July 10, 2005

Unhappy with MEAP results

In Michigan, sliding MEAP scores are scaring officials, who believe the state's chances of attracting more high-tech companies are in jeopardy:

...a shrinking percentage of high school students are exceeding the state's requirements in math and science. Results from Michigan Educational Assessment Program tests for the class of 2005, released this month, give the state a clearer picture of the challenge it faces in its effort to attract high-tech businesses with a better-educated work force...

Over the past six years, the percentage of high school students scoring well below what the state considers failing on standardized math tests grew from about 20 percent to 29 percent, which translates into almost 32,000 students in the class of 2005. In science, it grew from 20 to 26 percent. Meanwhile, the percentage of students scoring well above state standards fell from 22 percent to 9 percent in math and from 7 percent to 5 percent in science.

Testing experts and school officials caution that there could be many reasons for the trend in math and science other than students' skills getting worse. They say more kids are taking the test to meet federal requirements of the No Child Left Behind Act. Under the law, schools must test 95 percent of their students. These students, who previously may have avoided the test because they felt they wouldn't do well, could be bringing down scores, they say.

Certainly, an influx of underqualified examinees can reduce the percentages - but I'd be interesting in knowing if the raw numbers have changed. If the actual number of students scoring at proficient has stayed the same while the percentage dropped, that would be evidence of the more-underqualified-examinees theory. However, if the number of students who are doing well is dropping along with the percentages, then something else is happening here.

I wouldn't feel comfortable assuming that the numbers above were signs of random, meaningless fluctuation. Either the examinee population is changing, the tests aren't equated very well, or it's some combination of the two (perhaps along with other factors).

Posted by kswygert at July 10, 2005 11:27 AM
Sitemeter