The Rise of the Commuter Student: Changing Patterns of College Attendance for Students Living at Home in the United States, 1960–1980
by Dongbin Kim & John L. Rury - 2011
Background/Context: American higher education witnessed rapid expansion during the period between 1960 and 1980, as colleges and universities welcomed millions of new students. During the period, the proportion of 19- and 20-year-old students living in dormitories, rooming houses, or other group quarters fell from more than 40% to slightly less than a third. At the same time, the proportion of students in this age group living at home with one or two parents increased from about 35% to nearly 47%, becoming the largest segment of the entering collegiate population in terms of residential alternatives. During this period, while growing numbers of high school graduates each fall headed off to campus dormitories, even more enrolled in commuter institutions close to home, gaining their initial collegiate experience in circumstances that may not have differed very much from what they had experienced in secondary school. The increased numbers of commuter students, whether they attended two-year or four-year institutions, however, have received little attention from historians and other social scientists.
Purpose/Objective/Research Question/Focus of Study: This study focuses on students aged 19 and 20 who lived with parents and commuted from home during the years from 1960 to 1980, when commuters became the largest category of beginning college students. It also addresses the question of how this large-scale change affected the social and economic profile of commuter students in the United States. In this regard, this study can be considered an evaluation of policy decisions intended to widen access to postsecondary institutions. Did the growing number of students living at home represent a democratic impulse in higher education, a widening of access to include groups of students who had previously been excluded from college? The study approaches this question by examining changes in the characteristics and behavior of commuter students across the country. Recognizing the variation in enrollment rates and other educational indices by state or region, this study also focuses on how the individual behavior at the point of college entry is affected by these and other characteristics of the larger social setting, particularly from a historical perspective.
Research Design: To grasp the larger picture of historical trends in college enrollment during the period of study, particularly in the growth of commuter students, the first part of the study utilizes state-level data and identifies changes in the number of entering college students who were commuters. In the process, descriptive statistics and ordinary least squares regression are used to identify factors associated with the proportion of college students living with their parents across states. In the second stage of analysis, hierarchical generalized linear modeling, utilizing both state- and individual-level data, is used to consider different layers of contextual effects on individual decisions to enroll in college.
Data Collection and Analysis: At the individual level, the principal sources of information are from 1% Integrated Public Use Microdata Samples (IPUMS) for 1960 and 1980. These are individual-level census data that permit consideration of a wide range of variables, including college enrollment. State-level variables are drawn from the published decennial census volumes, from National Center for Education Statistics reports on the number of higher education institutions, and from aggregated IPUMS data.
Conclusions/Recommendations: This study finds that commuter students in the United States appear to have benefited from greater institutional availability, the decline of manufacturing, continued urbanization, and a general expansion of the middle class that occurred across the period in question. It was a time of growth for this sector of the collegiate population. Despite rhetoric about wider access to postsecondary education during the period, however, the nation’s colleges appear to have continued to serve a relatively affluent population, even in commuter institutions. Although making postsecondary institutions accessible to commuter students may have improved access in some circumstances, for most American youth, going to college appears to have remained a solidly middle- and upper-class phenomenon.
American higher education witnessed rapid expansion during the period between 1960 and 1980, as colleges and universities welcomed millions of new students (Schofer & Meyer, 2005). During this period, federal and state governments began to actively implement various policies and programs to expand college access for the individuals who were underrepresented in higher education, including low-income students, women, and African Americans (Karen, 1991). It also was the era of the baby boom; birthrates soared following World War II, boosting enrollments after 1962, when the first cohorts of these students reached the campuses. It was during this time that institutions changed, as more colleges were established, particularly urban two-year and four-year institutions, and existing schools evolved. This represented a new orientation in higher education, which made it possible for more students to commute to campus and to combine work with school, altering the collegiate experience in important ways (Brothers & Hatch, 1972; Kim & Rury, 2007).
According to U.S. Census data, the proportion of 19- and 20-year-old students living in dormitories, rooming houses, or other group quarters fell from more than 40% to slightly less than a third from 1960 to 1980. At the same time, the proportion of students in this age group living at home with one or two parents increased from about 35% to nearly 47%, becoming the largest segment of the entering collegiate population in terms of residential alternatives. While growing numbers of high school graduates each fall headed off to campus dormitories, even more enrolled in commuter institutions close to home, gaining their initial collegiate experience in circumstances that may not have differed very much from what they had experienced in secondary school (Astin, 1982a; Dougherty, 1987; Jencks & Riesman, 1968; Karabel, 1972).1
The increased numbers of commuter students, whether they attended two-year or four-year institutions, however, has received little attention from historians and other social scientists. We address this lack of consideration by investigating the growth of this aspect of postsecondary education and the types of students who came to comprise this category of enrollment. This is a study of shifting patterns of college access, focusing on students aged 19 and 20 who lived with parents and commuted from home during the years from 1960 to 1980, when commuters became the largest category of beginning college students. We also address the issue of how this large-scale change affected the social and economic profile of commuter students in the United States. In this regard, this study can be considered an evaluation of policy decisions intended to widen access to postsecondary institutions. Did the growing number of students living at home represent a democratic impulse in higher education, a widening of access to include groups of students who had previously been excluded from college? We approach this question by examining changes in the characteristics and behavior of commuter students across the country.
College enrollment has often been understood as an individual decision. Yet a number of historical studies have identified variation in enrollment rates and other educational indices by state or region (Goldin & Katz, 1998; Kim & Rury, 2007), suggesting that individual behavior can be understood in the context of larger social and economic forces (Baker & Velez, 1996; Perna & Titus, 2004, 2005). Little research has focused on how the individual behavior at the point of college entry is affected by these and other characteristics of the larger social setting (Goldin & Katz, 1999), particularly from a historical perspective. Instead, previous analyses of college enrollment have generally focused either on how individual characteristics influence student enrollment behavior (with little consideration of contextual effects), or ways that the characteristics of the larger community are related to the aggregated college enrollment rates of the community (e.g., state level analysis) separately, not at the same time (Beattie, 2002; Bishop, 1977; Clowes, Hinkle, & Smart, 1986; Hearn, 1991).
To overcome such limitations, we conduct analyses at both the state and individual levels in studying enrollment during this period. We begin by focusing on the state level to clarify contextual factors that affected changes in the numbers of commuter students. This provides a general overview of national trends in the growth of the commuter student body and ways that state social and economic characteristics influenced commuter enrollments at the time. Geographic regions, geographic aggregations of states, have also been associated with general levels of social and economic development and can be seen as proxies for cultural traditions or outlooks regarding education (Kim & Rury, 2007; Perna & Titus, 2005). Consequently, regional variables are an important consideration in this analysis. By utilizing a multilevel statistical framework that considers state, region, and individual-level factors simultaneously, in the second stage of statistical analysis, we seek to examine the effects of state and region, in addition to individual variables, on commuter student enrollment decisions.
Such an analytical perspective is particularly important in this study, given its focus on commuter students. By definition, after all, students living at home presumably were a group influenced considerably by local social and economic contextual factors in deciding to attend college. Taking this approach, we can offer a more comprehensive understanding of how individual college-going behavior was shaped not only by individual characteristics but also by characteristics of states and regions. For this scope of analysis, therefore, we use multiple sources of data, including individual-level variables drawn from 1% Integrated Public Use Microdata Samples (IPUMS) for 1960 and 1980, state-level variables drawn from the published decennial census volumes and from aggregated IPUMS data, and institutional data from the National Center for Educational Statistics (NCES, 1995).
POSTWAR ENROLLMENT GROWTH AND THE CHANGING COLLEGIATE POPULATION
The decades following World War II witnessed a remarkable expansion of American colleges and universities. Altogether, enrollments grew by 670%, and the number of institutions by 88% between 1940 and 1980 (NCES, 1995). Perhaps just as important, the assortment of institutions became more pronounced as a range of new schools were established. Of particular significance in this regard was the growth of community colleges and four-year urban public institutions, schools that came to be associated with students from less affluent backgrounds than the traditional collegiate population. Their development was part and parcel of a widening of access that had been called for at the start of the period in the Presidents Commission on Higher Education and echoed in the reports of state and regional planning documents in the years to follow (Brumbaugh & Sugg, 1955; Presidents Commission on Higher Education, 1947). The majority of these institutions served students who commuted from home.
Enrollment growth, however, was hardly uniform across the era. The general process of postsecondary expansion can be divided into two periods: before 1960 and after. The first of these saw significant growth among traditional-age college students (those entering immediately or shortly after high school), along with more dramatic enrollment increases associated with soldiers returning from World War II and the Korean War. The years following 1960 presented a different picture. With the arrival of the baby boom generation on campus, enrollments increased for all groups, but especially for women and minority students (Kim & Rury, 2007). It was a time when states, cities, and other government entities undertook massive expansion of postsecondary educational institutions, particularly community colleges and new urban universities, to broaden access (Brint & Karabel, 1986). It was this period, consequently, that produced some dramatic changes in the residential profile of the collegiate population.
In the minds of many contemporaries, most students attending these new and expanding institutions were believed to be relative newcomers to postsecondary education (Goodrich, 1971; Gusfield & Riesman, 1966). Researchers noted that the proportion of youth in the lower income quartile who planned to attend college more than doubled, from 20% to 46%, between the latter 1950s and the mid-1960s. Indeed, much of the increase in enrollments was believed to be due to these students, a majority of whom were the first in their families to attend college (Wegner, 1973). The expansion of federal financial aid in the 1960s and 1970s, especially with the Higher Education Act of 1965, provided substantial support for these developments. Initiated as part of President Johnsons Great Society agenda, the Higher Education Act created need-based grant and loan programs for students with financial hardship. In the years to follow, these programs became widespread in an effort to promote access and equal educational opportunity, and many student aid programs were established at the state level with a coordination of federal level as well (Gladieux, King, & Corrigan, 2005; Leslie & Fife, 1974; St. John, 2003). By the latter 1970s, large numbers of students were receiving this type of assistance to finance their collegiate education. Survey data at the time indicated that lower income students represented a substantial proportion of the collegiate population (Hansen, 1983), and college enrollment rates of high school graduates were significantly higher in 1975 than in 1970, particularly among African Americans (31.5% and 26.0%, respectively), although it is not clear that changes in financial aid had significantly increased access for those lower income or African American students (St. John, 2003).
Of course, not all students attended the same type of postsecondary institution; as the numbers of students grew, a wide variety of colleges beckoned to them. In 1960, so-called junior colleges (later, community colleges) claimed 23% of first-year students. By 1970, they enrolled nearly 30% of all college students, and in 1980, more than 40%, the vast majority in public institutions, which had tripled in number since the latter 1950s (Brint & Karabel, 1986; Dougherty, 1987; Karabel, 1972). This represented a critical shift in postsecondary policy, making it more widely accessible, and the percentage enrolled in these institutions has remained generally consistent up to the present (NCES, 2008).
By and large, the community college was linked in the perceptions of many contemporaries with lower income students, particularly those who could not afford to attend four-year institutions or who could not meet competitive admissions standards. Indeed, a commonplace view was that the two-year institutions helped to cool out working-class students who held unrealistic expectations of success in postsecondary education (Clark, 1960; Goodrich, 1971; Karabel, 1972). The community college was also associated with vocational education, which was typically linked to lower income students as well (Brint & Karabel, 1986; Munday, 1976). Generally speaking, in that case, it was reasonable to conclude that the rapidly rising numbers of students enrolled in two-year institutions represented this relatively new segment of the collegiate population.
In addition to the community colleges, new four-year institutions were built in many urban areas at the time, also drawing a largely commuter student body (Berube, 1978; Gusfield, Kronos, & Mark, 1970; Gusfield & Riesman, 1966). Some, like the University of Illinois at Chicago, the University of Alabama at Birmingham, and the University of California at Riverside, were launched as new institutions, whereas others had existed as teachers colleges or municipal institutions prior to the war and underwent significant expansion. Urban colleges and universities, both private and public, had been a feature of the educational landscape for decades, but the numbers of students attending such institutions increased significantly, eventually encompassing nearly half of the nations collegiate population (Berube, 1978; Klotsche, 1966).2 The expansion of these schools represented a significant policy shift, and substantial investment in making postsecondary education accessible. As Arnold Grobman (1988) noted, such institutions had been established in virtually all the nations principal metropolitan areas, with the vast majority being state supported. Their clientele reportedly included students from lower social and economic status backgrounds, as well as adults and growing numbers of women (Grobman, 1988). One campus leader characterized the images associated with these changes as such:
The nature of the student body influences the character of many urban universities. Streetcar college, subway university, and blue shirt institution convey a not always accurate description of institutions located in big cities. The commuting student who is half in and half out, half at college and half at home is common among undergrads. (Klotsche, 1966, p. 17)
The public discourse featured growing concern about the academic quality of urban institutions and the students attending them. This was probably most clearly evident in the case of the City University of New York (CUNY), which was established as a system in 1961. Comprising more than a dozen campuses and including both four- and two-year institutions, and free to New York residents, CUNY came to represent the nations most accessible model of urban public higher education. In 1970, a policy of open admissions led to a rapid rise in enrollments, although the numbers declined after 1975, when the system began charging tuition for the first time in its history. Thousands of students were required to take remedial courses upon enrolling, and the first-year dropout rate increased from less than 20% to more than a third in the early 1970s (Meyers, 1997; Wasser, 1973) Once again, the image connected with these developments was increased numbers of students from lower income backgrounds, most first-generation collegians, and many ill prepared for the demands of advanced study. In the case of CUNY, large numbers were from racial and ethnic minority groups, reflecting the changing characteristics of the urban population. And virtually all of them were commuters.
By the middle of the 1970s, urban public higher education had become widely associated with this new wave of students entering the collegiate population. Although the CUNY experience was somewhat unique, it figured prominently in national discussions about the challenges of making college accessible in the nations larger cities. (Marshak, 1975; Meyerson, 1975). As enrollments increased at urban institutions, including both four- and two-year schools, traditional images of college life began to shift as well. An urban university president assessed these changes in sweeping terms: In the last 40 years American education has changed significantly and dramatically. It has become less elitist, more democratic, more accessible, and less homogenous by paying more attention to non-traditional students (Shalala, 1985, p. 5). College was no longer considered an exclusive enclave of the well-to-do. Fewer students lived on campus and organized their lives around school, and a more complex picture of higher education began to emerge.
One reflection of this is evident in the statistics regarding students living at home. By 1980, these commuter students became the largest segment of students entering the postsecondary system (those younger than 20). At the same time, the proportion of students attending full time declined, according to census data, from 68% in 1960 to 59% in 1980, and it appears that growing numbers were combining school with work. The share of students employed or looking for work rose by one quarter, from 40% to 50%. This shift in the numbers of part-time, working, and commuter students represented a noteworthy change in the way that postsecondary education was planned and experienced (Kim & Rury, 2007).
The new orientation is captured in the dates framing this study. In 1960, the first of the two time points, White males were the predominant group of college students, and the vast majority of beginning collegians had been born before the end of the Second World War. As such, it was still the prebaby boom era in higher education. It also was a time when commuter students living at home represented slightly more than a third of the entering collegiate population. In all these respects, 1960 represented an earlier era in college enrollment patterns, a profile of attendance that might be described as historically more traditional than the one seen in the years to follow. The circumstances of American society were somewhat different in 1960 as well. At that time, some 23% of Americans lived in poverty, and nearly a third of American children resided in poor households (Hernandez, 1997). Given that higher education was then a largely middle- and upper-class prerogative, this placed a natural limit on enrollment levels.
The second point of time, 1980, represented an entirely different context. Nineteen- and twenty-year-olds at that time had been born at the height of the baby boom and benefited from the remarkable expansion of institutional capacity that characterized the previous two decades. They completed high school at a time when graduation rates approached historic peaks and when college attendance expectations were at an unprecedented high. In the same time frame, household poverty rates dropped in half, and more Americans joined the middle class. This added substantially to the numbers of potential college students (Danziger & Gottschalk, 2005). Also at this time, there is evidence that larger numbers of young adults were continuing to live at home, particularly as the job market became increasingly uncertain during the 1970s and early 1980s (Buchmann, 1989). But higher education continued to grow. Although it is true that the wage premium associated with college education had declined in the latter 1970s, it was still higher than it had been in 1950 and 1960 (Freeman, 1976; Goldin & Katz, 2008). Enrollment growth slowed in the 1970s, but the only group to register a reduced rate of expansion was White males, who had the highest levels of participation historically. Women and minority students enrolled in record numbers, with women outnumbering men by 1976 (Geiger, 1980; Goldin, Katz, & Kuziemko, 2006). It is hardly an exaggeration to state that it was a new stage in the history of American higher education. In the discussion to follow, we assess the impact of these changes on the expanding commuter student population.
ANALYTICAL METHOD AND DATA
As indicated earlier, we approach the question of individual and contextual factors influencing commuter enrollment patterns during this period in two stages of analysis. To grasp the larger picture of historical trends in college enrollment during the period of study, particularly in the growth of commuter students, in the first part of the study, we utilize state-level data to identify changes in the number of entering college students who were commuters. To do this, we employ descriptive statistics and ordinary least squares (OLS) regression to examine factors associated with the proportion of college students living with their parents. Descriptive statistics provide information on the changes by state and region. OLS regression is utilized to examine the effects of state or regional characteristics on growth in the numbers of college students commuting from home. In the second stage of analysis, we conduct a multilevel examination of enrollment patterns, utilizing both state- and individual-level data. A list of variables with descriptive information is provided in Appendixes A and B.
State-level variables are drawn from the published decennial census volumes, from NCES reports on the number of higher education institutions, and from aggregated IPUMS data. These sources have allowed us to identify a range of factors related to the social, economic, and educational condition of each state across the period in question. We use this information to examine how characteristics of states may have influenced the college enrollment decisions of students residing at home in these years.
The dependent variable in the state-level analysis is the 1980 number of college students in each state living with either a mother or father and currently attending school, expressed as a fraction of the states overall collegiate population. We have adjusted the latter figure for estimates of net student migration to account for students who were studying in other states and out-of-state students living on or around campuses. We focus on the enrollment rates of commuter students in 1980, controlling for enrollment levels in 1960. This approach is commonly employed to account for unobserved factors influencing the outcome variable to better assess the effect of other independent factors in the analysis (Wooldridge, 2002).3
Independent variables apart from the 1960 control variable include median personal income, the percentage of urban population, and the proportion of the adult labor force in manufacturing in 1970. Increases in the total number of higher education institutions and community colleges between 1960 and 1980 were also included in the analysis, along with the number of private institutions in each state. As indicated earlier, regional differences have long characterized American education, and the postsecondary sector in particular (Kim & Rury, 2007; Perna & Titus, 2005). To clarify whether region was a factor in patterns of commuter college enrollment, we utilize dummy variables to consider the impact of regional differences.
In the second stage of analysis, we utilize hierarchical generalized liner modeling (HGLM) to permit consideration of different layers of contextual effects on individual decisions to enroll in college. College enrollment in this analysis is measured as a binary variable representing whether students living with parents were enrolled in college. As noted, enrollment is not only influenced by individual variables, but also by the socioeconomic and educational characteristics of the larger community where students resided. For instance, individuals who live in the same state were affected by the economic circumstances of their immediate region and therefore may have been more similar to one another regarding enrollment than to individuals in other parts of the country. For purposes of our analysis, two separate analyses have been conducted for both 1960 and 1980, enabling us to identify similarities and differences in the factors that affected college enrollment of commuter students at each point in time.4
Multiple levels of data are required for HGLM analysis, individual and state level in this instance. At the individual level, the principal sources of information on college enrollment and a wide range of student characteristics are 1% IPUMS for 1960 and 1980 (Ruggles et al., 2008). These are individual-level census data that permit consideration of a wide range of variables, including college enrollment and whether a student was living on campus, at home, with her or his parents, or in other circumstances. As indicated, we have chosen to focus on 19- and 20-year-old individuals because these were the ages at which most youth entered college. Preliminary analysis for the years analyzed in this study indicated that the peak age of postsecondary enrollment was 21. Younger age groups, for which rates were climbing, therefore can be seen as representing points of initial access. Given that the primary goal of this study is to clarify how patterns of commuter enrollment shifted in response to a range of factors, focusing on a single age group allows us to compare behavior over time.
As a practical matter, the data for this study provide information on students parental and family background variables (such as home ownership and parental education or occupation) only for the individuals who lived at home. Prior research has consistently shown that students from high-income families or those with parents who had been to college were more likely to enroll than their counterparts from lower income backgrounds or without parental college experience (Bishop, 1977; Clowes et al., 1986; Hearn, 1991). By focusing on those who lived at home, we can estimate whether these and other socioeconomic background variables had similar effects on commuter student enrollment. This contributes to our understanding of who commuter students were and how they may have differed from the larger collegiate population.
Individual-level variables include gender, race (Black compared with non-Black), ethnicity (Hispanic), metropolitan status of residence (rural, central city, and suburb as compared with unknown), home ownership, and employment status (whether the individual is employed). Women, African Americans, and members of the working class are sometimes referred to as subordinate groups that historically have been underrepresented in higher education (Karen, 1991). It is therefore important to consider gender, race, and family socioeconomic background, insofar as characteristics have influenced commuter students enrollment behavior. Home ownership provides a proxy of family wealth, measured by whether the individuals household owned a residence and, if so, its value. It is of particular interest because home equity was taken into account for purposes of federal financial aid during this time and thus may have influenced college access (Long, 2007). Individuals whose families owned a relatively lower priced house (below the 50th percentile) and those with a relatively higher priced one (above the 50th percentile) are compared with those who did not live in a family-owned residence.
Apart from family wealth, household income (poverty) is also included in the analysis even though the two variables are somewhat interrelated. This variable measures a different aspect of well-being; it provides information on family income scaled to household size. This too was considered in federal financial aid formulas at the time. We have categorized the IPUMS household income variable into five categories based on the federal definition of poverty for each year. The high-income category is the reference point (above 400% of poverty level), with the variables in the analysis representing poverty (100% or below poverty level), low income (101% 200% of poverty level), lower middle income (201%300% of poverty level), and upper middle income (301%400% of poverty level). The lower/middle income category is of particular interest because it represents a group often associated with commuter college students in research from this period (Chickering, 1974; Klotsche, 1966).5 Parental educational attainment is a binary variable indicating whether at least one parent ever attended college, and parental occupation status is another binary factor indicating whether at least one was employed in a white-collar occupation (as defined by the 1950 census occupational classification code). Previous research generally supports the presumption that both parental education and occupation played a vital role in college access (Goldin et al., 2006; Goldrick-Rab & Pfeffer, 2009; Karen, 1991).6
At the state level, the HGLM analysis includes various economic, educational, and geospatial variables. Economists suggest that educational investment can differ by state or region because of the availability of jobs, wage levels, and monetary and nonmonetary returns to education (Goldin & Katz, 2008; Mitchener & McLean, 1999). Engels (1975), for instance, has described how such factors affected patterns of supply and demand for college-trained workers during the 1970s. Therefore, variables related to state economic development and region are included in the analysis. In addition, the number of higher education institutions, estimates of net student migration, and degree of urbanization are included as well. The same set of variables is used for both 1960 and 1980.
THE GROWING COMMUTER STUDENT POPULATION: A REGIONAL AND STATE-LEVEL ANALYSIS
As noted, one of the notable developments of this period of expansion in the nations collegiate population was the growing number of students living at home. Although there had been a significant number of such students dating to earlier decades in the 20th century, the numbers increased substantially in the two decades following 1960. Table 1 provides regional and state proportions of overall college enrollment of 19- and 20-year-olds represented by students living at home from weighted state-level U.S. census data. These figures have been adjusted for net migration rates of college students to better represent the state collegiate populations from which commuter students were drawn. The prevalence of these college students living at home in both years was highest in the North (defined to combine New England, Middle Atlantic, and East North Central States), the South, and the Pacific states along the nations West Coast. Two of these regions were characterized by a high degree of urbanization and relatively high levels of income. The North, which historically exported students to other parts of the country because of its immense population base, served a large number of commuter students across the period. This was also true of the Pacific states, which featured a large number of community colleges and other commuter institutions, especially in California. This reflected the effort to build a comprehensive system of postsecondary institutions in the so-called California Master Plan (Douglas, 2000). The Southwest also exhibited relatively high numbers of commuter students, and its population was highly urbanized as well. In other parts of the country, the picture was considerably different. The South, Plains (West North Central), and Mountain regions all exhibited relatively low proportions of college students living at home in 1960. These regions were characterized by somewhat lower levels of urbanization. Among these areas, the Mountain states had greater urbanization, but they also were a significant importer of students from other states, which may have created competition for space in the regions institutions for potential commuter students (Mitchener & McLean, 1999; Rury, 2004).
Source: IPUMS data, 19- and 20-year-old college students, adjusted for net migration.
The region that experienced the greatest growth in the proportion of entering college students living at home during this period was the South. Its rate of increase in the proportion of all college students who fell into this category was an astounding 68%, vaulting the South to third among all regions, right at the national mean (which reflected the densely populated North and Pacific regions) in 1980. The South was the only region to move up in the regional standings, displacing the Southwest, which dropped from third to fourth place, its commuter population having grown just 29% in relative size. The North and Pacific regions remained the leading areas of the country for commuting students, with nearly half or more of all beginning students living at home. But by 1980, the South had nearly caught up with the North in this respect. In no region was the proportion of beginning students represented by commuters less than a third. Thus, although growth in this segment of the collegiate population was substantial nearly everywhere, it was most notable in the South. Most other regions had a growth rate between 27% and 34% during this time.
Interestingly, the regions with the least growth in the number of institutions designed to serve these students, as reported in Table 2, also exhibited the smallest increases in commuter students. As noted earlier, the community college was widely associated with the growth of commuter students at this time. The region with the slowest growth in community colleges was the Plains states, which managed to post just a 17% increase in commuter students as a portion of all college enrollments, well behind all other major regions of the country. Although this is prima facie evidence of the importance of community colleges as a factor in the expansion of this segment of the collegiate population, it should not be taken to mean that it was by any means the most critical element.
Note: Rates and figures compiled from data in Education Directory, 196061, U.S. Department of Health, Education, and Welfare, and Digest of Education Statistics (1980), U.S. Department of Education.
There are always questions of scale and proportion in making comparisons among states and regions, and the regional figures in Table 1 are weighted by state population size (1980 high school graduates). Consequently, patterns of growth should be interpreted carefully. The relatively slow expansion of the commuter segment of the collegiate population in the Pacific states, for instance, was partly a function of the regions high starting point, given the long-standing leadership of California in establishing commuter campuses. Despite their relatively modest rate of change during the 1960s and 1970s, the Pacific states led the nation in overall numbers of commuter students, and large numbers of community colleges were established there, serving a predominantly commuter population (Brint & Karabel, 1986; Douglas, 2000).
Other parts of the country witnessed more dramatic increases in the proportion of students who commuted and in the numbers of community colleges, but this was partly a function of low starting points, as in the case of the South. These trends are evident in Table 2, which provides figures for institutional growth in each of the regions discussed herein. In the North and South, the number of community colleges roughly doubled during the period in question, and the proportion of commuter students increased by more than 30%. But the growth in institutions was greatest in the Mountain states, where change in the proportion of commuter students was rather modest. Thus, it is difficult to draw an immediate connection between the growth of commuter students and institutional growth, and particularly the development of the community college.
To identify factors affecting the increase in commuter students at the state level, we conducted an OLS regression, the findings of which are presented in Table 3. The dependent variable in this analysis is the proportion of each states collegiate population living at home with one or more parents, with adjustments made for student migration patterns. Results are reported as standardized regression coefficients expressed in standard deviation units (z scores). The first model is the baseline, which features commuter enrollment levels in 1960, defined in the same manner as the dependent variable, as a control factor, along with regional dummies. The coefficient for 1960 enrollments is quite robust and positive, clearly a good predictor of enrollment levels 20 years later, indicating that states with large commuter student populations prior to this period continued to feature them two decades later.
Table 3. OLS Regression, State-Level Predictors of Enrollment Rates
Dependent variable: Percentage of college students living at home in 1980 (adjusted for migration)
Weighted least squares regression (by 1980 State High School Pop), standardized coefficients (betas)
*** significant at .001 level. ** significant at .01 level. * significant at .05 level.
The coefficients for the three regions with the highest proportions of commuter students in 1980 (South, North, and Pacific states) are positive. Interestingly, however, effects associated with the South and Pacific are robust and statistically significant. As noted, the Pacific states exhibited the highest proportion of commuter students, and the South experienced the greatest growth in their numbers. Although the commuter student population in the North was historically large as well, both its growth rate and overall level at the end of the period were near the national average. Altogether, this baseline model explains more than 70% of the state-level variance in 1980 college enrollment levels of commuter students.
The next model introduces a number of social and economic factors from 1970. These state-level variables represent a period when change was occurring across the country, the time when factors such as these were likely to influence changing college enrollment rates for commuter students. The most important variable is the level of urbanization. This is not surprising given that commuters were associated with large population centers, where they could choose from local institutions to attend. Manufacturing employment historically has been associated with lower school enrollment rates (Rury, 2004), but the sign on this variable is positive, and it is not significant in the second model. The 1970s was a period of employment decline in the industrial sector of the economy, and it is possible that communities with large manufacturing centers realized enrollment gains later as displaced workers returned to school or as the sons and daughters of factory workers looked to colleges as an alternative to blue-collar employment. This would correspond to the working-class image of many commuter students from this era (Chickering, 1974; Grobman, 1988). State median personal income was not statistically associated with growth in commuter student enrollments.
In the second model, the value of the 1960 commuter rate factor was substantially reduced, suggesting that the 1970 social and economic factorsand urbanization in particularaccounted for some of the variation it had represented in the baseline model. On the other hand, the South and Pacific dummy variables were generally unaffected by the inclusion of these variables, indicating that even with such controls, these regions would have had a greater proportion of commuter students in their college populations than other parts of the country.
The final model in Table 3 adds several factors associated with the institutional configuration of higher education opportunity in each state at the end of the 1970s. The first two represent growth in the number of institutions, and the third is related to the representation of private institutions in 1980. The first of the institutional variables is the number of new postsecondary institutions that appeared in each state during this time, and the second is the rate of growth in community colleges. Interestingly, the number of new postsecondary institutions did not have a statistically significant effect on commuter enrollment, although the sign on the coefficient is positive. The effect of the community college growth variable, on the other hand, is positive, significant, and modestly robust. This finding suggests that states experiencing large increases in the number of community colleges also experienced increases in enrollment of commuters. In other words, substantial numbers of commuter students during the period appear to have attended new two-year institutions. This finding is hardly a surprise, but it confirms tendencies discussed earlier regarding regional differences in institutional change and enrollment growth. Finally, the private institutions variable, representing the proportion of each states total four-year institutions that were private in 1980, failed to reach statistical significance, although its sign was negative. This points to, but does not confirm, the idea that more students lived away from home in states with many private institutions.
Altogether, the 11 variables in the final model in Table 3 explain more than 80% of the variance in state levels of commuter enrollment. The 1970 urbanization variable was one of the significant variables, and as indicated earlier, commuter students wereand aremost commonplace in urban settings. The positive effect of community college growth was largely independent of this, suggesting that these institutions served students in a variety of settings. The growth of two-year institutions also appears to have accounted for the positive effect of the Pacific regional dummy variable, net of the 1960 commuter control factor. But in the final model, the new community colleges accounted for a relatively small amount of the state-level variation in 1980 commuter enrollments. Urbanization and prior enrollment patterns appear to have been more important.7
Beyond that, it is particularly worth noting that the South experienced unparalleled growth in its commuter student population during this time. It was a regional development substantially independent of a wide range of social, economic, and institutional factors that shaped the development of the nations commuter student body. As we have noted elsewhere (Kim & Rury, 2007), much of the South experienced the nations most rapid expansion of overall college enrollments during this time. As indicated in Table 2, however, the regions rate of institutional growth was only slightly above the national average. This suggests that existing Southern institutions must have grown at an unusually rapid pace, especially those serving commuter students. In any case, this is an unexpected finding; the images and major controversies linked to commuter students during this period were largely associated with the Northeast and California (Berube, 1978; Meyerson, 1975). Although recently there has been new interest in the history of higher education in the South (Michael, 2001; Wells, 2001), the story of this expansion in commuter student participation remains to be told.
HGLM ANALYSIS: COLLEGE ENROLLMENT PATTERNS FOR COMMUTER STUDENTS IN 1960 AND 1980
As indicated earlier, the second stage of the study considers the multilayered characteristics of college enrollment behavior that is influenced not only by individual characteristics but also by the characteristics of the state where the individuals are located. This is accomplished by conducting separate cross-sectional HGLM analyses for 1960 and 1980. The model is identical for both years, factors have been defined in the same manner for each to facilitate comparisons, and the dependent variable is college enrollmentor notof individuals living with their parents.8 The results from the 1960 and 1980 HGLM analyses are presented in Table 4. Only variables that have statistically significant effects on college enrollment are reported with odds ratios for better interpretability. Odds ratios represent how the likelihood of attending college (versus not attending) shifts when a unit changes in a given independent variable, all things being constant. An odds ratio greater than 1 indicates greater odds of attending college, whereas an odds ratio less than 1 indicates decreased likelihood of enrollment (Desjardins, 2001).
*** significant at .001 level. ** significant at .01 level. * significant at .05 level. significant at .10 level.
CIRCUMSTANCES AT THE START OF THE SIXTIES
The general profile of commuter students in 1960 revealed in this analysis presents few surprises. Age was a significant predictor of commuter student enrollment, indicating that individuals who were 20-year-olds had lower odds of attending college, about 77% that of individuals who were 19. This indicates that many of the 20-year-olds in the sample either may have dropped out of college or had already graduated from two-year institutions. It also suggests that among commuters, delayed college entry (taking time off before college) was not commonplace at this time. Female youth were significantly less likely to attend college than their male counterparts, all things being equal. Given what we know about overall trends in college enrollment at this time, this finding is hardly a revelation. On the other hand, being Black was not a significant factor in predicting college enrollment among commuter students in 1960 once other factors were controlled. This finding contradicts our earlier study (Kim & Rury, 2007), which found that Black high school graduates, regardless of gender, had significantly lower college enrollment rates than their White counterparts in 1960. However, these results suggest that lower college enrollment rates for Blacks were largely due to such other factors as family income, parental education, or holding a job. This was particularly true for individuals living at home; their employment rate was 47%. Having a Hispanic parent also does not seem to have affected the likelihood of an individual attending college while living at home. The odds for individuals who lived in a central city (122%) or suburb (53%) were substantially greater than individuals in the comparison group. On the other hand, the odds of attending college for the students living outside metropolitan areas were not statistically different from the comparison group.9 These results are consistent with the earlier finding that enrollments of commuter students were higher in urban areas. As expected, for individuals who were employed, the odds of enrolling in college were merely .182 (about 20%) that of individuals not employed, and this difference is statistically significant. For many youth living at home in 1960, going to college and holding a job appear to have been mutually exclusive activities.
Other factors in the model are consistent with prior research on college enrollment for the general college-age population. For example, the traditional indicators of social status are also significant predictors of commuter students college enrollment in 1960. The odds of enrolling in college were about 2.57 times higher for individuals with at least one parent who had attended college than for those whose parents had never been to college, and those with at least one parent holding a white-collar occupation had odds of enrolling in college that are nearly twice (odds ratio of 1.906) as high as counterparts whose parents did not enjoy this level of status. Home ownership (and value) was an indicator of family wealth, distinct from household income, and also was an important predictor of enrollment for commuter students in 1960. If an individuals parents owned a relatively higher priced house (above the 50th percentile in 1960), their odds of enrolling in college were about 75% higher than individuals whose parents did not own a home. Interestingly, however, if the house value was below the 50th percentile, an individuals odds of attending college were not significantly different from those whose parents were not homeowners. These results are generally consistent with the findings of previous studies of educational attainment (Jackson & Weatherby, 1975; Mare, 1981; Portes & Wilson, 1976; Stafford, Lundstedt, & Lynn, 1984; K. L. Wilson & Portes, 1975).
An important exception to this general pattern concerned the question of household income. As noted, individuals in the sample have been divided into five categories on this dimension of social status, with those at the highest level (above 400% of the poverty level) being the comparison group. As indicated in Table 3, youth residing in households in the poverty (below 100% of the poverty level) and low income (101 to 200% of the poverty level) categories exhibited a lower likelihood of college attendance than their high-income counterparts, all things being considered. However, those in the middle income group (two to three times the poverty level) actually had slightly higher odds of being a student than those in the highest income category, a difference significant at the .10 level. This, of course, conformed to popular images of the commuter student being of slightly lower social status than college students living on campus or in group quarters such as fraternities and sororities (Chickering, 1974). It is likely that among youth living at home, those from highest income households may have had reasons for staying there other than going to college, whereas many youth from the middle income group probably intended to minimize education-related expenses. Consequently, at this time, it was broadly accurate to think of the commuter student as being somewhat more blue collar than the traditional collegiate population.
At the state and regional levels, there appears to have been little in the social and economic environment that affected the behavior of commuter students in 1960. Logit coefficients (log odds) and standard errors for the continuous (nonbinary) variables in this stage of the analysis are expressed in standard units (z scores) for greater comparability, and odds ratios are calculated with the original data reported in Appendix B. Among the level 2 predictors that represent these conditions across states, only the percentage of manufacturing employment was significantly related to the odds ratio of commuter students enrolling in college. A decrease in manufacturing employment of 1% increased the odds of enrolling in college for students living at home by about 1.2%, so commuter students who lived in states with lower manufacturing employment had greater odds of enrolling in college. This suggests that for some students living at home, factory employment offered an alternative to college. It is consistent with the finding at the individual level that employment was negatively related to enrollment at this time. In 1960, states with high levels of manufacturing employment provided powerful disincentives to college enrollment for youth living at home. Having a job, particularly one with relatively good wages in a factory, meant that a young man or woman was unlikely to also be a college student.
A NEW SITUATION AT THE END OF THE SEVENTIES
As suggested earlier, the two decades following 1960 produced a number of important changes in American higher education, and the effects of several of these changes are evident in the results reported in Table 3. To start, by 1980, age was no longer a significant predictor of commuter students college enrollment behavior, indicating relatively little difference in the enrollment rates of 19- and 20-year-olds at that time. This may have been due to the wide range of institutions that commuter students attended, or simply because many high school graduates took time off before they went to college, or students took longer to complete their education. More important, the gender variable remained significant, but the sign changed from negative to positive, indicating that there was an advantage in commuter college enrollment associated with being female. Female youth in the sample had odds of enrolling in college nearly 50% greater (odds ratio = 1.471) than their male counterparts, all other things being equal. This finding corresponds with changes in national patterns, which saw a dramatic rise in college enrollment among women, such that they constituted a majority of college students in 1980 (Jacobs, 1996; Karen, 1991; Kim & Rury, 2007). It appears that this change occurred among commuter students as well.
In this light, it also is worth noting that race (specifically, being Black) was an important negative predictor of commuter student enrollment in 1980. Black students who lived with their parents had odds of attending college that were about .892 (about 89%) that of their non-Black (largely White) counterparts. In other words, non-Black individuals had odds of enrolling in college that were 11% (inverse odds ratio of .892) higher than Black youth. This was a change from the patterns observed in 1960 and something of a surprise because Black postsecondary enrollments had increased at a more rapid rate than non-Black enrollments during this period. The more pronounced increase in college enrollment rates for Blacks, however, did not necessarily mean that they had a higher college enrollment rate than non-Blacks among those living at home.
This is rather unexpected because of the widespread impression that African Americans were a substantial portion of the new students attending urban institutions in these years. All things being constant, Black youth living with their parents were less likely to attend college than Whites in similar circumstances at a time of improving equity indicators in higher education (Astin, 1982b). This may have been linked to changes in the characteristics of Black youth living at home not attending college. As indicated in Appendix A, a considerably larger portion of the at-home population in this age range was African American in 1980 as compared with 20 years earlier (10% vs. 6%), and most of these individuals were not attending college. It is possible that rising unemployment may have contributed to larger numbers of Blacks in this age range remaining in their parents homes, pulling their commuter enrollment rate downward. Evidence for this can be found in lower levels of labor force participation among Black youth; the IPUMS data reveal a 12% drop in the proportion of urban Black 19- and 20-year-olds living at home who were employed in this period (non-Black employment rates dropped half as much). Social scientists have tied this trend to the movement of industrial jobs out of urban areas and attendant increases in poverty, a process that was beginning to accelerate in the 1970s (W. J. Wilson, 1987). Rising poverty levels, of course, could also have affected enrollment rates, although the figures in Table 4 are net of these factors at the individual level. Given these developments, the somewhat lower odds of Black commuter college enrollment are a bit less remarkable and were perhaps a portent of even more dramatic relative decline in the decade to follow (Lang, 1992).
Residential and employment-related factors remained significant, but the effects of these variables changed somewhat between 1960 and 1980. The suburban residence variable was no longer a significant predictor of college enrollment, but the central city variable continued to be a positive and significant factor, indicating that individuals who lived in the central city areas had odds of enrolling in college that were 25% higher than their counterparts who lived in an unknown or unspecified residence area. This, of course, is consistent with the rise of urban institutions during this period. By 1980, the nonmetro residence variable had become a significant negative predictor, indicating that students living outside large urbanized areas had odds of being enrolled in college about 80% that of students in the comparison group. With respect to employment, youth living at home who had a job were substantially less likely to be enrolled in college, but the effect of this factor was considerably smaller at the latter date. This finding suggests that although working and college were still somewhat mutually exclusive, more youth living at home in 1980 were attempting to combine employment and study than in 1960.
Variables associated with socioeconomic status that were significantly linked to college enrollment for commuters in 1960 continued to be important 20 years later: parental occupation, parental education, income, and wealth (as indicated by home ownership and value). But there were telling shifts from one period to the next, especially concerning the economic profile of the commuter student population. Enrollment patterns for youth from poverty and low-income households were similar to those 20 years earlier, but youth representing the lower/middle-class income category had a lower likelihood of attending college in 1980 than their counterparts in the higher income comparison group, a reversal of the situation in 1960. This represents an important change; the relationship between enrollment and income as defined by these four categories, net of other factors considered, shifted from a slightly skewed bell-shaped or peaked distribution to a generally linear one.
This finding, of course, stands in stark contrast to the popular images of the era. The group with the highest propensity to commute to college from home was not the lower or middle-class students; it was those from the higher income category. This suggests that individuals from this latter group in 1980 stayed at home and attended college, different from the previous generation from this income category, who apparently preferred to attend residential four-year campuses. It also is possible that a decline occurred in the enrollment propensity of lower/middle-class youth, but this seems unlikely. As Roger Geiger (1980) has demonstrated, college-going slowed for White males in the 1970s, but it was a pattern evident for both high- and low-income groups. This finding of relatively high commuter enrollment among more affluent students probably reflected a changing social and educational environment wherein commuter institutions were increasingly acceptable as an alternative to residential campuses, even for higher status groups. This was a telling change from the situation at the start of the period. We revisit this issue in the conclusion.
At the state and regional level in 1980, a number of factors were significant, confirming the patterns found at the state-level analysis. Again, logit coefficients and standard errors for the continuous (nonbinary) variables are expressed in standardized terms. The level of manufacturing employment is insignificant, indicating that it did not act as an impediment to college enrollment for the youth living at home at that time, a finding broadly consistent with the positive sign on the manufacturing variable in the OLS analysis earlier. On the other hand, the total number of higher education institutions (including junior and four-year institutions), which was not a significant predictor in 1960 analysis, became significant in 1980. Nineteen- and twenty-year-olds living at home in states with 100 more institutions had odds of enrolling in college about 10% higher than their counterparts living in a state with 100 fewer institutions, holding other factors constant. Although there was a high correlation between the total number of institutions in 1960 and 1980 (r = .979, p < .000), this was a significant factor only for 1980 because of the wider range of institutions numerically across states in 1980. It appears that institutional availability in 1980 clearly influenced commuter students by providing more options to choose from. The negative association between the percentage of out-of-state students and commuter enrollment rates (odds ratio .066, p < .01) suggests that states with more student migration may have put a greater emphasis on traditional residential campuses, which could have contributed to lower commuter college enrollment rates.
In view of the state-level analysis discussed earlier, it is particularly noteworthy that the South regional variable in the level 2 analysis for 1980 is associated with greater odds of commuter college enrollment. This indicates that 19- and 20-year-olds living at home in the South were more likely to be enrolled than their counterparts who lived elsewhere, holding all other factors in the model constant. Specifically, the odds of being enrolled in college are 17% higher for the youth living in the South than their counterparts. This suggests that the South had higher commuter college enrollment levels than even the nation-leading Pacific region if all other individual- and state-level factors are assumed to be equal. This remarkable independent effect of region in college access at this time parallels the findings of the state-level OLS analysis conducted earlier and has not received attention from researchers.10 Yet it seems clear that the South became a distinctive part of the country with respect to the new plurality of commuter students.
CONCLUSION: COMMUTER STUDENTS IN A TIME OF CHANGE
In certain respects, this analysis has demonstrated the value of performing a multilevel analysis of a phenomenon such as college enrollment decisions at different points of time. As suggested by the state-level analysis presented in Table 3, many important changes in enrollment patterns for commuting students between 1960 and 1980 were readily evident at the state or regional level. Aggregate statistics reveal that these were years of massive change in the propensity of American youth to attend college while living at home with one or both parents. Their likelihood of doing this, however, varied substantially from one part of the country to another, depending on levels of urbanization, changes in manufacturing employment, and the availability of institutional options. Beyond that, there is the question of the South, which witnessed a process of change in the behavior of these commuting students even more dramatic than other parts of the country, independent of the factors that seem to have affected it elsewhere. This was a phenomenon that should be the object of further study.
At the individual level a basic status attainment model appears to have accounted for most of the variation in the likelihood of 19- and 20-year-olds living at home attending college. Those who lived with educated parents who were in high-status occupations and able to afford expensive homes exhibited a much higher tendency to enroll in college than their counterparts whose parents lacked college experience and who did not have high-status jobs or lived in less expensive homes. Across the period in question, the proportion of American adults over age 30 living in poverty fell appreciably (from 17% in 1960 to 10% in 1980), and the number of homeowners increased (from 66% to 74%); this doubtless contributed to rising enrollment levels. But fundamental status distinctions continued to be critical in determining who was most likely to go to college, even among the growing numbers of students who commuted from home. As indicated earlier, the slightly higher propensity for individuals in the lower middle class category to attend as commuter students in 1960 had changed by the end of this period. By 1980, there appears to have been a broadly linear relationship between household income and the likelihood of college attendance among students living at home, with the greatest odds evident for those in the highest income category. This result suggests that commuter institutions may have become something of an alternative to traditional residential campuses, even for the youth with higher status backgrounds during the latter period.
Of course, as indicated in Appendix A, greater numbers of young people were living with their parents in 1980 than in 1960, especially among the well-to-do. According to IPUMS data, the overall percentage of 19- and 20-year-olds living with parents, including those not in school, increased from 44% to 51% during this period. The results of the HGLM analysis suggest that many of these individuals were staying at home to attend school, but it also may have been the other way around. In an analysis that considers the same time frame, 1960 to 1980, Marlis Buchmann (1989) has suggested that many youth in the United States came to experience a postadolescence life stage characterized by prolonged economic reliance on their parents. Buchmann and others have argued that the growing uncertainty of the switch from school to work has contributed to the development of an extended period of dependency characterized by overlapping episodes of education, work, and training (Sanders & Becker, 1994). Buchmann also maintained that this pattern of destandardization in the transition to adulthood is most characteristic of youth from higher social class backgrounds whose families have the resources to support them. Thus, it is possible that at least a portion of the 19- and 20-year-olds in the 1980 sample living at home were attending college because they had no better options, at least with respect to work or vocation. For this segment of the new affluent commuter college student population, enrollment may have been part of a new indeterminate life stage, one of biding time with general preparation until a promising career direction appeared.
Along with the massive shift in student residential patterns, it appears that the statistical profile of individual commuter college students in the United States changed somewhat during this period. More of them were female and lived in central cities, and they were a bit older in 1980 than they had been 20 years earlier. It seems that students in some states may have had fewer options to enroll where there were large numbers of out-of-state students, a circumstance that may have reflected a form of institutional crowding. In addition to these changes, the effects of social and economic status variables were even more important in 1980 as they had been in 1960. From the standpoint of making college more accessible to those from the low-income, low educational attainment, and low occupational status categories of the national status hierarchy, little appears to have been accomplished during the period in question. Although some commentators suggested that greater numbers of working class youth attended college in these years, this analysis suggests that such students did not represent a greater fraction of commuter students in 1980 than 1960; in fact, they may have become an even smaller group in comparative terms (Wegner, 1973). This also appears to have been true for African American and Hispanic students, groups widely associated in the popular mind with the new urban institutions that expanded during the 1960s and 1970s. At the same time that larger numbers of American youth attended postsecondary institutions, and a growing number of them lived at home, the social and economic circumstances of being a commuter student appear to have changed rather little, at least with respect to widening access in this age group.
In the end, commuter students in the United States appear to have benefited from greater institutional availability, the decline of manufacturing, continued urbanization, and a general expansion of the middle class that occurred across the period in question. It was a time of growth for this sector of the collegiate population, and despite rhetoric about wider access to postsecondary education, the nations colleges appear to have continued to serve a relatively affluent population, even the commuter institutions. Although making postsecondary institutions accessible to commuter students may have improved access in some circumstances, for most American youth, going to college appears to have remained a solidly middle- and upper-class phenomenon.
We would like to acknowledge the assistance of Argun Saatcioglu, Editor Lynn Corno, and the anonymous reviewers who responded to successive drafts of this paper. Any problems that remain are the responsibility of the authors.
1. Researchers have raised the question of whether commuter students have the same opportunities for sustained interaction with diverse peers and other extracurricular activities enjoyed by their counterparts living on campus. Pascarella, Pierson, Wolniak, and Terenzini (2004) found a strong positive effect of living on campus in developmental terms, especially for White students. Focusing on the experiences of students from racial and ethnic minority groups, Renn and Arnold (2003) questioned the significance of this effect for all students. Most observers agree, however, that living at home typically provides fewer of the developmentally significant experiences that colleges can offer, especially for younger students (Pascarella et al., 2004).
2. This is a phenomenon that has not been systematically studied. Of 80 U.S. member institutions of the Coalition of Urban and Metropolitan Universities in 2007, all but four were state institutions transformed into universities (typically from state teachers colleges) or established during the postwar era. Although this group includes just a sampling of such institutions, its membership is broadly representative of general trends. Interestingly, more than 30% of the institutional membership of this organization is located in the South. See http://www.cumuonline.org/about/index.htm for additional information.
3. As Wooldridge (2002) noted, this approach is an intuitively appealing method of examining the impact of specified independent factors on a process of change (p. 66). Estimates for student migration are calculated from Gossman, Nobbe, Patricelli, and Schmid (1968) for 1960 and Petersen and Smith (1978) for 1980. Net migration rates for all college students are reported for 1958 in the former and for freshman in 1975 in the latter. The 1958 rates were adjusted to correspond to the freshman population in the Gossman et al estimates.
4. Given that the dependent variable in this analysis is binary, it is necessary to use hierarchical generalized linear modeling (HGLM), an extension of the generalized linear model (GLM). HGLM generates estimates of how various factors predict the probability of dichotomous outcome measures. In HLM/HGLM, observations are not assumed to be independent of each other as in OLS regression. Rather, the analysis takes into account the nested structure of the data (i.e., students are located within states, wherein each states unique characteristics influence students educational experience, and hence, college enrollment). In this manner, HLM/HGLM clarifies the extent to which the individual enrollment decision is influenced by individual and state contextual variables.
A distinction between unit-specific versus population-average models arises in the HGLM statistical framework. A unit-specific model provides estimated log odds for all level 1 and level 2 predictors with a random effect of zero and is useful to analyze the different effects of level 1 and 2 predictors across level 2 units (in our study, states). The population average model provides an estimate for all level 1 and level 2 predictors that are closer to the true population mean and is useful to examine the average predicted probability for all populations (Rumberger & Palardy, 2004). Given that this study is concerned with the unique effect of each states socioeconomic and education characteristics on individuals college enrollment behavior, the unit-specific estimates with robust standard error, rather than the population average estimates, are reported. In the HGLM statistical framework, only the measure of the intercept at student level is assumed to vary across states, and the effects of all student-level predictors are fixed to be the same for all states (Raudenbush, Bryk, & Congdon, 2000). For student-level predictors that are binary, no-centering technique was used. As a result, the intercept term may be interpreted as the adjusted college enrollment rate (or estimated college enrollment rate) for 19- and 20-year-old White males whose household did not own a house, whose residential areas were not known, who lived in a family below poverty level, who were not employed, and whose parents did not attend college and had a blue-collar job.
5. Although house value is a continuous variable that indicates the actual dollar amount of the house value, significant portions of the respondents did not own the house (e.g., 41.8% from 1960 and 36.0% from 1980). Therefore, by dividing the variable into three categories, we sought to maximize the number of cases in the analysis and to identify nonlinear relationships between house value and college enrollment. Additionally, we also divided the poverty variable into five categories to identify nonlinear relationships between income and college enrollment rates. Converting continuous factors to categorical variables can also help to reduce collinearity.
6. Research has consistently shown that parental education and occupational status have telling effects on students college enrollment behavior (Bishop, 1998; Goldrick-Rab & Pfeffer, 2009; Levine & Nidiffer, 1996; Nuñez & Cuccaro-Alamin, 1998). Instead of combining parental education and occupation into one variable, this study used two separate binary measures described in the text. This approach is consistent with prior research (Goldrick-Rab & Pfeffer) and also helps to distinguish the effects of these factors.
7. Defining these variables in this manner limits the level of multicollinearity in the model, which often is a problem in OLS regression with aggregate-level data such as these. Analyses were conducted with different types and combinations of these factors, and results did not differ substantially. Additionally, it should be noted that overall adult education levels in each state, a general measure of educational development, were used in other analyses and proved to be highly correlated with personal income. Substituting these factors did not change the results appreciably.
8. Many large national data sets, including IPUMS, use intricate sampling techniques that include overrepresenting of certain populations. Because each case is selected with a different probability and represents a different number of individuals in the population, it is important to use appropriate sampling weights to produce unbiased estimates (Lehtonen & Pahkinen, 1995). We employed relative weights, dividing the IPUMS sampling weight variable by its average in the sample to preserve the sample size for statistical testing. Thus, the total sample sizes were 11,921 youth for 1960 and 44,142 youth for 1980 in 48 states (excluding the District of Columbia, Alaska, and Hawaii). We are grateful to the IPUMS program and the University of Minnesota Population Center for making these data available.
9. The comparison group comprises individuals whose metropolitan locations have not been specified in each IPUMS sample. They have been classified in this manner because of concerns about the possibility of identification and other reasons, explanations of which can be found at the IPUMS Web site: http://usa.ipums.org/usa-action/variableGroups.do. Using this as a comparison group allows us to consider the independent effects of the other metropolitan categories in the IPUMS data. This groups characteristics in both years were quite similar to the general population characteristics of the entire sample. In 1960, it represented nearly 19% (18.9) of the sample, and in 1980, nearly 16% (15.8). Analyses run with nonmetro as a reference group produced similar results for the urban (central city) and suburban (metro, non central city) portions of the samples for each year.
10. Charles Clotfelter (2004) has noted the rapid expansion of two-year colleges in two Southern states at this time, Florida and Texas. Also see Katsinas (1994) for a case study of one Southern states expansion of two-year institutions.
Astin, A. W. (1982a). Contradictions in American higher education. Proceedings of the American Philosophical Society, 126(1), 610.
Astin , A. W. (1982b). Minorities in American higher education. San Francisco: Jossey-Bass.
Baker, T. L., & Velez, W. (1996). Access to and opportunity in higher education: A review. Sociology of Education, Special Issue, 82101.
Beattie, I. R. (2002). Are all adolescent econometricians created equal? Racial, class, and gender differences in college enrollment. Sociology of Education, 75, 1943.
Berube, M. (1978). The urban university in America. Westport, CT: Greenwood Press.
Bishop, J. (1977). The effect of public policies on the demand for higher education. Journal of Human Resources, 12, 285307.
Brint, S., & Karabel, J. (1986). The diverted dream: Community colleges and the promise of educational opportunity in America, 19001985. New York: Oxford University Press.
Brothers, J., & Hatch, S. (1972). Residence and student life: A sociological inquiry into residence in higher education. London: Tavistock.
Brumbaugh, A. J., & Sugg, R. S., Jr. (1955). Recent developments in state and regional planning of higher education. Annals of the American Academy of Political and Social Science, 301, 3240.
Buchmann, M. (1989). The script of life in modern society: Entry into adulthood in a changing world. Chicago: University of Chicago Press.
Chickering, A. W. (1974). Commuting versus resident students: Overcoming the educational inequities of living off campus. San Francisco: Jossey-Bass.
Clark, B. R. (1960). The cooling-out function in higher education. American Journal of Sociology, 65, 569576.
Clotfelter, C. (2004). After Brown: The rise and retreat of school desegregation. Princeton, NJ: Princeton University Press.
Clowes, D. A., Hinkle, D. E., & Smart , J. C. (1986). Enrollment patterns in postsecondary education, 19611982. Journal of Higher Education, 57, 121133.
Danziger, S., & Gottschalk, P. (2005). Diverging fortunes: Trends in poverty and inequality. In R. Farley & J. Haaga (Eds.), The American people: Census 2000 (pp. 4975). New York: Russell Sage.
Desjardins, S. L. (2001). A comment on interpreting odds-ratios when logistic regression coefficients are negative. AIR Professional File, 81.
Dougherty, K. (1987). The effects of community colleges: Aid or hindrance to socioeconomic attainment. Sociology of Education, 60, 86103.
Douglas, J. A. (2000). The California ideal and American higher education: From 1850 to the 1960 Master Plan. Stanford, CA: Stanford University Press.
Engels, R. A. (1975). A framework for assessing higher education supply and demand results for the southern region: 197080. Journal of Higher Education, 46, 287301.
Freeman, R. B. (1976). The overeducated American. New York: Academic Press.
Geiger, R. (1980). The changing demand for higher education in the seventies: Adaptations within three national systems. Higher Education, 9, 255276.
Gladieux, L. E., King, J. E., & Corrigan, M. E. (2005). The federal government and higher education. In P. G. Altbach, R. O. Berdahl, & P. J. Gumport (Eds.), American higher education in the twenty-first century (2nd ed., pp. 163196). Baltimore: Johns Hopkins University Press.
Goldin, C., & Katz, L. (1998). The origin of state level differences in the public provision of high education, 18901940. American Economic Review, 88, 303308.
Goldin, C., & Katz, L. F. (1999). The shaping of higher education: The formative years in the United States. Journal of Economic Perspectives, 13(1), 3738.
Goldin, C., & Katz, L. F. (2008). The race between education and technology. Cambridge, MA: Harvard University Press.
Goldin, C., Katz, L., & Kuziemko, I. (2006). The homecoming of American college women: The closing of the college gender gap. Journal of Economic Perspectives 20(4), 133156.
Goldrick-Rab, S., & Pfeffer, F. T. (2009). Beyond access: Explaining socioeconomic differences in college transfer. Sociology of Education, 82(2), 101125.
Goodrich, A. (1971). Education for the new student at community colleges: A legacy and a hope. Peabody Journal of Education, 48, 286293.
Gossman, C. S., Nobbe, C. E., Patricelli, T. J., & Schmid, C. F. (1968) Migration of college and university students in the United States. Seattle: University of Washington Press.
Grobman, A. (1988) Urban state universities: An unfinished national agenda. New York: Praeger.
Gusfield, J., Kronos, S., & Mark, H. (1970) The urban context and higher education: A delineation of issues. Journal of Higher Education, 41, 2943.
Gusfield, J., & Riesman, D. (1966). Academic standards and the two cultures in the context of a new state college. School Review, 74, 95116.
Hansen, W. L. (1983). Impact of student financial aid on access. Proceedings of the Academy of Political Science, 35(2) 8496.
Hearn, J. C. (1991). Academic and non-academic influences on college destinations of 1980 high school graduates. Sociology of Education, 64, 158171.
Hernandez, D. J. (1997). Poverty trends. In G. J. Duncan & J. Brooks-Gunn (Eds.), The consequences of growing up poor (pp. 1834). New York: Russell Sage.
Jackson, G. A., & Weatherby, G. B.(1975). Individual demand for higher education: A review and analysis of recent empirical studies. Journal of Higher Education, 46, 647650.
Jacobs, J. (1996). Gender inequality and higher education. Annual Review of Sociology, 22, 154156.
Jencks, C., & Riesman, D. (1968). The Academic revolution. New York: Doubleday.
Karabel, J. (1972). Community colleges and social stratification. Harvard Educational Review, 42, 1218.
Karen, D. (1991). The politics of class, race, and gender: Access to higher education in the United States, 1960-1986. American Journal of Education, 99, 208237.
Katsinas, S. G. (1994). George C. Wallace and the founding of Alabamas public two-year colleges. Journal of Higher Education, 65, 447472.
Kim, D., & Rury, J. (2007). The changing profile of college access: Enrollment patterns in the postwar era. History of Education Quarterly, 47, 302327.
Klotsche, J. M. (1966). The urban university: And the future of our cities. New York: Harper & Row.
Lang, M. (1992) Barriers to Blacks educational achievement in higher education: A statistical and conceptual review. Journal of Black Studies, 22, 510522.
Lehtonen, R., & Pahkinen, E. J. (1995). Practical methods for design and analysis of complex surveys. New York: Wiley.
Leslie, L. L., & Fife, J. D. (1974). The college student grant study: The enrollment and attendance impacts of student grant and scholarship programs. Journal of Higher Education, 45, 651671.
Levine, A., & Nidiffer, J. (1996). Beating the odds: How the poor get to college. San Francisco: Jossey-Bass.
Long, B. T. (2007). Do loans increase college access and choice? Examining the introduction of universal student loans (Working Paper No. 07-1). Boston: New England Public Policy Center.
Mare, R. D. (1981). Trends in schooling: Demography, performance and organization. Annals of the American Academy of Political and Social Science, 453, 99100.
Marshak, R. D. (1975). Problems and prospects of an urban public university. Daedalus, 104(1), 192201.
Meyers, B. (1997). The CUNY wars. Social Text, 51, 119130.
Meyerson, M. (1975). After a decade of the levelers in higher education: Reinforcing quality while maintaining mass education. Daedalus, 104(1), 304321.
Michael, D. (2001) Lessons in progress: State universities and progressivism in the New South, 18801920. Champaign: University of Illinois Press.
Mitchener, K. J., & McLean, I. W. (1999). U.S. regional growth and convergence, 18801980. Journal of Economic History, 59, 1016-042.
Munday, L. A. (1976). College access for nontraditional students. Journal of Higher Education, 47, 681699.
National Center for Education Statistics. (1995). Digest of educational statistics, 1995. Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement.
National Center for Education Statistics. (2008). Enrollment in postsecondary institutions, fall 2006; graduation rates, 2000 & 2003 cohorts; and financial statistics, fiscal year 2006. Washington, DC: Institute for Educational Sciences.
Nuñez, A., & Cuccaro-Alamin, S. (1998). First-generation students: Undergraduates whose parents never enrolled in postsecondary education (NCES 98-082). Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement.
Pascarella, E. T., Pierson, C. T., Wolniak, G. C., & Terenzini, P. T. (2004). First-generation college students: Additional evidence on college experiences and outcomes. Journal of Higher Education, 75, 249284.
Perna, L. W., & Titus, M. (2004). Understanding differences in the choice of college attended: The role of state public policies. Review of Higher Education, 27, 501525.
Perna, L., & Titus, M. A. (2005). The relationship between parental involvement as social capital and college enrollment: An examination of racial/ethnic group differences. Journal of Higher Education, 76, 485518.
Petersen, R. J., & Smith, C. R. (1978). Migration of college students: Preliminary analysis of trends in college migration. Washington, DC: National Center for Educational Statistics.
Portes, A., & Wilson, K. L. (1976). Black-White differences in educational attainment. American Sociological Review, 41, 414431.
Presidents Commission on Higher Education. (1947). Higher education for American democracy: A report of the Presidents Commission for Higher Education (Vol. 1, Ch. II). New York: Harper & Brothers.
Raudenbush, S., Bryk, T., & Congdon, R. (2000). HLM 6 hierarchical linear and nonlinear modeling. Lincolnwood, IL: Scientific Software International.
Renn, K. A., & Arnold, K. D. (2003). Reconceptualizing research on college student peer culture. Journal of Higher Education, 74, 261291.
Ruggles, S., Sobek, S. Alexander, T., Fitch, C. A., Goeken, R., Hall, P. K. et al. (2008). Integrated Public Use Microdata Series: Version 4.0 [Machine-readable database]. Minneapolis: Minnesota Population Center.
Rumberger R. W., & Palardy, G. J. (2004). Multilevel models for school effectiveness research. In D. Kaplan (Ed.), Handbook of quantitative methodology for the social sciences (pp. 235258). Thousand Oaks, CA: Sage.
Rury, J. (2004). Social capital and secondary schooling: Inter-urban differences in American teenage enrollment rates in 1950. American Journal of Education, 110, 293320.
Sanders, K., & Becker, H. A. (1994). The transition from education to work and social independence: A comparison between the United States, the Netherlands, West Germany, and the United Kingdom. European Sociological Review, 10, 135154.
Schofer, E., & Meyer, J. W. (2005). The worldwide expansion of higher education in the twentieth century. American Sociological Review, 70, 898920.
Shalala, D. (1985). Nontraditional students in higher education. Higher Education and National Affairs, 34(21), 5.
Stafford, K. L., Lundstedt, S. B., & Lynn, A. D. (1984). Social and economic factors affecting participation in higher education. Journal of Higher Education, 55, 602606.
St. John, E. P. (2003). Refinancing the college dream. Baltimore: Johns Hopkins University Press.
Wasser, H. (1973). An American university and universal higher education: The open admissions system at C.U.N.Y. Higher Education, 2, 151159.
Wegner, E. L. (1973). The effects of upward mobility: A study of working-status college students. Sociology of Education, 46, 263279.
Wells, A. (2001). Contested ground: Howard Odum, the Southern Agrarians, and the emerging university in the South during the 1930s. History of Higher Education Annual, 21, 79101.
Wilson, K. L., & Portes, A. (1975). The educational attainment process: Results from a national sample. American Journal of Sociology, 81, 343363.
Wilson, W. J. (1987). The truly disadvantaged: The inner city, the underclass, and public policy. Chicago: University of Chicago Press.
Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press.
Operational Definitions and Proportions of Individual-Level Variables in the HGLM Analysis
Operational Definitions and Descriptive Statistics for State-Level Variables in the HGLM Analysis