A sensible piece in the NYT looks at the use of growth scores to track student performance. There are some inevitable problems with this: chief among them that changes are statistically more variable than levels, and changes are sensitive to how the test is normed. But these aren’t insurmountable, and at the very least, reporting both changes and levels to administrators and parents would be valuable.
Advertisements
Andrew, a growth of N when starting at level L may not mean the same as a growth of N when starting at level L + M, plus there are other impacts on usual educational growth e.g. associated with puberty.
Reporting growth could be useful.
Not sure by what you mean by “…how the test is normed.” If US “standarized tests” are similar to what I know, then each US-wide standardized test would be put onto a single scale – I would guess that the 2007 standarized test would be put on the same scale as the 2006, and 2005, and 2004 standarized tests (and so on).
I like this line in the article: ‘Ms. Bolender said that many teachers question the reliability of the growth model data, calling it “voodoo math†because “you have to be a Ph.D. in statistics to even comprehend it.†‘
Ah, well, no – the concepts and mathematics involved in the reporting is probably extremely simple if it’s anything like what I work with.