When Washington state ushered in high-stakes tests as a graduation requirement, it also included a set of assessment alternatives. The most frequently utilized option is the Collection of Evidence, or COE. The theory was to provide some protection for students who do not ‘test well’ – who possess the level of knowledge necessary to meet academic standard but, for whatever reason, are not able to convey that knowledge effectively in a standardized testing environment.
What are they? Well, collections of evidence are what they sound like – “work samples based on classroom work prepared by the student with instructional support from a teacher” according to OSPIs official literature.
Collections represent a valuable alternative to students who struggle with standardized tests, but they are also costly. The legislature currently requires COEs in Reading and in Writing, but only this year have Math COEs been required for graduation. Because pass rates for the Math are relatively low, it is expected that more students will access the COE as an alternative to graduation. And since each COE comes with a state price tag of $400, it is likely that the legislature will have to give serious consideration to just how valuable COEs are. Are they worth the cost?
The chart below shows a projection of increased costs to the state associated with phasing in math and science end-of-course tests as graduation requirements over the next few years. The truth is, we don’t know what the actual costs will be, and existing estimates are widely diverging. But everybody agrees on one thing – the costs will go up considerably. Again, are they worth it?
In trying to answer that question, SBE staff member Linda Drake discovered something noteworthy in the data. While racial achievement gaps are quite prevalent in the results of the state’s standardized tests scores, they are far less prevalent in the COE results. The results below are for Reading and Writing. Note the wide gaps in the top graph (the HSPE) and the clustering of scores in the bottom graph (COE).
And most noteworthy is the results for the Transitional Bilingual students. In the top graph (HSPE) they are by far the lowest performing subgroup. Yet, on the bottom graph (COEs), they are right in the middle of the pack.
The results for the bilingual population in Reading should perhaps not surprise us. Clearly, the language barrier associated with a standardized test would be effectively mitigated by a COE in which tasks are performed collaboratively with an educator. But the racial gaps are somewhat more mysterious. Perhaps we could dismiss this data if the N size was extremely low, but its not, as shown in the chart below. The ‘n’ is consistently over 1,000 students.
Or, perhaps the passage rates for COEs is so high, that the assessment essentially fails to discriminate performance levels enough to reveal the gaps. But again, no – COE pass rates, as shown in the chart below, range from 60% – 85% in Reading and Writing. And although gaps still exist, particularly in Writing, they don’t follow the same conventional pattern as those revealed in the traditional
standardized tests. American Indian students, for example, go from one of the lowest performing subgroups in Writing on the standard HSPE, to one of the highest performing on the COE. And the overall picture on COEs is a racial mish-mash. The type of mish-mash we hope for – where your performance does not appear in any way to be influenced by the racial group you belong to.
Whether we could expect similar results for Math and Science COEs, and whether it justified the additional cost for the legislature, remains to be seen.
I look forward to your comments.