Tracking is a form of ability grouping in American secondary schools that has been viewed as the institutional mechanism by which students are selected or channeled for different educational experiences. It is often considered the principal means of academic stratification in high schools (Heyns 1974). Since earlier academic experiences affect subsequent educational and occupational experiences, students' location in a tracking system may have consequences for their career trajectories within and beyond the educational system.

This article reports findings on students' initial curriculum placement in U.S. schools using data from High School and Beyond (HSB). It focuses on the major criteria used for placement and on several structural properties that may specify the conditions under which some criteria may affect placement more than others. The main reason for this focus is that the criteria that affect the track location of students can do much to shape the quality of students' educational experiences, as well as the nature of the tracking system itself, and thus the subsequent amount and kind of mobility in it (Rosenbaum 1976, 1980; Sorensen 1970; Vanfossen, Jones, and Spade 1987). It is therefore important to identify the various criteria used to place students in the tracking systems of high schools to further understanding of tracking and its role in the schooling process in general. Given the significance of track placement, we concur with Hallinan (1992) that it is surprising that there is such a paucity of research on the determinants of track placement.

PREVIOUS RESEARCH

In the study presented in this article, our perspective differed from that of many previous studies (such as Alexander, Cook, and McDill 1978; Hauser, Sewell, and Alwin 1976; Rosenbaum 1980) on tracking that have approached track placement as part of the process of individual educational and status attainment and have ignored relevant organizational features of schools. Therefore, it contributes to the growing literature (see, for example, Gamoran 1992; Garet and Delany 1988; Hallinan 1992; Useem 1992) on the role of school characteristics and policies in shaping the tracking system and thus in affecting the opportunity structure that students face. In this literature, track placement, like educational attainment, is considered to be a consequence of the interaction between individual abilities and efforts, on the one hand, and the opportunities generated by the school, on the other hand (Hallinan and Sorensen 1983).

Schools are social organizations that, like other people-processing organizations, influence the individuals in them. As part of the opportunity structure of schools, tracking is a structural arrangement that is designed to accommodate the differences in the knowledge and skills that students bring to schools. The criteria and procedures used to place students in curricular tracks are shaped by schools' organizational characteristics and policies on tracking. Thus, the effect of students' characteristics on students' track location is constrained by the opportunity structure of the schools students attend. Therefore, an analysis of track placement in high schools should include school-level and individual-level characteristics. The school-level variables examined here are characteristics of tracking systems, or contextual variables. Understanding the relationship of these school-level variables to track placement is important for a complete knowledge of the process and consequences of stratification in schools (Gamoran 1989; Rosenbaum 1984).

Individual Determinants of Track Placement

Earlier research (see, for example, Alexander and Cook 1982; Alexander et al. 1978; Alexander and McDill 1976; Hauser et al. 1976; Heyns 1974; Rehberg and Rosenthal 1978; Rosenbaum 1980) suggested that several characteristics of students may be associated with students' location in the tracking systems of high schools. Among these characteristics are gender, race-ethnicity, ability (as measured by standardized tests), socioeconomic status (SES), achievement (grades and/or test scores), and educational expectations or aspirations. The exact estimates of the influence of each of these individual-level variables fluctuate from study to study, but in each instance these variables predict track placement. The reasons for the variations in the effects of these variables may be the impact of organizational or school-level variables that were not measured in the foregoing studies that focused on individual educational attainment. Indeed, several research reports in the early 1980s began to show the impact of school-level variables with individual level variables on track or ability-group placement and the consequences of grouping practices (Barr and Dreeben 1983; Hallinan and Sorensen 1983; ones, Vanfossen, and Spade 1985), and Rosenbaum (19841 argued that tracking systems in some types of schools may be .more responsive to ability and other individual characteristics than may others. Since then, more attention has been given to organizational variables in research on tracking (Gamoran 1987, 1992; Kilgore 1991).

Organizational Dimensions of Tracking Systems

An early theoretical treatment of the impact of the interaction of organizational variables with individual-level variables on both track placement and the cognitive and noncognitive consequences of tracking was articulated by Sorensen (1970, who focused on how schools may attempt to differentiate students and identified several organizational dimensions for the distribution of students in a tracking system. Specifically, Sorensen argued that schools may differ in terms of (1) inclusiveness, the extent to which several curricular tracks are available to students; (2) selectivity, the amount of differentiation intended by the school; (3) electivity, the extent to which students' choice is allowed to be a determining factor in track placement; and (4) scope, the extent to which students are kept together in classes or tracks throughout their high school careers.

Sorensen suggested that these dimensions may define the nature of the tracking system that a school devises to differentiate students. Schools may vary on these dimensions in ways that are likely to produce tracks with students with different characteristics. What is important is that the organizational dimensions then provide the context for the operation of individual-level variables and thus the interplay of organizational and individual-level variables in determining the location of students in different tracks. Two of these dimensions and another organizational characteristic are discussed in this article.

Inclusiveness. Inclusiveness refers to the number of different educational opportunities provided by a school. It is a measure of the extent to which a school attempts to give its students one or more types of education. A school that is high on inclusiveness attempts to give all or most of its students the same quality and quantity of education, usually academic or college preparation; thus, we would expect that all or most of the students would be in one curricular track. A school that is low on inclusiveness attempts to differentiate the quality and quantity of education among its students; it would have several tracks with sizable proportions of its students assigned to each.

The dimension of inclusiveness has been viewed as being most appropriately applied to European educational systems in which schools often specialize in one type of education, for example, academic or technical (Sorensen 1970). However, it may also be used to analyze American schools on this dimension in terms of the emphasis given to one or more types of education. The differential emphasis on one type of education, such as college preparation in the academic track, may be related to a school's tracking pattern.

It could be expected that students' characteristics relate more strongly to track placement in schools that are low on inclusiveness than in schools that are high on inclusiveness. Specifically, inclusiveness may affect the relationship of students' SES to track placement. Sorensen (1970) argued that the higher the inclusiveness of a school, the greater the chances that students will aspire to a high level of education. He also noted that since the family is an important source of influence on educational aspirations, "parents as well as students will adjust their educational aspirations to the inclusiveness of the educational system" (p. 364). Following Sorensen, we predicted that "high inclusiveness ... will weaken the effect of family background on aspirations, low inclusiveness will strengthen it." Therefore, we expected that students' characteristics would relate more strongly to track placement in schools that are low on inclusiveness than in schools that are high on inclusiveness.

Electivity. "By electivity is meant the degree to which students' own decisions are allowed to be a determining factor in the assignment to groups" (Sorensen 1970:361). This dimension refers to the extent to which students choose or are assigned to their curricular tracks. Schools that are characterized by higher electivity are those in which greater proportions of students choose tracks and only small proportions, if any, are assigned to them. These high-electivity schools with a high degree of student choice thus have an opportunity structure that is crated by students themselves, although the students undoubtedly receive much advice from school officials, parents, and others. In contrast, low-electivity schools are those in which smaller proportions of students choose their tracks; a school with this feature assigns or allocates most students to tracks. Low-electivity schools have an opportunity structure that is created by school officials with little or no participation by students in decisions about track placement.

However, it is unlikely that many schools either assign all their students to tracks or allow all their students to choose tracks. For example, Lee and Bryk (1988) reported that a higher proportion of students in Catholic schools than in public schools are assigned to tracks.

Sorensen (1970) also suggested that individual and school characteristics may interact to yield track-placement patterns that would not be discernible if either student or school variables were used to predict students' location in the it tracking system. Policies of high electivity may promote track-assignment patterns of students that are different from those promoted by policies of low electivity because the criteria for track placement vary among schools with different degrees of electivity. In tracking systems characterized by high electivity, the criteria for placement may be noncognitive and perhaps irrelevant to learning. That is, because students', parents', and peers' values, interests, and preferences are given free rein in this opportunity structure, ability, grades, or achievement may not be strongly associated with students' location in the tracking system, but SES, race, ethnicity, and gender may be differentiating criteria. Cognitive criteria are more likely to be applied in low-electivity schools, since school officials may be expected to view these criteria as more relevant to learning.

Sorensen's conceptualization of electivity yields the following hypothesis regarding track placement: In schools that assign larger proportions of their students to tracks (low-electivity schools), cognitive criteria, such as ability or grades, are more strongly associated with track placement, whereas in schools that allow larger proportions of students to choose their tracks (high-electivity schools), noncognitive criteria are more strongly related to track placement. In other words, the interaction of school electivity and students' ability will result in high-ability students in low-electivity schools being in a more goal-oriented curriculum, that is, in the college preparatory academic track or the vocational track. The interaction of school electivity and students' SES will result in high-SES students in high-electivity schools being in a curriculum that is less oriented to specific goals, that is, the general track.

School SES composition. The final school-level characteristic of interest here is the SES composition of the school. Although Sorensen (1970) did not identify it as an organizational dimension that affects tracking systems, many critics (see, for example, Apple 1982; Bowles and Gintis 1976; Rehberg and Rosenthal 1978) have argued that whether a school allocates students by social class and thus engages in class reproduction is an important issue to examine. Tracking is often cited as the mechanism that perpetuates patterns of society stratification. It is also possible that some schools may attempt to enhance, inhibit, or overcome class reproduction, perhaps because of the nature of the clientele and communities that they serve. Hence, depending on the SES of the clientele, a tracking system may be more structured or less structured to meet the educational needs of a school's students.

Two case studies of schools with students of different SES suggest that track-placement patterns and the criteria for placement in tracks may vary by the SES of schools. In the white-collar suburban schools studied by Rehberg and Rosenthal (1978), 60 percent of the students were initially located in the academic track, and this proportion increased as students progressed to their senior year: ability was the primary criterion for placement. In contrast, in the White working-class school studied by Rosenbaum (1976), more than one-third of the class of 1971 was found in the academic tracks initially, and this proportion decreased by the time the students were in their senior year; ability was only one of several correlates of track location.

The evidence from these two case studies suggests that schools that serve students of various SES backgrounds have tracking systems that differ in how students are distributed among tracks and in the criteria associated with these distributions. Furthermore, if schools of varying SES composition do have different tracking systems, then the effect of ability and students' SES (and other individual-level predictors) on track placement may also vary among schools. We examined these possibilities by investigating the effect of the interaction of school SES, individual ability, and individual SES on track placement.

We identified three organizational characteristics of schools that may shape the nature of the tracking system and the location of individual students within it. The remainder of this article explores whether these organizational conditions affect how students are tracked and whether the relationship of the individual-level variables of ability and SES to track placement vary under different school organizational conditions.

METHODS

The data for this research were based on the longitudinal data file for the sophomore cohort of the HSB study. They included information collected in 1980 when the respondents were sophomores and in 1982 when they were seniors. The original sample was a two-stage, stratified probability sample. Over 1,100 public and private schools were selected in the first stage, and 36 students were chosen from each school in the second stage. Except for certain special strata, which were oversampled, schools were selected with a probability proportional to estimated enrollment in their 10th and 12th grades. Responses from the student data file were weighted to hold constant the total sample size but to take into account the disproportionate sampling of specified subgroups.

Analytic Procedures

Our research questions concerned the importance of school characteristics in the determination of track placement. Since previous studies showed that individual characteristics, such as SES background, cognitive performance, and educational aspirations, are related to track placement, we examined the effect of school characteristics while controlling for these individual-level predictors.

Because track is a nominal variable, it was not possible to establish a linear order among the three tracks. Thus, the analytic strategy we used, multinomial logistic regression, is appropriate for this type of dependent variable. Multinomial logistic regression analysis makes it possible to assess the effect of various individual- and school-level variables on track placement without making any assumptions about whether the various tracks are linearly related. It provides estimates of the effects of the independent variables on the likelihood that students are located in the academic or vocational track compared to the general track with all variables held constant. Although it would be desirable to use a multilevel methodology that crosses the students and school levels, a multilevel analytic strategy for a multinomial logit, model is not readily available.

Multinomial logistic regression analysis allows one track to be identified as the point of comparison. We used the general track as the track with which we contrasted students' location in either the academic or the vocational track.

Variables

Track placement was measured by a question in the sophomore questionnaire: "Which of the following best describes your present high school program?" Responses were grouped into three categories: academic, general, and vocational. Academic was coded 1, vocational was coded 2, and general was coded 3.

Individual characteristics. Gender and the two race-ethnic variables were coded similarly. Gender was coded 1 for females and 0 for males. Hispanic was coded 1, and non-Hispanic was coded 0; Black was coded 1, and non-Black was coded O. Eighth-grade educational expectations were measured by a question that asked sophomores if they had expected to go to college when they were in the eighth grade. The responses were coded 1 = "yes," 2 = "not sure," 3 = "hadn't thought about it," and 4 = "no." Grades by the 10th grade were taken from a question in which students were asked which response best described their grades so far; the responses ranged from "mostly As" (coded 1) to "mostly Ds" (coded 8).

Family SES was indicated by a composite scale score based on family income; father's education; mother's education; father's occupation; and the average score for the presence of eight household items, such as newspapers, books, and a typewriter. Ability was indicated by a composite score of measured test performance, which is a combination of the scores on the mathematics, reading, and vocabulary tests taken in the sophomore year.

School characteristics. School electivity was calculated by aggregating by school the students' responses to whether they were assigned to or chose their track. The proportion of students who chose their track in each school was the value for this variable. The measure of school academic inclusiveness was the proportion of 10th-grade students in the academic track of the student's high school taken from the principals' questionnaire. Although there may be some advantages to using an aggregate measure of inclusiveness based on students' responses to the question about their high school curricular program, doing so may contaminate the dependent variable. Data taken from the school questionnaire avoids this possible problem. School SES was calculated by aggregating students' SES scores and deriving a mean for each school (for the means and standard deviations of the variables, see Table 1). (Table 1 omitted)

FINDINGS

Preliminary analyses revealed that the three school-level variables were highly correlated (.90+) to ability and the interaction terms containing the school and ability measures. To remove the possibility of multicollinearity, we used the centering procedure suggested by Cronbach (1987), which involves subtracting the mean of each of these variables from each student's score on them and then using the result as the respondent's score on each of the centered variables. Students' SES and ability (composite test score), electivity, inclusiveness, and school SES were centered, as were all six interaction terms. The centering process yielded no correlations among the affected variables higher that .57, thus eliminating the threat of multicollinearity.

Two types of findings are presented in Tables 2 and 3. (Tables 2 and 3 omitted) The figures in Table 2 are the logit coefficients and their standard errors; the log-odds ratios or odds are presented in Table 3. The data in Table 3 are odds ratios associated with a one-unit change in the dependent variable. The odds ratio is the antilog of the logit coefficient and is similar to the partial slope in multiple linear regression. It describes the net multiplicative impact on the odds of a unit increase of a predictor, controlling for all other effects in the model (Agresti 1990; DeMaris 1993).

Individual-level Predictors

The results of the multinomial regression analysis using only the student characteristics are presented in the first two columns of Tables 2 and 3. These are the base models for each contrast; Column 1 contains the logits/odds ratios for the comparison of the academic versus the general track, and Column 2 contains the logits/odds ratios for the comparison of the vocational versus the general track.

All the independent variables distinguish between students in either the academic or vocational track, on the one hand, and students in the general track, on the other hand. Students' chances of being in the academic track, rather than in the general track, rise as their grades, ability scores, SES origins, and educational expectations in the eighth grade increase and if they are non-Hispanic, Black, or female (Column 1, Table 3). The odds ratios indicate that Blacks, for example, are 2.10 times more likely than are non-Blacks to be in the academic track. But students' chances of being in the academic, rather than the general, track decrease if they are Hispanic, non-Black, or male and if they have lower grades, ability, SES, or educational expectations.

In the comparison of students in the vocational track to those in the general track, the probability of being in' the vocational track increases if students are Black or female and have lower educational expectations and grades. The odds of being in the vocational track decrease for students of lower SES origins and lower ability; specifically, the odds are .94 times as great as those for higher-ability students and only .80 times as great as those for higher-SES students. Also, Hispanic and higher-SES students are more likely to be in the general track than in the vocational track (see Column 2, Table 3).

School-level Predictors

Each of the school variables compares students in either the academic or the vocational track to those in the general track.

Inclusiveness. The higher the inclusiveness of the school, the greater the likelihood that students are in the academic or in the general track, rather than in the vocational track (Table 3, columns 7 and 81. More specifically, students in schools with higher inclusiveness are 1.29 times more likely to be in the academic than in the general track.

School SES. The higher the SES of the school, the more likely that students are in the academic track, rather than in the general track, and in the general, rather than in the vocational track (see columns 11 and 12). When controls are in place for all the individual-level variables, students attending higher SES schools are 1.12 times more likely to be in the academic track. However, the odds are somewhat lower that students will be in the vocational versus the general track.

Interaction of Students' SES and Ability with School-level Predictors

We also found that for each comparison, the effect of the original school variable retains its strength and significance when the interactions with students' ability and SES are included in the equations (see Table 2, columns 5, 6, 9, 10, 13, and 14). Since the interaction terms are also significant, this finding suggests that the influence of the school variable operates on track placement directly, as well as through its interactions with students' SES and ability. That is, the impact of school characteristics on track placement varies not only with the level of the school characteristic, but with the SES background and ability of the student.

Specifically, the higher the electivity of the school, the greater the chances that higher-ability students will be in either the academic or vocational than in the general track. Thus, when more students can choose their tracks, higher-ability students are more likely to choose a more goal-oriented track than when students are assigned to tracks (see Table 2, columns 5 and 6).

The results for the interaction of electivity and students' SES are the reverse: The higher the electivity of the school, the greater the chances that higher-SES students will be in the general track than in either the academic or vocational track (see Table 3, columns 5 and 6). That is, when schools let students choose, higher-SES students are more likely to be in the general than in the academic or vocational track than when more students are assigned to tracks. Conversely, when schools assign students, the lower the SES of the students, the greater the chances of an academic or vocational track placement.

The contrast in the interactions of students' ability and SES with school electivity suggests that the greater the tendency of schools to assign students to tracks (lower electivity), the greater the schools' reliance on students' ability as the criterion of placement and the lesser their tendency to favor higher-SES students. In lower-electivity schools, the odds are greater that lower-ability students are in the general track, but in these same schools, lower-SES students are more likely to be in the academic and vocational tracks. If a student's SES was the major criterion for assignment to a track, then we would expect that higher-SES students would be in the academic and vocational tracks in schools where more students are assigned to tracks.

Academic inclusiveness interacts positively with students' SES and negatively with ability in each comparison. The greater the inclusiveness of the school, the higher the odds that higher-SES students will be in either the academic or vocational tracks than in the general track and that higher-ability students will be in the general track than in either of the other two tracks (Table 3, columns 9 and 10). The effect of the inclusiveness-by-SES interaction term is the strongest among the interaction terms that include inclusiveness.

The interaction terms that include school SES are also significantly related to track placement (Table 3, columns 13 and 14). The higher the SES of the school, the greater the chances that higher-ability students will be in the academic and vocational tracks, but these effects are admittedly weak. The effects of the other interaction term differ for each comparison. In higher-SES schools, higher-SES students are more likely to be in the academic than in the general track or in the general than in the vocational track.

Finally, we examined the effects of each of the three organizational variables, controlling for the other two by adding them to the equation simultaneously (Table 2, columns 15 and 16), followed by the concurrent addition of all the interaction variables (Table 2, columns 17 and 18). Each of the three school variables significantly differentiates between academic and vocational versus general track placement with controls in place for each of the other two school-organizational variables. Furthermore, the effects of each of the school variables are the same as when they were entered in the equation singly for each comparison, with one exception: The higher the school SES, the greater the odds that students will be in the vocational track than in the general track, when the other school variables are controlled.

Another pattern evident in these final equations (Table 3, columns 15 and 16) is that different school variables affect each of the sets of track-placement comparisons. On the one hand, electivity has its greatest effect on the academic-general comparison: The greater the electivity of a school, the greater the odds of students being in the academic track, even with controls in place for the other school variables and the six interaction terms. On the other had, the strongest effect of inclusiveness is on differentiating students in the vocational general track comparison: The greater the inclusiveness of a school, the greater the odds of students being in the general track, also with controls in place for all other variables.

These patterns carry over into the last equations as well (Table 2, columns 17 and 18). In addition, the last models show that with the inclusion of all the interaction terms, all variables retain their significance to track placement in both sets of comparisons. The interaction terms that include ability have a consistently positive effect on track placement, regardless of which two of the school variables--electivity or school SES--makes up the other half of the interaction term; higher ability students in schools with the higher characteristic of the school variables are more likely to be in the academic or vocational track. The other interaction terms containing SES do not show a distinct pattern.

DISCUSSION AND CONCLUSION

The findings reported here revealed that the track placement of students is complex and dependent on properties of schools and students' characteristics. Four major points emerged from our analysis:

1. The organizational and compositional characteristics of schools do affect the track placement of students. School electivity, academic inclusiveness, and the socioeconomic composition of the school all are independently related to the location of students in the tracking system.

1. Sometimes, the effects of student variables on track placement are magnified through their interaction with the school variables.

3. The effects of student and school variables differ or placement in the three curricular tracks.

4. Students with similar individual characteristics may find themselves in different tracks, depending on the characteristics of the schools they attend.

Electivity

Some interesting and unexpected findings concerned school electivity. One is that the greater the tendency of schools 1 to assign students to tracks, (low electivity), the greater the likelihood that students will be placed in the general track, rather than in either the academic or vocational track. When students have greater choice or high electivity, then they are more likely to end up in a career- or goal-oriented track (academic or vocational).

Why would school officials place more students in the general track than would be the case if students chose their track? One reason may be the constraints incurred by organizational factors on the capacity of administrators and teachers to fit students and curricula. For example, it is conceivable that schools are high assigning because they do not have the resources to provide the curricula appropriate for all their students, given that they are probably operating under a charge to provide a comprehensive curriculum. That they simply may not be able to hire all the English, mathematics, science, and vocational teachers that they need and still offer viable general programs may increase the prospect that some students who are qualified for the academic track or interested in the vocational track may be placed in the general track. These and other related reasons are dealt with more extensively by DeLany (1991); Pallas, Natriello, and Reihl(1994); and Useem (1991).

Kilgore (1991) offered some additional reasons why schools' adaptive capacities may be limited. She suggested that schools may be high assigning because their small size would limit the number of students in each curriculum simply because shop teachers may not be qualified to teach academic-track courses, and vice versa. Union contracts may also restrict class sizes, which, with the particular configuration of teachers' qualifications to teach specific subjects, may limit the number of students in each track. A high degree of tension between teachers and the school administrators or board may also affect teaching assignments in ways that constrain the range of r curricular options for students. Finally, if the control of tracking decisions in high-assigning schools is in the hands of school administrators, who typically have less information about students than do teachers, placements in tracks other than those predicted on the basis of cognitive criteria are more likely to occur.

In the high-electivity schools, the odds are that students are more likely to choose the more goal-oriented tracks (academic and vocational). This finding is inconsistent with Sorensen's (1970) prediction that under the laissez-faire conditions of low assignment, students will choose the track that requires less effort. Some reasons why students are more likely than expected to choose the academic track than the general track come from Finley's (1984) study of tracking in a high school English department. Finley found that in schools in which students are not explicitly also signed to tracks, teachers with the most seniority may compete with other teachers for preferred students by actively encouraging the most able or motivated students. or the easiest or most rewarding to teach, into their academic-track classes. The process reverberates down the tracking system, from the honors or accelerated classes to the more basic or remedial-track classes. Thus, students who would otherwise he assigned to the general track may end up in the academic track because they were encouraged by teachers. In addition, schools may be characterized by high electivity for reasons that are the opposite of those cited for why schools may be identified by low electivity: that is. they may have fewer constraints on their capability to adapt to students' needs or qualifications. Consequently they can probably accommodate more students in the academic and vocational tracks than can the low-electivity schools.

Our analyses also provided some insights into what happens to different SES and ability groups of students in schools with different levels of electivity. Electivity interacts with both individual SES arid ability in each comparison. In higher-electivity schools the already strong effect of individual SES is reversed in that higher-SES students are more likely than expected to be in the academic or vocational track. This finding raises the possibility that school officials may use cognitive criteria to place students, rather than their knowledge of the SES origins of students. The effects of the interaction of electivity and students' ability are different: In higher-electivity schools, students of higher ability are more likely than expected to be in the academic or vocational track than in the general track. In the vocational-general comparison, students attending higher-electivity schools have a better chance of being in the vocational track than their individual ability predicts.

More generally, our findings on electivity provide an empirical illustration of the importance of this school variable to the process of track placement. As a school variable, electivity contributes to the structural contour of the opportunities furnished by tracking systems.

Inclusiveness

The results of our research also substantiate the utility of the concept of academic inclusiveness for studies of track placement, as was reported by Lee and Bryk (1988) and Gamoran )1987). We did not find support for the final hypothesis (derived from Sorensen 1970) that high inclusiveness will weaken the effect of family background (SES) on track placement while low inclusiveness will strengthen it. Rather, our results suggest that the organizational policies followed by the school may not minimize or maximize the differentiating characteristics of social-class origins on track placement. Perhaps if different measures of the concepts we studied were operationally defined differently the results would provide support for Sorensen's hypothesis.

Our findings on the effect of inclusiveness on track placement are all the more salient, given Gamoran's (1992) findings on the impact of this variable on achievement. Gamoran reported that inclusiveness is nonlinearly related to the between-track inequality of mathematics achievement and that mean school verbal and mathematics are also affected by inclusiveness. Together, these two sets of findings suggest that the academic inclusiveness of the school affects the initial track placement of students and subsequently affects educational achievement as well, when individual background and ability measures are controlled. Thus, the school-level variable, inclusiveness, warrants inclusion in future research on both track placement and the effects of tracking on achievement.

School SES

The finding that school SES is related to track placement provides some support for the argument that schools are engaged in reproducing the stratification system of the larger society. The higher the SES of the school, the greater the likelihood that students will be in the academic track than in the general tack. In lower-SES schools, students are more likely to be in the vocational track than in the general track. Furthermore, school SES interacts with both students' ability and SES to increase the odds that higher-ability and higher-SES students will be in the academic, rather than the general, track. In the comparison of the vocational and general tracks, school SES interacts with students' SES showing that (1) students with higher-SES origins are more likely to be in the academic track and (2) lower-SES students are more likely to have a vocational track placement in lower-SES schools. Thus, lower-SES schools may specialize in and expand their vocational track offerings, encouraging larger numbers of students (particularly lower-SES students) to go into vocational courses than is the case with middle- or upper-SES schools.

The fact that this research revealed systematic differences in track placement among types of schools suggests that track placement is more complex and more dependent on organizational dimensions of schools than previous research has shown. It also indicates that measures of organizational differentiation should be included in research on track placement and perhaps in more general studies of tracking systems and processes. To the extent that electivity academic inclusiveness, and the socioeconomic composition of the school can be viewed as dimensions of the school opportunity structure, our findings indicate that different opportunity structures frequently yield equally distinctive patterns of curricular track placement. Thus, schools do not track students in the same way, and a variety of tracking arrangements exist in our nation's schools, as Rosenbaum (1984) contended.

REFERENCES

Agresti, Alan. 1990. Categorical Data Analysis. New York: John Wiley & Sons.

Alexander, Karl L. and Martha Cook. 1982. "Curricula and Coursework: A Surprise Ending to a Familiar Story." American Sociological Review 47:626-40.

Alexander, Karl L., Martha Cook, and Edward L. McDill. 1978. "Curriculum Tracking and Educational Stratification: Some Further Evidence." American Sociological Review 41:963-80.

Alexander, Karl and Edward L. McDill. 1976. "Selection and Allocation Within Schools: Some Causes and Consequences of Curriculum Placement." American Sociological Review 43:47-66.

Apple, Michael W. 1982. Education and Power. Boston: Routledge & Kegan Paul.

Barr, Rebecca and Robert Dreeben. 1983. How Schools Work. Chicago: University of Chicago Press.

Bowles, Samuel and Herbert Gintis. 1976. Schooling in Capitalist America. New York: Basic Books.

Cronbach, L. 1987. "Statistical Tests for Moderator Variables: Flaws in Analysis Recently Proposed." Psychological Bulletin 102:414-17.

DeLany, Brian. 1991. "Allocation, Choice, and Stratification Within High Schools: How the Sorting Machine Copes." American Journal of Education 99:182-207.

DeMaris, Alfred. 1993. "Odds versus Probabilities in Logit Equations: A Reply to Roncek." Social Forces 71:1057-65.

Finley, Marilee K. 1984. "Teachers and Tracking in a Comprehensive High School." Sociology of Education 57:233-43.

Gamoran, Adam. 1987. "The Stratification of High School Learning Opportunities." Sociology of Education 60:135-55.

--. 1989. "Measuring Curriculum Differentiation." American journal of Education 97:129-43.

--. 1992. "The Variable Effects of High School Tracking." American Sociological Review 57:812-28.

Garet, Michael S. and Brian DeLany. 1988. "Students, Courses, and Stratification." Sociology of Education 61:61-77.

Hallinan, Maureen T. 1992. "The Organization of Students for Instruction in the Middle School." Sociology of Education 65:114-27.

Hallinan, Maureen T. and Aage B. Sorensen. 1983. "The Formation and Stability of Instructional Groups." American Sociological Review 65:569-76.

Hauser, Robert M., William H. Sewell, and Duane F. Alwin. 2976. "High School Effects on Achievement." Pp. 309-41 in Schooling and Achievement in American Society, edited by William H. Sewell, Robert M. Hauser, and David L. Featherman. New York: Academic Press.

Heyns, Barbara. 1974. "Social Stratification Within Schools." American Journal of Sociology 79:1434-51.

Jones, James D., Beth E. Vanfossen, and Joan Z. Spade. 1985. "Curriculum Placement: Individual and School Effects Using the High School and Beyond Data." Paper presented at the annual meeting of the American Sociological Association, Washington, DC.

Kilgore, Sally. 1991. "The Organizational Context of Tracking in Schools." American Sociological Review 56:189-203.

Lee, Valerie E. and Anthony S. Bryk. 1988. "Curriculum Tracking as Mediating the Social Distribution of High School Achievement." Sociology of Education 61:78-94.

Pallas, Aaron M., Gary Natriello, and Carolyn Riehl. 1994. "Tweaking the Sorting Machine: The Dynamics of Students' Schedule Changes in High School." Paper presented at the annual meeting of the American Sociological Association, Los Angeles.

Rehberg, Richard A. and Evelyn R. Rosenthal. 1978. Class and Merir in the American High School. New York: Longman.

Rosenbaum, James E. 1976. Making Inequality: The Hidden Curriculum of High School Tracking. New York: John Wiley & Sons.

--. 1980. "Track Misperceptions and Frustrated College Plans: An Analysis of the Effects of Tracks and Track Perceptions in the National Longitudinal Survey." Sociology of Education 53:74-88.

--. 1984. "The Social Organizational of Grouping." In The Social Context of Instruction edited by P.L. Peterson, L.C. Wilkinson, and M. Hallinan. Orlando, FL: Academic Press.

Sorensen, Aage B. 1970. "Organizational Differentiation of Student and Educational Opportunity." Sociology of Education 43: 355-56.

Useem, Elizabeth L. 1991. "Student Selection into Course Sequences in Mathematics: The Impact of Parental Involvement and School Policies." Journal of Research on Adolescence 2:231-50.

--. 1992. "Getting on the Fast Track in Mathematics: School Organizational Influence on Math Track Assignment." American Journal of Education 100:325-53.

Vanfossen, Beth E., James D. Jones, and Joan Z. Spade. 1987. "Curriculum Tracking and Status Maintenance." Sociology of Education 60:104-22.

James D. Jones, Ph.D., is Professor of Sociology, Department of Sociology, Anthropology, and Social Work, Mississippi State University, Mississippi State. His main fields of interest are sociology of education, social stratification, and complex organizations. He is now completing a study of the impact of community and school structural variables associated with initial placement in high school curricular programs using NELS:88 data.

Beth Vanfossen, P.D., is Director, Institute for Teaching and Research on Women, and Affiliate Professor of Sociology, Towson State University, Towson, Maryland. Her main fields of interest are social stratification, gender inequality, sociology of education, and women and development. She is currently using a longitudinal database on 120 nations to research the impact of diverse trade strategies on women's participation in the formal labor force.

Margaret E. Ensminger, Ph.D., is Associate Professor, Department of Health Policy and Management, School of Hygiene and Public Health, Johns Hopkins University, Baltimore. Her main fields of interest are life-course development, impact of social structure on individuals' lives, and sociology of mental health. She is conducting a follow-up study of a cohort of inner-city children who are now in their early 30s to determine the pathways to health, employment, and drug abuse.

The research reported in this article was supported, in part, by Grant No. SES-8310687 from the National Science Foundation. Address all correspondence to Dr. James D. ones, Department of Sociology, Anthropology, and Social Work, Drawer C., Mississippi State University, Mississippi State, MS 39762, or by e-mail at jdj1(at)msstate.bitnet.