Rural-Urban Disparity

Rural-Nonrural Differences in Educational Attainment:
Results from the National Educational Longitudinal Study of 1988-2000

Soo-yong Byun, Judith L. Meece, and Matthew J. Irvin University of North Carolina at Chapel Hill

April, 2010

Running Head: Rural-Nonrural Differences in Educational Attainment Revisited

Word count: 7,890

*This paper was presented at the annual meeting of the American Educational Research Association, May 3, 2010, Denver, CO. The analyses of the National Education Longitudinal Study of 1988-2000 national database were supported by a grant from the U.S. Department of Education’s Institute of Education Sciences Grant (R305A00406) awarded to the National Research Center on Rural Education Support at the University of North Carolina-CH. Opinions reflect those of authors and do not necessarily reflect those of the granting agencies. Direct correspondence to Soo-yong Byun, National Research Center on Rural Education Support/Center for Developmental Science, the University of North Carolina at Chapel Hill, 100 E. Franklin Street, Suite 200, Campus Box 8115, Chapel Hill, NC 27599-8115. Email: [e-mail]; Phone: [phone].

Rural-Nonrural Differences in Educational Attainment:
Results from the National Educational Longitudinal Study of 1988-2000

ABSTRACT
This study examines differences in educational attainment between rural and nonrural youth, using data from the National Educational Longitudinal Study. Given preexisting differences in family and student backgrounds between rural and metro areas, a series of statistical models including propensity sore matching methods are used to address selection bias in the estimates of the effects of rurality on educational attainment. Results show that there are few differences in postsecondary attendance and attainment between rural and metropolitan youth when students are matched on preexisting background characteristics. Significant gaps in educational attainment exist between those students who come from advantaged versus disadvantaged backgrounds for both rural and metropolitan youth. Policy implications are discussed.

Rural-Nonrural Differences in Educational Attainment:
Results from the National Educational Longitudinal Study of 1988-2000
U.S. adolescents, regardless of their socioeconomic and demographic background, have higher educational aspirations than ever before (Aldeman 2006; Ingels and Dalton 2008; Schneider and Stevenson 1999). This is also the case for rural youth who have tended to have lower educational aspirations than nonrural youth (Cobb, McIntire, and Paratt 1989; Haller and Vickler 1993; Hu 2003; Rojewski 1996). A recent study shows that almost 9 out of 10 rural adolescents aspired to attend college (Meece et al. 2010). Indeed, over the past three decades, there has been an increase in educational attainment for young adults in rural areas (Gibbs 2003). In 2000, approximately 16 percent of nonmetro adults age 25 and older graduated from a four-year college, which is more than double the 1970 rate (Gibbs 2003). Yet the rural-nonrural gap in educational attainment persists especially with respect to postsecondary enrollment and attainment. Young adults in rural areas are less likely than their counterparts in nonrural areas to attend college and thus earn a college degree (Gibbs 2003; Provansik et al. 2007). However, previous research documenting the rural-nonrural gap in postsecondary attainment has major limitations, especially given their reliance on census data such as the Current Population Survey (Gibbs 2003) and American Community Survey (Provansik et al. 2007). In not all but many cases, rural young adults, especially those talented adolescents, need to leave their home communities to seek educational and employment opportunities (Corbett 2000; Crockett, Shanahan, and Jackson-Newsom 2000). Outmigration of youth from rural communities may thus have consequences for estimating the level of educational attainment of young adults in both rural and nonrural areas (Howley and Gunn 2004). Specifically, it may lead to underestimating postsecondary degree completion rates in rural areas, while overestimating those in metro areas. In that regard, to assess rural-nonrural differences in educational attainment, it is important to use longitudinal data that follow youth from high school to postsecondary education. Indeed, research using longitudinal data reveals a somewhat differing pattern especially with respect to rural-nonrural differences in postsecondary attainment. In their analysis of the National Longitudinal Survey of Youth (NLSY), Blackwell and McLaughlin (1998) found that although rural youth aspired to fewer years of education than their urban counterparts in 1979 (age 14 to 17), they came closer to achieving their goals than urban youth by 1990 (age 25 to 28). Using data from the National Education Longitudinal Study of 1988 (NELS:88), Bozick and Deluca (2005) and Adelman (2006) also found few rural-urban differences in bachelor’s degree attainment. Although these studies using longitudinal data offer important insights into a better understanding of rural-nonrural differences in educational attainment, it is still unclear whether rurality (i.e., residential or school location) indeed matters in educational attainment or observed rural-nonrural differences simply reflect preexisting differences in background characteristics between rural and nonrural adolescents. To control for preexisting differences, prior studies (e.g., Bozick and Deluca 2005; Blackwell and McLaughlin 1999; Roscigno and Crowley 2001) have used regression-based methods with a limited number of background variables. However, such statistical adjustment by itself cannot ensure removal of selection bias unless additional procedures are taken for modeling the selection process (Rosenbaum and Rubin 1983, 1984). While there are a variety of alternative methods to control for selection bias (see Schneider et al. 2007), one useful way is to mimic random assignment to treatment conditions by ensuring that students have equivalent chances of being in one particular group or in another, often referred to as the propensity score matching (PSM) method (Rosenbaum and Rubin 1983, 1984). The PSM technique uses the predicted probability (i.e., propensity score) from logistic regression to match students from the treated and control groups on their relative probability of being assigned to the treatment. Because treated students are matched to control students with the same (or similar) propensity score, direct comparisons of the outcomes of the treated student groups and the control student groups yield unbiased estimates of the treatment effect. This way, the PSM method allows for the drawing of a causal inference from non-experimental settings (Rosenbaum and Rubin 1983, 1984). In this study, by implementing this alternative approach (i.e., PSM), we revisit rural-nonrural differences in educational attainment, focusing on postsecondary entry and completion. We draw on data from the NELS, which are very well suited for addressing the issues of selection bias as well as outmigration of rural youth previously described, given their detailed student and parent components and longitudinal nature. Our rationale for choosing NELS data over other more recent datasets such as the Education Longitudinal Study (ELS) is that although the ELS has followed a more recent cohort since 2002, its second follow up (2006) does not yet provide data that can be used to examine postsecondary persistence and attainment. In other words, no U.S. longitudinal high school sample but the NELS data that follows students from high school to postsecondary education is currently available. We aim to extend previous research by more rigorously assessing the role of rurality in educational attainment with longitudinal data and more sophisticated statistical models. However, the current study does not seek to provide a definitive answer about the causal effects of rurality on educational attainment. Rather, we hope to address limitations of prior studies that have examined rural-nonural disparities in educational attainment.

REVIEW OF LITERATURE
Determinants of Youth’s Educational Attainment
Research has identified a host of factors as sources of differences in educational attainment, including family background, demographic background (e.g., gender, race/ethnicity), and school resources (e.g., curriculum intensity) (see Goldrick-Rab, Carter, and Wagner 2007 for the literature review; also see Deil-Amen and Turley 2007). Of these varying factors, socioeconomic status (SES) measured by parental education and family income is one of the most powerful determinants of educational attainment (Adelman 1999, 2006; Bozick 2007; Cabrera, La Nasa, and Burkum 2001; Goldrick-Rab 2006; Sewell and Hauser 1972; Walpole 2003). Numerous studies document that children from high-SES families being more likely than their counterpart from low-SES families to complete high school (Coleman 1988; McNeal 1999; Smith, Beaulieu, and Seraphine 1995), go on to college (Bozick and Deluca 2005; Cabera and La Nasa 2001; Kim and Schneider 2005), and earn a college degree (Adelman 1999, 2006; Cabrera, La Nasa, and Burkum 2001; Walpole 2003). Beyond SES, another influence of family characteristics on educational attainment includes family structure and the number of siblings (Coleman 1988; Downey 1995; Powell and Steelman 1989). Children from two-parent households are more likely than children from sing-parent households to complete high school (Coleman 1988; Smith, Beaulieu, and Seraphine 1995) and enroll in a postsecondary institution (Bozick and Deluca 2005; Kim and Schneider 2005). Children with a greater number of siblings show relatively lower school performance (Dowley 1995) and are more likely to drop out of high school (Coleman 1988). Numerous studies of educational attainment also document demographic disparities (see Goldrick-Rab et al. 2007; Deil-Amen and Turley 2007 for the literature review). With respect to race/ethnicity, research suggests that minority students, especially black and Hispanic adolescents, are less likely than white students to enroll in college and complete college degree (Adelman 2006; Goldrick-Rab 2007). Exceptions are Asian American youth: Children from Asian immigrants show higher levels of educational achievement and attainment than white children (see Kao and Thompson 2003 for the literature review). Among NELS eight graders, 95 percent of Asian students entered college, compared to 77 percent of white and black students and 70 percent of Hispanics (Ingels et al. 2002). Several studies reveal the significant gender gap in educational attainment. A growing trend of the gender gap is the reversal from a favoring of males to a favoring of females in college entry as well as college completion (Buchmann and DiPrete 2006; U.S. Department of Education 2004). Finally, a good deal of research suggests that high school factors such as tracking and academic coursework matter in the transition to college and subsequent performance (Adelman 1999, 2006; Goldrick-Rab 2007). Using data from the High School and Beyond (HS&B) Survey, Adelman (1999) identified a “toolbox” of high school courses considered crucial in math, science, and foreign language. He found that students who took advanced levels of these courses are disproportionally likely to enter a four-year institution and perform better in college (Adelman 1999). Using data from the NELS, Adelman (2006) confirmed his previous finding.

The Role of Ruality in Youth’s Educational Attainment
While a great deal of research has examined rural-nonrural differences in educational aspirations (Howley 2006; Hu 2003; Rojewski 1996) and school performance (Fan and Chen 1999; Howley and Gunn 2004; Lee and McIntire 2000), limited research has explicitly tested rural-nonrural disparities in educational attainment on a large scale (Blackwell and McLaughlin 1998; Roscigno and Crowley 2001). Furthermore, little research has explicitly investigated rural-nonrural differences in postsecondary attainment using longitudinal data. Consequently, relatively little is known about the role of ruality in youth’s postsecondary attainment. Nevertheless, previously described evidence concerning the role of family background and school resources have important implications for potential rural-nonrual differences in postsecondary attainment, given the socioeconomic and educational challenges of many rural areas across the U.S. (Conger and Elder 1994; Hobbs 1994; Lichter and McLaughlin 1995; Provasnik et al. 2007; Roscigno and Crowley 2001). Specifically, while rural poverty rates are significantly higher than those in nonrural settings (O’Hare 2009; Rogers 2005), discrepancies in rural and urban poverty have been increasing since the last 1990s (O’Hare 2009). In addition, rural children were once relatively more likely than nonrural children to live in traditional family arrangements, but it is no longer true (O’Hare and Churilla 2008). Furthermore, students in rural schools have tended to have the least opportunity to take AP courses (Graham 2009). Rural schools have long faced a shortage of teachers with advanced degrees (Monk 2007; Provasnik et al. 2007). Likewise, many rural communities have experienced a dramatic job loss with regards not only to technical, business, and other professional careers, but also to many service, labor, or agricultural occupations, which have been mainstay of rural communities for generations (Conger and Elder 1994; Gibbs, Kusmin, and Cromartie 2005; Hobbs 1994; Lichter and McLaughlin 1995). All these rural socioeconomic and educational features may constrain and shape educational aspirations and attainment of rural youth, leading to the rural-nonrural gap. Indeed, while documenting the rural disadvantages in educational aspirations and attainment (Cobb et al. 1989; Haller and Vickler 1993; Hu 2003; Rojewski 1996; Roscigno and Crowley 2001), a large body of research has attributed the lower educational aspirations and attainment of rural youth to higher rates of poverty, poorer schooling conditions, lower parental and teacher expectations, and lower school achievement (see Howley and Gunn 2004 for the discussion; also see Howley 2006). Further, it is believed that rural students receive education that is inferior to that of their counterpart students in nonrural settings (Edington and Koehler 1987). However, this rural deficit model is not fully supported by empirical evidence. With respect to school performance, for example, in their examination of the NELS, Fan and Chen (1999) found that rural students perform as well as their counterpart students in metropolitan schools. The authors concluded that, things being equal, rural adolescents “do not suffer disadvantage simply as the result of their residence in rural areas or their attendance at rural schools” (Fan and Chen 1999:31). In their investigation of the National Assessment of Educational Progress (NAEP) assessment scores from 1992-1996, Lee and McIntire (2000) even found a rural advantage in math achievement, with rural students outperforming their nonrural counterparts. This supports a ‘rural strength model’ (Edington and Koehler 1987). Regardless of whether evidence supports the rural deficit or strength model, most previous research has major limitations that preclude drawing strong conclusions about the causal effects of rurality on educational outcomes. The central problem lies in its reliance on observational data, which are often considered inappropriate for drawing a causal inference due to the lack of randomization (Rosenbaum and Rubin 1983, 1984). In a nonexperimental setting, direct comparisons between the treatment and control groups may be misleading because individuals exposed to one treatment tend to differ systematically from individuals exposed to the other treatment (Rosenbaum and Rubin 1983, 1984). This is clearly the case in research about the rural-nonrural achievement gap, given systemic differences in background characteristics between rural and nonrural adolescents previously described. To hold student background characteristics constant and examine the effects of rurality on educational outcomes, random assignment to families that vary in residence only (rural vs. nonrural) is needed but impossible. Although lack of randomization in a nonexperimental setting makes it difficult to estimate the causal effects of rurality, it does not necessarily mean that drawing causal inference from observational studies is impossible. Rather, it highlights the importance of addressing the selection issue in observational studies in order to establish the causal relationship between rurality and educational attainment. In this study, to address the selection bias inherent in observational data, we employ a series of statistical models including the PSM technique, which allows us to draw a causal inference about the role of rurality in educational attainment.

RESEARCH QUESTIONS
Guided by previous literature on the rural-nonrural achievement gap and its limitations, the current study addresses the following research questions:
1. To what extent do rural-nonrural differences exist in achieving their educational goals (i.e., high school completion, college enrollment, and college degree)?
2. If there are observed rural-nonrural differences in educational attainment, does this result prove robust when preexisting differences in socioeconomic and demographic background between rural and non-rural youth are taken into account?

DATA AND METHODS
Sample
We drew on data from the National Education Longitudinal Study, a large, nationally-representative data set. In 1988, the National Center for Education Statistics (NCES) drew random samples of approximately 25 eight graders in each of about 1,000 randomly selected middle schools. NELS followed these students through high school in 1990 and 1992, and beyond in 1944 and 2000 (at age 26 or 27).The NELS: 88-00 panel consists of approximately 12,100 students. All of our measures came from the 88-00 postsecondary transcript data containing more accurate information about postsecondary attainment. The exceptions are family structure and number of siblings, which we extracted from the 1994 wave and merged with the 88-00 transcript data. Due to small sample sizes, American Indian/Alaska Native and multiracial students were excluded. Missing rurality identifiers and weights resulted in the final analytic sample of 11,700 with rural youth being approximately 30 percent. Table 1 shows the percentage distribution of the high school graduation status, postsecondary participation, and highest degree (completed by 2000) for the 1992 high school seniors by rurality. Results show few differences in high school completion status between rural and nonrural students. Of approximately 3,630 rural students, 83.1 percent earned an academic diploma; 8.1 percent obtained a GED or other; and 7.8 percent failed to complete high school by 2000. The corresponding rates for nonrural students are 82.4 percent, 8.3 percent, and 8.0 percent, respectively. On the other hand, results of Table 1 show differences in postsecondary participation between rural and nonrural students. Among rural youth, 70.1 percent had enrolled in a postsecondary institution by 2000. Among nonrural youth, 78.6 percent had enrolled in a postsecondary institution eight years after high school graduation (when students were on the right track), showing 8.5 percent point higher than the rate of rural students.
TABLE 1 ABOUT HERE Results also reveal differences in the highest degree between rural and nonrural youth. Among rural youth, 4 percent and 6.5 percent earned a certificate or an associate degree, respectively. The corresponding rates are 3.5 percent and 5.1 percent, respectively, showing a slightly lower than those rates of rural youth. On the other hand, 25.6 percent of rural youth earned a bachelor’s degree or above, whereas 30.7 percent of nonrural youth did so, showing approximately a five percentage point difference. It is interesting to note that a relatively greater proportion of nonrural youth (31.6%) who had enrolled in a postsecondary institution remained incomplete, as compared to that of rural youth (28.1%). For multivariate analysis, we focused on the postsecondary entrance status and highest degree status, given no observable differences in high school completion between rural and nonrrual youth. For the analysis of the highest degree status, we excluded those interminable cases (7.2%) (see Table 1). With exceptions for educational attainment, rurality, gender, we imputed missing data through an alternative algorithm suggested by King and colleagues (2001). In the multivariate models, we accounted for the initial sample clustering of students within schools using the NELS school identifier, and utilized the longitudinal second follow-up to fourth follow-up panel weight.

Measure Educational Attainment. We are interested in two postsecondary attainment variables: postsecondary participation and postsecondary degree attainment. Postsecondary participation was measured by whether or not the respondent had ever enrolled in a postsecondary institution as of 2000. Postsecondary degree attainment was measured by the highest degree attained by 2000 with four categories: (1) no postsecondary enrollment, (2) certificate/associate degree, (3) bachelor’s degree, and (4) no postsecondary degree. No postsecondary degree refers to students who had ever enrolled in college at any time after high school graduation but had not earned a college degree by 2000. Both measures of postsecondary enrollment and degree attainment were based on postsecondary transcript data. Rurality. The operational definition for “ruality” or rural residence was measured by school location where schools are outside of Metropolitan Statistical Areas defined by Quality Education Data and NELS (Lippman, Burns, and McArhur 1996). Original responses denoted (1) rural, (2) suburban, and (3) urban, but we collapsed those into the dichotomous category (i.e., rural vs. nonrural). Controls. As described previously, socioeconomic and demographic backgrounds of youth play an important in determining their educational attainment. Moreover, as will be seen in Table 2, there are significant differences in these backgrounds between rural and nonrural youth. Accordingly, without controlling for these background variables associated with educational attainment and rurality, it is not possible to obtain the credible relationship between educational attainment and rurality. Therefore, drawing on literature, we included a number of measures of background variables that also may shape educational attainment of youth as controls. All background variables were measured at grade 12 (1992). For family background, we included (1) parental education, (2) family income, (3) family structure, and (4) family size. Parental education was measured by the highest level of education that parents reported. Original responses were 1=less than high school graduation, 2=high school graduation, 3=some college, 4=bachelor’s degree, 5=master’s degree, 6=doctorate or other professional degree. We collapsed these responses into the dichotomous categories: (1) some college or less and (2) bachelor’s degree and above. Some college or less is reference category. Family income was based on the parent’s report of family income at grade 12. Original responses were 1=$75,000 or more, 2=$50,000 ~ $74,999, 3=$35,000 ~ $49,999, 4=$25,000 ~ $34,999, 5=$15,000 ~ $24,999, and 6=less than $15,000. We inversed and collapsed these categories into trichotomous categories: (1) less than $25,000, (2) $25,000 ~ $49,999, and (3) $50,000 or more. Less than $25,000 is the reference category. Family structure denoted whether students lived in two-parent families (=0) or in other forms of families (=1). Parents reported the number of siblings that a student had at grade 12. For individual student characteristics, we included (1) gender, (2) race/ethnicity, and (3) academic achievement. Gender was measured by the student’s sex (female=1 vs. male=0). Race/ethnicity was measured by students’ self-reported race/ethnicity (Asian, Hispanic, black, and white). White students served as the reference group. Academic achievement was measured by the math/reading composite score from the standardized test administered by the NELS during the 12th grade. Table 2 provides weighted descriptive statistics for all indictors. Because we already examined the unadjusted differences in outcome variables between rural and nonrural youth (see Table 1), we focus on rural-nonrural differences in background variables. Results of Table 2 clearly show differences in socioeconomic background between rural and nonrural youth. With respect to parental education, only 21 percent of rural youth had parents who had a bachelor’s degree or more, whereas 32 percent of nonrural youth did so, showing more than 10 percentage point differences. With respect to family income, only 25 percent of rural youth were from families whose annual income was $50,000 or more, whereas the corresponding rate of nonrural youth is 36 percent. In contrast, 41 percent of rural youth were from families whose annual income was less than $25,000, whereas the corresponding rate of nonrural youth is only 30 percent. Differences in family structure and family size were minimal between rural and nonrural youth. Results of Table 2 also clearly show differences in racial/ethnic background between rural and nonrual youth. 84 percent of rural youth were white, whereas 68 percent of nonrural youth were. 7 percent of rural youth were Hispanic and black, whereas the corresponding rates of nonrural youth were 13 percent and 14 percent, respectively, both of which are roughly double rates of rural youth. Only 1 percent of rural youth were Asian, whereas 5 percent of nonrural youth were Asian. No observable rural-nonrural differences were detected in gender. Finally, results show the rural-nonrural gap in academic achievement with nonrural youth outperforming rural youth (51.7 vs. 48.6). These differences in the background characteristics between rural and nonrural youth indicate that there are selection effects. Accordingly, this study uses a series of models that attempt to address this selection bias, including PSM.

Analytic Strategies
We implemented two analytic strategies for each of the outcome variables. For the analysis of postsecondary participation, using conventional logistic regression, we first analyzed the likelihood of being enrolled in a college (vs. no college enrollment). For logistic regression, we introduced two models. The first model included the rurality variable only to determine whether there was a significant difference in postsecondary enrollment between rural and nonrural youth. The second model added parental education, family income, and race/ethnicity, along with the controls mentioned previously. The aim was to examine whether observed rural-nonrural differences, if any, would exist when other factors were taken into account. We used STATA to utilize its survey (svy) commands, which allow for adjusting for stratification, clustering and individual weighting (Broene and Rust 2000). Second, using the PSM technique[1], we revisited the relationship between rurality and postsecondary enrollment. Our intent was to determine whether the observable relationship between rurality and postsecondary enrollment prove robust when students were matched on preexisting background characteristics. Note that we defined the propensity score as the probability of attending a rural high school, conditional on observed backgrounds. For PSM analysis, following prior literature (Guo and Fraser 2009; Rosenbaum and Rubin 1983, 1984), we first conducted logistic regression to generate propensity scores, using the controls shown in Table 1 as well as other covariates (not shown but available on request) associated with the likelihood of attending a rural high school. After estimating propensity scores using logistic regression, we created matched data in which students are as similar as possible conditional on observed backgrounds by implementing the one-to-one caliper matching procedure. In this matching process, we used the caliper size of a quarter of a standard deviation of the estimated propensity scores, following Rosenbaum and Rubin (1985). To check the balance of the propensity scores in the matched sample, we conducted a t-test or chi-square test for each of covariates, depending on their scales, and found no systematic difference in the covariates between rural and nonrural youth (not shown). This suggested that the estimated propensity scores remove most of the selection bias for the covariates in the matched sample. Finally, using the optimally matched samples, we replicated the logistic regression described above. Likewise, for the analysis of postsecondary attainment, using conventional multinomial logistic regression, we first examined differences in the odds that youth attained different levels of postsecondary degree (i.e., certificate/associate degree, bachelor’s degree), compared with no postsecondary enrollment. For multinomial logistic regression, we also introduced two models: The first model added the rurality variable only, while the second model the controls. We then revisited the effects of rruality on postsecondary attainment, using the PSM technique. Note that as the sample sizes differ by the outcome variables, those also vary across PSM analyses. The following section presents the results, beginning with postsecondary participation.

RESULTS

Rural-nonrural Differences in Postsecondary Participation

Is there the relationship between rurality and postsecondary enrollment? If there is, does the relationship between rurality and postsecondary enrollment hold when we control for a host of other factors that also may shape students’ odds of participating in postsecondary education? Table 3 addresses these questions. Note that we present odds ratios for ease of interpretation. A ratio greater than one represents increased odds, whereas a ratio less than one represents decreased odds, of falling into the comparison category rather than the reference category.
TABLE 3 ABOUT HERE As already seen in Table 2, results of Model 1 that included the rurality variable only reveal rural-nonrural differences in postsecondary enrollment. Rural youth are 36.2 percent less likely than nonrural youth to participate in postsecondary education. Significant differences remain in Model 2 which added the controls to Model 1, suggesting that rural adolescents are less likely than their metropolitan adolescents to participate in postsecondary education after background characteristics are controlled. Before we turn to PSM results, we briefly mention results related to the control variables. Students whose parents had a bachelor’s degree or more are 3.6 times more likely than students whose parents had some college or less to participate in postsecondary education. Family income significantly predicts the odds that children attend college. Students from nontraditional families are far less likely than students from traditional families to participate in postsecondary education. An increase in the number of siblings significantly decreases the odds that students are enrolled in college. Female students are 43.8 percent more likely than male students to participate in postsecondary education. Minority students, except for Hispanic students, are more likely than white students to participate in postsecondary education. Academic achievement is a significant predictor of the odds that students participate in postsecondary education.
TABLE 4 ABOUT HERE We now turn to the question of whether the relationship between rurality and postsecondary enrollment proves robust when students were matched on preexisting background characteristics. Table 4 presents results from PSM analysis. Results of Table 4 reveal a somewhat different pattern found in Table 3. Specifically, in the PSM analysis we find no significant differences in postsecondary participation between rural and nonrural youth. The finding suggests that there are few differences in postsecondary attendance between rural and metropolitan youth when they are matched on preexisting background characteristics.

Rural-nonrural Differences in Postsecondary Attainment
Now we examine rural-nonrural differences in postsecondary attainment. Table 5 presents odds ratio from multinomial logistic regression predicting the highest degree achieved. The first two columns estimate the likelihood of obtaining each of the two types of the highest level of educational attainment versus no postsecondary enrollment, the reference category. The last column estimates the likelihood of obtaining a bachelor degree versus a certificate or associate degree, the reference category. We present the multinomial logistic regression results of the comparison of no postsecondary degree with the other categories in Appendix A.
TABLE 5 ABOUT HERE Model 1 including the rurality variable shows rural-nonrural disparities especially in a bachelor’s degree or more versus no postsecondary and a bachelor’s degree versus a certificate/associate degree. Rural adolescents are less likely than nonrural adolescents to earn a bachelor’s degree, while they are more likely than nonrual adolescents to earn a certificate/associate degree. However, Model 2 that controlled for other factors demonstrate few differences in postsecondary attainment between rural and nonrural youth. Noteworthy in Model 2 is that socioeconomically advantaged students are far more likely than disadvantaged students to earn a college degree, especially a bachelor’s degree.
TABLE 6 ABOUT HERE Now, given selection bias inherent in the NELS data, we revisit the effect of ruality on postsecondary attainment using the PSM approach. Table 6 presents the PSM results of the comparison of no postsecondary degree with the other categories in Appendix B. PSM results confirm that there are no significant differences in postsecondary attainment between rural and nonrural youth when they are matched on preexisting background characteristics. The finding suggests that observed differences in postsecondary attainment between rural and nonrural youth are attributable to preexisting differences in their background characteristics, rather than to their residence.

DISCUSSION
Many educators, legislators, and the general public tend to believe that rural students receive an inferior education, given their relatively lower educational achievement and attainment compared to their nonrual counterparts (Edington and Koehler 1987; Howley and Gunn 2003; Howley 2006). While prior studies offer mixed evidence on the rural-nonrural achievement gap, their conclusions have been questioned because of the selection issues not being adequately addressed (Edington and Koehler 1987; Howley and Gunn 2003). To address this issue, we have attempted to more rigorously assess the role of rurality in educational attainment by using longitudinal data and more sophisticated statistical models such as PSM, both of which allow us to draw a causal inference about the effect of rurality. Our results showed that there are few differences in postsecondary participation and attainment between rural and metropolitan youth when background characteristics are taken into account. In other words, observable rural-nonrural differences in postsecondary enrollment and attainment are largely due to preexisting differences in the socioeconomic and demographic backgrounds between rural and nonrural youth. This finding supports neither the rural deficit model nor the rural strength model. Yet our findings are consistent with past research (Adelman 2006; Bozick and Deluca 2005). Another key finding is that significant gaps in educational attainment exist in postsecondary entry and degree attainment between those students who come from advantaged versus disadvantaged backgrounds. Consistent with previous research (Adelman 2006; Cabrera et al. 2001; Walpole 2003), our results demonstrated that students who have a disadvantaged family background are far less likely than students who have an advantaged one to participate in postsecondary education and earn a college degree. Thus, educational policy and programs designed to increase college enrollment and completion in rural and urban areas alike would be more effective if they targeted the youth from disadvantaged backgrounds. The present study has several limitations that need to be addressed in the future. First, some students whose postsecondary attainment was incomplete within the data collection timeframe of this study may eventually earn a college degree, although it may take them a number of years to do so. These students are then likely to be counted as college completers. Thus, our estimates of the level of educational attainment may be subject to change, depending on the timespan of any longitudinal study carried out in future. A longitudinal study with a longer timespan may increase our understanding of the complex patterns of college persistence and completion. Second, despite the importance, we did not fully examine the pathways in which adolescents achieve their postsecondary education goals and how they differ, depending on rurality. Recent research documents the increasing complexity of the postsecondary participation pattern pathways, with words like rebounding, reserve transfer, transfer swirl (Goldrick-Rab et al. 2007). Research also suggests that these diverse pathways differ by students’ socioeconomic background (Goldrick-Rab 2006, 2007). Little is known about rural-nonrural differences in pathways through which students take for postsecondary education. This should be empirically addressed in the future. Finally, future research should examine the role that school and community factors may play in shaping youth’s educational attainment. Literature suggests that school and community resources promote not only the transition from high school to college but college completion (Goldrick-Rab et al. 2007; Deil-Amen and Turley 2007). Examining the role of school resources in youth’s educational attainment is particularly important, given that rural schools have long faced challenges in recruiting and retaining certified and highly qualified teachers and leaders (Monk 2007; Provasnik et al. 2007; Roscigno and Crowley 2001). Therefore, future studies investigating the role of school and community factors should offer important insights to policymakers for the better design of both school and community programs in rural areas.

REFERENCES
Adelman, Clifford. 1999. Answers in the Tool Box: Academic intensity, Attendance Patterns, and Bachelor’s degree attainment. Washington, DC: U.S. Department of Education
----------------------. 2006. The Toolbox Revisited: Paths to Degree Completion from High School Through College. Washington, D.C.: U.S. Department of Education. Retrieved March 6, 2010 (http://www2.ed.gov/rschstat/research/pubs/toolboxrevisit/toolbox.pdf).
Blackwell, Debra L. and Diane K. McLaughlin. 1998. “Do Rural Youth Attain Their Educational Goals?” Rural Development Perspectives 13(3):37–44.
Broene, Pam and Keith Rust. 2000. Strengths and Limitations of Using SUDAAN, Stata, and WesVarPC for Computing Variances from NCES Data Sets. Working Paper No. 2000-03. Washington, CD: U.S. Department of Education, National Center for Education Statistics. (Also available athttp://nces.ed.gov/pubs2000/200003.pdf).
Bozick, Robert. 2007. “Making it Through the First Year of College: The Role of Students’ Economic Resources, Employment, and Living Arrangements.” Sociology of Education 80:361-285.
Bozick, Robert and Stefanie Deluca. 2005. “Better Late than Never? Delayed Enrollment in the High School to College Transition.” Social Forces 84(1): 531-54.
Buchmann, Claudia and Thomas A. DiPrete. 2006. “the Growing Female Advantage in College Completion: The Role of Family Background and Academic Achievement.” American Sociological Review 71(4):515-41.
Cabrera, Alberto F. and Steven M. La Nasa. 2001. “On the Path to College: Three Critical Tasks Facing America’s Disadvantaged.” Research in Higher Education 42(2):119-49
Cabrera, Alberto F., Steven M. La Nasa, and Kurt R. Burkum. 2001. “Pathways to a four-year Degree: The Higher Education Story of One Generation. WI, Madison: Wisconsin Center for the Advancement of Postsecondary Education. Retrieved March 23, 2010 (http://www.wiscape.wisc.edu/calendar/edpolicies/pathways.pdf)
Cobb, Robert A., Walter G. McIntire, and Phillip a. Pratt. 1989. “Vocational and Educational Aspirations of High School Students: A Problem for Rural America.” Research in Rural Education 6(2):11-6.
Coleman, James S. 1988. “Social Capital in the Creation of Human Capital.” American Journal of Sociology 94(Supplement):S95-S120.
Conger, Rand D. and Glen H. Elder. Jr. 1994. Families in Troubled Times: Adapting to Change in Rural America. New York: Walter de Gruyter.
Corbett, Michael John. 2000. “Learning to Leave: The Irony of Schooling in a Coastal Community.” PhD dissertation, University of British Columbia, Canada.
Crockett, Lisa J., Michael J. Shanahan, Julia Jackson-Newsom. 2000. “Rural Youth: Ecological and Life Course Perspectives." In Adolescent Diversity in Ethnic, Economic, and Cultural Contexts: Advances in Adolescent Development, Volume 10, edited by Raymond Montemayor, Gerlad R. Adams, and Thomas P. Gullotta. Thousand Oaks, CA: Sage Publications.
Deil-Amen, Regina and Ruth Lopez Turley. 2007. “A Review of the Transition to College Literature in Sociology.” Teachers College Record 109(10):2324-66.
Downey, Douglas B. 1995. “When Bigger is Not Better: Family Size, Parental Resources, and Children’s Educational Performance.” American Sociological Review 60(5):746-61.
Edington, Everett D. and Lyle Koehler. 1987. “Rural Student Achievement: Elements for Consideration.” ERIC Digest, (ERIC Document Reproduction Service No. ED 289658).
Gibbs, Robert, Lorin Kusmin, and John Cromartie. 2005. “Low-skill Employment and the Changing Economy of Rural America.” Economic Research Report 10. Washington, DC: U.S. Department of Agriculture, Economic Research Service. Retrieved March 4, 2010 (http://www.ers.usda.gov/publications/err10/err10.pdf).
Gibbs, Robert. 2003. “Rural Education at a Glance.”Rural Development Research Report Number 98. Washington, DC: U.S. Department of Agriculture, Economic Research Service. Retrieved March 4, 2010 (http://www.ers.usda.gov/publications/rdrr98/rdrr98_lowres.pdf). Goldrick-Rab, Sara. 2006. “Following Their Every Move: An Investigation of Social-class Differences in College Pathways.” Sociology of Education 79:61-79.
----------------------. 2007. “The Context of “Choice” in College Pathways.” Working Paper Series 008. Madison, WI: Wisconsin Center for the Advancement of Postsecondary Education.
Goldrick-Rab, Sara, Deborah Faye Carter, and Rachelle Winkle Wagner. 2007. “What Higher Education Has to Say About the Transition to College.” Teachers College Record 109(10):2444-81.
Graham, Suzanne E. 2009 “Students in Rural Schools Have Limited Access to Advanced Mathematics Courses.” Durham, NH: Carsey Institute, University of New Hampshire. (Also available at http://www.carseyinstitute.unh.edu/publications/FS-Rural-married-couple-families-08.pdf)
Guo, Shenyang, and Mark W. Fraser. 2009. Propensity Score Analysis: Statistical Methods and Applications. Thousand Oaks, CA: Sage Publications.
Haller, Emil J. and Sarah J. Virkler. 1993. “Another Look at Rural-nonrural Differences in Students’ Educational Aspirations.” Journal of Research in Rural Education 9(3):170-78.
Hobbs, Daryl. 1994. “Demographic Trends in Nonmetropolitan America.” Journal of Research in Rural Education 10(3):149-60.
Howley, Caitlin W. 2006. “Remote Possibilities: Rural Children’s Educational Aspirations.” Peabody Journal of Education 81(2):62-80.
Howley, Craig B. and Erik Gunn. 2003. “Research about Mathematics Achievement in the Rural Circumstance.” Journal of Research in Rural Education 18(2):86-95.
Hu, Shouping. 2003. “Educational Aspirations and Postsecondary Access and Choice: Students in Urban, Suburban, and Rural Schools Compared.” Education Policy Analysis Archives 11(14). Retrieved March 3, 2010 (http://epaa.asu.edu/epaa/v11n14/).
Ingels, Steven J. and Ben W. Dalton. 2008. Trends Among High School Seniors, 1972–2004 (NCES 2008-320). Washington, DC: U.S. Department of Education, National Center for Education Statistics. Retrieved March 6, 2010 (http://nces.ed.gov/pubs2008/2008320.pdf).
Ingels, Steven. J., Thomas R. Curtin, Phillip Kaufman, Martha Naomi Alt, Xianglei Chen. 2002. Coming of Age in the 1990s: The Eighth-grade Class of 1988 12 Years Later. Washington, DC: U.S. Department of Education, National Center for Education Statistics. Retrieved March 6, 2010 (http://nces.ed.gov/pubs2002/2002321.pdf).
Kao, Grace and Jennifer S. Thompson. 2003. “Racial and Ethnic Stratification in Educational Achievement and Attainment.” Annual Review of Sociology 29:417-42.
Kim, Doohwan and Barbara Schneider. 2005. “Social Capital in Action: Alignment of Parental Support in Adolescents’ Transition to Postsecondary Education.” Social Forces 84(2):1181-206.
King, Gary, James Honaker, Anne Joseph, and Kenneth Scheve. 2001. “Analyzing Incomplete Political Science Data: An Alternative Algorithm for Multiple Imputation.” American Political Science Review 95: 49-69.
Lee, Jaekyung and Walter G. McIntire. 2000. “Interstate Variation in the Mathematics Achievement of Rural and Nonrural Students.” Journal of Research in Rural Education 16(3):168-181.
Lichter, Daniel T. and Diane K. McLaughlin. 1995. “Changing Economic Opportunities, Family Structure, and Poverty in Rural Areas.” Rural Sociology 60(4):688-706.
Lippman, Laura., Shelley Burns, and Edith McArthur. 1996. Urban schools: The challenge of location and poverty. Washington, DC: U.S. Department of Education, National Center for Education Statistics.
McNeal, Ralph B. Jr. 1999. “Parental Involvement as Social Capital: Differential Effectiveness on Science Achievement, Truancy, and Dropping Out.” Social Forces 78(1):117-44.
Meece, L. Judith, Matthew Irvin, Soo-yong Byun, Thomas W. Farmer, Bryan C. Hutchins, and Margaret Weiss. “Preparing for Adulthood: A Recent Examination of Rural Youth’s Future Educational and Vocational Aspirations.” Manuscript submitted for publication.
Monk, David H. 2007. “Recruiting and Retaining High-Quality Teachers in Rural Areas.” The Future of Children 17(1):155-74.
O’Hare, William P. 2009. The Forgotten Fifth: Child Poverty in Rural America. Durham, NH: Carsey Institute, University of New Hampshire. (Also available at http://www.carseyinstitute.unh.edu/publications/Report-OHare-ForgottenFifth.pdf).
O’Hare, William P. and Allison Churilla. 2008. Rural Children Now Less Likely to Live in Married-Couple Families. Durham, NH: Carsey Institute, University of New Hampshire. (Also available at http://www.carseyinstitute.unh.edu/publications/FS-Rural-married-couple-families-08.pdf)
Provasnik, Stephen, Angelina KewalRamani, Mary McLaughlin Coleman, Lauren Gilbertson, Will Herring, and Qingshu Xie. 2007. Status of Education in Rural America (NCES 2007-040). Washington, DC : U.S. Department of Education, National Center for Education Statistics, Institute of Education Sciences.
Rogers, Carolyn C. 2005. “Rural Children at A Glance.” Economic Information Bulletin Number 1. Washington, DC: U.S. Department of Agriculture Economic Research Service. Retrieved March 4, 2010 (http://www.ers.usda.gov/publications/eib1/eib1.pdf).
Rojewski, Jay W. 1999. “Career-Related Predictors of Work-Bound and College-Bound Status of Adolescents in Rural and Nonrural Areas.” Journal of Research in Rural Education 15(3):141-56.
Roscigno, Vincent J. and Martha L. Crowley. 2001. “Rurality, Institutional Disadvantage, and Achievement/Attainment.” Rural Sociology 66(2):268-93.
Rosenbaum, Paul R. and Donald B. Rubin. 1983. “The Central Role of the Propensity Score in Observational Studies for Causal Effects.” Biometrika 70:41-55.
----------------------. 1984. “Reducing Bias in Observational Studies Using Subclassification on the Propensity Score.” Journal of the American Statistical Association 79:516-24.
----------------------. 1985. “Constructing a Control Group Using Multivariate Matched Sampling Methods that Incorporate the Propensity Score.” American Statistician 39:33-38.
Schneider, Barbara and David Stevenson. 1999. The Ambitious Generation: America’s Teenagers, Motivated but Directionless. New Haven: Yale University Press.
Schneider, Barbara, Martin Carnoy, Jeremy Kilpatrick, William H. Schmidt, and Richard J. Shavelson. 2007. Estimating Causal Effects Using Experimental and Observational Designs. Washington, DC: American Educational Research Association.
Sewell, William H. and Hauser, Robert M. 1972. “Causes and Consequences of Higher Education: Models of the Status Attainment Process.” American Journal of Agricultural Economics 54(5):851-61.
Smith, Mark H., Lionel J. Beaulieu, and Anne Seraphine. 1995. “Social Capital, Place of Residence, and College Attendance.” Rural Sociology 60:363-80.
Steelman, Lala Carr and Brian Powell. 1991. “Sponsoring the Next Generation: Parental Willingness to Pay for Higher Education.” American Journal of Sociology 96(6):1505-29.
U.S. Department of Education, National Center for Education Statistics. 2004. Digest of Education Statistics. Washington, DC: U.S. Government Printing Office.
Walpole, Marybeth. 2003. “Socioeconomic Status and College: How SES Affects College Experiences and Outcomes.” The Review of Higher Education 27(1):45-73.

TABLES

|Table 1. Percentage Distribution of High School Graduation Status, Postsecondary Participation, and Degree Completion of the 1992 High |
|School Seniors by Rurality |
| |Total |Rural |Nonrural |χ2 |
|All students |100.0 |30.2 |69.8 | |
| | | | | |
|Educational Attainment | | | | |
| High School Completiona | | | | |
| Academic diploma |82.6 |83.1 |82.4 |1.47 |
| GED or other |8.2 |8.1 |8.3 | |
| Did not complete |8.0 |7.8 |8.0 | |
| Interminable (no transcript information) |1.2 |1.1 |1.3 | |
| | | | | |
|Postsecondary entrya |76.1 |70.1 |78.6 |97.76*** |
| | | | | |
|Highest degreea | | | | |
| No postsecondary enrollment |23.9 |29.9 |21.4 |126.90*** |
| Certificate |3.7 |4.0 |3.5 | |
| Associate |5.5 |6.5 |5.1 | |
| Bachelor 's or above |29.2 |25.6 |30.7 | |
| No postsecondary degree |30.5 |28.1 |31.6 | |
| Interminable (no transcript information) |7.2 |6.0 |7.7 | |
|Data Source: NELS:88-00 | | | | |
|Note: Numbers are proportions. Data are weighted. Unweighted N=11,700. |

|Table 2. Variables Included in Analyses | | | | |
| |Mean or % |SD |Mean or % |SD |
|b. Indeterminable cases were excluded. Unweighted N = 11,050 | |

|Table 3. Odds Ratio from Logistic Regressions Predicting Postsecondary Participation |
| |1 |2 |
|Rurality |.638 |*** |.779 |* |
| | | | | |
|Controls | | | | |
|Parent had a bachelor 's degree or more (vs. some college or less) | |3.645 |*** |
|Family income (less than 25,000 omitted) | | | | |
| More than 25,000 and less than 50,000 | | |1.487 |*** |
| 50,000 or more | | |1.815 |*** |
|Nontraditional family (vs. traditional family) | | |.609 |*** |
|Number of siblings | | |.851 |*** |
| Female (vs. male) | | |1.438 |** |
|Race/ethnicity (white omitted) | | | | |
| Asian | | |2.340 |*** |
| Hispanic | | |1.386 | |
| Black | | |2.021 |*** |
|Academic achievement at grade 12 | | |1.034 |*** |
|Data source: NELS: 88-00 (unweighted N=11,700) | | | | |
|Note: Data are weighted. | | | | |
|*** p

References: Adelman, Clifford. 1999. Answers in the Tool Box: Academic intensity, Attendance Patterns, and Bachelor’s degree attainment. Washington, DC: U.S. Department of Education ---------------------- Blackwell, Debra L. and Diane K. McLaughlin. 1998. “Do Rural Youth Attain Their Educational Goals?” Rural Development Perspectives 13(3):37–44. Bozick, Robert. 2007. “Making it Through the First Year of College: The Role of Students’ Economic Resources, Employment, and Living Arrangements.” Sociology of Education 80:361-285. Bozick, Robert and Stefanie Deluca. 2005. “Better Late than Never? Delayed Enrollment in the High School to College Transition.” Social Forces 84(1): 531-54. Buchmann, Claudia and Thomas A. DiPrete. 2006. “the Growing Female Advantage in College Completion: The Role of Family Background and Academic Achievement.” American Sociological Review 71(4):515-41. Cabrera, Alberto F. and Steven M. La Nasa. 2001. “On the Path to College: Three Critical Tasks Facing America’s Disadvantaged.” Research in Higher Education 42(2):119-49 Cabrera, Alberto F., Steven M Cobb, Robert A., Walter G. McIntire, and Phillip a. Pratt. 1989. “Vocational and Educational Aspirations of High School Students: A Problem for Rural America.” Research in Rural Education 6(2):11-6. Coleman, James S. 1988. “Social Capital in the Creation of Human Capital.” American Journal of Sociology 94(Supplement):S95-S120. Conger, Rand D. and Glen H. Elder. Jr. 1994. Families in Troubled Times: Adapting to Change in Rural America. New York: Walter de Gruyter. Corbett, Michael John. 2000. “Learning to Leave: The Irony of Schooling in a Coastal Community.” PhD dissertation, University of British Columbia, Canada. Deil-Amen, Regina and Ruth Lopez Turley. 2007. “A Review of the Transition to College Literature in Sociology.” Teachers College Record 109(10):2324-66. Downey, Douglas B. 1995. “When Bigger is Not Better: Family Size, Parental Resources, and Children’s Educational Performance.” American Sociological Review 60(5):746-61. Edington, Everett D. and Lyle Koehler. 1987. “Rural Student Achievement: Elements for Consideration.” ERIC Digest, (ERIC Document Reproduction Service No. ED 289658). Goldrick-Rab, Sara. 2006. “Following Their Every Move: An Investigation of Social-class Differences in College Pathways.” Sociology of Education 79:61-79. ----------------------. 2007. “The Context of “Choice” in College Pathways.” Working Paper Series 008. Madison, WI: Wisconsin Center for the Advancement of Postsecondary Education. Goldrick-Rab, Sara, Deborah Faye Carter, and Rachelle Winkle Wagner. 2007. “What Higher Education Has to Say About the Transition to College.” Teachers College Record 109(10):2444-81. Haller, Emil J. and Sarah J. Virkler. 1993. “Another Look at Rural-nonrural Differences in Students’ Educational Aspirations.” Journal of Research in Rural Education 9(3):170-78. Hobbs, Daryl. 1994. “Demographic Trends in Nonmetropolitan America.” Journal of Research in Rural Education 10(3):149-60. Howley, Caitlin W. 2006. “Remote Possibilities: Rural Children’s Educational Aspirations.” Peabody Journal of Education 81(2):62-80. Howley, Craig B. and Erik Gunn. 2003. “Research about Mathematics Achievement in the Rural Circumstance.” Journal of Research in Rural Education 18(2):86-95. Hu, Shouping. 2003. “Educational Aspirations and Postsecondary Access and Choice: Students in Urban, Suburban, and Rural Schools Compared.” Education Policy Analysis Archives 11(14). Retrieved March 3, 2010 (http://epaa.asu.edu/epaa/v11n14/). Kao, Grace and Jennifer S. Thompson. 2003. “Racial and Ethnic Stratification in Educational Achievement and Attainment.” Annual Review of Sociology 29:417-42. Kim, Doohwan and Barbara Schneider. 2005. “Social Capital in Action: Alignment of Parental Support in Adolescents’ Transition to Postsecondary Education.” Social Forces 84(2):1181-206. King, Gary, James Honaker, Anne Joseph, and Kenneth Scheve. 2001. “Analyzing Incomplete Political Science Data: An Alternative Algorithm for Multiple Imputation.” American Political Science Review 95: 49-69. Lee, Jaekyung and Walter G. McIntire. 2000. “Interstate Variation in the Mathematics Achievement of Rural and Nonrural Students.” Journal of Research in Rural Education 16(3):168-181. Lichter, Daniel T. and Diane K. McLaughlin. 1995. “Changing Economic Opportunities, Family Structure, and Poverty in Rural Areas.” Rural Sociology 60(4):688-706. Lippman, Laura., Shelley Burns, and Edith McArthur. 1996. Urban schools: The challenge of location and poverty. Washington, DC: U.S. Department of Education, National Center for Education Statistics. McNeal, Ralph B. Jr. 1999. “Parental Involvement as Social Capital: Differential Effectiveness on Science Achievement, Truancy, and Dropping Out.” Social Forces 78(1):117-44. Monk, David H. 2007. “Recruiting and Retaining High-Quality Teachers in Rural Areas.” The Future of Children 17(1):155-74. O’Hare, William P. 2009. The Forgotten Fifth: Child Poverty in Rural America. Durham, NH: Carsey Institute, University of New Hampshire. (Also available at http://www.carseyinstitute.unh.edu/publications/Report-OHare-ForgottenFifth.pdf). Rogers, Carolyn C. 2005. “Rural Children at A Glance.” Economic Information Bulletin Number 1. Washington, DC: U.S. Department of Agriculture Economic Research Service. Retrieved March 4, 2010 (http://www.ers.usda.gov/publications/eib1/eib1.pdf). Rojewski, Jay W. 1999. “Career-Related Predictors of Work-Bound and College-Bound Status of Adolescents in Rural and Nonrural Areas.” Journal of Research in Rural Education 15(3):141-56. Roscigno, Vincent J. and Martha L. Crowley. 2001. “Rurality, Institutional Disadvantage, and Achievement/Attainment.” Rural Sociology 66(2):268-93. Rosenbaum, Paul R. and Donald B. Rubin. 1983. “The Central Role of the Propensity Score in Observational Studies for Causal Effects.” Biometrika 70:41-55. ----------------------. 1984. “Reducing Bias in Observational Studies Using Subclassification on the Propensity Score.” Journal of the American Statistical Association 79:516-24. ----------------------. 1985. “Constructing a Control Group Using Multivariate Matched Sampling Methods that Incorporate the Propensity Score.” American Statistician 39:33-38. Schneider, Barbara and David Stevenson. 1999. The Ambitious Generation: America’s Teenagers, Motivated but Directionless. New Haven: Yale University Press. Schneider, Barbara, Martin Carnoy, Jeremy Kilpatrick, William H. Schmidt, and Richard J. Shavelson. 2007. Estimating Causal Effects Using Experimental and Observational Designs. Washington, DC: American Educational Research Association. Sewell, William H. and Hauser, Robert M. 1972. “Causes and Consequences of Higher Education: Models of the Status Attainment Process.” American Journal of Agricultural Economics 54(5):851-61. Smith, Mark H., Lionel J. Beaulieu, and Anne Seraphine. 1995. “Social Capital, Place of Residence, and College Attendance.” Rural Sociology 60:363-80. Steelman, Lala Carr and Brian Powell. 1991. “Sponsoring the Next Generation: Parental Willingness to Pay for Higher Education.” American Journal of Sociology 96(6):1505-29. U.S. Department of Education, National Center for Education Statistics. 2004. Digest of Education Statistics. Washington, DC: U.S. Government Printing Office. Walpole, Marybeth. 2003. “Socioeconomic Status and College: How SES Affects College Experiences and Outcomes.” The Review of Higher Education 27(1):45-73.