Source Provenance: https://ged.com/wp-content/uploads/assessment_guide_for_educators_science.pdf Authority Tier:
TIER_1_OFFICIAL| License:PROPRIETARY_SPEC| SHA-256:cac1fe5cff5354fe...
Assessment_Guide_For_Educators_Science (Page 23)
23 Assessment Guide for Educators: Science Chapter 3: The Development Cycle of a Short Answer Item Developing Short Answer Scoring Guides Each SA item on the GED® test - Science is scored on a three-point scale. For some items, the three points are accumulated when the test-takers identify or analyze up to three specific details or correct answers. This type of SA can be scored analytically, meaning that one point is given for each correct part in a test-taker’s response. Other items are scored holistically. In these holistic items, each score point reflects a range of possible test-taker responses, with varying levels of correctness. This style of SA is particularly effective at measuring a test-taker’s skills with regard to summarizing or synthesizing information. Because each item has its own rules for scoring, scoring guides are developed alongside the item itself in two stages. The completed guides contain as broad a selection of responses as necessary to convey the types of answers that can receive full and partial credit. In the first stage of scoring guide development, test developers list possible correct answers that reflect the content target that the item intends to measure. For instance, if the SA item asks test- takers to identify three textual details that support a main idea in a passage, the preliminary scoring guide will list as many relevant phrases or sentences from the passage as necessary. The second stage of SA scoring guide development takes place during the rangefinding process. The primary purpose of rangefinding is to select exemplars at each score point level from a representative sample of responses. An exemplar is an optimal sample response for each score point. These exemplars, which are drawn from the pool of responses created when the items are field tested, serve to help train both human scorers and the automated scoring engine. Because test-takers can be both creative and insightful, they may come up with correct responses to a given SA item that the test developers did not anticipate when drafting the preliminary scoring guides. Therefore, SA scoring guides are often updated and completed during the rangefinding process, incorporating answers or common phrasings that have been directly observed in test-taker responses during the field test. Scoring guides broadly represent the variety of answers found in the sample pool. After finalization, the scoring guides are used with the exemplars to train human scorers. Once all items from field testing have been scored by humans, the scoring guides and exemplars are used to train the automated scoring engine. The