Printer Friendly

Longitudinal analysis of time, engagement, and achievement in at-risk versus non-risk students.

Longitudinal Analysis of Time, Engagement, and Achievement in At-Risk Versus Non-Risk Students

In the early 1980s, my colleagues and I reported a significant difference in the academic engagement of high- versus low-socioeconomic-status (SES) student groups (Greenwood, Delquadri, et al., 1981; Stanley & Greenwood, 1983). On a daily basis, high-SES students spent 11 min or 5% more time per day engaged in writing, reading, and talking about academic matters than did their low-SES counterparts. At this daily rate, we reported that low-SES students needed to attend school for as much as one and a half months during summer vacation to obtain an equivalent amount of engaged time in 1 year (Stanley & Greenwood, 1983).

Based on Carroll's time-based model of school learning (Carroll, 1963; 1985) and subsequent research on time variables (e.g., Berliner, 1988; Greenwood, Delquadri et al., 1981; Hall, Delquadri, Greenwood, & Thurston, 1982), we hypothesized that these low-SES students were at risk for academic delay and retardation because of their lower daily engagement rates. Carroll's model stated simply that the degree of learning was a function of (a) the time spent learning and (b) the time needed to learn. The time spent learning has most frequently been defined as the time allocated for instruction and the time within it during which the student was engaged (e.g., Denham & Lieberman, 1980). Our data illustrated that the time spent engaged in learning tasks was clearly reduced for low-SES students.

In addition, time variables (i.e., time allocated and time engaged) have been reported to be (a) positive correlates of academic achievement measures (e.g., Berliner, 1988; Brophy, 1979; Gettinger, 1985; Greenwood, Delquadri, & Hall, 1984; Rosenshine & Berliner, 1978; Rosenshine & Stevens, 1986) and (b) component processes of effective teaching practices, those producing increased levels of student achievement (e.g., Brophy, 1986; Stallings & Stipek, 1986).

Based on these findings, we sought to develop an instructional intervention that would at least increase students' levels of academic engagement and that teachers could adapt to the existing curricula (see Delquadri, Greenwood, Carta, Whorton, & Hall, 1986; Delquadri, Greenwood, Stretton, & Hall, 1983; Greenwood, Carta, & Hall, 1988). Findings from these studies and replications by other investigators indicated that classwide peer tutoring (CWPT) did increase both general and special education students' (a) time spent with specific instructional materials (e.g., readers, paper and pencils, worksheets, Greenwood, Delquadri, & Hall, 1984), (b) their levels of engagement (Greenwood, Dinwiddie, et al., 1984), (c) their daily and weekly measures of curriculum-based achievement (Greenwood, Delquadri, & Hall, 1984; Greenwood, Dinwiddie, et al., 1987; Maheady & Harper, 1987), and (d) their grades (Maheady & Harper, 1987; Maheady, Sacca, & Harper, 1988).

During 1983-1987, we conducted a longitudinal, experimental/control group study of the effects of CWPT in urban schools (Greenwood, Delquadri, & Hall, 1989). The CWPT program was implemented with the same inner-city cohort during Grades 1-4. Results from this study replicated and expanded the prior findings. After the 4-year tutoring intervention, the low-SES experimental group had made statistically significant and educationally important gains on the Metropolitan Achievement Test when compared with an equivalent control group not receiving the tutoring program (Greenwood, Delquadri, & Hall, 1989). Academic achievement effect sizes in this study ranged from 0.38 in mathematics to 0.60 in language.

The significant difference in engagement between high- versus low-SES students in this study also replicated the earlier findings of Greenwood, Delquadri, et al., 1981. High-SES students spent significantly more time engaged in academic behaviors during conventional instruction compared with low-SES students whose teachers also employed conventional instructional methods. In addition, it was demonstrated that low-SES spent more time in nonacademic activities (Greenwood, Delquadri, & Hall, 1989). However, because these data were summative--that is, they were aggregated over the entire investigation--it was impossible to observe trends in these data over grades and measurement occasions.

The goal of this investigation was to conduct additional analyses of these longitudinal data. The purpose was to examine the trends and timing of differences in specific processes of instruction in urban versus suburban schools and their relationship to academic products by the end of Grades 2 and 3. The primary questions addressed in these analyses were as follows:

1. What were the trends over time and differences between groups in (a) the time students received instruction in academic subjects; (b) the time students engaged in academic responding versus task management or competing, inappropriate behaviors; and (c) the time spent learning, defined as the product of time received and time engaged.

2. Are the differences between low- versus high-SES groups predicted by the work of Stanley and Greenwood (1983) evident earlier than Grade 4, and what is the timing of these differences over grades?

3. For low-SES students, what was the effect of CWPT on their time spent in instruction and engagement?

4. Do differences in classroom processes covary with students' gains in standardized achievement at the end of Grades 2 and 3?

METHOD

Design

The general design involved three groups, a control group (low-SES), and experimental group (low-SES), and a comparison group (high-SES), for whom classroom process and product variables were assessed. The experimental-group teachers implemented CWPT, whereas the control and comparison groups received the conventional, teacher-designed instructional program. All students received annual pretest and posttest product assessments. In addition, a subsample of students was randomly selected to also receive day-long process observations.

Sample

Subject Pool. A total pool of 35 schools was reduced to 25 after eliminating 10 due to the district's busing policy to achieve racial balance. Seventeen of these qualified as Chapter 1 (low-SES) and eight qualified as non-Chapter 1 (high-SES) schools. Random selections were made within these SES-status groups. The final sample consisted of six Chapter 1 schools and three non-Chapter 1 schools. Because two low-SES school faculties declined participation, additional random selection were made to replace them.

To limit experimental contamination between teachers assigned to different conditions, all teachers and students within a school were assigned to the same condition (e.g., experimental group). This meant that students within a school continued within the same condition over the four grades. They did not, however, necessarily remain in the same classroom groupings from year to year.

Teachers. Sixty-nine teachers (23, 22, and 24 in Grades 1 through 3, respectively), participated in this investigation. Their years of teaching experience ranged from 1 to 35 years, with a mean of 10 years, and did not differ across groups. After briefings by both district officials and project investigators, the faculties in the schools agreed to participate. For the first-grade teacher, participation was immediate, whereas for second- and third-grade teachers, participation meant waiting a year or more before direct involvement. In each year of the project, all teachers, regardless of their assigned condition, received either 3 hr of paid university credit or the dollar equivalent for their participation.

Students. The participants in this investigation included 416 first-grade students. A subsample of 115 students, 28% of the total, were randomly selected to receive classroom observations (i.e., n = 5 per first-grade classroom) in addition to standardized achievement test measures. The mean age of students in months was 80.5, ranging from 71 months to 104 months, and did not differ by groups, F(2,403) = 1.91, p = 0.15. Chi-square tests on gender and racial composition indicated no difference on gender, [X.sup.2](2, N = 416) = 2.05, p = 0.340, but a difference on race was indicated, [X.sup.2](8, N = 416) = 113.83, p = 0.001. The proportion of minority-group students (e.g., Blacks, Hispanics, and Orientals,), was 0.98 in the control group, 0.86 in the experimental group, and 0.02 in the comparison group.

Student Attrition

Of the original 416 first graders, complete data were available for 241 (58%) by the end of third grade. The percentages of students that had left the original nine schools were 35.0%, 48.3%, and 40.0% for control, experimental, and comparison groups, respectively. These rates of attrition were not significantly different across groups, [X.sup.2](2, N = 416) = 5.47, p = 0.065. Analyses were conducted of the comparability of the original versus final samples in terms of measured IQ using the BMDP programs (Dixon, 1988). Overall, differences between sample estimates were nonsignificant on the order of only 2 to 5 IQ points. Students were lost to the study primarily as a result of relocation out of the nine participating schools. The somewhat higher loss within the experimental group was due to the closing of a school not in the current investigation and the subsequent reassignment of students to nearby schools to achieve a better population balance in all schools in the district.

A loss of similar magnitude occurred within the 115 observation target subsample: only 56 students, or 49% of the original target sample, remained with complete data. As with the larger sample, this loss of observation target students was also balanced across all schools and all experimental conditions. The number of remaining target students were 16, 21, and 19 for the control, experimental, and comparison groups, respectively.

Measures

Process Measures. A direct observation system was used to measure classroom processes: The Code for Instructional Structure and Student Academic Response (CISSAR; Stanley & Greenwood, 1981; Greenwood & Delquadri, 1988). The CISSAR is a 53-code system containing both classroom ecology and student behavior event categories (see Table 1). It is organized into six subcategories of ecological variables (i.e., Academic Activities, Nonacademic Activities, Task, Structure, Teacher Position, and Teacher Behavior), and three subcategories of student behavior (i.e., Academic Response, Task Management, and Competing Response).

Momentary time sampling was used to pace observers' recording of classroom ecology and student behavior events (Powell, Martindale, & Kulp, 1975). Selected events were recorded sequentially every 10 seconds (s), with priority given to faster changing variables. Hence in a 70-s cycle, one recording of the activity, task, and structure events was made, whereas six recordings of the teacher's position, teacher's behavior, and student behavior were each made. This one-to-six sampling pattern was repeated throughout an entire observation, which lasted for the entire school day.

Students were observed during "prime" academic instruction time (excluding recess, lunch, and noninstructional times of the day). Six observations were completed for each student (i.e., fall and spring, in each of Grades 1 through 3). The mean number of intervals that prime time observations lasted was 973 per day overall or about 3 hours (hr) (range = 228 in Phase 2 to 1,537 in Phase 3). The mean intervals observed in order by Phase were 877; 807; 1,131; 1,093; 974; and 954. To ensure that variations in the time students were observed did not affect estimates of time spent in academic activities, Pearson correlations were computed between these variables at each phase. These correlations ranged from -0.24 to 0.25 across phases and were not statistically significant. Thus, variations in time observed were independent of the estimates of academic activities.

For each student, six observation scores were available for each of the 53 codes within the nine subcategories of the CISSAR. CISSAR scores were the percentage of each observation that each separate event occurred. The first observation was obtained in the fall of first grade before implementation of CWPT (i.e., baseline); the remaining five observations reflected the subsequent phases in which CWPT was implemented by the experimental group teachers. The CISSAR variables used in this investigation were the Academic Activities subcategory, its composite and separate codes, and the Student Behavior subcategory composites. The Academic composite was the sum of the separate percentages of occurrence for reading, math, spelling, handwritting, language, science, and social studies. Similar composites were formed for the Student Behavior subcategories as defined in Table 1.

Engagement in this study was defined as active, academic responding as previously employed in our research (e.g., Greenwood, Delquadri, & Hall, 1984). Engagement was a composite of the seven separate behaviors described in Table 1. In contrast to most definitions of engagement, however, it does not include students' attention, defined as looking at the teacher or at peers.

To examined time as a single combination of time spent in academic activities and time engaged in academic behavior, their product was computed as [(time taught) (time engaged)]. This time spent-time engaged composite variable ranged in value from 0 to 1.00. If both time taught and time engaged were high, then the resulting value was also relatively high, and vice versa. If one time variable was high and the other low, a moderate value resulted.

Interobserver-agreement checks were randomly sampled by observers once during each observation phase in each year. Agreement observations occurred for standard 14-min checks. The Kappa statistic which controls for chance agreement was used to compute an index of inter-observer agreement (Hollenbeck, 1978). The mean Kappa for ecological events was 0.64, and ranged from 0.0 to 0.94. The mean Kappa for student behaviors was 0.66, and ranged from 0.00 to 0.91. Values of Kappa above 0.60 are generally considered to be adequate levels of agreement (e.g., Hartmann & Woods, 1982, p. 126).

Ability Measure. The Otis-Lennon School Abilities Test-Primary I, Form R (Otis & Lennon, 1979) was used to obtain measured IQs. Reliability of scores based on the Kuder-Richardson Formula 20 is reported to be 0.88. The test was group-administered according to test instructions to all students in March of the first grade.

Product Measure. The Metropolitan Achievement Test-Basic Battery (Prescott, Balow, Hogan, & Farr, 1978) was used to measure academic growth between fall semester of first grade and spring semester of third grade. The test was selected to facilitate interpretation of the importance of the current findings relative to those of other intervention studies published in the literature in which the same test was employed (e.g., the national follow-through evaluation study, Becker, 1977).

The Preprimer (Grade 1 pretest), Primary II (Grade 2 posttest), and Elementary (Grade 3 post-test) levels of the test were used. Subtests included within each level of the test sampled the domains of reading, mathematics, and language. Kuder-Richardson Formula 20 reliabilities for the subtests were generally high ranging from 0.77 for mathematics (Primary II) to 0.96 for reading (Elementary) across levels of the test.

Classwide Peer Tutoring

CWPT is an instructional system in which tutor-tutee pairs work together on a classwide basis (Carta, Greenwood, Dinwiddie, Kohler, & Delquadri, 1987). At the beginning of each week, all students in a class are paired for tutoring. These pairs are then assigned to one of two competing teams. Tutees earn points for their team by responding to the tasks presented to them by their tutors. The winning team is determined daily and weekly based on each team's point total. Tutor and tutee roles are highly structured to ensure that tutees receive rapid-response trials in a consistent format and that a standard error-correction procedure is applied (e.g., see reviews by Delquadri et al., 1986; Greenwood, Carta, & Hall, 1988).

Teachers organize the academic content to be tutored into daily and weekly units and prepare materials to be used within the CWPT format. Tutoring occurs simultaneously for all tutor-tutee pairs involving the entire class at the same time. This leaves the classroom teacher free to supervise and monitor students' responding during tutoring sessions.

CWPT Teacher Training. Each year, the experimental group teachers were trained to implement the CWPT procedures. Teachers initially read a CWPT program manual that described the procedures (Carta et al., 1987; Greenwood, Delquadri, & Carta, 1988), and then discussed with their consultant the necessary changes in current classroom practices to be made. After the necessary planning and preparation of materials, consultants helped the teachers initiate the program in each classroom. Teachers were considered trained when they produced a CWPT implementation score of at least 85% on a checklist used by consultants. Consultants used the checklist to monitor teachers' CWPT implementation and discussed any deviations with the teachers. An additional standard component of the CWPT program included monitoring of students' weekly curriculum-based measures of achievement (i.e., weekly spelling tests, math tests, and reading-rate checks).

CWPT Classroom Implementation. The implementation plan called for CWPT to be first introduced to students during their regularly scheduled 30-min spelling period. CWPT was then introduced into a second (i.e., math), and later a third subject-matter area (i.e., reading), after both the teacher and students had established the required criterion level of implementation and were also satisfied with their progress. When fully implemented, CWPT was intended to be in place for 90 min per day in spelling, math, and reading. In actuality, however, variations on this plan occurred across teachers and years of the project. In each year, CWPT implementation by the experimental group teachers ranged from only the spelling period to all three academic periods; and the extent of implementation in each area was positively correlated with students' achievement gains (see Greenwood, Delquadri, & Hall, 1989).

Teacher-Designed Instruction

Teachers in the control and comparison groups employed their own instructional practices, based on district guidelines. Reading procedures included use of basal texts and a typical three-reading-group structure. In math and spelling, teachers' procedures also included use of a text, lecture and teacher-student discussion formats, and seatwork assignments. Texts used were the Macmillan Series R (1980), Elementary Mathematics (1983) or Health Elementary Mathematics (1983), and HBJ Spelling (1983).

Low-SES students who had been identified by district personnel as academically at risk left their regular classroom program for periods of the day to attend Chapter 1 instruction in reading or math. These programs consisted of lower-teacher-pupil ratios, programmed instructional materials, and individual study opportunities. Chapter 1 instructional programs were not available to students in the high-SES comparison group.

Data Analyses

Factorial analyses of variance (ANOVAs) and covariance (ANCOVAs) with repeated measures were used. Where multiple dependent measures or covariates were involved, multivariate analyses of variance (MANCOVAs) were conducted. Separate analyses were applied to the process and product data.

Process Analyses. A 3 x 6 factorial design, with repeated-measures involving groups (i.e., Control versus Experimental versus Comparison) and phases (1 through 6), was used to analyze the Academic Activities subcategory, the three Student Behavior subcategory composites, and the Time Spent composite. A univariate analysis was run on each composite. In the case of Academic Activities, the univariate analysis was followed by a multivariate analysis to isolate group differences in time spent in specific instructional activities.

Product Analyses. Because tests of initial group differences in first-grade IQ and reading achievement were significant, a multivariate factorial design was used that employed IQ and pretest reading achievement scores as covariates. A 3 x 2 factorial design, with repeated measures involving groups (i.e., control, experimental, and comparison) and phases (Grade 2 versus Grade 3), was used to analyze the product data. To estimate the practical significance of the observed outcomes, effect sizes were computed for both the experimental and comparison groups relative to the control group, based on differences in covariate-adjusted mean scores.

RESULTS

Classroom Processes

Time Spent in Academic Activities. ANOVAs performed on the Academic Activity composite indicated a significant effect for Groups and for Phases (see Table 2). The group effect means were 85.2% (SD = 4.0), 85.2% (SD = 3.4), and 89.9% (SD = 3.4) for the control, experimental, and comparison groups, respectively. Post hoc tests indicated that differences occurred between the comparison group and both the control and the experimental groups. Overall, the comparison group students spent 4.7% more of prime time per day in academic instruction, compared with either the control or the experimental group.

Over phases and grades, the time spent in Academic Activities by all students was generally high and stable. The Phase means, in order, were 86.8% (SD = 10.3), 85.5% (SD = 8.0), 87.7% (SD = 8.5), 83.&% (SD = 9.1), 88.0% (SD = 5.9), and 89.2% (SD = 5.2) for the six phases.

MANOVA was used to locate the specific differences in groups' time spent in particular academic activities (see Table 2). The multivariate groups effect was significant. Univariate differences were noted for reading, science, and handwriting. Post hoc tests indicated that the comparison group significantly exceeded the other two groups in reading and science. The time spent in reading instruction was 34.9% (SD = 6.6), 32.7% (SD = 6.4) and 38.6% (SD = 7.3) for the control, experimental, and comparison groups, respectively. The time spent in science instruction was 2.1% (SD = 2.3), 2.2% (SD = 2.1), and 4.0% (SD = 3.0), respectively, for these same groups.

Student Behavior. The ANOVA for the academic response composite indicated a significant main effect for Groups, Phases, and Groups X Phase. Post hoc tests of the Groups X Phase effect indicated several important patterns in student engagement. Within each group over phases, there was a generally increasing trend in engagement (see Figure 1).

At Phase 1, the experimental group (M = 32.2%) fell below both the control (M = 35.9%) and comparison groups (M = 36.6%). At Phase 2, however, there was a significant gain, 10.9%, in academic responding associated with the onset of CWPT in the experimental group (M = 43.1%), compared to both control (M = 35.4%) and comparison group (M = 35.8%). The experimental group exceeded the controls in all subsequent phases except for Phase 4. Most interesting was the fact that the comparison groups exceeded both experimental and control groups except at Phase 4. At Phase 4, the controls (M = 53.0%) performed better than or as well as either the experimental (M = 49.6%) or comparison (M = 54.9%) groups.

The ANOVA for the task-management composite indicated a significant main effect for Phases and a significant Group X Phase effect (see Table 2). Within groups over phases, there was a general declining trend in task-management behaviors. For the experimental group, this trend began in Phase 2, increased slightly at Phase 4, and declined thereafter (see Figure 2). For both the comparison and control groups, there was an increase in task management in Phase 2 of Grade 1 followed by a systematic decline at Phase 3 and thereafter. The control group was most variable, however, showing a significant increase in task management at Phase 5 (M = 43.1%) and 6 (M = 38.9%), compared to its lowest point at Phase 4 (M = 33.4%).

The ANOVA for the competing-response composite indicated a significant main effect for Group and for Phases. The groups effect means were 14.7% (SD = 9.33), 8.3% (SD = 4.09), and 8.6% (SD = 4.74) for the control, experimental, and comparison groups. Overall, the controls spent nearly twice as much time engaged in competing behaviors as did the other two groups. The phase means were 18.26% (SD = 11.8), 9.4% (SD = 9.7), 9.7% (SD = 12.4), 8.2% (SD = 10.3), 7.4% (SD = 6.7), and 8.5% (SD = 10.9), in order by phases. All groups declined significantly after the first semester of Grade 1.

To more clearly show these changes in students' academic engagement, task management, and competing behaviors, means for each variable are plotted by groups in Figure 3. These data indicated that only in the experimental and comparison groups was there an increasingly large separation over phases between (a) level of engagement and (b) levels of task management and competing behavior. It was also clear that these changes in the experimental group coincided with the onset of CWPT in first grade and with the onset of Grade 2 in the other two groups.

Time Spent/Time Engaged Composite. The ANOVA for the time spent/time engaged composite indicated a significant main effect for Groups, Phases, and the Group X Phase interaction effect (see Table 2). The trends in these data replicated those noted earlier for engagement (see Figure 1). Evident in these data in Figure 4, compared with Figure 1, was the impressive overall advantage of the comparison group over both experimental and control groups because of its higher levels of both time spent and time engaged.

Achievement Products

Significant MANCOVA main effects were found for Groups, Phases, and the Groups X Phase (see Table 2). ANCOVA main effects for groups were significant in reading and language. Significant phase effects were noted for reading, mathematics, and language. A significant Groups X Phase was noted for mathematics.

Post hoc analyses of the groups main effect in reading and language indicated that significant differences occurred between covariate adjusted means of the control group (40.52, reading, and 48.37, language) and both the experimental (45.95, reading, and 54.48, language) and comparison (54.48, reading, and 57.29, language) groups. The difference between the experimental and comparison groups was significant in reading (45.95 versus 54.48) but not in language (54.48 versus 57.29). In language, the experimental group had gained as much as had the comparison group. Effect sizes (ES) for the groups main effect were as follow: in reading, 0.36 for the experimental and 0.71 for the comparison groups; in language, 0.33 and 0.48 for the experimental and comparison groups, respectively.

In mathematics, post hoc tests indicated an increasing trend within each group between Grade 2 and Grade 3. The experimental group significantly exceeded the control group at the end of both second ([M.sub.adj] = 46.23 versus [M.sub.adj] = 39.4) and third grades ([M.sub.adj] = 57.9 versus [M.sub.adj] = 52.2). The comparison group also significantly outgained the control group at the end of Grade 2, ([M.sub.adj] = 45.0 versus [M.sub.adj] = 39.4), and Grade 3, [M.sub.adj] = 64.0 versus [M.sub.adj] = 52.7). The comparison group outgained the experimentals in mathematics only at the end of Grade 3 ([M.sub.adj] = 64.0 versus [M.sub.adj] = 57.9).

DISCUSSION

A number of interesting and important observations were made in this study. First, high-SES students in suburban schools spent on the order of 5%, or 15 minutes, more of their prime instruction time per day in academic-oriented instruction than did low-SES controls. The activities to which this extra instructional time was devoted varied by phase and grade but included more time in reading and language. Second, for all groups we observed an increasing trend in levels of engagement and decreasing trends in students' levels of task management and competing, inappropriate behaviors over grades.

Furthermore, when time spent and time engaged were expressed as a single time-to-learn variable in the spirit of Carroll (1963), the overall superiority of the comparison group, who differed significantly on both variables, was even more clearly apparent. Third, the significant difference in academic engagement between the high-SES comparison group and the low-SES control group in this study extended our earlier fourth-grade findings (Stanley & Greenwwood, 1983) to Grades 2 and 3. There wre no engagement differences between groups at Grade 1. Fourth, CWPT resulted in increased levels of engagement for low-SES students at its onset at first grade and consistently over time. Fifth, after initial IQ and pretest achievement levels were statistically controlled, the experimental and comparison groups performed significantly higher than did controls on the Metropolitan Achievement reading, language, and mathematics subtests at the end of second and third grades. The effect sizes for the experimentals ranged from 0.29 to 0.36, and for the comparison group they ranged from 0.48 to 0.70. These were moderately large effects and were all educationally significant (e.g., Bloom, 1984; Walberg, 1986). Sixth, these data are from a longitudinal study carried out over the first three grades, representing 25% of each student's time within public education.

It was impossible to know with certainty how comparison-group teachers managed to increase their students' time spent in academic instruction or their levels of engagement. Because of the presence of a uniform district policy concerning daily schedules and expected time allocations to subject-matter instruction, this systematic difference in time spent was not due to differences in policy. Rather, we speculate that it was a subtle function of the interaction between these students' initially greater academic skills at school entry (measured IQ and reading achievement) and their more rapidly developing ability to handle academic assignments independently for longer periods of the daily prime time, compared with the skills and abilities of controls. Another possible explanation of this finding might be differential

effects of pull-out, Chapter 1 programs on the regular classroom ecology. Comparison-group students did not experience the additional transitions required for students to attend Chapter 1 instruction, which could explain their greater time spent in instructional time.

Clearly, attrition was a major threat to the internal validity of this longitudinal study. The rate of participant attrition over 3 years was on the order of 40% to 50%. However, tests indicated that the attrition rate did not significantly differ across the three groups, and that estimates of students' measured IQ between the original and remaining samples were highly similar. Thus, we concluded that the results based on the sample remaining after 3 years were representative of those that would have been obtained for the entire sample.

The current findings added information concerning what is known about process-product relationships. The current study demonstrated that low-SES students in urban, Chapter 1 schools were not only at increased risk for academic delay compared with their suburban agemates, but that this risk is in part a function of time devoted to instruction and students' engagement in lessons. The current data supported the assertion that the conventional instructional practices employed by the teachers of students in the control group were partly responsible for the students' increased risk for academic delay. This was tenable because the experimental group immediately increased their engaged time relative to controls as a result of the use of CWPT. Engaged time was maintained over subsequent phases and grades as the CWPT program continued to be implemented. Use of CWPT, however, did not increase the time spent in academic instruction for this group.

It is important that future research attempt to replicate these time-spent and time-engaged differences between suburban and urban classrooms. It is essential that similar analyses be conducted of the special education programs in these settings. If it is widely established that students in urban schools are obtaining less instructional and engaged time in basic academic subjects, special education teachers must be aware of this problem and must correct it. Clearly, it is known that poverty affects school processes (Final Report on the National Assessment of Chapter 1, 1987) and the schools "mirror their neighborhoods" in this regard (Meyer, 1984). However, other than school-input variables (e.g., budgets and resources), it has not been widely considered how poverty might actually affect the opportunity to learn within the classroom from an instructional point of view. In this study, control-group children remained delayed and vulnerable after after 3 years of schooling wereas experimental-group students at equal risk were less vulnerable because of increased academic outcomes. Thus, CWPT was validated, as were other effective instructional practices such as direct instruction (Becker & Gersten, 1982; Gersten, Keating, & Becker, 1988) because of their capability of improving both students' engagement and rate of academic progress. Based on these current process-product data, time spent in academic instruction and time engaged were prognosticators of actual academic outcome and of the presence of effective instructional practices.

REFERENCES

Becker, W. C. (1977). Teaching reading and language to the disadvantaged: What we have learned from field research. Harvard Educational Review, 47, 518-543.

Becker, W. C., & Gersten, R. (1982). A follow-up of follow through: The later effects of the direct instruction model on children in fifth and sixth grades. American Educational Research Journal, 19, 75-92.

Berliner, D. C. (1988). The half-full glass: A review of research on teaching. In E. L. Meyen, G. A. Vergason, & R. J. Whelan (Eds.), Effective instructional strategies for exceptional children (pp. 7-31). Denver, CO; Love Publishing.

Bloom, B. S. (1984). The 2 sigma problem: The search for methods of group instruction as effective as one-to-one tutoring. Educational Researcher, 13, 4-16.

Brophy, J. E. (1979). Teacher behavior and its effects. Journal of Educational Psychology, 71, 733-750.

Brophy, J. E. (1986). Teacher influences on student achievement. American Psychologist, 41, 1069-1077.

Carroll, J. B. (1963). A model of school learning. Teachers College Records, 64, 723-733.

Carroll, J. B. (1985). The model of school learning: Progress of an idea. In C. W. Fisher & D. Berliner (Eds.), Perspectives on instructional time (pp. 29-58). New York: Longman.

Carta, J. J., Greenwood, C. R., Dinwiddie, G., Kohler, F. W., & Delquadri, J. (1987). The Juniper Gardens Classwide Peer Tutoring Programs for Spelling, Math, and Reading. Kansas City, KS: The Juniper Gardens Children's Project, Bureau of Child Research, University of Kansas.

Delquadri, J., Greenwood, C. R., Stretton, K., & Hall, R. V. (1983). The peer tutoring game: A classroom procedure for increasing opportunity to respond and spelling performance. Education and Treatment of Children, 6, 225-239.

Delquadri, J., Greenwood, C. R., Whorton, D., Carta, J., & Hall, R. V. (1986). Classwide peer tutoring. Exceptional Children, 52, 535-542.

Denham, C., & Lieberman, A. (Eds.). (1980). Time to learsn. Washington, DC: National Institute of Education.

Dixon, W. J. (Ed.). (1988). BMDP statistical software manual (Vol. 1). Berkeley, CA: University of Califonia Press.

Elementary mathematics. (1983). Menlo Park, CA: Addison-Wesley.

Final report on the national assessment of Chapter One. (1987). Washington, DC: Office of Research, Office of Educational Research and Improvement, U.S. Department of Education.

Gersten, R., Keating, T., & Becker, W. C. (1988). The continued impact of direct instruction. Education and Treatment of Children, 11, 318-327.

Gettinger, M. (1985). Time allocated and time spent relative to time needed for learning as determinants of achievement. Journal of Educational Psychology, 77, 3-11.

Good, T. L., & Grouws, D. A. (1979). The Missouri mathematics effectiveness project: An experimental study in fourth-grade classrooms. Journal of Educational Psychology, 71, 355-362.

Greenwood, C. R., Carta, J. J., & Hall, R. V. (1988). The use of peer tutoring strategies in classroom management and instruction. School Psychology Review, 17, 258-275.

Greenwood, C. R., & Delquadri, J. (1988). Code for Instructional Structure and Student Academic Response (CISSAR). In M. Hersen & A. S. Bellack (Eds.), Dictionary of behavioral assessment (pp. 120-122). New York: Pergamon.

Greenwood, C. R., Delquadri, J., & Carta, J. J. (1988). Classwide peer tutoring (CWPT). Delray Beach, FL; Education Achievement Systems.

Greenwood, C. R., Delquadri, J., & Hall, R. V. (1984). Opportunity to respond and student academic performance. In W. Herward, T. Heron, D. Hill, & J. TrapPorter (Eds.), Behavior analysis in education (pp. 58-88). Columbus, OH: Charles E. Merrill.

Greenwood, C. R. Delquadri, J., & Hall, R. V. (1989). Longitudinal effects of classwide peer tutoring. Journal of Educational Psychology, 81, 371-383.

Greenwood, C. R., Delquadri, J. C., Stanley, S., Sasso, G., Whorton, D., & Schulte, D. (1981). Allocating opportunity to learn as a basis for academic remediation: A developing model for teaching. Monograph in Behavioral Disorders, Summer, 22-32.

Greenwood, C. R., Dinwiddie, G., Bailey, V., Carta, J. J., Dorsey, D., Kohler, F., Nelson, C., Rotholz, D., & Schulte, D. (1987). Field replication of classwide peer tutoring. Journal of Applied Behavior Analysis, 20, 151-160.

Greenwood, C. R., Dinwiddie, G., Terry, B., Wade, L., Stanley, S., Thibadeau, S., & Delquadri, J. (1984). Teacher- versus peer-mediated instruction: An ecobehavioral analysis of achievement outcomes. Journal of Applied Behavior Analysis, 17, 521-538.

Hall, R. V., Delquadri, J., Greenwood, C. R., & Thurston, L. (1982). The importance of opportunity to respond to children's academic success. In E. Edgar, N. Haring, J. Jenkins, & C. Pious (Eds.), Serving young handicapped children: Issues and research (pp. 107-140). Baltimore: University Park Press.

Hartmann, D. P., & Woods, D. D. (1982). Observational methods. In A. S. Bellack, M. Hersen, & A. E. Kazdin (Eds.), International handbook of behavior modification and therapy (pp. 109-138). New York: Plenum.

HBJ spelling. (1983). New York: Harcourt Brace Jovanovich.

Heath elementary mathematics, (1983). Lexington, MA: Heath.

Hollenbeck, A. R. (1978). Problems of reliability in observational research. In G. P. Sackett (Ed.), Observing behavior: Data collection methods (Vol. 2, pp. 79-98). Baltimore: University Park Press.

Hummell, T. J., & Sligo, J. R. (1971). Empirical comparisons of univariateand multivariate analysis of variance procedures. Psychological Bulletin, 76, 49-57.

Macmillan series R. (1980). New York: Macmillan.

Maheady, L, & Harper, G. (1987). A classwide peer tutoring program to improve the spelling performance of low-income, third-, and fourth-grade students. Education and Treatment of Children, 10, 120-133.

Maheady, L., Sacca, M. K., & Harper, G.F. (1988). Classwide peer tutoring program on the academic performance of mildly handicapped students. Exceptional Children, 55, 52-59.

Meyer, L. A. (1984). Long-term academic effects of the direct instruction project follow through. The Elementary School Journal, 84, 380-394.

Otis, A. S., & Lennon, R. T. (1979). Otis-Lennon School Abilities Test-Primary I. New York: The Psychological Corp.

Powell, J., Martindale, A., & Kulp, S. (1975). An evaluation of time-sample measures of behavior. Journal of Applied Behavior Analysis, 8, 463-470.

Prescott, G. A., Balow, I. H., Hogan, T. P., & Farr, R. C. (1978). Metropolitan Achievement Tests. New York: The Psychological Corp.

Rosenshine, B., & Berliner, D. C. (1978). Academic engaged time. British Journal of Teacher Education, 4, 3-16.

Rosenshine, B., & Stevens, R. (1986). Teaching functions. In M. C. Wittrock (Ed.), Handbook of research on teaching (3rd ed., pp. 376-391). New York: Macmillan.

Stallings, J. A., & Stipek, D. (1986). Research on early childhood and elementary school teaching programs. In M. C. Wittrock (Ed.), Handbook of research on teaching (3rd ed., pp. 727-753). New York: Macmillan.

Stanley, S. O., & Greenwood, C. R. (1981). Code for Instructional Structure and Student Academic Response (CISSAR): Observer's manual. Kansas City, KS: Juniper Gardens Children's Project, Bureau of Child Research, University of Kansas.

Stanley, S. O., & Greenwood, C. R. (1983). Assessing opportunity to respond in classroom environments: How much opportunity to respond does the minority, disadvantaged student receive in school? Exceptional Children, 49, 370-373.

Walberg, H. J. (1986). Synthesis of research on teaching. In M. C. Wittrock (Ed.), Handbook of research on teaching (3rd ed., pp. 214-229). New York: Macmillan.

ABOUT THE AUTHOR

CHARLES R. GREENWOOD (CEC Chapter #436) is the Research Director of the Juniper Gardens Children's Project at the University of Kansas, Kansas City.

This research was supported by grant No. 03144 from the National Institute of Child Health and Human Development. The author would like to thank the project staff: Carmen Arreaga-Mayer, Voris Bailey, Kathy Banks, Judith J. Carta, Richard Couch, Paul Diedrich, Joe Delquadri, Granger Dinwiddie, Don Dorsey, Marleen Elliott, Rebecca Finney, Verona Hughes, Frank Kohler, Betsy Leonard, Esther Lerner, Dennis Madrid, Chris Nelson, David Rotholz, Barbara Terry, Davida Sears, Dale Walker, Debra Whorton; office staff: Carmen Root, Betty Smith, Alva Beasley, Bernadine Roberts; and Mary Todd, Dan Schulte (observer team director), and the observer team. I also thank the school staff, including Don Moritz, director of research, and the principals, teachers, parents, and students of the Kansas City, Kansas District USD #500.

I also thank Hyman Hops for his contribution to the initial design of this investigation and Judith Carta and Jane Doherty for their comments on earlier drafts of this manuscript.

Reprints of this article can be obtained from Charles R. Greenwood, Juniper Gardens Children's Project, 1614 Washington Blvd., Kansas City, KS 66102.

Manuscript received February 1989; revision accepted November 1989.
COPYRIGHT 1991 Council for Exceptional Children
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:at risk of low academic achievement
Author:Greenwood, Charles R.
Publication:Exceptional Children
Date:May 1, 1991
Words:6516
Previous Article:Direct instruction in math word problems: students with learning disabilities.
Next Article:Strategy usage among deaf and hearing readers.
Topics:


Related Articles
Self-concept and academic performance in gifted and academically weak students.
Predictors of global self-worth and academic performance among regular education, learning disabled, and continuation high school students.
Reflections on a Research Career: Perspective on 35 Years of Research at The Juniper Gardens Children's Project.
Advising indicators: operating from an empirical perspective. (On-going Topics).
Longitudinal effects of kindergarten.
Motivation and academic resilience: developing a model for student enhancement.
Motivational goals and school achievement: Lebanese-background students in south-western Sydney.
Time perspective and school membership as correlates to academic achievement among African American adolescents.
Teaching quality and student outcomes: academic achievement and educational engagement in rural Northwest China.
Academic achievement differences between aerobically active versus inactive advanced degree students.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters