Monday, March 17, 2014
The Problem With a Well-Standardized Test
I recently read some information on the American Psychological Association's website about the relationship between SES (particularly low-SES) and education. Most of the statistics were familiar. We know how kids living in poverty perform relative to their more affluent counterparts. But what do we DO with that knowledge in assessment/analysis short of tut-tutting about the difficult realities that some of our kids face? Similarly, we are supposed to take into account a child's racial/ethnic background when choosing standardized tests--make sure they were part of the standardized sample, etc.
Well, what about when 95% of your students live in poverty and only 20% of the norm group of the test were from low-SES backgrounds--because that was the makeup of the US when standardization was done? What about when 95% of your students are non-White but only 30% of the norm group was?
Yes, it's true that my students are living in the US and therefore theoretically a US-based norm sample should reflect their reality. But I would argue that it doesn't really. My students do not leave their section of the city. Except for school and an occasional ethnic restaurant trip, they are racially, linguistically and culturally isolated. They live in generational poverty. Their world is small--certainly not the world of national norm samples. When they get a standard score of 75, and C-/D+ grades, how worried should I be that there is a true disorder? Are we over-identifying students in poverty based on standardized test scores? How can we be more certain that a child's lack of skills is not due to lack of exposure, but a true disorder? It is rare that I have an initial assessment these days, but when I do 3-year reassessments and I get a standard score of 80, should I keep them in? Is this still a significant enough discrepancy from peers...?
For curiosity's sake, I would love to give the OWLS (because I like it and it's quick and easy to administer) to a random representative sample of kids in my school. It would be interesting to see what the standard scores would look like compared to the national norm group. Can anyone here say "sabbatical.....?!"
Well, what about when 95% of your students live in poverty and only 20% of the norm group of the test were from low-SES backgrounds--because that was the makeup of the US when standardization was done? What about when 95% of your students are non-White but only 30% of the norm group was?
Yes, it's true that my students are living in the US and therefore theoretically a US-based norm sample should reflect their reality. But I would argue that it doesn't really. My students do not leave their section of the city. Except for school and an occasional ethnic restaurant trip, they are racially, linguistically and culturally isolated. They live in generational poverty. Their world is small--certainly not the world of national norm samples. When they get a standard score of 75, and C-/D+ grades, how worried should I be that there is a true disorder? Are we over-identifying students in poverty based on standardized test scores? How can we be more certain that a child's lack of skills is not due to lack of exposure, but a true disorder? It is rare that I have an initial assessment these days, but when I do 3-year reassessments and I get a standard score of 80, should I keep them in? Is this still a significant enough discrepancy from peers...?
For curiosity's sake, I would love to give the OWLS (because I like it and it's quick and easy to administer) to a random representative sample of kids in my school. It would be interesting to see what the standard scores would look like compared to the national norm group. Can anyone here say "sabbatical.....?!"
Subscribe to:
Posts (Atom)