The item writing process was not ‘spec-driven’. Unfortunately, item writers did not posses a guiding document that would allow them to generate items. Therefore, the need for more examples of item formats in order to develop better items, a proper identification of materials for reading and listening, and mainly the lack of detailed item test specifications was imminent. For example, despite the fact that Oregon item-writers produced test items in summer 2004, they identified some deficiencies as they followed the process. Therefore, the following thesis reports on a process of the development of this assessment as I was at some point one of the members in the test development team and then drew upon teachers’ opinions and judgments to compensate for a gap in this process.

Language testing plays a significant role in the collection of systematic information about English Language Learners’ ability and progress in K-12 public schools in Oregon. Recommendations were made for future research in line graph interpretation in mathematics and science education and for improving instruction in this area. This study added to our understanding of the role of science content knowledge during line graph interpretation, highlighted the importance of heuristics and mathematics procedural knowledge, and documented the importance of perception attentions, motivation, and students’ self-generated questions.

Finally, using Roth and Bowen’s (2001) two-stage semiotic model of reading graphs, representative vignettes showed emerging patterns from the study. A second set of findings corroborated how science background knowledge affected graph interpretation: correct science knowledge supported students’ reasoning, but it was not necessary to answer any question correctly correct science knowledge could not compensate for incomplete mathematics knowledge and incorrect science knowledge often distracted students when they tried to use it while answering a question. In addition, students answered the TOGS in one of three ways: as if they were mathematics word problems, science data to be analyzed, or they were confused and had to guess. One student with lower prior achievement motivated herself when she thought the questions were too difficult. In some cases, if students with lower prior achievement had thought about their answers in the context of their science knowledge, they would have been able to recognize their errors. Students with lower and moderate prior achievement favored one heuristic even when it was not useful for answering the question and rarely asked their own questions. Most used their science knowledge spontaneously to check their understanding of the question and the adequacy of their answers. They also monitored their understanding of the question and the adequacy of their strategy and answer by asking themselves questions. Students with higher prior achievement used a greater number and variety of heuristics and more often chose appropriate heuristics. Key findings from this analysis included differences in the use of heuristics, self-generated questions, science knowledge, and self-motivation. Data were analyzed using Roth and Bowen’s semiotic process of reading graphs (2001). The assessment rubric was validated by Padilla and a veteran mathematics and science teacher. Factors were categorized using Bertin’s (1983) three graph question levels.

Student performance on the TOGS was assessed using an assessment rubric created from previously cited factors affecting students’ graphing ability. A purposive sample of 14 6th-grade students engaged in think aloud interviews (Ericsson & Simon, 1993) while completing an excerpted Test of Graphing in Science (TOGS) (McKenzie & Padilla, 1986). This study examined how intermediate elementary students’ mathematics and science background knowledge affected their interpretation of line graphs and how their interpretations were affected by graph question levels.

Ib command terms