Standardized Achievement Tests

Standardized achievement tests provide important information to both the School and parents. However, we should be clear about what are and are not the valid implications of the information. Additionally a comprehensive assessment of the School’s curriculum and instructional program must include a variety of techniques, including standardized tests, analysis of a student’s portfolio of her/his actual school work, direct observation of teaching, and curriculum reviews benchmarked against state and national standards.

Summary Results of 2008 ERB Scores
The SFDS students consistently score higher than the group norm of students in independent schools. This year and in years past, on average SFDS students score between the 65^th and 75^th percentile compared to the ERB norm for students in independent schools. In other words, most SFDS students score above the 50^th percentile compared to independent school norms; about half of SFDS students score between the 50^th percentile and the 75^th percentile, and half score above the 75^th percentile compared to the independent school norm. A small percentage of SFDS students score below the 50^th percentile.

• This summary conclusion was observed by reviewing each subtest score for this year for each grade. I then compared this year’s scores to the scores of the grade level cohort from the previous two years (same students a year and two years younger) and the same grade level from each of the previous two years. The ERB provides an administrative summary providing average percentile scores for SFDS students and compares them to national, suburban, and independent school norms. Historically, suburban students score higher than the national sample, and independent school students score higher than suburban norms.

A Note about Percentiles and Statistical Reliability
A critical question is “What is a meaningful difference in percentile scores?” The answer to this question is a statistical calculation called the “Standard Error of Measurement (SEM),” and is defined as the extent to which the students’ scores would tend to differ if they were retested with a different set of questions measuring the same skills at the same level of difficulty. ERB publishes the Standard Error of Measurement for each subtest. When nearly all the student scores are bunched at the upper range of number of correct items – most of the students only miss a few items - then the SEM is large. Because nearly all scores in the independent school norming group are very close to each other, when the statisticians calculate the percentiles, small differences in items answered correctly create large differences in percentile scores.

Consider the following chart taken from this year’s 7^th Grade ERB scores in Reading Comprehension.

Student	A	B	C	D	E
Total Items = 37
Items Correct	30	31	32	34	35
Percentile	58%	65%	73%	87%	93%

The difference between the 58^th percentile and the 93^rd percentile is 5 items out of a total of 37. Student C who answered 2 more items correct than Student A is rated 15 percentile points higher. According to the Standard Error of Measurement, there is little meaningful difference between the 58^th percentile and the 73^rd percentile. If the student were to retake the test, he/she could score as many as 20 percentile points higher or lower.

In terms of classroom performance, there is little difference between the student who scored 58^th percentile and the student who scored 73^rd percentile. They both can read and understand the same level of texts.

What are the Appropriate Uses of ERB Tests?
Given the overall high performance of SFDS students and the statistical issues inherent in any standardized test, how do teachers and administrators use the test results?

1. We review the individual results to identify students who are scoring below the 50^th percentile of the independent school norms. Those students should be receiving supplemental instruction to enable them to continue to master the basic skills tested by the ERB’s. If they are not currently receiving assistance, we will contact the parents and discuss with them options for supplemental instruction.

2. We look for student scores that differ in a significant way from their classroom performance. Those can be students who score high on tests but who are having difficulties with classroom learning tasks, as well as students who score low, but are performing very well on classroom tasks. In each case we will review the student’s situation to determine how to respond appropriately.

3. We examine grade level trends on each of the subtests to determine if specific aspects of SFDS curriculum and instruction may need revision and improvement.

Basic Skill Mastery and Critical Thinking Skills
The ERB tests only measure basic reading, writing and math skills. They do not measure the ability to apply basic skills to solve complicated problems or reach a sophisticated understanding of a complex phenomenon. At SFDS, of course, we strive for both basic skill mastery and the subtle use of thinking strategies to understand human, social, and physical realities.

Consider the following learning task from the 5^th grade curriculum: At the end of a multi-week unit on Egyptian gods and goddesses, the students participate in a Council debating the issues plaguing the modern world: the war in Iraq, global warming, apple moth spraying, and education funding cuts. The students are asked to take on the identity of their god or goddess. They have been researching a specific god or goddess for several weeks and have written essays describing their character, religious purpose, and special powers. They have learned that all gods and goddesses often have contradictory character traits, e.g., love and jealousy. They must represent how their god or goddess might resolve the four current issues and not their own personal opinion. They must understand the current social and scientific dynamics of the current issues, craft an actual solution, and also be true to the character of their god. Finally they must reach a consensus decision with different gods and goddesses with different personalities and solutions.

Clearly being able to read a paragraph and accurately answer a question about its content is necessary to be able to gather information, but hardly sufficient to complete the learning task of the Council of the Gods and Goddesses.

Thank you for your attention to these somewhat technical descriptions. Please know that in many ways the teachers are attending to the learning capabilities of your students and providing thoughtful instruction to meet their needs.