| Home |
 |
Chapter 14 |
|
Theory into Practice |
|
Ch. 14, p. 534
Interpreting Standardized Test Scores
This section presents a guide to interpreting test reports for one widely used standardized test of academic performance, the Terra Nova, published by CTB/McGraw-Hill (1997). Other widely used nationally standardized tests (such as the CAT, the Iowa, and the Stanford) use similar report formats.
Class record sheet. Figure 14.6 on pages 536–537 shows portions of an actual Terra Nova (CTBS/5) pre-post class record sheet for children (whose names have been changed) in a Title I second-grade reading class. The main information on the form is as follows.
Identification Data
Look first at the top of the form. It identifies the tests taken at the end of the previous year (pre) and at the end of the current year (post). The grade (2.7) indicates that at the time of post-testing, students were in month seven of second grade (April; September is month zero). Information at the bottom left shows testing dates, school, district, test norm, and “quarter month” (i.e., weeks since school began).
Scores
Under each column, test scores are shown in two metrics. NP refers to national percentiles; NCE, to normal curve equivalent. For example, look at the fifth child, Marvin Miller. At the end of first grade, his national percentile score in reading was 49, indicating that he scored better than 49 percent of all first-graders. By second grade, his percentile score had increased to 76. In NCEs, however, he increased from 49 to 65. The test form shows a gain of 16 NCEs; NCE scores can be added and subtracted, because they are on an equal interval scale, whereas percentile scores cannot. Now look at the second student, Brittany Duphily. In reading, her percentile scores (and NCEs) dropped from first to second grade. Does this mean that she knows less in second grade than she did in first? Not at all. However, she did perform less well in second grade compared to other second graders. To understand this, consider a girl who is the third fastest runner in the fourth grade, but a year later is the twelfth fastest in the fifth grade. The girl has not slowed down, and probably can run faster than before, but other runners are making better progress.
Summary
At the bottom of the form is a summary of the test scores for second graders in the entire district (a class or school summary would look the same). The scores are presented as median percentiles (the score of the middle child in the district) and the national percentile of the median, which indicates how well the district is doing among all districts. In this district, for example, the middle second-grader is scoring better than 35.3 percent of all second-graders in reading. The last set of numbers shows the mean NCEs for all second-graders and the difference between first-grade NCEs (42.5) and second-grade NCEs (41.6). The difference, a loss of 0.9 NCEs, is very small, essentially indicating that children in this district score at about the same reading level in first and second grades, in comparison to children in other districts.
Individual profile report. Like most standardized tests, the Terra Nova provides a detailed analysis of the test performance of each child. Figure 14.7 (page 538) shows an example for a third-grader, Maria Olthof (a real report, but not her real name). The form gives the following information:
Norm-Referenced Scores
At the top of the report is a list of Maria’s scores on 14 scales listed six ways. The first is grade equivalent. In reading, Maria’s grade equivalent is 2.0, indicating that her score is like that which would be obtained by an average child just starting second grade. Her NCE of 35 also indicates that she is performing significantly below grade level. (In general, an NCE of 50 is considered “at grade level.”) Skip over scale score, which is not interpretable. Maria’s local percentile indicates that she is reading extremely poorly in comparison to other children in her class, school, or district (however “local” was defined). A percentile of 1 is the lowest possible score.
Number correct is self-explanatory. In reading, Maria’s national percentile indicates that she is scoring better than only 24 percent of all third-graders in the United States. “NP range” indicates the likely range of national percentile scores that Maria might receive if she took the same test many times. That is, there is always a range of scores a student might get, depending on luck, inadvertent errors, testing conditions, motivation, and so on—all factors that could vary each time a student took a test even if his or her level of knowledge or skill stayed the same. The chart on the upper right shows this national percentile range with a diamond indicating the actual percentile score. The shading between the 25th and 75th percentiles indicates the “normal range”; Maria’s reading score is below that range, although her own “NP range” suggests that on a very good day she might score within the normal range.
Note that at the bottom of the national percentile chart is a scale indicating stanines. Recall that stanine scores range from 1 to 9, with a score of 5 indicating the national average.
Performance on Objectives
The remainder of the individual profile report breaks Maria’s test down into subskills in each area. This breakdown can provide some useful information to explain overall scores. For example, look at Maria’s mathematics scores. She scores very well on an “objectives performance index” in every subscale of math but one: problem solving. For Maria, this one low score could be due to her reading problems; or she might need additional work with this skill. However, subscale analyses of this kind should be interpreted very cautiously. The small number of items involved and the lack of a clear connection to the material Maria is studying mean that classroom assessments, perhaps supplemented by more fine-tuned diagnostic tests in mathematics, would give a much better indicator of Maria’s strengths, weaknesses, and instructional needs.
|