10:00-10:50 a.m. READING (Discussion Session) Salon A

Presider: Abraham Andero, Alabama State University

THE EFFECT OF PHONEMIC AWARENESS INSTRUCTION IN FIRST GRADE ON THE READING SCORES OF RURAL PRIMARY STUDENTS

Linda H. Thornton and Rebecca Vinzant, Harding University

             The purpose of this study was to determine whether second grade students in a rural elementary school who had received a color-differentiated, rhythmic program of instruction in phonemic awareness and orthographic patterns with practice in student-selected trade books in first grade scored higher on the reading subtest of the Stanford Achievement Test than second grade students who had been instructed using a basal reading program. Using a t-test for independent samples, it was found that the students (n=42) who had been taught using the phonemic awareness/orthographic pattern/literature program achieved significantly higher scores on the reading subtest of the Stanford Achievement Test, 9th edition than did the students (n=58) who had been taught using a basal reading program, t (98) = 3.24, p < .01. It was concluded that the phonemic awareness/orthographic pattern/literature program was effective in raising the reading achievement level of the participating students.

INCREASING SILENT READING COMPREHENSION RATES VIA REPEATED READINGS

Jennifer T. Freeland and Bertha Jackson, Mississippi State University, and Christopher H. Skinner, The University of Tennessee-Knoxville

             Previous researchers have found that repeated reading could be used to increase rates of accurate oral reading or words correct per minute. Furthermore, researchers have found some evidence that repeated readings procedures may also increase the amount of material students comprehend when they are reading aloud. However, researchers have not directly measured the effects of repeated readings on rates of silent reading comprehension (a more functional skill than oral reading as individuals usually read silently). The purpose of this study was to begin investigating procedures designed to directly assess and increase silent reading comprehension rates.
             A multi-element design was used to compare the effects of the treatment with a control condition across three secondary students diagnosed with a specific learning disability in reading. During the control condition students were timed while they silently read a passage. Students were then asked to answer five factual questions and five inferential questions. During the repeated readings, condition students read the passage aloud twice to the experimenter and then answered the questions. The number of questions answered correctly and rate of comprehension were measured and graphed. Rate of comprehension was defined by multiplying the percentage of comprehension questions answered correctly by 60 and dividing this by the number of seconds required to complete a reading passage by 100.
             Results showed that repeated readings increased factual comprehension levels and factual reading comprehension rates. No differences were found across conditions on inferential comprehension levels or rates. Discussion focused on empirically validating reading interventions using rates of silent reading comprehension and theoretical implications related to enhancing inferential and literal comprehension and fluency.

AN INVESTIGATION OF THE SUCCESS FOR ALL READING PROGRAM

Lauren Rabb Wells, Lauderdale County (TN) Public Schools, and Jack Blendinger and Diane Greene, Mississippi State University

             This case study investigated the Success for All (SFA) reading program at two Mississippi elementary schools. Approximately 3,000 teachers, students, and parents served as subjects for the study. In particular, the study investigated: (1) whether SFA was being implemented as advocated by its developers, (2) what were students' reading scores on standardized achievement tests, (3) how the program was monitored, (4) teachers, students, and parents attitudes toward reading, and (5) what key informants said about the program. Mixed methods--qualitative and quantitative--were used to collect, analyze, and interpret data. Document review, field observation, surveys, and interviews were used to collect data.
             The investigation yielded a number of interesting findings about the SFA reading program. The findings revealed that adherence to the program's protocols, as designed by the developers, was only partial. Standardized achievement test scores increased for some students, but the results were inconclusive: the scores of children in some grades increased, while the scores of children in other grades decreased.
             The program's prescriptive nature (i.e., teacher behaviors are carefully scripted) controls teaching methods, children's learning experiences, and curriculum content. Therefore, it was not surprising to find that teachers were less positive in their attitudes toward the program than were students and parents. Younger children (grades K-2) were much more positive about the program and reading in general than older children. Parents were the most positive of the three groups.
             Teachers commented that the training they received to teach the program was not adequate and needed to be improved. Reading facilitators were more aware of the details of the program and how it affected students, teachers, and parents than were the principals of the two schools studied. Because the program is so controlling, facilitators and principals both reported that SFA greatly impacts teaching and learning in a school.

10:00-10:50 a.m. STUDENT ACHIEVEMENT (Discussion Session) Salon B

Presider: Mary O'Phelan, Western Kentucky University

PERCEPTUAL TYPOLOGIES OF SUCCESSFUL AND UNSUCCESSFUL STUDENTS IN THE CONTEXT OF GARDNER'S MULTIPLE INTELLIGENCE THEORY

Wade C. Smith, Jr., Tennessee State University

              The purpose of this research was to use XXXX High School as a research site to assess the impact of Gardner's Multiple Intelligences (MI) on students' academic successes in 10th-grade English, Social Studies, Mathematics, and Science classes. This research used a two-part minimally intrusive data collection protocol. The student population of XXX's 10th grade was stratified into two academic groups: (1) honors group and (2) regular group. From these two populations 60 students from each research group were randomly assigned to the research participation database. This resulted in 60 randomly assigned students in the Honors research group and 60 students in the Regular research group. Each student was surveyed to ascertain which multiple intelligence(s) they have used in English, Social Studies, Mathematics, and Science classes. This required each student to complete the survey instrument, Student Multiple Assessment Reporting Test (SMART), four times. Each survey was completed in approximately 10 minutes. The entire data collection process was completed in 40 minutes. Students' semester, first quarter, and second quarter grades were collected. Stepwise multiple regression with hierarchical clustering was used to determine the typologies of successful and unsuccessful students in the core subjects of Language Arts, Mathematics, Science, and Social Studies. There were significant differences between successful and unsuccessful students in all subject areas.

THE IMPACT OF MUSIC EDUCATION AND ATHLETIC PARTICIPATION ON STUDENT ACHIEVEMENT

Tim Schneider and Jack Klotz, The University of Southern Mississippi

             This paper reviewed the literature reporting the relationship between student participation in formal music education activities and its impact on student academic performance during the high school experience. Specifically, the paper reported findings of a recent research study of a stratified random sample of 110 high school seniors selected from a total population of 175 musicians and 200 athletes (non-musicians) and 865 non-musicians, non-athletes who graduated in 1998. The subjects of this study were drawn from high school students within a large southern metropolitan school district, having a total student population of 40,000 students and serving its high school age population within seven high school attendance centers. The findings of the study reported student achievement results within the identified student groups from a baseline data set that reported student participant achievement scores in the fifth grade and also achievement scores for the participants from testing points at grades six through nine. Additionally, the study employed a causal-comparison design, which allowed for analysis of how the different identified study groups behaved over time using both continuous and categorical data. Reported results were based upon a multivariate analysis of variance that was employed to determine if significant differences existed in the composite means of the identified study population.

THE RELATIONSHIP BETWEEN PERSONAL CHARACTERISTICS AND PERCEPTIONS OF NATIONAL HONOR SOCIETY STUDENTS

Jerri H. Bullard and Joe W. Wilson, University of North Alabama

             This study was conducted to identify patterns of characteristics among National Honor Society students to provide a more accurate description of this academically successful group. This study has particular significance because the identification of characteristics associated with achievers can then be encouraged, or manipulated, by schools to create a more positive attitude toward learning or a more supportive learning environment. Both secondary and postsecondary school systems would benefit by having information to support measures that might lead to increased retention and lower dropout rates.
             Of specific interest to these researchers were the following research questions: (1) What are the common characteristics of National Honor Society students (students who are academically successful)? (2) What personal characteristics and habits are associated with the demographic characteristics of National Honor Society students? (3) What personal perceptions are associated with the demographic characteristics of National Honor Society students? and (4) Are the personal characteristics and habits of National Honor Society students associated with their personal perceptions?
             The data were collected via a survey instrument developed to elicit information related to Honor Society student's background characteristics, personal characteristics and habits, and personal perceptions reflecting self-esteem and self-motivation. All personal characteristic questions and questions related to personal perceptions were scored using a five-point Likert scale ranging from (1) strongly disagree to (5) strongly agree.
             The survey was administered during fall 1999 to National Honor Society students from two public schools within a city school system. A total of 78 usable questionnaires constituted the data-producing sample. The two high schools differed in terms of urban-rural classification. Data analysis revealed a number of statistically significant relationships between the demographic variables, students' personal characteristics, and students' perceptions.

10:00-10:50 a.m. TEACHER EDUCATION (Discussion Session) Meeting Room 1

Presider: Jane Nell Luster, Louisiana State University

MODEL BUILDING AND PRESERVICE TEACHER DEVELOPMENT: RETHINKING REFLECTIVE PRACTICE IN TEACHER EDUCATION

H. Michael Crowson, The University of Alabama

             Research and theory on preservice teacher development has often emphasized the development of both domain-specific knowledge (Borko & Putnam, 1996) and general reflective capabilities (Mumby & Russell, 1993) as central components of teacher education. It is often assumed that the development of domain-specific knowledge is necessary for addressing specific, well-structured nature of teaching, while the development of more general reflective capabilities are sufficient for managing the more ill-structured (King & Kitchener, 1994) nature of teaching. Although research on these two aspects of teacher development has provided a wealth of information on teacher expertise and has provided avenues for improving teacher education, it has tended to ignore more central aspects of the knowledge construction and revision process that preservice teachers undergo during training. To be sure, the literature on teacher development has failed to adequately address issues pertinent to preservice teachers' overall cognitive developmental capabilities, particularly those related to the development of reflective capabilities. This has resulted in a restricted understanding of teacher development as it occurs within the context of the overall lifespan. This paper examined those processes and contents associated with reflection that may facilitate the construction of more complex mental models (see Johnson-Laird, 1983) of teaching among preservice teachers. Furthermore, it suggested that preservice teacher education should adopt a developmental focus that emphasizes the skills and knowledge that make up individual reflective activity in preservice teachers. Mental models theory (Johnson-Laird, 1983) and research in epistemic knowledge (Kruglanski et al., 1991; Kruglanski, & Webster, 1996; Schommer, 1991, 1998) and statistical reasoning (Nisbett & Ross, 1981; Klaczynski et al., 1997) were recommended as potential avenues for improving preservice teacher development. The paper culminated with a model that incorporates the relationships among statistical reasoning, epistemic knowledge, and motivation in preservice teachers' construction and revision of their mental models of teaching.

INSTRUCTIONAL DESIGN: FORGING LINKS WITH PRESERVICE TEACHERS

James R. Craig, Barbara Kacer, and Theron Thompson, Western Kentucky University

             Do preservice teachers create linkages between instructional theory and practice? If preservice teachers do create linkages between instructional theory and practice, what is the process of knowledge construction? During field work experience, do students recognize content as a vehicle for instructional design?
             John Dewey (1900) called for a "linking science" that connects learning theory and instructional practice. In 2000, curriculum developers are those "linkers." McIntyre, Byrd, and Foxx (1996) lament the "disjointedness of teacher education programs" (e.g., programs that to not work toward developing those linkages). What role (if any) field experiences play in the connection between instructional theory and practice? In other words, how do teachers also become linkers?
             Using Posner (1996) as a guide, a set of questions was constructed. During field work, students responded to those questions. Journals were analyzed for critical incidents and emergent categories. Subjects were students from an introductory education class, an introductory methods class, and a science methods class.
             Preservice teachers revert to control issues rather than a balanced instructional design. Preservice teachers use the "teacher as manager" metaphor. Preservice teachers assume that their knowledge of content is new and therefore "content will take care of itself." Preservice teachers revert to personal school experiences when reflecting upon what they observed and/or participated in during their field experiences. Growth over time (developing "linkages") often does not occur for preservice teachers.
             Results help inform field experience approaches in teacher education classes. Defining perspectives and identifying developmental levels of preservice teachers are critical components in delivering quality teacher education programs. Voices of preservice teachers provide data that should impact curriculum alignment, instructional design, and preparation for fieldwork. The research results delivered pragmatic suggestions for the revision of teacher education programs.

A THIRD ANALYSIS OF THE PERCEPTIONS AND STRATEGIES OF TEACHERS AS REFLECTIVE PRACTITIONERS

John Light and Janice Myhan, University of North Alabama, and Lynn Gillaspie, Eastern Kentucky University

             The purpose of this study was to determine if graduates of a teacher education program were using reflective practices in their classrooms and professional lives. Since the College of Education knowledge-base was founded on a reflective model, faculty members wanted to examine how teachers perceived the importance and degree of usage of reflective practices on a day-to-day basis.
             In 1995, data were obtained from 145 graduate students through the field testing of an instrument entitled Teacher Survey of Reflective Practices. Items were designed to assess information related to the following areas: (1) identification of reflective practices, (2) use of professional practices to modify teaching strategies, (3) use of activities and behaviors to teach others, (4) factors affecting reflective practices; and (5) forces influencing teaching practices.
             Data collected were analyzed via ANOVA to discern the degree of the actual use of reflective practices in contrast to whether students agreed such practices should be used in modifying classroom instruction. Fifty-one percent reported that they had received training in the use of reflective practice. Seventy-nine percent stated that the use of reflective practices was of "medium" to "high" importance in their current positions.
             In 1997, this initial instrument was modified from the previous study. Graduate students were surveyed and the results analyzed. Students reported they had received training in the use of reflective practice and were using these practices in their classrooms.
             In 1999, a third study was conducted to establish a baseline for determining program effectiveness. One hundred twenty-six graduate students enrolled in summer classes responded to the survey. Sixty-three percent of respondents indicated they had received training in the use of reflective practices. Eighty-seven percent stated that the use of reflective practices is of medium to high importance.

10:00-10:50 a.m. STATISTICS (Discussion Session) Meeting Room 2

Presider: Scott Bauer, University of New Orleans

A REVIEW OF THE LITERATURE ON MISSING DATA

Jesus Tanguma, Texas A & M University

             Most studies contain some missing data. The reasons for the missing data are many and varied. Respondents did not provide complete information. Observers failed to record all pertinent information. Participants did not participate throughout the duration of the study. Data were not properly coded/transferred.
             Four commonly used methods (listwise deletion, pairwise deletion, mean substitution, and regression substitution) for dealing with missing data were illustrated by means of hypothetical examples.
             Listwise deletion, being the default option in many statistical software packages (e.g., SPSS, and SAS), is the one most commonly used, also by default. However, because listwise deletion eliminates all cases from a participant missing data on any predictor or criterion variable, it is not the most effective method.
             Pairwise deletion uses those observations that have non-missing values to compute the correlations. Thus, it preserved information that would be lost when using listwise deletion. In mean substitution, the mean for a particular variable, computed from available cases, is substituted in place of missing data values on the remaining cases. This allows the researcher to use the rest of the participant's data.
             When using a regression-based procedure to estimate the missing values, the estimation takes into account the relationships among the variables. Thus, substitution by regression is more statistically efficient.

THE PITFALLS OF IGNORING MULTILEVEL DESIGN IN NATIONAL DATASETS

J. Kyle Roberts, Baylor College of Medicine

             Almost half a century ago, Robinson (1950) discovered the need for multilevel techniques while performing regression analyses at different levels of variables (i.e., regressions with students and regressions with schools). Later termed the Robinson effect, these different level regressions "[showed] that analyses executed at different levels of the hierarchy do not necessarily produce the same results" (Kreft & de Leeuw, 1998, p. 3). Although these regressions often gave opposite results when measured at different levels, no existing statistical method could overcome this problem.
             Robinson's problem was the inability to describe data that have group regressions with both random slopes (differences among schools) and random intercepts (differences among students). This problem occurs in many large-scale data sets (Seltzer, 1994). The challenge is two-fold. It is necessary not only to recognize the need for multilevel techniques, but also to utilize the potential value of multilevel techniques to broaden the types of questions that can be addressed (Seltzer, 1991).
             Despite the apparent promise of multilevel design, few researchers have used these techniques to study complex problems, especially in school-effects models. Just as Robinson (1950) first noted, companions of more commonly used methodologies invoking multilevel methods might reveal differences in findings across units of analysis that have implications for both policy and practice.
             To illustrate the pitfalls of ignoring data structure, this paper provided a brief overview of the national datasets available from the National Center for Education Statistics and illustrated multilevel techniques within the NELS:88 dataset. Multilevel analysis was conducted on 50 variables from the NELS:88. In addition, results from both weighted and unweighted samples were presented. From these results it was argued that, as the intraclass correlation for a given variable becomes larger, the more a researcher needs to utilize multilevel techniques over normal ordinary least squares.

AN APPLICATION OF GROWTH MIXTURE MODELING TO A STATEWIDE READING INITIATIVE DATASET

Frank R. Lawrence and Marcia R. O'Neal, The University of Alabama at Birmingham

             Advances in statistical procedures have spawned new data analysis techniques that can be applied to time series measures. Growth curve modeling (GCM) is one of the relatively new longitudinal data analytic techniques. Data analysts employ GCM to study individual differences over time. Individual differences are captured in the random coefficients, i.e. growth parameters that vary across individuals. The random coefficients may be viewed as continuous latent variables.
             Another technique for modeling growth over time is the hierarchical linear growth model (HLGM). The HLGM is particularly useful for modeling individual growth trajectories when those growth trajectories arise in naturally occurring clusters. The HLGM is able to account for variation among individuals and among clusters of individuals, thereby providing the analyst with more precise and accurate information.
             An assumption underlying GCM and HLGM is that the subjects came from the same population. Furthermore, these modeling techniques assume that each covariate affects the growth curve parameters in the same way. Yet, covariates do not always affect all individuals in the same way. Hence, it is not realistic to make that assumption. Conventional growth curve models cannot accurately reflect qualitatively different developmental trajectories.
             Growth mixture models (GMM) reportedly accommodate qualitatively different growth trajectories. GMM accommodates qualitatively different development though latent class variables. The latent class variable is used to cluster subjects into classes or subpopulations. GMM both estimates mean growth curves for each class and models individual variation around the curves.
             This research compared the performance of CGM, HLGM, and GMM by modeling the development of children who participated in a statewide reading improvement program. It may be useful to policy analysts because it shows unique relationships of antecedent variables to growth trajectories. Intervention policy designed around these unique relationships will likely be more cost effective.