conventional moral reasoning).

Some Final Thoughts on Change During College

In our previous synthesis, we made a number of general observations about documented changes during the undergraduate years. We believe these observations also apply to the present synthesis and are worth reiterating, at least briefly. First, there is a remarkable agreement across four comprehensive literature reviews (i.e., Feldman & Newcomb, 1969; Bowen, 1977; Pascarella & Terenzini, 1991; and the present synthesis) with regard to the changes that occur during college. Taken as a total body of evidence, all four syntheses suggest that a reasonably consistent set of cognitive, attitudinal, value, and psychosocial changes have occurred among college students over the past 50 years. In our previous synthesis (Pascarella & Terenzini, 1991, pp. 563-64), we concluded that:

Students learn to think in more abstract, critical, complex, and reflective ways; there is a general liberalization of values and attitudes combined with an increase in cultural and artistic interests and activities; progress is made toward the development of personal identities and more positive self-concepts; and there is an expansion and extension of interpersonal horizons, intellectual interests, individual autonomy, and general psychological maturity and well-being. Thus, it can be said that the nature and direction of freshman-to-senior changes appear to be reasonably stable and to some extent predictable.

Although there are understandable differences in the estimated magnitude of changes across the different syntheses, we uncovered little in our present synthesis that would lead us to fundamentally alter our 1991 conclusion.

Second, throughout this book we have, where possible, estimated the magnitude of the changes that happen during the undergraduate years. While we believe that this can be useful information, we would caution against an overzealous focus on quantitative estimates of change. A potential danger in focusing only on quantitative estimates of change, such as effect size, is the assumption that growth or development happens on a continuum where all changes are of equal importance. A number of developmental theorists would argue that growth does not always happen in such an even and continuous fashion, and that not all changes are of equal importance (e.g., King & Kitchener, 1994; Kohlberg, 1981; Perry, 1970; Rest, 1986). Some shifts are particularly critical to development irrespective of whether or not they are reflected in a large quantitative change on some continuous scale. For example, the shift during college from reasoning based on beliefs to one relying on evidence in making judgments represents a key prerequisite to dealing effectively with ill-structured problems. Similarly, the shift from conventional to post-conventional or principled reasoning during college represents a major qualitative advance in moral development. On both of these dimensions of development, the qualitative nature of the change may be of greater consequence than the quantitative estimate of the change.

Third, we would also argue that the magnitude of the change on any particular variable, or set of variables, during the undergraduate years may not be as important as the pronounced breadth of interconnected changes that occur. The evidence indicates not only that individuals change on a broad developmental front during college, but also that the changes are of a mutually consistent and supporting nature. There may be insufficient empirical grounds to speak of changes in one area causing or permitting changes in other areas. However, it is clear from the evidence we reviewed in both our previous (1991) and present syntheses that the changes coinciding with college attendance typically involve the whole person and proceed in a largely integrated manner. (Link with developmental theories?)

Finally, there are at least three nontrivial problems endemic to the study of change during college. The first stems from the fact that the evidence is based largely on studies estimating typical or average change in some sample (longitudinal studies) or typical or average differences between samples (cross-sectional studies). By focusing on average group shifts or differences, the findings of such studies tend to mask the presence of pronounced individual differences in patterns of change.

A second problem, one that we have emphasized throughout this book, is that freshman-to-senior change during college does not necessarily represent the impact of college. Nearly all the studies of change we reviewed in our synthesis lacked a control group of students who did not attend college. Consequently, it is extremely hazardous to conclude that a one standard deviation gain on some variable during college is really the result of college attendance. Part or even all of the gain could be attributable to such confounding influences as simple maturation or the practice effect associated with completing the same measurement instrument twice. Similar individuals not attending college might well exhibit the same size gains as college students over the same period of time. Furthermore, just as the presence of change does not necessarily indicate the impact of college, so too the absence of changes does not necessarily indicate the absence of college impact. As suggested by Feldman and Newcomb (1969), one consequence of exposure to college may be to stabilize development on some dimensions at a certain level and prevent reversion or regression. If such were the case on a specific trait, little or no freshman-to-senior change would be evidenced. However, those not attending college might well regress or change in a negative direction on the same trait.

Third, it is important to reiterate that change and development are not synonymous. Change simply means that some fact or condition at Time₂ is different from what it was at Time₁, whereas development implies ordered, predictable, even hierarchical shifts or evolution have taken place in fundamental, intra-individual structures or processes. In many areas of observed change during college, it is tempting simply to conclude that observed change reflects some form of internal growth or development in the individual, that an inner restructuring has taken place, and that the senior is functioning with an advanced set of inner rules or perspectives not present in the typical freshman. This is a particular temptation when the changes that occur are consistent with those posited by developmental models or theories. The danger inherent in this assumption is that what we commonly refer to as development may in large measure be the result of an individual’s response to the anticipated norms of new social settings or social roles. Different categories of people may be socialized to think and behave differently in society, and a substantial part of this categorization may have its basis in educational level. Thus, for example, college-educated men and women may have certain psychosocial traits and values and may think about controversial issues in certain ways not necessarily because of some inner developmental restructuring but because they have been socialized to behave and think in ways consistent with dominant cultural norms for educated adults.

This is not to say that the changes that occur during college merely represent the learning of social or cultural norms instead of important developmental steps. Rather, it is to suggest that we need to be wary of the tendency to equate the learning of social or cultural norms with development. It behooves us to bear in mind that change during the college years is produced by multiple influences, some internal (and perhaps ontogenetic) and others external to the individual. Theories can overly restrict as well as focus vision.

Net Effects of College

Because self-selection, as opposed to random assignment, determines who attends and who does not attend college, studies that seek to estimate the net (or unique) impact of college (as distinct from normal maturation or other extraneous sources of influence on change) employ some rather creative research designs or, more typically, statistical controls. Clearly, the causal inferences one can make from such studies are not of the same order of certitude as those made from randomized, “true” experiments. Nevertheless, we can make a reasonably valid set of tentative conclusions about the changes or outcomes observed that are attributable to college attendance and not to rival explanations. As with our 1991 synthesis, there is more extensive and consistent evidence supporting the net impact of postsecondary education on learning and cognition, moral reasoning, and career and economic returns than in the areas of attitudes, values, and psychosocial characteristics. One should not necessarily conclude, however, that postsecondary education has a stronger net impact on the former than on the latter outcomes. Some of these differences could reflect variations in the extent and quality of the available evidence across different areas of inquiry rather than major differences in the actual impact of exposure to college.

Learning and Cognitive Changes

In our 1991 synthesis, we concluded that on nearly all of the dimensions on which there is demonstrated freshman-to-senior change, a statistically significant part of that change was attributable to college attendance, not to rival explanations. Exposure to postsecondary education not only appeared to significantly enhance general verbal and quantitative skills as well as oral and written communication, but it also had a statistically significant, positive net effect on general intellectual and analytic skills, critical thinking, the use of reason and evidence in addressing ill-structured problems (reflective judgment/thinking), and intellectual flexibility. These effects could not be explained away by maturation or differences between those who attended and those who did not attend college in tested intelligence, academic ability, or other precollege characteristics. What was less clear from our previous synthesis was the magnitude of the net impact of college. The only learning and cognition outcomes for which we were able to estimate an effect size were general verbal skills (.26 to .32 of a standard deviation); general quantitative skills (.29 to .32 of a standard deviation); and critical thinking (.44 of a standard deviation) during the first year of college.

Our present synthesis did not uncover a large body of evidence on the net effects of college on learning and general cognitive skills. However, the evidence we did uncover is not inconsistent with the conclusions from our previous synthesis. What is new from the research of the 1990s is our ability to estimate the magnitude of the net college effect for each outcome. We estimate the magnitude of those net effects to be as follows:

Dimension	Net Effect Size
General verbal and quantitative skills	.25 sd
English	.59 sd
Mathematics	.32 sd
Science	.47 sd
Social studies	.46 sd
Critical thinking	.55 sd (For first 3 years of college)
Reflective judgment/thinking	.90 sd

As a rule, these estimated effects cannot be explained away by such rival hypotheses as academic ability, sex, race, or, with the possible exception of reflective judgment/thinking, maturation.

Two additional points are worthy of consideration. First, our estimate of the net impact of three years of postsecondary education on gains in critical thinking skills (.55 sd) is actually larger than our estimate of freshman-to-senior change (.50 sd). This may, to some extent, reflect the confounding influence of research conducted with different critical thinking measures. However, it also suggests that our estimate of freshman-to-senior change in the present synthesis is overly conservative. Second, it is also distinctly possible that our estimates of the net college effects on English, mathematics, science, and social studies err on the conservative side. The instruments on which these net effects are based focus on assessing general education competencies that are typically taught in the first two years of college. Consequently, our estimated net effects in these content areas may fail to capture the full net effects of postsecondary education.

(Sections on Values, Attitudes, & Psychosocial Development)

Moral Development

In our previous 1991 synthesis, we concluded that individuals exposed to postsecondary education demonstrated significantly greater growth in the use of principled moral reasoning to judge moral issues than individuals whose formal education ended with high school. This difference persisted even in the presence of controls for maturation and for differences between those who attended and those who did not attend college in level of precollege moral reasoning, intelligence, and socioeconomic status. It was not possible, however, to estimate the magnitude of postsecondary education’s net impact on growth in the use of principled moral reasoning.

The presence of a statistically significant net impact of college on moral behavior was less clear. On the basis of two separate bodies of evidence, however, we hypothesized a positive indirect effect in which college increases the likelihood of principled behavior by stimulating the growth of principled moral reasoning.

The conclusions from our present synthesis are essentially identical to those from our 1991 synthesis. Postsecondary education appears to have a statistically significant and positive net effect on growth in the use of principled moral reasoning in addressing moral problems, but it is difficult to estimate the magnitude of that effect. Consistent with our previous synthesis, we also found extensive evidence of a positive relationship between level of principled moral reasoning and the likelihood of principled behavior in a wide range of settings. Thus, growth in principled reasoning during college should, at least indirectly, increase the probability of principled action. Principled reasoning alone, however, is probably insufficient to insure principled behavior because it is only one of several influences on moral action. Other influences include such things as moral sensitivity, moral motivation, and moral character. Unfortunately, compared to principled moral reasoning, we have almost no systematic evidence concerning postsecondary education’s impact on these other precursors of moral action.

Net Long-Term Effects of College

The vast majority of the substantial body of research concerning the long-term effects of college focuses on estimating the enduring impact of attending versus not attending college. Consequently, we depart from the typical pattern of the preceding chapters and summarize the evidence on the net long-term effects of college here rather than near the end of this chapter.

Consistent with our 1991 review, our present synthesis indicates that postsecondary education has a rather broad range of enduring or long-term impacts. These include not only the more obvious impacts on occupation and earnings, but also influences on cognitive, moral, and psychosocial characteristics, as well as on values and attitudes, and various indexes of the quality of one’s life. Furthermore, there is also evidence to suggest that the impacts of postsecondary education extend beyond the individuals who attend college to the nature of their children=s lives.

It is clear that the net long-term impact of postsecondary education manifests itself in at least two ways. One part of the long-term impact can be traced directly back to college attendance or degree attainment (e.g., effects on job status and earnings). A second part of this impact, however, is likely the indirect result of the socioeconomic positioning and the kinds of interests, experiences, and opportunities made more likely by being a college graduate. Put another way, a major part of the long-term effect of college arises out of the distinctive kinds of lives led by people who attend and graduate from college. Such indirect routes of influence are an important consideration in understanding the full long-term impact of college. Indeed, a substantial number of college effects tend to persist in large measure as the result of living in postcollege environments that support and further stimulate those effects.

Socioeconomic Outcomes

The evidence uncovered in our previous synthesis limited the conclusions we could make about the socioeconomic impacts of postsecondary education to the net impact of having a bachelor’s degree versus a high school degree. We concluded that a bachelor’s degree conferred average net advantages over a high school degree of about one standard deviation in occupational status (job desirability); 20-40% in earnings; and 9.3-11% in return on personal investment in postsecondary education (private rate of return). These advantages persisted in the presence of statistical controls for such confounding influences as socioeconomic origins, tested intelligence, aspirations, and in the case of private rate of return, costs of education and foregone earnings. Moreover, having a bachelor’s degree continued to confer a statistically significant net advantage over a high school diploma in terms of both stability of employment and career mobility and attainment. The actual magnitude of these advantages, however, was difficult to determine.

The evidence we uncovered in our present synthesis would not lead to major revisions of our 1991 conclusions. What is new in the research of the 1990s, however, is our ability in some areas to estimate separate net effects by gender and for levels of postsecondary education less than a bachelor’s degree. Following are estimates of the net effects of postsecondary education on socioeconomic outcomes as indicated by the research of the 1990s:

Dimension	Net Effect Size
Occupational status:
Bachelor’s degree	.95 sd advantage over a high school diploma
Associate’s degree	.24 - .44 sd advantage over a high school diploma
Vocational degree or license/certificate	.12 - .22 sd advantage over a high school diploma
Earnings (men):
Bachelor’s degree	37% advantage over a high school diploma
Associate’s degree	17.5% advantage over a high school diploma
Earnings (women):
Bachelor’s degree	39% advantage over a high school diploma
Associate’s degree	27% advantage over a high school diploma
Credentialing/program effects (men)
Bachelor’s degree	15% earnings advantage over 4 years of college credits but no degree
Associate’s degree	9% earnings advantage over 2 years of college credits but no degree
Credentialing/program effects (women)
Bachelor’s degree	12% earnings advantage over 4 years of college credits but no degree
Associate’s degree	11% earnings advantage over 2 years of college credits but no degree
Private rate of return	Bachelor’s degree confers about a 12% return on investment (over a high school diploma), with a typical range of 9-16%
Stability of employment	Increases with amount of post-secondary education, but size of effect is unclear

In addition to our estimates of the net economic premiums conferred by bachelor’s and associate’s degrees, four additional points need to be made with respect to the impacts of sub-baccalaureate postsecondary education. First, the earnings returns to an associate’s degree appear to be essentially the same for experienced workers who return to school as they are for continuing high school graduates. Second, completion of one-year vocational certificates can increase earning power, particularly for women, but the average net earnings returns to such certificates (versus a high school diploma) appear to be less certain, and likely smaller, than the average net returns to associate’s degrees. Third, individuals can potentially increase their earnings by obtaining modest amounts of postsecondary education without completing a degree of certificate; but the average economic premium appears to be less certain and smaller in magnitude than the average return yielded by an associate’s degree or certificate. Finally, the economic payoff of completing a year of academic credits at a community college appears to be at least equal to the payoff of completing the same number of credits at a four-year college.

Although there are discernible between- and within-college effects (which we summarize later in this chapter), it is clear that the average occupational and economic returns to a bachelor’s degree remained substantial in the research produced in the 1990s. Similarly, while periodic fluctuations are to be expected, the private rate of return on investment in a bachelor’s degree continued to be quite competitive with benchmark rates for alternative ways of investing one’s money. Of equal importance, the average occupational and economic benefits attributable to the completion of an associate’s degree also appear to be appreciable, though obviously smaller in magnitude than those linked to completing a bachelor’s degree.

The ways in which postsecondary education, and particularly a bachelor’s degree, positions one occupationally and economically represents an important long-term effect in and of itself. But this socioeconomic positioning has additional implications for other long-term impacts. One stems from the simple fact that the jobs typically held by the college educated tend to be characterized by a relatively high level of earnings. This provides greater discretionary income that can be used to acquire a range of material and nonmaterial resources and opportunities (e.g., books, magazines, computers, travel, cultural experiences, household maintenance, medical care, additional education) that have potential impact on other long-term outcomes. A second implication stems from the fact that the college-educated tend to be employed in jobs characterized by relatively higher levels of social interaction and self direction than those whose education ends with high school. Such job traits may themselves provide an important continuing influence on trends in cognitive and noncognitive development initially shaped during the college years.

Learning and Cognitive Development

In our 1991 synthesis, we concluded that, despite some obvious methodological weaknesses, the evidence from national samples indicated that college graduates had a substantially larger general knowledge base across a wide range of topics than did individuals whose education ended with high school. Similarly, in evidence from national surveys, alumni were consistent in reporting that college had a major positive influence both on their specific and their general knowledge base, and on their ability to think critically, analytically, and clearly.

Although the literature from the 1990s is somewhat different in form and in some cases less extensive than the literature supporting our previous synthesis, it nevertheless provides substantial reinforcement for our 1991 conclusions. The evidence is clear in suggesting not only that the college-educated are more knowledgeable and more proficient at becoming informed than those with a high school education, but also that the undergraduate experience provides information and cognitive skills that increase one’s capacity for life-long learning and continuing intellectual development. The trends in learning and intellectual growth shaped by the undergraduate experience tend to continue along the same trajectory after college; but continued learning and intellectual growth after college also depends, to a substantial extent, on the degree of intellectual stimulation and availability of learning opportunities in one’s post-college life.

Compared to those with a high school education, the college-educated have a number of advantages in terms of available learning opportunities and continuing intellectual stimulation in their lives. First, because of the nature of their occupations and level of earnings, the college-educated are likely to have greater access to resources and experiences that provide continued intellectual stimulation and opportunities for learning (e.g., books, magazines, computers, travel, concerts, continuing education courses). However, simply having more discretionary income or more intellectually stimulating work doesn’t tell the whole story. The college-educated are more likely than those with a high school education to participate in activities that enrich their funds of knowledge and provide intellectual stimulation, such as using a public library, even when income, occupation, and other confounding influences are taken into account. This suggests that, in addition to fostering the capacity for life-long learning, an important long-term impact of postsecondary education is its role in crystallizing one’s personal disposition for life-long learning and intellectual development.

(Sections on values, attitudes, & psychosocial changes)

Moral Development

In our 1991 synthesis, we concluded that there was strong evidence for an enduring impact of postsecondary education on the use of principled moral reasoning, at least through the first six years after graduation. Individuals attending college not only made greater gains in the use of principled reasoning during college than those individuals whose formal education ended with high school, but the gap between the two groups continued to widen in the years subsequent to college. These different patterns of change could not be accounted for by initial group differences in moral development or differential regression artifacts.

We did not find the kind of longitudinal evidence in the research of the 1990s that permitted long-term comparisons of trends in principled moral reasoning growth among individuals completing different amounts of postsecondary education. However, the scant evidence we did uncover reinforces the conclusions of our previous synthesis in suggesting that the developmental trends in principled moral reasoning put in motion by the postsecondary experience do not retrogress after college. Rather, level of principled reasoning at the end of the senior year tends to persist or modestly increase for college graduates during the years immediately following graduation. In part, this tendency may be attributable to an environment of continuing intellectual stimulation that often characterizes the occupational and lifestyle choices typical of, or more available to, college graduates.

Quality of Life Indexes

In our 1991 synthesis, problems in research design and the inability to control important confounding influences made causal attributions about the long-term impact of postsecondary education on various quality of life indexes tenuous. Nevertheless, college-educated individuals consistently ranked higher than those with less education on a range of quality of life indexes. Compared to those with less education, the college-educated tended to have better overall health and a lower mortality rate, have smaller families, and spend a greater portion of time in child care activities of a developmentally enriching nature (e.g., teaching, reading, talking). The college-educated also tended to be more efficient in making consumer choices, saved a greater percentage of their income, made more effective long-term investment of discretionary resources, and spent a greater proportion of discretionary resources and time on developmentally enriching activities (reading, participation in arts and cultural events, involvement in civic affairs, and the like). However, while college-educated individuals ranked higher on a broad array of quality of life indexes, they did not, on average, indicate appreciably greater overall satisfaction with their lives than did those with less education.

Although it does not provide evidence on all the quality of life dimensions addressed in our previous synthesis, the research from the 1990s yields findings in the areas of health, child welfare, community/civic involvement, subjective well being, and job satisfaction that are generally supportive of our 1991 conclusions. What is different about the literature from the 1990s is that on several quality of life indexes, the evidence permits a somewhat better identification of the potential causal mechanisms underlying the significant, positive link with level of formal education. It is also the case, however, that the majority of studies treat education as a continuous variable (e.g., years of formal education completed). Thus, it is often difficult to estimate the magnitude of the net effect of different amounts of postsecondary education. Rather, about the best one can do is to infer, or extrapolate, the impact of postsecondary education from the overall effect of education.

The late 1980s and the 1990s produced an extensive body of methodologically sound research indicating that educational attainment has a positive causal impact on good health. The evidence is reasonably clear that, in the presence of statistical controls for salient confounding influences, increased educational attainment significantly lowers:

1. the probability of mortality at any particular age;

2. the likelihood of specific health problems, such as disability or frailty;

3. the probability of mortality from cancer and cardiovascular disease; and

4. the probability of having risk factors for cardiovascular and other diseases.

Although it is difficult to estimate the magnitude of the effects, those studies that permit identification of the net influence of different levels of formal education indicate that, compared to those with no exposure to postsecondary education, individuals who attend or graduate from college (a bachelor’s degree) have significantly lower risk profiles (i.e., blood pressure, cholesterol levels, cigarettes smoked per day) for both coronary heart disease and cancer. Moreover, even with this risk profile and age taken into account, those who attend or graduate from college also have a significantly lower risk (than those with no postsecondary education) of actual mortality from all causes and from coronary heart disease.

A number of mechanisms underlying the likely causal influence of educational attainment on health have been identified. However, they are numerous and their mutual interaction is probably complex. They reflect both the direct and indirect effects of education and include work and economic conditions, health lifestyle, access to better health information, producing better health decisions from available information, and time preference for the future. The search for a single causal mechanism to explain the link between educational attainment and health may be a largely fruitless exercise.

For example, one of the most clearly demonstrated impacts of educational attainment on health is realized through health lifestyle. Net of important confounding influences, educational attainment in general tends to have significant negative effects on cigarette smoking, alcohol abuse/dependency, and cholesterol level, and significant positive effects on aerobic exercise, a healthy diet, and dietary fiber intake. As an example, compared to those with a high school education, individuals with a bachelor’s degree are substantially less likely to smoke and substantially more likely to quit smoking if they had ever smoked. Part of the explanation for this is that the college-educated have better access to health information than high school graduates. (Recall, for example, that postsecondary education increases the likelihood of library use, irrespective of occupation and income.) However, having a bachelor’s degree or higher appears to have a role in reducing bad health habits (e.g., smoking) and promoting good health habits (e.g., aerobic exercise) beyond simply being knowledgeable concerning the impact of such behaviors on health. This finding suggests two other causal mechanisms that may be at work. First, it may be that the college-educated are able to produce better health decisions from the same information than high school graduates. Second, postsecondary education may reinforce a future orientation in which the individual is better able to understand the long-term health consequences of present day behaviors.

Though not as methodologically rigorous as the body of research on educational attainment and health, there is nevertheless correlational evidence from the 1990s that supports and expands the conclusion from our previous synthesis that parental education is positively linked to indicators of child welfare. For example, parents’ formal education is positively associated with a newborn child receiving good prenatal care, parental involvement in a child’s school, reading to a child, help with a child’s homework, and a child’s access to computer resources. Conversely, it is negatively linked to the risk of childhood death by age two, and to the probability of teenage pregnancy.

The evidence of the 1990s also extends the conclusions from our 1991 synthesis by suggesting that educational attainment in general, and a bachelor’s degree in particular, lead to significantly higher levels of community and civic involvement. Net of other factors, including prior levels of involvement, individuals with a bachelor’s degree (compared to those with a high school diploma) were 1.8 times as likely to be frequently involved in political activities, 2.4 times as likely to be an active participant in community welfare groups, 1.5 times as likely to be frequently involved in political discussions, 1.8 times as likely to be highly committed to community leadership, and 2.5 times as likely to vote in a national, state, or local election. Those individuals with less than a bachelor’s degree, but at least some exposure to postsecondary education, were also between 1.6 and 1.7 times as likely to vote as their counterparts with a high school diploma.

Finally, with respect to the long-term impacts of postsecondary education on job satisfaction and life satisfaction (or subjective well-being), our conclusions from the research of the 1990s are generally, if not totally, consistent with the conclusions of our previous synthesis. Overall, we found the net total effect of education on both outcomes to be quite small. This is probably because level of formal education has direct and indirect impacts on both job satisfaction and life-satisfaction that tend to be conflicting. For example, having a college education tends to have a positive indirect effect on job satisfaction through its impact on such factors as job prestige and earnings, job autonomy, and nonroutine work. However, the direct effect of a college degree on job satisfaction tends to be negative, possibly because postsecondary education creates a critical perspective and functions to raise occupational expectations. Similarly, educational attainment has positive indirect impacts on life-satisfaction or subjective well-being by means of its enhancement of such factors as economic affluence and sense of personal control over one’s life. Conversely, the direct effect of education is often negative because of education’s enhancement of one’s capacity to make measured, comprehensive, and critical judgments. Increased education may also lead one to interpret subjective well-being or life happiness in more complex and qualitatively different terms.

Intergenerational Effects of College

An often overlooked element of the long-term impact of postsecondary education is the intergenerational transmission of benefits. In our 1991 synthesis, we found evidence indicating that the net benefits of a college education are not restricted to the individual who attends college, but are also passed along to his or her sons and daughters. Net of important confounding influences, such as parental income and one’s race, gender, and aspirations, having college-educated parents modestly enhanced one’s educational attainment, job status, early career earnings, and, if one is a woman, the likelihood of entering a more financially lucrative, male-dominated occupation.

The literature of the 1990s was largely silent on the intergenerational impacts of college on the socioeconomic achievements of offspring. However, it did provide new evidence to suggest that postsecondary education may provide an intergenerational legacy in terms of children’s knowledge acquisition. Net of such confounding influences as race, sex, test scores, and parental socioeconomic status, having parents who attended college had a positive influence on a son’s or daughter’s high school science and mathematics achievement, as well as on his or her college-level reading comprehension. This effect is likely attributable to the fostering of greater “learning capital” in the sons or daughters of the college educated, much of which is realized indirectly through the nature of the home environment. There was little compelling support, however, for a similar intergenerational impact of parental postsecondary education on children’s general cognitive skills.

Between College Effects

In our 1991 synthesis, we concluded that, across all of the outcomes we considered, where one attended college tended to have smaller impacts than either the net impact of attending (versus not attending college) or differences among individuals’ experiences at the college attended (within-college effects). The body of evidence underlying the present synthesis would tend to reinforce this generalization. Clearly there are substantial differences among the United States’ 3,000+ postsecondary institutions in a range of areas, including size and complexity, public or private control, mission, financial and educational resources, the scholarly/teaching orientation of faculty, reputation and prestige, and the characteristics of the students enrolled. Yet, with some notable exceptions, the weight of evidence from the 1990s casts considerable doubt on the premise that the substantial structural, resource, and qualitative differences among postsecondary institutions produce correspondingly large differences in net educational effects on students. Rather, it may be that the vast majority of postsecondary institutions have surprisingly similar net impacts on student growth, although the “start” and “end” points may be very different across different institutions. Consistent with our 1991 synthesis (as well as with Bowen’s 1977 review), the research from the 1990s leads to the conclusion that similarities in between-college effects would appear to vastly outweigh the differences.

Of course, this is a general conclusion based on the total body of evidence. Although no specific institutional characteristic or set of characteristics has a consistent impact across outcomes, there are, in fact, some statistically reliable between-college effects in certain areas. These between-college effects are more pronounced in the areas of career and economic attainment after college than they are in the developmental changes that occur during college (e.g., knowledge acquisition, cognitive development, values and attitudes, psychosocial development). This is perhaps not overly surprising since in the areas of career and economic achievement the status allocating aspects of a college, and what a degree from that college signals to prospective employers about the characteristics of its students, may count as much if not more than the actual quantity of the education provided. Moreover, the estimation of between-college effects on the developmental changes that occur during college is seriously complicated by recent national demographic trends in student participation in postsecondary education. Evidence from nationally representative samples suggests that since the late 1980s, about 50% of students who initially enroll at a four-year college attend two or more undergraduate institutions. This migration among institutions makes the estimation of between-college effects on developmental changes complex and problematic. Indeed, the findings of the research in this area may not generalize to the substantial numbers of U.S. students who attend more than one institution as undergraduates.

This final caveat aside, we uncovered considerably more evidence pertaining to between-college effects in the 1990s than we did for our previous synthesis. We turn now to a summary of those between-college effects. The evidence is summarized within the following categories: two-year versus four-year colleges, college quality, college type, college size, college racial and gender composition, and college environments.

Two-Year Versus Four-Year Colleges

In our 1991 synthesis, we concluded that students who began their postsecondary education at a two-year community college were about 15% less likely to complete a bachelor’s degree during the same time period as similar students who started at a four-year college. Because of this negative impact on bachelor’s degree completion, initial attendance at a two-year college appeared to have a negative, indirect impact on occupational status, and possibly, earnings. However, for community college students who successfully transferred to a four-year college and completed their bachelor’s degree in the same period of time as four-year college students, any negative impact of initial two-year college attendance on occupational status or earnings appeared to be small and perhaps trivial. Similarly, when individuals of equal educational attainment were compared, there was little to indicate that those starting out at two-year colleges were penalized in terms of job stability, unemployment rate, or job satisfaction.

While its findings do not seriously conflict with the conclusions of our 1991 synthesis, the research of the 1990s was clearly more comprehensive and addressed several uncharted areas with respect to the relative impact of initial attendance at a two-year versus a four-year college. It suggests the following conclusions:

1. When precollege ability, motivation, and other confounding influences are taken into account, there is an essential parity between students at two- and four-year colleges in first-year gains made in reading comprehension, mathematics, and critical thinking, and in gains made across two years of college in science reasoning and writing skills.

2. Pat: effects on openness to diversity, and other value, attitude, psychosocial dimensions, and educational aspirations/attainment.

3. Net of other factors, initial attendance at a two- versus a four-year college appears to decrease the likelihood that high-ability minority students will persist in mathematics, science, and engineering careers.

4. Even when educational attainment is taken into account, initially attending a two-year college appears to have a very modest negative effect on subsequent occupational status. However, for similar individuals of equal educational attainment, initial attendance at a two-year college does not appear to confer a significant earnings penalty.

5. Net of other factors, students who initially enroll in a community college are able to transfer to more academically selective four-year institutions (defined by average entering student SAT score) than they would have attended directly out of high school. This effect was most pronounced for students who came from poor families, were of low tested ability, or performed poorly in high school.

College Quality

In our 1991 synthesis, we concluded that the net impact of “college quality” indexes (most typically defined in terms of student body selectivity or prestige/reputation) depended in large measure on the type of outcome being considered. Indexes of institutional quality had less extensive impacts on developmentally-oriented outcomes (learning, cognitive development, values, psychosocial change, and the like) than on socioeconomic outcomes (educational attainment, occupational status, career mobility, earnings, and the like).

In the area of developmentally-oriented outcomes, there was some weak to moderately consistent evidence that college selectivity had a small, positive impact on aesthetic, cultural, and intellectual values; political and social liberalism; and secularism. Beyond these small effects, however, there was little consistent evidence that college selectivity, prestige, or educational resources had any important net impact in such areas as learning, cognitive and intellectual development, the vast majority of psychosocial changes, the development of principled moral reasoning, or shifts in other attitudes and values.

There was clearer evidence to support the importance of institutional quality in one’s career and socioeconomic attainments. Attending a selective or prestigious college had small, positive net effects on educational aspirations, plans for graduate or professional school, choice of an academic career, and choice of sex-atypical careers among women. In addition to influencing aspirations and plans, attendance at an elite institution also modestly enhanced a range of actual socioeconomic attainments. These included bachelor’s degree completion, attending graduate or professional school, entering sex-atypical careers if one is a woman, level of managerial attainment, occupational status in professional careers such as law and medicine, and earnings. The seemingly broad-based impacts of college selectivity/prestige on socioeconomic outcomes, however, were seriously tempered by two facts. First, a substantial portion of the evidence was inconsistent. Second, the magnitude of the effect that could actually be attributed to college quality measures and not other influences was quite small.

As with our previous synthesis, the evidence from the 1990s suggests that college “quality” has less important implications for one’s intellectual and personal growth during college than for one’s career and socioeconomic achievements. Net of important confounding influences, including tested precollege ability or cognitive level, we once again found little consistent support for the contention that attending an academically selective institution enhances one’s knowledge acquisition or general cognitive development in more than very small and statistically chance ways. Similarly, --Pat (values, attitudes, etc.)

Evidence with respect to the impact of college quality measures on socioeconomic-oriented outcomes is clearly more pronounced, though far from unequivocal. The following generalizations appear warranted:

1. Pat (effects on ed. aspirations and ed. attainment and the like here).

2. Attending a selective or prestigious institution appears to have little more than a trivial direct impact on one’s overall occupational status. However, attendance at an institution in the upper one or two percent of the selectivity distribution enhances occupational attainment in specific high status professions such as medicine or law.

3. Selective colleges confer a modest advantage in job attainment and career mobility. However, the advantage appears to be limited to promotion in the middle organization ranks. Graduating from a selective college had little impact at the highest ranks. The evidence from this research further suggests that college selectivity signals the possession of intellectual and related traits that are important for job performance. However, it is not clear if individuals attending selective institutions acquire these traits from their college experience or simply enter college already advantaged in them.

4. The weight of evidence suggests that, in the presence of controls for tested academic ability, socioeconomic origins, undergraduate major, and the like, measures of institutional quality, and particularly student body selectivity, have statistically significant, positive impacts on subsequent earnings. When quality is defined as selectivity, attending a college with a 100 point higher SAT score or ACT equivalent is associated with a net increase of about 2 to 4% higher reported earnings in later life. However, the effect of selectivity may not be linear, and only those elite schools at the very top of the selectivity distribution may confer an appreciable earnings premium. Beyond institutional selectivity, it is difficult to find indexes of institutional quality that have consistent, positive effects on earnings across different studies. Moreover, we suspect that the overall body of evidence probably provides an inflated estimate of the earnings premium attributable to a bachelor’s degree from a selective college. Adequate specifications for individual ambition are almost universally absent in studies of the impact of college quality on earnings; and there is replicated evidence that when ambition is taken into account, along with other factors, the impact of college selectivity on earnings tends to become chance. Thus, it is possible that a substantial part of the net earnings premium attributed to a degree from a selective/prestigious college is really due to the talents of the students they enroll.

College Type

In our previous synthesis, we concluded that estimating the net impact of different college types was complicated by the fact that type itself is often confounded by other institutional characteristics. For example, selective liberal arts colleges tend to combine private control, a liberal arts curricular emphasis, and selective admissions. Thus, it is often difficult to articulate the probable causal mechanisms underlying observed effects. There was, nevertheless, modest evidence to suggest that even when student body selectivity is held constant, private institutions tended to have small positive effects on educational aspirations and educational attainment, and liberal arts colleges tended to enhance the likelihood that women would choose sex-atypical majors and careers. Not surprisingly, liberal arts institutions tended to stimulate gains during college in the value attached to a liberal education and intrinsic occupational rewards, while church-related colleges tended to have negative net effects on gains in secularism and a positive influence on gains in humanitarian and altruistic social values. Beyond these few small impacts, however, institutional categorizations such as the Carnegie classification revealed little difference in between-college impacts.

The evidence from the 1990s parallels that from our previous synthesis in suggesting that institutional type or Carnegie classification are not particularly productive as frameworks for understanding between-college effects. The literature of the 1990s is virtually silent concerning the net impact of institutional type on knowledge acquisition or general cognitive development. However, there was moderately credible evidence to suggest that small, private liberal arts colleges, including those that are religiously affiliated, may be particularly effective in fostering growth in principled moral reasoning.

Pat (College type effects on values, attitudes, ed. attainment, etc.

There is little empirical evidence to suggest that attending a private (versus public) institution has anything more than a trivial net influence on occupational status. Across all studies that provided requisite information, we estimate an average net earnings premium associated with graduating from a private college (irrespective of its level of selectivity) of about 3%. Were differential tuition costs taken into account, however, this earnings premium would in all likelihood be substantially smaller, at least in one’s early career.

When academic selectivity is taken into account, it is questionable that either an institution’s Carnegie classification or its doctoral/research orientation have a consistent impact on one’s earnings. The one exception to this appears to be graduation from a Carnegie-type specialized institution; and this probably occurs because such institutions frequently focus on preparing individuals for occupational fields characterized by high earnings.

College Size

Our 1991 synthesis concluded that the net influence of institutional size (that is enrollment) varied with the outcome considered. On the one hand, attending a large institution tended to have small, positive direct effects on both occupational status and earnings, even when selectivity was taken into account. Conversely, institutional size had small, negative impacts on bachelor’s degree completion, educational attainment, and the development of social self-image during college. These impacts also persisted irrespective of student body selectivity, but tended to be realized through indirect routes. Attending a large institution tended to inhibit a student’s level of social involvement (extracurricular activities, interaction with faculty, and the like) during college, and social involvement was an important determinant of such outcomes as educational attainment and self-concept.

In contrast to the conclusion of our 1991 synthesis, the evidence from the 1990s suggests that institutional size has only a trivial and statistically nonsignificant net impact on occupational status. However, consistent with our previous synthesis, the weight of evidence from the 1990s indicates that attending a large institution confers a small, but statistically significant earnings advantage, even when such factors as student body selectivity and private/public control are taken into account. This positive effect of graduating from a large institution probably stems from economies of scale in providing diverse programs and major fields of study, as well as a wide range of links with occupational and economic groups in society.

Pat (conclusions re: ed attainment values, attitudes, etc.

College Racial and Gender Composition

Although the evidence was not particularly compelling, we concluded in our 1991 synthesis that attending a predominantly Black (versus a predominantly white) college appeared to have a modest positive net impact on the cognitive development and educational attainment of African-American students in general, and a small positive impact on the occupational status and on both the academic and social self-images of African-American women. These effects persisted even in the presence of controls for academic aptitude, socioeconomic origins, educational or occupational aspirations, college selectivity, and the like. Similarly, there was also moderately strong evidence to suggest that single-sex colleges tended to enhance students’ socioeconomic aspirations and career attainments. This was particularly true for women. Net of college selectivity and individual background factors, attending a women’s college appeared to enhance educational aspirations and attainment, choice of sex-atypical (male-dominated) careers, and the achievement of prominence in a field. Graduates of women’s colleges were strongly overrepresented in the high-status, male-dominated occupations of medicine, scientific research, and engineering. However, there was little support for net impacts of women’s colleges on occupational status and earnings in general.

Our present synthesis offers partial support for the conclusions of our previous synthesis. Overall, the evidence indicates that, despite relative disadvantages in financial and educational resources, historically Black colleges appear to be as proficient as primarily White institutions, if not more so, in fostering the knowledge acquisition (e.g., reading comprehension, mathematics, science reasoning) and general cognitive growth (e.g., critical thinking) of African-American students. There was little evidence, however, that women’s colleges had a stronger net impact on either the knowledge acquisition or general cognitive growth of women than did coeducational institutions.

Pat: Impacts of College Race and Sex on Ed Attainment, Values, Attitudes, etc.

Overall, it is difficult to form an unequivocal conclusion from the literature of the 1990s about the impact of attending an historically Black college on African-Americans’ career and economic success. Historically, Black colleges appear to enhance the career aspirations of African-American students; and there is some evidence that a bachelor’s degree from an HBC is linked with one dimension of career eminence among African-American women. However, evidence of the influence of HBCs on African-Americans’ occupational status, career mobility, and earnings is inconsistent and, on occasion, conflicting.

Interestingly, there is consistent evidence indicating that nonAfrican-American students, and particularly men, may derive potential benefits on a racially diverse campus that translate into subsequent earnings advantages. This may be due to the tendency for employers to recognize the importance of, and reward recent graduates’ experiences with, diverse populations; and these experiences are most likely to occur at schools with racially diverse undergraduate student bodies.

There is single-study evidence to suggest that graduating from a women’s institution positively influences both women’s subsequent career eminence and, interestingly, the likelihood of marrying a high-earning spouse. However, the 1990s provided only mixed and inconsistent evidence with respect to the net impact of women’s colleges on the acquisition of career-related skills, the likelihood of choosing nontraditional majors and careers, and workplace participation. Similarly, and consistent with our 1991 synthesis, the overall weight of evidence concerning the net impact of women’s colleges on women’s occupational status and earnings is inconsistent and unconvincing.

College Environments

Our 1991 synthesis led to the following general conclusions about the net impact of institutional environments:

1. Knowledge acquisition on standardized measures such as the Graduate Record Examination was facilitated by institutional environments that stressed frequent student-faculty interaction and curricular flexibility.

2. General cognitive growth on such dimensions as critical thinking and adult reasoning skills was enhanced by a general education emphasis in the curriculum.

3. The environmental factors that maximized persistence and educational attainment included a peer culture in which students developed close on-campus friendships, participated frequently in college-sponsored activities, and perceived their college to be highly concerned about the individual student, as well as an emphasis on support services (including advising, orientation, and individualized general education courses that developed academic survival skills).

4. Decreases in authoritarianism and increases in general psychosocial adjustment and maturity appeared to be maximized by environments which emphasized intrinsic motivations, student involvement in classroom discussions and course decision making, and general student involvement with faculty in an academic community.

5. The most consistent college environmental impact on career choice was that of “progressive conformity.” Progressive conformity hypothesizes that career choice will be influenced in the direction of the dominant peer groups in a college. The weight of evidence tended to support this hypothesis. Irrespective of initial career choice, there was a small but persistent tendency for seniors to plan and subsequently enter careers consistent with the most typical academic majors at their institutions.

The research of the 1990s frequently developed operational definitions of institutional environments that had little in common with those employed in the research underlying our 1991 synthesis. Nevertheless, the following conclusions warranted by our present synthesis have a few commonalities with those of our previous review.

1. Although the specific findings do not replicate those of the research in our previous synthesis, the literature of the 1990s is consistent with previous literature in suggesting that institutional environments may be a more useful approach to understanding between-college effects on learning and cognitive development than a school’s structural characteristics (e.g., selectivity, private/public control, size, single-sex versus coeducational, Carnegie-type, and the like). We found consistent evidence that both learning and general cognitive growth were enhanced by institutional environments with a scholarly/analytical emphasis, and such an emphasis was not simply determined by, or merely a proxy for, institutional selectivity. Replicated evidence also suggests that growth in critical thinking, analytic competencies, and general intellectual development are fostered by college environments that stress close relationships and frequent interaction between faculty and students, and faculty concern about student growth and development during college. Moreover, this environmental emphasis appeared to have an influence on dimensions of general cognitive growth that operated independently of an institution’s scholarly environment. Finally, there is evidence indicating that environments supportive of diversity and relatively free of racial bias may facilitate the learning of students of color.

2. Irrespective of size and selectivity, institutions with honor codes or honor systems that are enforced by students have significantly less anonymous self-reported academic dishonesty than institutions without such honor codes. It may be that such honor codes foster a student culture of academic integrity that functions to reduce cheating behavior; although it is also possible that schools with visible honor codes tend to attract students for whom academic integrity is a higher personal priority when they enter college.

Pat (Your sections here)

3. Consistent with the conclusion of our 1991 synthesis, we once again found strong evidence that most consistent college environmental influence on career choice is that of “progressive conformity.” Irrespective of initial career choice, a student’s major field of study and career choice were influenced in the direction of the dominant academic major(s) at his or her institution.

Some Final Thoughts on Between-College Effects

If one focuses on developmental changes that occur during college (e.g., learning, cognitive growth, moral reasoning, values, psychosocial growth, and the like), even the statistically significant between-college effects in the most methodologically sound studies tend to be quite small and of questionable practical importance. It is entirely possible that the aggregation of structural characteristics (e.g., selectivity, size, private/public control) or even environmental stimuli at the institutional level yields indexes that are simply too remote from the actual social and intellectual forces that shape intellectual and personal growth during college. With some exceptions, such as small homogeneous liberal arts colleges, the vast majority of institutions in the American postsecondary system have important social and intellectual subenvironments and peer cultures with more immediate and powerful impacts on individual students. Because of these powerful subenvironments and subcultures, we should probably anticipate a substantially greater diversity of impacts within than between institutions.

Where an individual attends college appears to play a consistently more visible role in socioeconomic achievement. Clearly, attendance or graduation from a selective or prestigious institution may confer statistically significant advantages in various dimensions of career mobility and earnings. With the possible exception of earnings, however, the magnitude of these advantages is not always consistent or clear; and even if we include earnings, part of what appears to be an institutional effect may still be the result of failing to adjust for ambition, aspiration, and other individual traits that determine socioeconomic achievement. Moreover, there is evidence to suggest that only a relatively small number of the most elite and selective institutions can be counted upon to consistently confer economic and career advantages on their graduates. From the standpoint of incremental improvements in one’s chances for success, where one attends college probably confers an average advantage that is less pronounced and consistent than either the average advantage attributable to obtaining a bachelor’s degree versus a high school diploma, or the average advantage attributable to within-college experiences such as major field of study.

Within-College Effects

Consistent with our 1991 synthesis, the research literature from the 1990s on within-college effects is massive. Not surprisingly, the evidence as a whole is not based on a common set of conceptual or theoretical themes. Consequently, we have organized the synthesis of this large body of research around our own reading of the common threads running through the evidence. First, we offer several observations or conclusions about the evidence as a whole, focusing on commonalities and contrasts with our previous synthesis. Second, we offer our conclusions about the major determinants of within-college effects as organized under the categories of residence, major field of study, the academic experience, interpersonal involvement, extracurricular involvement, and academic achievement.

General Conclusions

The body of evidence from the 1990s on within-college effects has a number of features in common with the literature from our previous synthesis. First, the types of within-college experiences that maximize impact are not totally independent of the kind of college attended. For example, a social environment that facilitates frequent student-faculty informal contact is more likely at small, residential colleges than at large universities with a mix of residential and commuter students. However, there is likely considerable variability in student-faculty contact among small, residential colleges (as well as among other types of institutions for that matter), and nearly all of the important within-college impacts persist irrespective of the institutional context in which they occur.

Second, many of the experiences that maximize impact are not totally independent of the characteristics of the students who engage in them. For example, students who are most likely to engage in diversity experiences during college (e.g., interact with racially diverse peers, attend racial/cultural awareness workshops) are also likely to be most open to diversity when they enter college. The net impact of their involvement in diversity experiences during college would be to even further strengthen their openness to diversity. Thus, many within-college effects are essentially the accentuation of initial student characteristics. Certain experiences tend to attract students with certain traits or dispositions and, in turn, tend to accentuate the traits or dispositions that drew those students to the experiences in the first place.

Third, one of the most unequivocal conclusions we can make from the combined literature of both our previous synthesis and the decade of the 1990s is that the impact of college is largely determined by the individual’s effort and involvement in the academic, interpersonal, and extracurricular opportunities afforded by postsecondary education. Students are not merely the passive recipients of institutional efforts to “educate” or “change” them. Rather, students, themselves, bear a major responsibility for what they take away from the postsecondary experience. This is not to say that individual campus policies, programs, and ethos are unimportant. Indeed, we would once again strongly argue the contrary. If individual effort or engagement is the critical determinant of college impact, then a central question focuses on the ways in which an institution can shape its academic, interpersonal, and extracurricular offerings in ways that encourage increased student engagement.

The research of the 1990s on within-college effects also differs from the research underlying our previous synthesis in a number of respects. Perhaps the most dramatic difference is that the research of the 1990s placed a substantially greater emphasis on estimating the impacts of the academic experience. The 1990s produced a vast body of evidence on different approaches to instruction and to teaching and learning which either did not exist, or was in its nascent stages, when we published How College Affects Students in 1991. Two major themes woven through a large percentage of these new pedagogies are active student engagement in learning and learning in collaboration with faculty and one’s peers.

A second major addition of the literature of the 1990s has been the effort to estimate the impacts of specific extracurricular, interpersonal, or nonacademic experiences. Although based on a body of research that is nowhere near as voluminous as that concerning instructional methods, we nevertheless have a more complete understanding than we did a decade ago about the potential of such activities as fraternity/sorority membership, intercollegiate athletic participation, work during college, and interaction with racially diverse peers.

Pat (anything you wish to add here)

Residence

In our 1991 synthesis, we concluded that living on campus (versus living off campus and commuting to college) was perhaps the single most consistent within-college determinant of impact. Net of important background traits and other confounding influences, on-campus residence had statistically significant, positive impacts on increases in aesthetic, cultural, and intellectual values; a liberalization of social, political, and religious values and attitudes; the development of more positive self-concepts; increases in intellectual orientation, autonomy, and independence; gains in tolerance, empathy, and ability to relate to others; and increases in the use of principled reasoning to judge moral issues. Residing on campus also significantly increased the likelihood of both persistence in college and bachelor’s degree completion, and the positive influence on persistence was further enhanced by on-campus residence in a living-learning center. There was little compelling evidence, however, to suggest that either knowledge acquisition or general cognitive growth were significantly linked to living on or off campus.

It appeared that many of the impacts of living on campus manifest themselves indirectly. Living on campus maximized the opportunities for social, cultural, and extracurricular engagement; and it was, in turn, this increased engagement that largely accounted for the impact of residential living on student change.

Pat: You may have to supply most of 1990s evidence on residence. I can offer the following paragraph.

Consistent with the conclusions from our previous synthesis, we found little consistent evidence to suggest that living on campus (versus living off campus and commuting to college) directly influences either knowledge acquisition or more general cognitive growth. We suspect that living on campus may well exert an indirect positive influence on these outcomes, and particularly on general cognitive growth, by means of its facilitation of students’ academic and social engagement. However, we uncovered no empirical test of this hypothesis.

Major Field of Study

The overall evidence from our 1991 synthesis indicated that one’s undergraduate academic major had its most visible impacts on cognitive and career outcomes. Unsurprisingly, the cognitive impact of major field of study was selective. Students tended to demonstrate the highest levels of learning in subject matter most congruent with their major. Similarly, they tended to demonstrate the greatest proficiency on measures of general cognitive development when confronted with problems or examples taken from the content most typical of their academic major or the disciplinary emphasis of their coursework. Beyond these selective impacts, however, there was little consistent evidence that one’s major had more than a trivial net impact on one’s general intellectual or cognitive growth.

Major field of study had potentially important implications for the occupation one entered and economic rewards received from one’s work. Majoring in areas such as engineering, business, some preprofessional programs, and some natural sciences significantly increased the probability that one would enter a job with skill requirements consistent with one’s academic training; that women would enter high status, male-dominated occupations; and that one would enjoy advantages in early career earnings. The occupational impacts of one’s academic major, however, appeared to be strongest in the early stages of one’s career. As an individual’s career matured, the impact of undergraduate field of study decreased in importance, and was replaced in importance by general intellectual skills and the ability to learn on the job. Similarly, over the long term, career mobility and occupational attainment levels of liberal arts majors in the private business sector appeared to equal (though not exceed) those of business or engineering majors.

The impact of academic major on other outcomes was substantially less apparent. Some evidence indicated that majoring in the natural sciences, mathematics, or technical fields had small positive effects on academic self-concept; but there is little evidence of changes in other areas of psychosocial growth, attitudes or values that were attributable to academic major. The interpersonal climate and value homogeneity within a department may be more important than the structural characteristics of the discipline in shaping student psychosocial and attitudinal changes.

While a substantial part of the evidence from the 1990s on the effects of major field of study is in accord with the conclusions of our 1991 synthesis, other evidence suggests the need to revise some of our previous conclusions. There is also evidence on the impact of college major that is new to the 1990s.

1. Consistent with our previous synthesis, undergraduates tend to make the greatest knowledge gains and attain the highest level of academic skills in subject matter areas consistent with their major field of study and in disciplines where they take the most courses. There is additional evidence that the intellectual training in different fields of study leads to the development of different reasoning skills which is also consistent with our 1991 conclusion that academic major has a selective impact on cognitive growth.

2. As with our 1991 conclusion, we found little consistent evidence to suggest one’s major field of study, in and of itself, leads to different effects on general measures of critical thinking. However, when one considers specific coursework taken, there is replicated evidence indicating that exposure to natural science courses positively influences growth in critical thinking skills. Evidence with respect to other coursework concentrations was either inconclusive or awaits replication.

3. Undergraduate major appears to have a significant net impact on the probability of getting a job and securing employment at a level appropriate to a bachelor’s degree early in one’s career. The clearest advantages in these areas accrue to students majoring in fields having the most direct functional linkages with specific jobs or occupational sectors (e.g., computer science, engineering, social work, nursing, accounting).

4. In contrast to the conclusion from our 1991 synthesis, evidence from the 1990s indicated that academic major did, in fact, have a significant influence on occupational status. Net of other factors, individuals majoring in fields traditionally dominated by men (e.g., engineering, mathematics, physical science, and technical/preprofessional areas) tended to be overrepresented in high status occupations.

5. A small body of evidence challenges our previous conclusion that undergraduate major may have little net impact on career mobility in business. However, it is difficult to form a clear pattern from the evidence. The impact depends on the type of company being considered and its unique cultural norms and values, the particular time in one’s career, the sector of the company in which individuals with certain majors tend to be placed, and the level of promotion or advancement being considered. Consistent with our 1991 synthesis, there was little to suggest that undergraduate major plays a significant independent role in promotion to the highest levels of corporate leadership.

6. Single study evidence suggests that the degree of congruence between one’s academic major and one’s job has a positive influence on job satisfaction that was independent of salary, and similar in magnitude to salary’s impact on job satisfaction.

7. Consistent with our 1991 synthesis, undergraduate major has a substantial impact on subsequent earnings. The evidence from the 1990s, however, permitted one to estimate the magnitude of this impact. With important confounding influences taken into account, there is typically a difference in the earnings of individuals in the most and least lucrative majors of between 25 and 35%. The largest earnings premia continue to accrue to majors characterized by a relatively specific and well-defined body of content knowledge and skills, an emphasis on quantitative or scientific methods of inquiry, a generally close and direct functional link to occupations with relatively high average earnings, often an applied orientation, and a history of being dominated by men (e.g., engineering, business/accounting, physical sciences, mathematics and computer science, and preprofessional majors in health science areas). Although the impact of major appears to be most pronounced early in one’s career, the same general pattern appears to hold later in one’s career.

8. A possible exception to the general pattern of major field of study effects on earnings noted above may occur at particularly selective/prestigious institutions. In these institutions, majoring in the liberal arts may provide an “option value” of potential graduate or professional study that functions to enhance longer-term earnings.

9. Economic returns to associate’s and bachelor’s degrees overlap--largely due to an individual’s major field of study. For example, women can generally get a greater earnings return from an associate’s degree in business or health than from a bachelor’s degree in humanities or education. Men can generally realize a larger economic premium from an associate’s degree in engineering, public service, or vocational/technical areas than from a bachelor’s degree in the humanities or education. At both the baccalaureate and sub-baccalaureate level, however, starting salary and early career earnings are enhanced by the extent to which one’s undergraduate major is related to, or congruent with, his or her job.

10. Generally, men tend to be overrepresented in academic majors that are closely linked to the highest paying occupations, while the opposite tends to be true for women. Moreover, women are more likely than men to enter nonlucrative fields of study even after important background characteristics are taken into account. However, differences in academic major chosen by men and women fail to account for all of the gender gap in earnings among the college-educated.

Pat: Values, attitudes, psychosocial growth here

The Academic Experience

Our 1991 synthesis of research supported five general conclusions about the impact of the academic program.

1. The strongest conclusion was the least surprising. Other things being equal, the greater the student’s involvement or engagement in academic work or in the academic experience of college, the greater his or her level of knowledge acquisition and general cognitive growth. Though less extensive, evidence also indicated that academic engagement enhanced declines in authoritarianism and dogmatism and increases in autonomy and independence, intellectual orientation, and the use of principled moral reasoning.

2. There were instructional and programmatic interventions that not only increased active engagement in learning and academic work but also enhanced knowledge acquisition and some dimensions of both cognitive and psychosocial growth. Instructional strategies such as peer teaching and various individualized learning approaches (e.g., personalized system of instruction, computer-based instruction) are based to a large extent on increasing active engagement in learning, and each appeared to enhance knowledge acquisition under experimental conditions. Similarly, there was also evidence to indicate that inductive learning based on active involvement in concrete activities fostered growth in abstract reasoning, while critical thinking was enhanced by instruction that stressed active student discussion at a relatively high cognitive level and instruction that engaged students in active problem solving.

3. Change in a wide variety of cognitive areas was stimulated by academic experiences that purposefully provided for challenge and/or integration. These academic experiences included cognitive-developmental instruction and curricular experiences that required students to integrate learning from separate courses around a critical theme.

4. Student learning was unambiguously linked to instructor or teacher classroom behaviors. Two dimensions of teacher behavior were most salient in predicting student learning. These were instructor skill (particularly clarity of presentation) and course structure/organization (e.g., class time is structured and organized efficiently). Elements of both instructor behaviors were learnable by faculty.

5. There was tentative evidence suggesting that, irrespective of academic ability, the pattern and sequence of courses taken as an undergraduate influenced not only the subject matter content of what was learned but also more general cognitive abilities. This research, however, was in its initial stages. Thus, the mapping of consistent and replicable findings was unresolved.

The research of the 1990s focusing on the academic experience produced what may be the largest single body of new knowledge we uncovered. The number of well-conducted experimental, quasi-experimental, and correlational investigations addressing the different aspects of a student’s academic experience were so varied and numerous as to defy any tight organizational structure. Nevertheless, we believe the following general conclusions are warranted by this vast body of evidence:

1. The evidence from the 1990s suggests a modest revision in our previous conclusion that subject matter knowledge is acquired with equal proficiency in large as well as small classes. When learning is measured by course grade, the weight of evidence is reasonably clear that class size has a negative impact. When learning is assessed by a standardized measure, however, there is little consistent evidence that class size has a negative influence, at least in the field of economics.

2. Although it is not always clear how the “control” or “traditional” method of instruction is operationally defined, we uncovered reasonably consistent experimental and quasi-experimental evidence that a range of innovative pedagogical approaches appears to improve subject matter learning. These approaches and the estimated average improvement in subject matter learning they produce over “control” or “traditional” methods (expressed as a part of a standard deviation) are listed below:

Pedagogical Approach	Net Effect Size
Learning for mastery	.41 sd to .68 sd
Computer-assisted instruction	.31 sd
Active learning	.25 sd
Collaborative learning	Unclear
Cooperative learning	.51 sd
Small-group learning	.51 sd
Supplemental instruction	.39 sd (compared to nonparticipation)
Constructivist-oriented approaches	.14 sd to .40 sd (based on 2 studies)

3. There are also a number of more focused classroom instructional techniques that appear to be effective tools for enhancing student learning of course content. These techniques include peer tutoring, reciprocal teaching, attributional retraining, concept/knowledge maps, and the one-minute paper. Methodologically sound research on problem-based learning and learning communities is in its nascent stages. However, both approaches demonstrate promise as interventions that foster improved learning.

4. A massive body of evidence suggests that students who study via distance education approaches master course content with about the same level of proficiency as their counterparts in conventional on-campus settings. However, this research is plagued by threats to internal validity inherent in the fact that students, almost without exception, self-select themselves into on-campus and remote instructional sites.

5. Replicated experimental or quasi-experimental evidence indicates that learning a computer program language can provide an advantage of .35 of a standard deviation in general cognitive skills such as planning, reasoning, and metacognition. Similarly, consistent correlational evidence indicates that critical thinking and general reasoning skills are fostered by: coursework requiring students to use computers; using computers in learning activities such as data analysis, creating visual displays, and Internet searches for course material; and using electronic mail to practice argumentation.

6. The clear weight of experimental and quasi-experimental evidence indicates that students learning in cooperative groups acquire an average advantage in problem-solving skills over their counterparts not learning in a cooperative format of .47 of a standard deviation. Moreover, net of other factors, cooperative or group learning experiences appear to positively influence self-reported student growth in career-related skills such as leadership abilities, public speaking ability, ability to influence others, and ability to work effectively in groups.

7. The weight of evidence suggests that critical thinking can be taught. Students receiving purposeful instruction and/or practice in critical thinking and/or problem solving skills gain an average advantage of .23 of a standard deviation over similar students not receiving such instruction. The general absence of a consensus across studies in what constitutes “instruction in critical thinking” may, at least partially, account for the modest magnitude of the effect.

8. Although the total body of evidence is relatively small, and may not be particularly robust, it would appear that growth in postformal reasoning may be facilitated by three loosely related innovative instructional approaches: reflective judgment/developmental instruction, active learning/team problem-solving instruction, and deliberate psychological instruction. On average, students exposed to these instructional approaches gain an advantage in measures of postformal reasoning of about .65 of a standard deviation over similar students not receiving this type of instruction.

9. While not unequivocal, the weight of evidence from both quasi-experimental and correlational research indicates that service-learning experiences enhance both course learning and dimensions of general cognitive development. Quasi-experimental evidence concerning the impact of service-learning on principled moral reasoning is mixed. However, service-learning or service-involvement positively influence related outcomes such as the importance students place on social justice, sense of civic responsibility, and importance of service to the community. In addition, service-learning experiences appear to positively influence such dimensions of career development as self-ratings of leadership skills, the importance of a helping career, occupational identity processing, and salient career development tasks. The most effective service-learning approaches appear to be those that integrate service experiences with course content and provide for reflection about the service experience through discussion or writing.

10. In our previous synthesis, moral development interventions focusing on moral dilemma discussion and personality development were found to be particularly effective in stimulating the use of principled moral reasoning. The role of the instructor in these interventions was more as a facilitator than information provider. Quasi-experimental evidence from the 1990s, however, suggests that growth in the use of principled moral reasoning is even further enhanced when students are exposed to direct instruction in philosophical methods of ethical analysis, as well as to dilemma discussion and personality development.

11. Evidence from two independently conducted longitudinal studies suggests that purposefully integrating ethical content into an undergraduate professional curriculum in nursing may foster growth in principled moral reasoning. Similarly, single-sample evidence indicates that the use of principled reasoning is increased by a general education curriculum that integrates moral and ethical decision making throughout a multidisciplinary six-course sequence.

(Pat: Evidence on instructional effects on values, attitudes, etc.)

12. Generally consistent experimental and quasi-experimental evidence indicates that career development courses or interventions (several of which are computer-based) can significantly enhance dimensions of students’ career development and maturity.

13. Modest quasi-experimental evidence supports the hypothesis that health knowledge and good health habits can be taught during college. Alumni exposed to a one-semester health and physical education course during college that combined classroom instruction with physical activity sessions had significantly higher levels of health knowledge and were significantly more likely to practice good health habits (e.g., diet, exercise, nonsmoking) than alumni not exposed to the course.

14. Consistent with our previous synthesis, a large body of correlational research in the 1990s indicates that differences in teacher behaviors have important consequences for the acquisition of course subject matter by students. Such factors as teacher preparation/organization, teacher clarity and understandableness, teacher availability and helpfulness, quality and frequency of teacher feedback to students, and teacher concern for and rapport with students continued to have significant positive correlations with student mastery of course content. Furthermore, although it awaits replication, there is also correlational evidence, paired with extensive statistical controls, to indicate that extent of teacher organization/preparation in the overall instruction received at an institution positively influences general measures of learning and cognitive development not tied to specific courses. These general measures include reading comprehension and critical thinking.

15. In our previous synthesis, we presented evidence indicating the experimental validation of the effects of teacher clarity on student mastery of course content. The literature of the 1990s presents additional experimental evidence validating the positive impact of both teacher expressiveness/enthusiasm and teacher organization on student content acquisition. What is perhaps most important is that each of the three teacher behaviors shown to facilitate student knowledge acquisition in a course may be learnable by college faculty.

16. Consistent with our previous synthesis, we found substantial evidence indicating that, quite apart from courses taken and instruction received, both knowledge acquisition and general cognitive growth depend in large measure on an individual’s level of academic effort and engagement. Other things being equal, the more a student is psychologically engaged in activities such as use of the library, reading nonassigned books, individual study, writing papers, course assignments, and the like, the greater his or her knowledge acquisition and general intellectual growth. If the literature of the 1990s says anything, it is that colleges can fashion an undergraduate academic experience characterized by a plethora of effective learning opportunities. However, it is the extent to which the student becomes personally engaged in these opportunities and fully exploits them that largely determines the personal benefits derived.

Interpersonal Involvement

In our 1991 synthesis, we concluded that a large part of the impact of college was determined by the extent and content of one’s interactions with major agents of socialization on campus, namely, faculty members and students. The influence of interpersonal interaction with these groups was manifest in intellectual outcomes as well as changes in attitudes, values, aspirations, and a number of psychosocial characteristics. Net of student background characteristics, the extent of informal contact with faculty was positively linked with a wide array of outcomes. These outcomes included perceptions of intellectual growth during college, increases in intellectual orientation, liberalization of social and political values, growth in autonomy and independence, increases in interpersonal skills, gains in general maturity and personal development, educational aspirations and attainment, orientation toward scholarly careers, and women’s interest in and choice of sex-atypical (male-dominated) careers. The most influential interactions appeared to be those that focused on ideas or intellectual matters, thereby extending and reinforcing the intellectual goals of the academic experience.

Interactions with peers also had a strong influence on many aspects of change during college. Included were such areas as intellectual development and intellectual orientation; political, social, and religious liberalism; positive academic and social self-concept; interpersonal skills; principled moral reasoning; general maturity and personal development; and educational aspirations and educational attainment. The degree of peer influence varied across outcomes, with some evidence suggesting that fellow students exerted greater influence on change in attitudinal and psychosocial areas than in learning or cognitive areas, where faculty influence appeared greater. The impact of interaction with peers was greatest when students were exposed to peers who challenged currently held beliefs, attitudes, and values, and who forced introspection, reflection, and reevaluation.

While not as extensive as the evidence supporting our previous synthesis, the evidence of the 1990s further underscores the importance of interaction with faculty and peers as a determinant of growth and change during college. The weight of evidence suggests that student-faculty nonclassroom interactions that tend to reinforce and extend the intellectual ethos of the classroom or formal academic experience, or that tend to focus on issues of student personal growth positively influence dimensions of general cognitive development such as postformal reasoning, analytic ability, and critical thinking skills. Similarly, there is single-sample evidence that out-of-class interaction with faculty has a net positive influence on growth in principled moral reasoning. Consistent with our 1991 conclusions, student-faculty out-of-class interaction may also play a role in students’ careers. For example, student-faculty interaction appears to have a positive influence on the likelihood of students choosing academic and scientific research careers, although there is some ambiguity with respect to causal direction. Specifically, does student-faculty interaction increase the likelihood that one will choose an academic or scientific research career, or are students who have decided on those careers simply more likely to seek out interaction with faculty?

Pat: Evidence re: values, attitudes, etc., and student-faculty contact

The evidence from the 1990s with respect to the importance of interactions with peers is also quite consistent with the evidence from our 1991 synthesis. However, a major additional contribution of the research of the 1990s has been a better understanding of the kinds of peer interactions that are most influential. The interactions with peers that matter most appear to be those that expose the student to diverse racial, cultural, social, value, and intellectual perspectives. Put another way, students derive the greatest developmental benefits from engagement in peer networks that expose them to individuals different from themselves. Net of confounding influences, interactions with diverse peers have modest but consistently positive impacts on knowledge acquisition, dimensions of general cognitive development such as critical thinking and thinking complexity, principled moral reasoning, self-rated job skills, involvement in community service activities in the years following college _ _ _ _ _ _

Pat: Evidence re: values, attitudes, ed attainment.

Extracurricular Involvement

Our 1991 synthesis uncovered only a small body of evidence concerning the net impact of extracurricular involvement. We concluded that, in the presence of controls for important confounding influences, extracurricular involvement had modest, positive effects on institutional persistence and educational attainment, women’s choice of sex-atypical careers, and the development of a positive social self-concept.

The evidence from the 1990s is decidedly more focused on the impacts of several specific types of extracurricular involvement. These include athletic participation, Greek affiliation, and work during college. The total body of evidence is not particularly large, but it does suggest that the impact of involvement in each of these extracurricular experiences is complex. For example, intercollegiate athletic participation, particularly for men in revenue-producing sports such as football and basketball, appears to have an inhibiting influence on both general measures of learning (e.g., reading comprehension and writing skills) and general measures of cognitive development (e.g., critical thinking). However, there was little evidence to suggest that participation in intercollegiate athletics had more than a chance impact on either principled moral reasoning or academic dishonesty; and there was modest support for the contention that athletic participation may actually enhance growth in openness to diversity.

Pat: Evidence on athletic participation and attitudes, values, etc.

The impact of Greek-affiliation is also complex and depends to some extent on the outcome being considered and when it is considered. For example, fraternity membership would appear to inhibit the general knowledge acquisition and critical thinking growth of men during the first year of college. After the first year of college, however, this negative effect essentially disappears. Modest evidence suggests that Greek affiliation, in general, may inhibit growth in principled moral reasoning and increase the likelihood of both academic dishonesty and binge drinking during college. Interestingly, however, the effect of Greek affiliation on binge drinking during college does not extend to the postcollege years. When prior drinking behavior is taken into account, Greek affiliation has little impact on postcollege drinking levels for either men or women. There is modest evidence to support the contention that fraternity or sorority membership can assist in one’s career. Net of other factors, Greek affiliation during college has a positive impact on the development of career-related skills.

Pat: Evidence re: Greeks and values, attitudes, etc.

Given the substantial number of undergraduates who work during college, it is particularly appropriate that the research of the 1990s has begun to estimate the impact of this experience. The small body of evidence we uncovered suggests that the impact of work depends on the outcome being considered. For example, we conclude that there is little compelling evidence to suggest that on-or off-campus work in general has more than a trivial net impact on either general measures of knowledge acquisition or general measures of cognitive growth during college. A small body of evidence indicates that off-campus employment during college has a negative influence on both principled moral reasoning and involvement in community service. However, work or internship experiences during college appear to have a positive net impact on the development of career-related skills and the likelihood of being employed immediately after college, particularly when one’s work experience is related to one’s major field of study. Evidence with respect to the impact of work during college on earnings or earnings growth is inconsistent.

Pat: Evidence re: work and values, attitudes, etc. Also, evidence re: general extracurricular involvement.

Academic Achievement

In our 1991 synthesis, we concluded that academic achievement, as indicated by grades during college, reflects a number of personal traits that had implications for job productivity and success. These included requisite intellectual skills, personal motivation and effort, and the willingness and ability to meet organizational norms. The independent impact of undergraduate grades on indexes of occupational success was modest, but persistent. Net important confounding influences, undergraduate grades had a positive influence on the status or prestige of the job one entered, career, mobility, and earnings. Although we could not accurately estimate its magnitude, part of this effect was direct, but part was also indirect, being transmitted through the strong impact of grades on educational attainment.

With one exception, the evidence of the 1990s supports or extends the conclusions from our previous synthesis. Net of important confounding influences, undergraduate grades appear to have a modest positive impact on the probability of being employed full time and of being employed in a job appropriate to a bachelor’s degree in the early stages of one’s career. On the other hand, the evidence that grades have a causal impact on either job satisfaction or job mobility is unconvincing. The latter finding calls into question the conclusion from our 1991 synthesis. Consistent with our previous synthesis, the evidence from the 1990s indicates that college grades have a positive net impact on both occupational status and earnings; and the form of the evidence permitted us to estimate the magnitude of the impact. We estimate the net total impact of grades on occupational status at between .10 and .20 of a standard deviation. Independent of important background traits and college experiences, including undergraduate major, the net earnings premium associated with an increase in one grade group was estimated at about 6.8%, while the total earnings premium (the net direct effect plus the indirect effect through educational attainment) was estimated at between 8 and 9%. Thus, other things being equal, good undergraduate grades afford an individual modest but discernible career advantages.

Conditional Effects of College

In our 1991 synthesis, we concluded that relatively little attention had been paid to the assessment of conditional effects. However, there were isolated exceptions to this general conclusion. The presence of conditional effects was most pronounced in two areas: learning and cognitive development and the socioeconomic outcomes of college. In the area of learning and cognitive development, there was reasonably strong evidence that certain kinds of students benefitted more from one instructional approach than another. Instruction appeared to interact with both personality traits and level of cognitive development. Students high in need for independent achievement or internal locus of control appeared to learn more when instruction stressed independence, self-direction, and participation. Conversely, students high in the need for conforming or dependent achievement or who had an external locus of control appeared to benefit more from more highly structured, teacher-directed instructional formats.

In terms of socioeconomic outcomes, the clearest conditional effects concerned race. Men of color, or black men in particular, derived somewhat greater incremental occupational status benefits from a bachelor’s degree than did white men. Similarly, between 1970 and 1990 men of color, or black men in particular, appeared to be receiving somewhat greater incremental earnings benefits from a bachelor’s degree than white men. Gender effects were less clear and for private rate of return were conditional on race. Of all groups, women of color appeared to receive the greatest economic return on investment from a bachelor’s degree. In terms of incremental effects on earnings, a bachelor’s degree was probably more valuable to a woman than to a man.

A final set of conditional effects concerned the influence of college selectivity on occupational status and earnings. College selectivity had a positive impact on occupational status in professional careers (e.g., medicine and law) but was of questionable value for business or managerial careers. In terms of economic returns, college selectivity had a stronger impact on earnings for men from relatively high socioeconomic backgrounds than for men from relatively low socioeconomic origins.

The research of the 1990s placed a substantially greater focus on estimating conditional effects than the research summarized in our 1991 synthesis. Accordingly, we uncovered many more statistically significant conditional effects in our current synthesis. These conditional effects appeared to cluster into three general types: 1) student characteristics x net effects of college, 2) student characteristics x between-college effects, and 3) student characteristics x within college effects. Unfortunately, as we indicated in How College Affects Students, specific conditional effects do not always replicate well, particularly in correlational or nonexperimental research. In fact, a clear majority of the larger number of conditional effects we uncovered in the literature of the 1990s are based on a single finding and await replication. Because of their great number, we reasoned that it was impossible to report and describe every statistically significant conditional effect we uncovered in our literature review. In selecting what to report, we placed greatest emphasis on: 1) conditional effects that were replicated, 2) areas of research where there has been a reasonably concerted effort to determine the presence of conditional effects, and 3) conditional effects having either strong theoretical rationales and/or potentially important policy implications.

Student x Net Effects of College

The evidence from the 1990s clearly suggests that the net effects of college differ in magnitude by student sex and race.

1. There is partially replicated evidence that women may derive smaller learning or knowledge acquisition benefits from college than men. In the areas of science reasoning and mathematics, this difference remains even with statistical controls for patterns of coursework taken.

2. The weight of evidence suggested that the net earnings premium from a bachelor’s degree (versus a high school diploma) was about the same for men and women; a finding somewhat at odds with our 1991 conclusion that a bachelor’s degree benefited women more. However, the earnings premium linked to an associate’s degree was about 1.5 times as large for women as for men, and it also appears that women receive larger earnings premia than men from vocational certificates and the completion of vocational training.

3. The advantage in occupational status attributable to a bachelor’s degree (versus a high school diploma) was about 1.7 times as large for men as for women, while the corresponding advantage in occupational status linked to an associate’s degree was about twice as large for men as for women.

4. Single-sample evidence suggests that African-American students may make smaller gains in critical thinking skills through the first and third years of college than do their white counterparts. Similarly, Latino students may derive smaller critical thinking benefits from the first year of college than do white students.

5. Generally consistent with the conclusion from our 1991 synthesis, we estimate the net earnings premium for African-American bachelor’s degree recipients to be about 1.2 times as large as the corresponding premium for white recipients. To some extent, this may be attributable to large returns for African-American women.

6. The intergenerational effect of parental education on the career choices of children may vary by race. Having a mother who was a college graduate had stronger positive effects on African-American men and women majoring in mathematics and science fields than it did for their white counterparts.

Pat: Evidence of conditional effects here

Student x Between-College Effects

Although most of it is based on single-sample findings, there is ample evidence from the 1990s to indicate that the impacts of attending different kinds of postsecondary institutions vary in magnitude for different kinds of students.

1. There is single-sample evidence suggesting that students of color derive larger first-year reading comprehension and mathematics benefits at two-year (versus four-year) colleges, while their white counterparts benefited more on these dimensions of learning from attendance at a four-year college. Similarly, students of color who were relatively older and from low socioeconomic backgrounds benefited more in writing skills from attendance at a two-year college, while white students who were relatively younger and from high socioeconomic backgrounds benefited more in writing skills from four-year college attendance. Thus, the kinds of students appearing to derive the greater learning from attendance at a community college tended to be those students of color, older students, and less affluent students who were most likely to attend a community college (versus a four-year institution) in the first place.

2. Single-sample evidence indicates that women made significantly smaller first-year critical thinking gains at two-year colleges than did men. Conversely, women made significantly larger first-year critical thinking gains at four-year colleges than men.

3. Single-sample evidence suggests that Latino students benefit more in terms of first-year critical thinking gains than other students from an institutional environmental emphasis on being critical, evaluative, and analytical. Conversely, elements of covert discrimination in an institutional environment had a stronger negative effect on critical thinking gains for Latinos than for other students.

4. The evidence from the 1990s indicates that students of color, or African-American students in particular, benefit economically from institutional quality or selectivity about as much as, or perhaps more than, white students. The impact of college quality measures on earnings was about the same for men and women.

5. Somewhat at odds with the conclusion of our previous synthesis, attending a selective college had its strongest positive impact on earnings for students with relatively low academic ability or from relatively low socioeconomic origins.

6. College selectivity may have a positive, net effect on earnings for students who transfer to selective institutions, but may have little or no impact on earnings for students entering selective institutions directly out of high school. Similarly, institutional quality measures may have different impacts on the earnings of different major fields of study. Such evidence suggests that any economic benefits linked to institutional quality or selectivity do not accrue homogeneously to all students. Thus, blanket statements about the impact of institutional quality or selectivity on earnings may be unwarranted.

Pat: Evidence re: values, attitudes, etc.

Student x Within-College Effects

By far, the most numerous conditional effects we uncovered in the literature of the 1990s involved the interaction of student characteristics and various within-college experiences. This body of evidence suggests the following general conclusions.

1. The most consistent evidence of conditional effects was in the area of student learning style, and it essentially extended the findings of our 1991 synthesis into a more general conclusion about matching student learning style to the appropriate instruction or learning experiences. Replicated experimental and quasi-experimental evidence clearly indicates that college students demonstrate significantly higher levels of knowledge acquisition when they receive instruction that matches their preferred learning style than when they receive instruction that does not. Across all studies with college-level samples, we estimate that students receiving instruction matched to their learning style demonstrated an advantage of .91 of a standard deviation over their counterparts who did not receive instruction accommodating their preferred learning style.

2. Although based largely on single-sample findings, the research of the 1990s suggests that many within-college effects vary in magnitude by sex, race, academic ability, and parental education.

Sex: Engaging in volunteer work during college and coursework in the natural sciences and humanities may have stronger positive effects on measures of learning (e.g., reading comprehension) and general cognitive development (e.g., critical thinking) for men than for women. Men, however, appear to incur a greater deficit in first-year critical thinking growth from fraternity membership than do female sorority members. Women may derive greater general cognitive growth (e.g., reflective thinking and critical thinking) than men from work experiences and living on campus during college; and they appear to derive greater economic returns from a bachelor’s degree in engineering, mathematics, or the physical sciences, as well as from good undergraduate grades, irrespective of major. The positive impact on earnings linked to an engineering major (versus other majors) for women was about 1.5 times as large as the impact for men. Similarly, the positive economic impact linked to mathematics/physical science majors for women was about 1.75 times as large as the impact for men. Thus, while women may be less likely to enter these majors than men, they appear to enjoy greater incremental economic advantages when they do.

Race: The positive learning effects of studying with peers and using computers, as well as the positive economic effects of good undergraduate grades appear to be somewhat larger for African-American students than for white students. Students of color may also derive greater knowledge acquisition benefits from cooperative learning approaches than do their white counterparts; and while fraternity membership appears to negatively influence gains in first-year knowledge acquisition and critical thinking for white men, it appears to have little or no impact on the learning and cognitive growth of men of color in fraternities. Conversely, African-American students appear to receive smaller critical thinking benefits from coursework in the natural sciences (versus other majors) than white students and smaller economic returns from majoring in the social sciences (versus other majors) than do other students. Similarly, the positive effects on critical thinking of involvement in diversity experiences (e.g., attending racial/cultural awareness workshops, making friends with someone of a different race) were significantly stronger for white students than for students of color.

Tested Academic Ability: The relative impacts of a variety of within-college experiences also may vary in magnitude for students who enter postsecondary education with different levels of tested academic ability. For example, replicated evidence suggests that high ability students may be able to convert the use of different information technologies (e.g., hypertext and unstructured e-mail use) into greater learning and cognitive gains than students of relatively low ability. Similarly, there is evidence to suggest that high ability students derive greater learning and cognitive benefits from social interaction with peers than do students with lower ability. On the other hand, instructional approaches such as supplemental instruction, knowledge maps, and cooperative learning with knowledge maps may provide significantly larger learning benefits for students with relatively low academic ability or prior performance than for their high ability or high prior performance counterparts.

Parental Education: A small body of single-sample evidence also suggests that certain within-college experiences may be more important for first-generation college students than for students whose parents attended college. For example, first-generation college students derived greater learning (reading comprehension) and general cognitive development (critical thinking) benefits than other students from extent of study effort and full-time enrollment.

3. Single-sample evidence indicates that the positive effects of cooperative learning may be more pronounced for complex levels of cognitive functioning than for less complex levels of cognitive functioning.

4. Although undergraduate grades have a generally positive net effect on subsequent earnings, the magnitude of the effect may vary according to one’s academic major. Grades appear to be of greater consequence for the subsequent earnings of students majoring in business, education, and science/mathematics than for their counterparts majoring in engineering, social sciences, or health-related fields.

Some Final Thoughts

While it has not always produced evidence that differs substantially from that uncovered in our 1991 synthesis, we believe that the research of the 1990s has set in motion a number of new lines of inquiry that are crucial to a more complete and accurate portrayal of the impact of postsecondary education on students. First, the research has begun to seriously acknowledge the rapidly increasing diversity of students in the American postsecondary system. To be sure, there is still a natural tendency to focus inquiry on the students who are easiest to study (e.g., those who attend college full time, who live on campus, and, primarily, who are white). But the literature of the 1990s has also made major strides in providing a better understanding of the impact of the postsecondary experience for individuals who are first-generation college students, who commute to college, who attend college part-time, who work during college, and who are students of color. This is an effort that not only needs to be continued but also expanded, particularly in light of growing evidence that student body diversity in an institution is itself a potentially powerful educational influence on a range of important cognitive and noncognitive outcomes.

In addition to paying greater attention to students who had been largely marginalized in previous research, the decade of the 1990s also acknowledged the existence of institutions, namely community colleges, that had been essentially ignored until then. The impact of community colleges on students is no longer an empirical “black hole.” We have learned much about the impact of such institutions on cognitive, noncognitive, career, and economic outcomes. Given that community colleges and other sub-baccalaureate institutions enroll almost 4 of every 10 students in American postsecondary education, this is clearly another line of research that needs to be continued and expanded.

A second important line of research has been an expanded effort to understand the impact of the academic program, and particularly teaching and instructional approaches. Concern with the impact of classroom experiences was certainly visible in our previous synthesis, but one could reasonably argue that it bloomed into a full fruition during the decade of the 1990s. Compared to evidence available through the end of the 1980s, we now have a much clearer understanding of the cognitive impact of a wide range of different instructional approaches and teaching behaviors. These approaches include active learning, collaborative learning, cooperative learning, small-group learning, supplemental instruction, service-learning, contructivist-oriented approaches, distance learning, attributional retraining, reciprocal teaching, concept maps, teacher expressiveness, and teacher organization. In short, the decade of the 1990s has provided experimental or quasi-experimental validation of an extensive repertoire of approaches to teaching and classroom instruction that postsecondary faculty can call upon to enrich an institution’s learning environment.

A third important contribution of the 1990s has been the initiation of research on the effects of new information technologies. In addition to extending and refining the findings on the cognitive effects of computer-assisted instruction, the 1900s provided initial evidence on the cognitive impacts of hypertext, learning a computer language, computer use in the classrooms, and various unstructured and classroom uses of electronic mail. Given the immense potential of computer and information technologies to fundamentally change the nature of teaching and learning, it is likely and appropriate that this initial inquiry will be the precursor of a substantial body of future research. How computers and information technologies influence students’ cognitive processing, the role of the instructor, the psychosocial climate of teaching and learning, and the extent and nature of a student’s interaction with peers and faculty, may be particularly salient lines of inquiry.

A fourth important contribution of the research of the 1990s has been a pronounced effort to determine the presence of conditional or interaction effects of college on students. To some extent, this may be a response to dramatic increases in the diversity of students entering the American postsecondary system. In a very real sense, a concern with estimating conditional effects is an acknowledgment that student diversity may play an important role in shaping the impact of college. As the undergraduate student population becomes more diverse in terms of race, socioeconomic background, native language, age, academic preparation, and the cultural and intellectual capital they bring to college, we should probably expect that fewer educational experiences will have similar effects for all the students engaged in them. Indeed, the substantial number of statistically reliable conditional effects we uncovered in the literature of the 1990s suggests that this increasing heterogeneity of college effects shaped by individual differences among students may already be well underway. Perhaps the important question to ask in future inquiry about the impact of postsecondary education is not which experiences are most influential, but rather which experiences are most influential with which particular type of student. Of course, conditional effects have the potential to make an explanation more complex and less tidy than general effects--where it is assumed that an experience has the same effect for everyone. But sometimes the explanation closest to the truth is just that--complex and untidy.

A fifth and final contribution of the research of the 1990s has been an increased emphasis on reporting the magnitude of college effects, not just their statistical significance. Obviously there remain many areas of research where we could not estimate an effect size from the findings. However, compared to our previous (1991) review, our ability to estimate the magnitude of an effect from the literature we reviewed in our current synthesis increased substantially. Since statistical significance is so often at the mercy of factors such as sample size and unexplained variance, the reporting of effect sizes is one way to put research findings from different studies on a reasonably common and comparable metric. Having this common and comparable metric not only sharpens the conclusions that one can draw from the findings, it can also play an important role in facilitating policy formulation and the effective allocation of resources. Knowing the estimated magnitude of an intervention’s effect on some outcome is likely to be of greater utility to policy makers than simply knowing that the improvement linked to the intervention was not due to chance. We would argue that, where appropriate, all future research on college impact should make a concerted effort to estimate the magnitude as well as the statistical significance of effects.

Pat: Add anything you want here

In addition to these new or expanded research initiatives in the research of the 1990s, we think that there are several other important considerations for future research. First, we would argue that less relative effort needs to be spent on studies that merely document change or growth during the undergraduate years; and, relatively greater effort should be directed to documenting the net or unique impacts of undergraduate education. Since net effect constitutes that part of change during college that is attributable to the postsecondary experience and not other influences, the net effects question is considerably more difficult to answer than the question of whether or not students change during college. At the same time, there is probably little debate that knowing the value added by an undergraduate education is far more important than simply knowing if student change coincides with their undergraduate years. Yet our documentation of student change that occurs during college substantially exceeds our documentation of the net effects or value-added of college. It behooves the scholarly community that studies college impact on students to mount a concerted effort to correct this situation. Estimating the net effects of postsecondary education should become a high priority direction for future inquiry.

Second, we would argue that, with a few exceptions, the study of between-college effects has not been particularly productive or informative. In the studies with the best research designs and strongest statistical controls for important individual student characteristics, between-college effects tend to be quite small and inconsistent. Increased use of more advanced multilevel statistical procedures such as hierarchical linear modeling may at least partially speak to this problem. However, we expect that a far more significant issue may be the global characteristics of institutions that tend to be studied. Institutional characteristics such as selectivity, private/public, Carnegie-type, average faculty salary, and educational expenditures/students have two major and related problems. First, they tend to mask a great deal of variability in subenvironments and subcultures within institutions; and second, they are simply too distal from the classroom and nonclassroom experiences that shape impact in major ways. Although the analogy is overwrought, the typical study of between-college effects is a little like tossing a boulder into Lake Erie at Cleveland, Ohio, and trying to measure the ripple in Buffalo, New York. If the ripple comes at all, it is unlikely to be very dramatic. This is certainly not to say that there is little in the way of important work to be done on between-college effects. However, in terms of incremental payoff with regard to our understanding of college’s impact on students, a focus on within-college effects will likely yield far greater returns than will between-college effects.

Finally, we reiterate the call we made in 1991 for research on college impact to place a greater dependence on naturalistic and qualitative approaches. To be sure, the decade of the 1990s saw a marked increase in qualitative studies of college effects, and much of this inquiry made important contributions to our understanding. Yet the simple fact is that traditional quantitative approaches continued to almost totally dominate the wide range of literature we reviewed. Certainly, quantitative approaches provide a powerful set of tools for estimating the impact of college on students. But they are perhaps most useful in painting the broad outlines of the portrait. Rendering tone, tint, texture, and nuance may require the finer brush strokes that characterize naturalistic and qualitative approaches. The impact of postsecondary education on students is not only an immense field of study, it is also an extremely complex one. It is unlikely that this complexity will be adequately captured by any single approach to inquiry. If there was ever a time when we needed multiple perspectives to guide inquiry on the impact of college on students, it is now.