Tuesday, February 17, 2015

Predicting Intervention Effectiveness from Oral Reading

Selecting interventions to meet student needs is important for any teacher whose role it is to provide Response to Intervention, Title I reading or special education services. Only limited amounts of research are available that help discriminate between evidence based approaches. Isadora Elizabeth Szadokierski attempted to identify if there were differences in students who would respond best to particular interventions, what is the relationship between intervention effectiveness and baseline data, and how well did baseline measures predict the effectiveness of two fluency interventions. She described her work in her dissertation, Predicting Intervention Effectiveness from Oral Reading Accuracy and Rate Measures through the Learning Hierarchy/Instructional Hierarchy (2012), for the University of Minnesota. The Learning Hierarchy is described below with an adaption of a chart from the text.

In the study
Learn how to correctly perform a skill. Require explicit instruction.
Accuracy of response
·         Demonstration
·         Models
·         Cues
·         Routine drills
Fluency/ proficiency
Can perform the skill with adequate accuracy but need to practice to get faster at it. May require reinforcement to complete drills.
·         Repeated novel drills
·         Reinforcement
Can perform the skill accurately and quickly but needs to do it in multiple situations or discriminate when to perform the skill.
Novel stimulus
·         Discrimination training
·         Differentiation training
Needs to learn how to modify the skill to meet the needs of various situations.
Adapted response
·         Problem solving
·         simulations

Her study involved 49 second and third grade students selected by their teachers as struggling readers. These students were required to have median baseline rates below the 50th percentile for the grade level at the time of year, read at least 11 words correct per minute, be native English speakers, and not be receiving special education for moderate/severe cognitive delays, autism or a visual disability. The author did not identify whether the selected students had been identified as having a learning disability.

Students received the intervention for five days over the course of two weeks in a one-on-one setting. For a  baseline, they were asked to read 10 passages at their grade level initially with no intervention (these were the passages used in the later interventions). The interventions were alternated in order to identify the more effective intervention. Then the more effective intervention was administered after a reversal procedure. The first intervention was modeling-error correction intervention package (M-EC) in which students were read the passage (listening passage preview), then were asked to read the passage with phrase error drill correction, followed by a final timed read without correction. The second intervention was repeated reading-reward intervention (RR-R) in which the students read the passage twice through for 90 seconds without feedback, they were offered a student selected reward if they met a goal (20% higher than the baseline reading of the passage), and then they participated in a timed reading.

Students who read at a faster rate and with few errors were more likely to benefit from the RR-R method than the M-EC intervention package. When the author normalized the reading rate by establishing difference from expected rate at grade level at that time of year, the least predictive data was developed. Unfortunately, since normalization of the data led to less predictive information, these cut score results have limited value to students outside of second and third grade. What we can learn from this study is that students who appear to be learning to decode are probably better served by fluency interventions that include modeling, and direct instruction of decoding.

No comments:

Post a Comment